#last post did incentivize me hi. i had .3 seconds of a break so i wanted to doodle real quick before bed

[todayisafridaynight]

[pahrak-the-sinnoh-slizer]

I beat NEOTWEWY!  That was the most JRPG-ass final boss I’ve fought in years, love to see it.  Need to dig into the post-game content before I try to compare it overall to the first game, but the ending at least...I’d probably say that’s better than the first game’s, and being able to say that just makes me giddy.  A long list of other, super spoilery thoughts below the cut.

-It’s pretty cool that every family of Noise has its own symbol this time!  Makes it a lot more clear when you��re hunting a specific Noise.

-Pachy Noise seem a little less annoying?  Maybe?  I still dread them but not the most.

-Puffers take a bit too long to explode and that’s annoying, but in the grand scheme of things that’s pretty minor.

-Those fucking Chameleons, though.  The vanish and snipe routine I could tolerate, but combined with that counter blast every time you hit ‘em...yikes.

-I still think Rex Noise are cool af, but the Maximazaurus kicked my shit in and it didn’t feel entirely fair.  How was I supposed to know that roar attack covered the entire battlefield?  Can it even be dodged, or are you just supposed to kill it before it can use it?

-Fuck Plague Noise

-RIP Drake Noise and Progfox :(

-I do find it a little weird that some Noise have a regular and boss version that use the exact same name.  Makes me wonder why they didn’t just recolor and rename the boss version.  Meh, it’s not that important.

-So it seems “mutating” a Pin just means evolving it but it requires certain conditions, and those conditions seem to be which character it’s equipped to when it finishes leveling.  Very glad there’s a skill on the social network that makes their conditions clear.

-RIP Shutdown PP :(

-Music still slaps.  I really like that one, “Breaking Free” I think?  It embodies that early 2000s angst, plus the final lyrics are “the world ends with me” and I’m a sucker for that shit.

-The direction they went with Kanon didn’t quite land with me personally.  Like it was still pretty good, I did feel a bit sad to see her go, but her introduction just really rubbed me the wrong way I guess.

-I was unsure about the VIP system at first, but shopkeepers do still warm up to you even if it isn’t as trackable.  And...I never did a whole lot with the brand chart anyway, I guess.  The VIP level being a permanent thing rather than resetting every time you leave the area is a strong point in its favor.

-I really like the social network!  Gives you even more info about the background characters and helps drive home the themes of connection, and it’s heavily incentivized by the wide variety of rewards you get from it.  Five of my six pins can be Uber now.

-There was a minute there I thought I wouldn’t be able to progress without grinding Style, but then I realized it only restricts abilities, not what you can and can’t even equip like Bravery did.  Yeah that’s a step up.

-OH, and Pin Mastery!  Mastered pins counting even if you evolve or sell them!  That’s good.  That’s very, very good.

-I like how most of the playable characters have their own unique Psych used in the overworld to make up for their interchangeability in battle.  Do sort of wish Minamimoto and Neku had their own, but it’s not a sticking point by any means.

-Also how fucking hype was fighting Leo Cantus Armo and then seeing Neku bust in with Twister playing, that was so fucking cool.

-Speaking of Hype...was kind of expecting Tsugumi to do more?  In and of itself I don’t think there’s really an issue, I 100% think it’s a matter of her being central to the sequel buildup in both Solo and Final Remix.  She’s super sweet, though. ^_^

-As for Tsugumi’s Noise form...well, lolz told me to bring Stone pins and those really trivialize the fight.  The first time, at least.

-Someone I do think was underutilized is Ayano.  We didn’t get enough from her prior to her infection and that limited the impact.  The flashbacks offered during that scenario definitely helped, and I really liked Rindo going through the day trying to talk with Shoka and mentally prepare her, but...they definitely could’ve done more with her, and her being one of the characters who actually dies doesn’t help.

-Neku’s so happy in this game!!

-SHIKI FACE REVEAL!!!!

-I lost my shit when Beat took off his mask, adding him relatively early was a Good Move.

-HACKER RHYME

-I feel like Joshua showing up right out of nowhere would be confusing to new players, but obviously I can’t say that with certainty.  Plus, there’s still the Secret Reports, those might explain quite a few things.

-Right, Secret Reports!  I actually got one before even beating the game (Report 3), it was after my social network got to 70% completion so I think that was the unlock.  I am...nervous about not having a checklist of the unlock requirements this time, but it seems they might be easier to get overall (I got Report 2 just from fighting Go-Go Beringei’s symbol on Easy) so we’ll see.

-Don’t think I see game time anywhere so that sucks, especially since the Switch’s gameplay records are so damn vague.

-Still a tiny bit salty the game didn’t tell me about what the “-” button does right away.

-All the books seem to be in one shop and I quite like that, as well as the whole organization/aesthetic of the Collection menu.  I really want to see if I can 100% this game.

-The Graffiti wall is also a cool spin on achievements, even moreso since you can actually see the wall in Udagawa.

-Susukichi’s completed Noise form is tough.  You really need to time your dodges when he punches, took me a minute to get that, and taking cover from that massive lightning attack is a pretty cool gimmick.  I got massive Ovis Cantus vibes.

-Can we talk about how fucking packed with spoilers that last trailer was?  I am of the opinion spoilers don’t necessarily ruin things but that’s a personal choice and holy shit Square.

-Really like how they subtly mislead you in regards to Swallow’s identity.  The way scenes cut made me think for the longest time that they’d be the leader of the Ruinbringers, helped by the fact that Susukichi and Tsugumi start just a bit out from Rindo on the social network, leaving just enough space for one character to link them.  Then they start making you think it’s Rhyme.  Then you get surprised by the final reveal, but it’s a surprise that makes sense looking back.  Love it.

-The whole bit with Rindo meeting anOther is also a really interesting juxtaposition with Neku meeting CAT.  Both of their character flaws can be traced back to their idols, but in Neku’s case it’s down to his own misinterpretation, while Rindo is just the type of follower Motoi wanted to create.  Meeting his idol is a positive experience for Neku, but a negative one for Rindo, yet they both grow as a direct result of the encounter.  Super cool.

-In fact, it’s actually Fret whose backstory ends up being closest to Neku’s, despite their wildly different dispositions, that’s nuts and I love it.

-After her speech in Week 3, Nagi might just be the best character.

-They did a hell of a lot with Shoka and I love all of it.

-I said it before but I was not expecting to love Kaie as much as I do.  Definitely in the running for favorite Shinjuku Reaper.

-I didn’t expect much from Hishima due to his nondescript character design, but he’s actually pretty cool.

-For Susukichi and Shiba both, they’re complete assholes in different and entertaining ways, but both manage to have a healthy amount of depth as well, I really like that.

-And fucking Kubo!  Hated him from the start, but he managed to grow on me as things went, and his final reveal left a hell of an impression.  I do sort of wish we got to fight him ourselves, but he got a brutal death and that’s what really matters.  Xander Mobus sounds like he’s having fun.

-Haz was a very interesting addition.  We’ve been wondering who he is since that final trailer (where they showed the last scene he, a Walking Spoiler, was a part of, what the hell) and I still wasn’t sure what to expect once we did get his name.  He definitely gave off the same vibes as Joshua so I thought he might be a Composer (either replacing Joshua in Shibuya or maybe coming from Shinjuku), but the reveal of what he really is was pretty cool.

-Uzuki mellowed out just the right amount, I like her a lot more now.

-Kariya’s pretty much the exact same character and I love that for him.

-Would’ve liked to see Hanekoma but oh well, it sounds like he’s still writing the Secret Reports so that’s cool.

-Oh yeah, Coco.  Mmm...I mean I still don’t trust her?  It is a little weird that Neku does, but not to a jarring extent?  I dunno, I feel like she’s still got a scheme going on...

-Again, fantastic ending, last scene very similar to the last scene of Steins;Gate so that’s a plus in my book.

-And it’s got its own Another Day!!  The mention of Tin Pin makes me wonder if it’s the same timeline as the first game’s Another Day, but not sure.  All three of the checkpoint bosses kicked my ass so I left to go hunt Secret Reports and fill out my Noisepedia, but eventually I hope to get further into it.

-I wasn’t super diligent with Pig Noise but there’s only one variety missing in my Noisepedia, plus I missed the second level Rex Noise.  Other than that and the spots past Phoenix Cantus I just need to try to fill in drops.

-Blond kid spending one of his seven days hunting for seven urban legends really took me back to the KHII prologue, and I’m one of the folks who liked that so I mean that as a good thing.

-”The pigs’ll come sniffing!”  Beat knows what’s up.

-The Chapters menu seems to count Scramble Slam rewards towards completion...I am quite nervous about that, seeing as I only got like halfway to the lowest prize every time…

-Commentator Reaper has my whole heart though

-There was one sidequest...W3D1, I think?  It said someone in Udagawa needed help, but I ran up and down Udagawa like 5 times and couldn’t see anyone.  Did I miss it?  Is it post-game?  Is the game bugged?  Hm.

-But I have done almost all of the sidequests, and I...usually did well on Dive missions.  There’s a decent chunk of stuff for me to do in revisits but not too much.

-Where I am right now, I still have some unanswered questions, but a lot of that comes down to specific details I might not mind if they don’t clarify.  Still hope they do, of course.  I crave knowledge.  Time to get back to finding those Reports...

[amusewithaview]

there’s no “I” in “team” (and where I’m going, there isn’t a “we” either) - MCU gen fic

I was about to crash and then @phoenix-173 asked me to break her heart and I saw this post in rapid succession.  It was like somebody gave my muse a beverage with 13 espresso shots.

“It doesn’t work!” Jane shrieks, hands clenched around Goon #6′s forearm where he’s got it wrapped across her front.  She’s kicking and twisting, fighting to be set loose, but she’s about as effective as a kitten against a bear.  “Please, you have to give me more time,” she begs.

“You’ve had time,” Goon #2 points out.  “Time, money, all the resources we can give you, and yet...”  She tilts her head to one side, her mouth a moue of disappointment.  “The Boss’s generosity only goes so far,” she says, spreading her hands and shrugging.  She seems genuinely apologetic.

Darcy hates Goon #2 the most.

Goon #1 is grinning at the spectacle, his teeth are white and shiny and so straight that they’ve got to be fakes.  He seems like the kind of guy who gets punched a lot, and Darcy’s pretty sure that’s not just her own personal antipathy talking.  He’s popping his gum so loudly that she can hear it even over Jane’s not-so-muttered curses and shouts, she can hear it over the rapid tattoo of her own heartbeat.  Her pulse is pounding in her ears, at her throat, she can even feel it in her clenched fists.  Her hands are slick with sweat, the rest of her feels clammy and uncomfortably damp even in the aggressively climate-controlled lab.

Darcy has been fighting off the impending panic attack ever since the Goons burst in, three hours ahead of schedule.

Two hours ahead of the planned rescue.

“Please,” Jane says again, eyes locked on Goon #2, begging with expression and words, frantically stalling.  “Please, just - let me have a few minutes!  Just some last calculations, I think I can-”

Goon #1 rolls his eyes.  “Look, the Boss-”

Goon #2 holds up a hand, looking hopeful.  “Are you close?”

“Yes!  I mean, I think-”

The Goons exchange a speaking look.

Goon #2 sighs, stepping back.

Goon #1 cracks his knuckles and steps forward.  “Too late,” he says.  “Too bad, so sad, hope you’ve made progress since your last little test.”

Darcy swallows, throat so dry it makes a faint clicking sound.

The last test had ended in an explosion.  It took two days for the smell of burnt hair and muscle tissue - all that remained of the rabbit’s remains - to fully fade from the lab.  The test before that was even worse.

“I need her!” Jane insists.  “She’s-”

“Expendable,” Goon #1 interjects.  He adopts the same faux-apologetic look that Goon #2 is sporting.  “It is regrettable that we must incentivize your research in this manner, but progress must be made,” he intones.  “...did you like that?  That’s almost a direct quote from the Boss, you know.  He’s been watching your progress, or lack thereof, very closely.”

“Please.”

Jane is crying now, and it forces Darcy past her own terror into a state of near-calm.  If someone else is panicking, then she has to be strong.  If Jane is breaking down, then she has to stand tall.  She grabs that flimsy justification with all her might and pushes the tears and her fears far down into her stomach, tightening her core around that repressed emotion.

She can panic later.

If there is a ‘later.’

For the first time, Goon #1 turns to her.  “My momma always told me there was no ‘I’ in ‘Team,’” he says.  “You’ve been freeloading for weeks, time to pay your keep.  After all, there’s a ‘lab’ and ‘rat’ in ‘laboratory, so how’s about you crawl into that transporter like a good little intern?”

“She’s not my intern,” Jane spits.  “She’s my friend.”

“That will only make this harder for you,” Goon #2 says sorrowfully.

“Don’t you wish you’d worked a little harder?” Goon #1 says, playing bad cop with relish and gusto.  He eyes Darcy for a second and shakes his head, turning to address Goon #3.  “Lucky, get on the radio and tell the boys in maintenance to break out their mops, the intern is a damn sight bigger than that rabbit was.”

I’m about to die and he’s making fat jokes, Darcy thought.  “You know, nasty little weasels like you-”  She staggers to the right, the force of the unexpected blow almost enough to knock her to the ground.

“Rabbits don’t talk,” Goon #1 says.  “Get in the transporter.”

Jane is openly sobbing now, fighting so hard that Goons #4 and 5 have moved to stand beside Goon #6, just in case.  Darcy sees Goon #6 shift her so he can cover her mouth with one meaty hand, muffling her shouts and screams.

Goon #3 shifts to her left and she feels something hard and cold poke at the small of her back.  Darcy tries to turn back to Jane, wanting to say something, anything to make the other woman feel better - offer some sort of closure.  They both know there’s no coming back from this.

The Goons don’t give her time, #3 herds her up the ramp, stops her when she tries to turn, wanting one last look at Jane, maybe she can try and communicate something-

Darcy hears the shrill whine of the power cells, a click more physical than audible, and then her world is consumed in vibrant light.

[kristinsimmons]

Commissioning Healthcare Policy: Hospital Readmission and Its Price Tag

By ANISH KOKA MD 

The message comes in over the office slack line at 1:05 pm. There are four patients in rooms, one new, 3 patients in the waiting room. Really, not an ideal time to deal with this particular message.

“Kathy the home care nurse for Mrs. C called and said her weight yesterday was 185, today it is 194, she has +4 pitting edema, heart rate 120, BP 140/70 standing, 120/64 sitting”

I know Mrs. C well. She has severe COPD from smoking for 45 of the last 55 years. Every breath looks like an effort because it is. The worst part of it all is that Mrs. C just returned home from the hospital just days ago.

The youngest of six children, Mrs. C was born with many embedded disadvantages. Being born black in a poor West Philadelphia neighborhood in the 1960s is a story that too often writes itself with a bad ending.  But Mrs. C avoided the usual pitfalls that derail young women in the neighborhood early. No drugs. No alcohol. No teenage pregnancies. Finished high school. Mrs. C. worked for the hospital as a unit clerk, had her own place, health benefits, and even a retirement plan.

Certain life habits, however, carry a heavy price. George Burns, the comedian never pictured without a cigar who died past his hundredth birthday, may have been immune to the effects of tobacco.  Mrs. C was not. She started smoking when she was 16. She doesn’t recall why. Her dad smoking didn’t help perhaps. Nausea racked her body after that first drag. It eased up after. Too bad.

That measly cigarette became the great addiction of her life. Day by day, the exquisitely thin membranes of the lungs that mediate gas exchange were destroyed. By the time the disease manifests with shortness of breath and bluish tinged lips, it’s too late. Short of the very few who qualify for a lung transplant, the efforts of doctors at this point are for mitigation rather than cure.

Complicating things further, in Mrs. C’s case, the normally low pressure vascular circuit of her lungs became a high pressure circuit that places ever increasing demands on the normally thin-muscled right ventricle of the heart.  This jeopardizes the ability of her heart to handle changes in blood volume.

A little extra fluid and the right side of the heart ends up causing unbearable swelling in her legs.  A little dehydration and severe disabling dizziness on standing ensues. Adding to that that her tenuous lung function decompensates with the slightest respiratory infection, that chronic steroid treatment to decrease her wheezing suppresses her immune system, and that the young man down the street helpfully drops off Newports at her home for a few extra dollars, and it’s easy to see why the hospital is her second home.

The most recent admission to the hospital was for kidney failure related to taking too much fluid off with diuretics. What was to be a short stay for gentle hydration turned into a longer stay when a pneumonia complicated the matters (though a trip to the intensive care unit and a ventilator was barely but fortunately avoided).  She was treated by the pulmonology team and sent home on a lower dose of diuretics.

The situation I am now confronting puts me in a quandary.  Her edema and weight are up markedly just a few days after returning home.  Could her fluid overload be because her kidneys are shutting down? Or does she just need more aggressive diuresis?

Should I guess? Knowing her present renal function would be helpful. But even if the Theranos lab I could appeal to for help wasn’t fictional, I would have to get her to my office everyday or every other day while adjusting her diuretic dose.

And so it comes to be that, days removed from a hospital admission, I’m sending her back to the hospital to be readmitted.  According to some, this is not supposed to happen.

A policy on readmission

In 2008, the commission that advises Medicare – the Medicare Payment Advisory Commission (MEDPAC), issued a report that focused on hospital readmissions.

The focus on hospital readmissions had been of great interest to the health policy community for some time. At the core of this interest lies the belief that hospitals and physicians are incentivized to treat patients rather than prevent admissions.

The MEDPAC report wanted to discourage readmissions like Mrs. C’s.  And so it wasn’t a terrible surprise that, rolled into the 2010 Affordable Care Act, was a section called the Hospital Readmission Reduction Program (HRRP) which created a system for Medicare to penalize hospitals with ‘high’ readmission rates.  The program was rolled out in 2013.

At first, the program seemed to work like a charm.  Hospitals significantly ramped up their efforts at care coordination.  Teams of nurses and aids were assembled to make sure patients would get their medications as prescribed upon discharge and to check on patients once they got home.

Hospital readmission rates suddenly dropped and Medicare started saving money.  A staggering 81% of all hospitals suffered penalties in 2018, which translates to ~$500 million or 0.3% of total Medicare payments to hospitals.

A complex analysis

But there’s more to this too good to be true story.

The HRRP penalty schemes are risk-adjusted based on administrative claims data. Risk-adjustment is a statistical procedure to take into account the diversity in complexity and severity of disease among patients so they can be compared.

Physicians know that risk-adjusted claims data are of dubious value because they themselves are often the reluctant data entry clerk in the byzantine scheme that starts with adding diagnostic items to the medical chart and ends with generating a coded billing claim for Medicare.  Needless to say, there’s a huge potential disconnect between what a claim attempts to convey and the actual condition of a given patient.

Yet another major problem is that the risk-adjustment employed by the HRRP does not take socioeconomic status into account, when that is arguably the single biggest driver of poor outcomes and of hospital readmissions.  The creators of the HRRP program seem to believe that a hospital located in poor area shouldn’t get a break for having high readmission rates, perhaps because they believe that hospital systems in general should be mindful of health inequities and address “care gaps” (differences in care provided to poor vs. affluent patients)  in their neighborhoods no matter what.

Finally, the initial out of the gate benchmark for readmission rates on which the HRRP would adjudicate the need for a penalty was a national average.  In such a scheme, a Johns Hopkins Hospital serving inner city Baltimore could be pitted against a regional hospital in rural Montana with an entirely different patient demographics.  This made the regional hospital in Montana very happy.

Gaming the metrics

Regardless of these technical considerations, it is an adage of social science that any metric will be gamed, and healthcare is unfortunately not immune to that law.

One tool increasingly used by hospitals to comply with Medicare payments rules is to admit patients to short stay units, under so-called “observation status.”  Another is to put pressure on emergency departments to avoid readmitting certain types of patients.

So, instead of primarily functioning as a triage operation where sick patients would be turned over to the care of the cardiologist in the hospital, the ER has been increasingly housing and managing heart failure patients to save the hospital money.

But the ER physician or the hospitalist supervising the short stay unit and who just meet a patient in the setting of an acute illness are poorly equipped to know which heart failure patient to discharge after a diuretic dose and which to keep for advanced heart failure therapies.

Source: MedPAC’s June 2018 Report (http://bit.ly/2CXD0fJ)

Nowadays, the cardiologist is increasingly insulated from those decisions.  I have personally experienced with alarming frequency instances where I learn only after the fact that a complex patient of mine has been treated for heart failure in the ED.

And my experience seems to be shared by many of my cardiology colleagues, especially among cardiologists who work in academic centers that are most affected by the policy.  Luckily, some of them are also clinician-scientists that can do more than just whine to colleagues about the new policy.  They can also study its outcomes.

What do the outcomes data show?

In a pivotal study, a group of cardiologists (Gupta et al.) saw that the drop in readmissions that followed the introduction of HRRP was unfortunately accompanied by a reversal in the decade long downward trend in heart failure mortality. This reversal suggested a serious potential harm from the policy.

But the possibility of harm was quickly challenged by another group of researchers led by one of the biggest names in health policy: Harlan Krumholz, a cardiologist who directs the influential Center for Outcomes Research and Evaluation at Yale University.

The Krumholz group analyzed millions of Medicare claims data from 2007 to 2016.  They also found that mortality for heart failure patients increased, but concluded that no causal link between decreasing readmission and increased mortality could be established.

Krumholz at al. noted that mortality rates for heart failure started climbing before the HRRP program was announced and they noted no inflection point in mortality rates with the policy announcement in 2010.  The evidence for their claim is highlighted in the table below:

As can be seen in the boxed row, Krumholz’ team concludes that the increasing mortality slope post-HRRP is no different from that pre-HRRP because the change did not reach statistical significance at the obligatory and arbitrary P<0.05 level.  The actual P-value was 0.11 and the confidence interval for the positive increase in mortality slope of 0.006 is (-.002 to .015).  Even poor students of epistemology would be loathe to conclude this result excludes a signal of harm.  It seems entirely plausible that, with all the limitations of the data set in question, mortality may in fact have accelerated after the institution of the HRRP.  Yet Krumholz insists that no signal of harm is to be considered.

But this did not stop another group of cardiologists (Wadhera et al.) from adding their contribution to the HRRP literature. Using the same data-set that the Krumholz group used — Medicare claims data — these researchers found once again that accelerating mortality coincided with the announcement of the HRRP.  More troubling, they also demonstrated that mortality rose primarily among patients not readmitted to the hospital.

A messy science

Admittedly this whole business of analysis is incredibly messy, with a number of moving parts.

My brief summary doesn’t do justice to a variety of maneuvers taken by the various groups to account for many of the limitations inherent in this type of study. Two of the competing analyses (Krumholz, Wadhera) used Medicare claims data while the other (Gupta) used a more limited voluntary registry.

During the time period in question, there were also other policy changes such as the introduction of new hospital billing codes (MS-DRG) that sought to adjust hospital payment rates to patient complexity. Better patient coding meant higher reimbursement from Medicare.  Armies of “documenters” were then employed by hospitals to capture more revenue.

This means that the claims data gathered by the researchers might look significantly different from one time period to the next even if the patients themselves were ostensibly the same. As the readmission rate is risk-adjusted, it is eminently plausible and likely that systematically upcoding patient risk could actually have been the primary driver of the drop in hospital readmission rate.

The other program playing a confounding role is the Recovery Audit Contractor (RAC) program begun in 2010 to reduce payments for inappropriate hospital admissions. Hospitals responded to denials for inpatient admissions by expanding ‘observation status’ stays. Which was the biggest driver for expanded observation stays? RAC or the HRRP? Once again an exact attribution is impossible.

Denying the obvious?

Despite the messiness of the data and the variety of analytic methods used, a consistent and uncontested observation remains: Heart failure mortality has increased in the last decade. The question being hotly contested is Why?

Oddly, Dr. Krumholz is steadfast in denying the possibility that the policy may have caused harm, even though the independent and contradictory conclusions from the other research groups at least raises a reasonable doubt.  And Dr. Krumholz has been quick to cast shade on research that does not conform to his conclusion.

By tweet he appears to ask for a level of detail his own papers lack, and he questions the legitimacy of another group’s data-set, all the while resisting any calls to put the program on hold despite the paucity of evidence showing benefit, the signal for harm, and perhaps most importantly, the concern of clinicians who see a mechanism for harm.

Greatly admire @rwyeh and his group…appreciate his focus on readmission &public policy. For such high-profile article, really need more info about statistical weighting. Methods should be sufficient so others can reproduce results. Can’t do that here. Look forward to more info.

— Harlan Krumholz (@hmkyale) December 21, 2018

Can you account for why the registry you used had, on average, such a small number of patients per site. Did you determine how many patients coded with heart failure by CMS were in the Registry? Just curious about the selection. It may not explain your results…but is a question. https://t.co/WSvJAN8Jjd

— Harlan Krumholz (@hmkyale) June 2, 2018

Dr. Krumholz also places much weight on an independent analysis carried out by MEDPAC which concluded there was no link between policy and the uptick in mortality. This particular conclusion rests heavily on the assertion that heart failure patients in 2016 were much sicker than patients in 2010. Recall that this coincides with a period of more intensive coding over the same time frame, so it is impossible to say this with any confidence.

The MEDPAC conclusion also relies on an analysis that finds no correlation between hospital level readmission and mortality rates.

While technically true, that conclusion overlooks that a large number of hospitals exhibited reduced readmission rates and increased mortality. Perhaps MEDPAC feels that patients dying at low readmit/high mortality hospitals should be mollified by the knowledge that somewhere there’s a low-readmit/low mortality hospital to balance things out?

That Krumholz and MEDPAC display such certainty about the direction of the signal they observe, and take pains to discount other possibilities seems strange and suggests that pre-existing biases may be at work. What might those biases be?

Conflicts of interest: You get what you pay for

In a world where heads roll for undisclosed personal financial conflicts of interest, it is remarkable that the current dispute, while full of scintillating exchanges about “propensity weighting” and other arcane points of statistics, does not reference any other potential conflicts at work that might affect the conclusions being reached.

Medicare’s decision to start the HRRP program didn’t come in a vacuum. It was inspired by years of research from Dr. Krumholz himself, who suggested that preventing admissions should be a goal for any policy that would aim to move the system from one paying for “volume” to one paying for “value.”

As far back as 2003, Krumholz held the view that:

“Hospitals may not support programs that improve the quality of care delivered to heart failure patients because these programs lower readmission rates and empty beds, and therefore further diminish already-declining revenues.”

If Krumholz’s unfavorable and crudely simplistic view of the operations and motivations of hospitals (and of the still relatively independent physicians staffing those hospitals) informs his position on health policy, it stands to reason that serious blinders would prevent him from seeing any evidence of harm in a particular policy that promotes the same view.

But that’s not all.  Krumholz’s group at Yale received grants from CMS under the auspices of the Measure and Instrument Development and Support (MIDS) program to study and produce the metrics and instruments needed to devise the readmission measures.

The MIDS program supports the “development and use of clinical quality measures which remains a critical healthcare priority and the tool of choice for improving quality of care at the national, community and facility levels” and it allocates $1.6 billion dollars to this purpose.

Thanks to a bipartisan act of Congress, a helpful little website, usaspending.gov, provides contract level detail about payments made to the Krumholz’s group from the MIDS program.  Those payments can be seen in the table below:

The numbers are staggering. I know little about how to interpret these data about federal contracts, but it sure appears that the Yale-New Haven Health Service group led by Krumholz has received $144 million dollars since 2008.

Yet the only clue to these payments in Krumholz’ published analyses of the HRRP program comes in one disclosure sentence in a footnote, as seen here:

It seems to me that the disclosure is hardly proportional to the amount of funding that his group receives and understates the inherent pressures it must be under to demonstrate that the policy did not actually result in higher mortality.

And recall that MEDPAC’s “independent” analysis that also rejected a policy-mortality link came from the organization that recommended the policy to begin with.  The bottom line is this: There’s a tremendous amount of face to lose and a massive source of institutional funding at risk if the policy is found to be harmful.

It now becomes more clear why, in the following tweets, Dr. Krumholz feels that only he can say anything definitive about readmission rates and mortality:

Um, @JAMA_current, even the authors say they cannot say their findings are causal… "but whether this finding is a result of the policy requires further research.” Why do you promote the paper as proving harm? Need to treat twitter like you do any of your Editorial comments. pic.twitter.com/xugasHvPWY

— Harlan Krumholz (@hmkyale) December 26, 2018

6/Hospitals that have improved their readmission rates tended to improve their mortality rates. Published in @JAMA_current… https://t.co/R7FXLwTMPp

— Harlan Krumholz (@hmkyale) December 22, 2018

A diversity of biases

Biases are ubiquitous.  When I was a cardiologist-in-training, spending hours on the hospital consult service for a fixed salary, I vividly recall looking for ways to avoid doing any work I considered unimportant or banal: The minor cardiac enzyme leak in a patient with a widespread infection; The extra heartbeats on the ECG that the ER physician didn’t like the look of; etc.  “Are you sure you need an official consult?”  “The chances we’re going to recommend doing anything about a small enzyme leak in an 80-year-old with a severe lung infection are very low… “ I was even successful sometimes.

Contrast these comments to my demeanor in private practice where I am acutely aware that my income relies on such consults: “I just need the patients name or room number…” “I’ll take care of it!…” “I can put in the orders if you’d like!”

But there are other biases and incentives that motivate human beings, apart from personal financial incentives. Do they pale relative to the financial ones as is so often claimed? How does one begin to quantify them?

When it comes to the HRRP policy, no individual person’s bank account ballooned every time a patient didn’t get admitted. And yet this is a story of ideological bias that drove the design of policy and now claims ‘success’ for its own program.  The HRRP saga is illustrative of the importance of non-financial bias and of the dangers of blinding ourselves to that bias.

The story also highlights the downsides of tweaking healthcare systems that were built to deliver more care.

Clearly, I personally have a direct financial conflict of interest to provide more care. Since I haven’t talked anyone out of a consult in 8 years, I’m probably guilty of participating in a system that detractors appropriately criticize for promoting overuse of healthcare.

But the problem is that some of those consults I was trying to avoid as a fellow ended up really needing a cardiologist. There was the 55-year-old Cambodian woman admitted to the medical intensive care unit with pneumonia who went into atrial fibrillation. I recall rolling my eyes and thinking that the ICU could certainly handle this without a cardiologist. It turned out she didn’t have a pneumonia. It was pulmonary edema from heart failure related to undiagnosed rheumatic mitral valve disease. She had been in the wrong unit. She needed diuresis, heart rate control, and eventual surgery to replace her valve, not antibiotics. Less isn’t always more.

Attempt to reduce inappropriate hospital admissions? Get ready to pay a price.  To contradict Dr. Krumholz, it is entirely probable that we are underestimating the upside of our current system when we contemplate changing the status quo.

We are underestimating the downside of our current system when we contemplate change. We need to take some risks to do better. #abimf2013

— Harlan Krumholz (@hmkyale) August 5, 2013

Final Thoughts

Ironically, the HRRP quagmire offers a number of clarifying lessons.

Empiricism in social policy is a subjective enterprise. The often parroted conclusion is that cold, hard, unbiased evidence trumps the biased, unmeasurable judgment of clinicians. Yet, frequently, real world data-sets are complex, the choice of analytic paths can be highly variable, and the instruments to measure success are often imperfect. As it relates to HRRP, which analysis should we trust? The choice requires faith. And if the currency here is faith, perhaps the concern of clinicians at the bedside has more value than advertised.

Metrics won’t save us. The narrative of metrics is an appealing one that promises hard and objective accountability. The problem comes when the metric (readmission) becomes disconnected from outcomes that actually matter (death). False and blind prophets are good descriptors for those who claim to be unable to see without metrics. The fools in this enterprise are easy to identify as those who think the answers lie with ever better metrics.

Conflicts of interest: Going beyond the simple narrative.  Focusing on biases induced by personal financial interests is a mistake. Personal enrichment is just one bias in a sea of conflicts. In the healthcare context, financial disclosures—while clarifying in themselves—may simply give cover to other, more perverse biases, unless those other biases are equally disclosed.  It requires diligence to ascertain the impact, and direction of bias.  Rarely do we get the opportunity to observe the direction of bias in policy research.  In the case of the HRRP, the presence of bias was made evident because research groups with opposing biases (clinician-scientists versus policy wonks) have reached conclusions that would be expected on the basis of those pre-existing biases.  How often is the problem of such bias examined in the design, implementation, and analysis of health policy?

Beware of technocrats with all the answers.  I am reasonably sure that if practicing clinicians would have been asked to devise a rule to reduce heart failure readmissions for the whole population they would have refused. It seems too challenging a task to get right.  Even if clinicians can be induced to participate in the design of such a policy, most would readily acknowledge the likelihood that it could harm some patients. It requires a special type of hubris to design a policy and refuse to acknowledge its potential for harm. Unfortunately, hubris within a public health community that believes only they can give us a better health system is more feature than bug.

As for Mrs. C, she has been home for 16 days. All fingers on both hands are currently crossed.

Anish Koka is a cardiologist in private practice in Philadelphia.  He can be followed on Twitter @anish_koka. This post originally appeared here on The Accad & Koka Report. 

Commissioning Healthcare Policy: Hospital Readmission and Its Price Tag published first on https://wittooth.tumblr.com/

[isaacscrawford]

Commissioning Healthcare Policy: Hospital Readmission and Its Price Tag

By ANISH KOKA MD 

The message comes in over the office slack line at 1:05 pm. There are four patients in rooms, one new, 3 patients in the waiting room. Really, not an ideal time to deal with this particular message.

“Kathy the home care nurse for Mrs. C called and said her weight yesterday was 185, today it is 194, she has +4 pitting edema, heart rate 120, BP 140/70 standing, 120/64 sitting”

I know Mrs. C well. She has severe COPD from smoking for 45 of the last 55 years. Every breath looks like an effort because it is. The worst part of it all is that Mrs. C just returned home from the hospital just days ago.

The youngest of six children, Mrs. C was born with many embedded disadvantages. Being born black in a poor West Philadelphia neighborhood in the 1960s is a story that too often writes itself with a bad ending.  But Mrs. C avoided the usual pitfalls that derail young women in the neighborhood early. No drugs. No alcohol. No teenage pregnancies. Finished high school. Mrs. C. worked for the hospital as a unit clerk, had her own place, health benefits, and even a retirement plan.

Certain life habits, however, carry a heavy price. George Burns, the comedian never pictured without a cigar who died past his hundredth birthday, may have been immune to the effects of tobacco.  Mrs. C was not. She started smoking when she was 16. She doesn’t recall why. Her dad smoking didn’t help perhaps. Nausea racked her body after that first drag. It eased up after. Too bad.

That measly cigarette became the great addiction of her life. Day by day, the exquisitely thin membranes of the lungs that mediate gas exchange were destroyed. By the time the disease manifests with shortness of breath and bluish tinged lips, it’s too late. Short of the very few who qualify for a lung transplant, the efforts of doctors at this point are for mitigation rather than cure.

Complicating things further, in Mrs. C’s case, the normally low pressure vascular circuit of her lungs became a high pressure circuit that places ever increasing demands on the normally thin-muscled right ventricle of the heart.  This jeopardizes the ability of her heart to handle changes in blood volume.

A little extra fluid and the right side of the heart ends up causing unbearable swelling in her legs.  A little dehydration and severe disabling dizziness on standing ensues. Adding to that that her tenuous lung function decompensates with the slightest respiratory infection, that chronic steroid treatment to decrease her wheezing suppresses her immune system, and that the young man down the street helpfully drops off Newports at her home for a few extra dollars, and it’s easy to see why the hospital is her second home.

The most recent admission to the hospital was for kidney failure related to taking too much fluid off with diuretics. What was to be a short stay for gentle hydration turned into a longer stay when a pneumonia complicated the matters (though a trip to the intensive care unit and a ventilator was barely but fortunately avoided).  She was treated by the pulmonology team and sent home on a lower dose of diuretics.

The situation I am now confronting puts me in a quandary.  Her edema and weight are up markedly just a few days after returning home.  Could her fluid overload be because her kidneys are shutting down? Or does she just need more aggressive diuresis?

Should I guess? Knowing her present renal function would be helpful. But even if the Theranos lab I could appeal to for help wasn’t fictional, I would have to get her to my office everyday or every other day while adjusting her diuretic dose.

And so it comes to be that, days removed from a hospital admission, I’m sending her back to the hospital to be readmitted.  According to some, this is not supposed to happen.

A policy on readmission

In 2008, the commission that advises Medicare – the Medicare Payment Advisory Commission (MEDPAC), issued a report that focused on hospital readmissions.

The focus on hospital readmissions had been of great interest to the health policy community for some time. At the core of this interest lies the belief that hospitals and physicians are incentivized to treat patients rather than prevent admissions.

The MEDPAC report wanted to discourage readmissions like Mrs. C’s.  And so it wasn’t a terrible surprise that, rolled into the 2010 Affordable Care Act, was a section called the Hospital Readmission Reduction Program (HRRP) which created a system for Medicare to penalize hospitals with ‘high’ readmission rates.  The program was rolled out in 2013.

At first, the program seemed to work like a charm.  Hospitals significantly ramped up their efforts at care coordination.  Teams of nurses and aids were assembled to make sure patients would get their medications as prescribed upon discharge and to check on patients once they got home.

Hospital readmission rates suddenly dropped and Medicare started saving money.  A staggering 81% of all hospitals suffered penalties in 2018, which translates to ~$500 million or 0.3% of total Medicare payments to hospitals.

A complex analysis

But there’s more to this too good to be true story.

The HRRP penalty schemes are risk-adjusted based on administrative claims data. Risk-adjustment is a statistical procedure to take into account the diversity in complexity and severity of disease among patients so they can be compared.

Physicians know that risk-adjusted claims data are of dubious value because they themselves are often the reluctant data entry clerk in the byzantine scheme that starts with adding diagnostic items to the medical chart and ends with generating a coded billing claim for Medicare.  Needless to say, there’s a huge potential disconnect between what a claim attempts to convey and the actual condition of a given patient.

Yet another major problem is that the risk-adjustment employed by the HRRP does not take socioeconomic status into account, when that is arguably the single biggest driver of poor outcomes and of hospital readmissions.  The creators of the HRRP program seem to believe that a hospital located in poor area shouldn’t get a break for having high readmission rates, perhaps because they believe that hospital systems in general should be mindful of health inequities and address “care gaps” (differences in care provided to poor vs. affluent patients)  in their neighborhoods no matter what.

Finally, the initial out of the gate benchmark for readmission rates on which the HRRP would adjudicate the need for a penalty was a national average.  In such a scheme, a Johns Hopkins Hospital serving inner city Baltimore could be pitted against a regional hospital in rural Montana with an entirely different patient demographics.  This made the regional hospital in Montana very happy.

Gaming the metrics

Regardless of these technical considerations, it is an adage of social science that any metric will be gamed, and healthcare is unfortunately not immune to that law.

One tool increasingly used by hospitals to comply with Medicare payments rules is to admit patients to short stay units, under so-called “observation status.”  Another is to put pressure on emergency departments to avoid readmitting certain types of patients.

So, instead of primarily functioning as a triage operation where sick patients would be turned over to the care of the cardiologist in the hospital, the ER has been increasingly housing and managing heart failure patients to save the hospital money.

But the ER physician or the hospitalist supervising the short stay unit and who just meet a patient in the setting of an acute illness are poorly equipped to know which heart failure patient to discharge after a diuretic dose and which to keep for advanced heart failure therapies.

Source: MedPAC’s June 2018 Report (http://medpac.gov/docs/default-source/reports/jun18_ch1_medpacreport_sec.pdf?sfvrsn=0)

Nowadays, the cardiologist is increasingly insulated from those decisions.  I have personally experienced with alarming frequency instances where I learn only after the fact that a complex patient of mine has been treated for heart failure in the ED.

And my experience seems to be shared by many of my cardiology colleagues, especially among cardiologists who work in academic centers that are most affected by the policy.  Luckily, some of them are also clinician-scientists that can do more than just whine to colleagues about the new policy.  They can also study its outcomes.

What do the outcomes data show?

In a pivotal study, a group of cardiologists (Gupta et al.) saw that the drop in readmissions that followed the introduction of HRRP was unfortunately accompanied by a reversal in the decade long downward trend in heart failure mortality. This reversal suggested a serious potential harm from the policy.

But the possibility of harm was quickly challenged by another group of researchers led by one of the biggest names in health policy: Harlan Krumholz, a cardiologist who directs the influential Center for Outcomes Research and Evaluation at Yale University.

The Krumholz group analyzed millions of Medicare claims data from 2007 to 2016.  They also found that mortality for heart failure patients increased, but concluded that no causal link between decreasing readmission and increased mortality could be established.

Krumholz at al. noted that mortality rates for heart failure started climbing before the HRRP program was announced and they noted no inflection point in mortality rates with the policy announcement in 2010.  The evidence for their claim is highlighted in the table below:

As can be seen in the boxed row, Krumholz’ team concludes that the increasing mortality slope post-HRRP is no different from that pre-HRRP because the change did not reach statistical significance at the obligatory and arbitrary P<0.05 level.  The actual P-value was 0.11 and the confidence interval for the positive increase in mortality slope of 0.006 is (-.002 to .015).  Even poor students of epistemology would be loathe to conclude this result excludes a signal of harm.  It seems entirely plausible that, with all the limitations of the data set in question, mortality may in fact have accelerated after the institution of the HRRP.  Yet Krumholz insists that no signal of harm is to be considered.

But this did not stop another group of cardiologists (Wadhera et al.) from adding their contribution to the HRRP literature. Using the same data-set that the Krumholz group used — Medicare claims data — these researchers found once again that accelerating mortality coincided with the announcement of the HRRP.  More troubling, they also demonstrated that mortality rose primarily among patients not readmitted to the hospital.

A messy science

Admittedly this whole business of analysis is incredibly messy, with a number of moving parts.

My brief summary doesn’t do justice to a variety of maneuvers taken by the various groups to account for many of the limitations inherent in this type of study. Two of the competing analyses (Krumholz, Wadhera) used Medicare claims data while the other (Gupta) used a more limited voluntary registry.

During the time period in question, there were also other policy changes such as the introduction of new hospital billing codes (MS-DRG) that sought to adjust hospital payment rates to patient complexity. Better patient coding meant higher reimbursement from Medicare.  Armies of “documenters” were then employed by hospitals to capture more revenue.

This means that the claims data gathered by the researchers might look significantly different from one time period to the next even if the patients themselves were ostensibly the same. As the readmission rate is risk-adjusted, it is eminently plausible and likely that systematically upcoding patient risk could actually have been the primary driver of the drop in hospital readmission rate.

The other program playing a confounding role is the Recovery Audit Contractor (RAC) program begun in 2010 to reduce payments for inappropriate hospital admissions. Hospitals responded to denials for inpatient admissions by expanding ‘observation status’ stays. Which was the biggest driver for expanded observation stays? RAC or the HRRP? Once again an exact attribution is impossible.

Denying the obvious?

Despite the messiness of the data and the variety of analytic methods used, a consistent and uncontested observation remains: Heart failure mortality has increased in the last decade. The question being hotly contested is Why?

Oddly, Dr. Krumholz is steadfast in denying the possibility that the policy may have caused harm, even though the independent and contradictory conclusions from the other research groups at least raises a reasonable doubt.  And Dr. Krumholz has been quick to cast shade on research that does not conform to his conclusion.

By tweet he appears to ask for a level of detail his own papers lack, and he questions the legitimacy of another group’s data-set, all the while resisting any calls to put the program on hold despite the paucity of evidence showing benefit, the signal for harm, and perhaps most importantly, the concern of clinicians who see a mechanism for harm.

Greatly admire @rwyeh and his group…appreciate his focus on readmission &public policy. For such high-profile article, really need more info about statistical weighting. Methods should be sufficient so others can reproduce results. Can’t do that here. Look forward to more info.

— Harlan Krumholz (@hmkyale) December 21, 2018

Can you account for why the registry you used had, on average, such a small number of patients per site. Did you determine how many patients coded with heart failure by CMS were in the Registry? Just curious about the selection. It may not explain your results…but is a question. https://t.co/WSvJAN8Jjd

— Harlan Krumholz (@hmkyale) June 2, 2018

Dr. Krumholz also places much weight on an independent analysis carried out by MEDPAC which concluded there was no link between policy and the uptick in mortality. This particular conclusion rests heavily on the assertion that heart failure patients in 2016 were much sicker than patients in 2010. Recall that this coincides with a period of more intensive coding over the same time frame, so it is impossible to say this with any confidence.

The MEDPAC conclusion also relies on an analysis that finds no correlation between hospital level readmission and mortality rates.

While technically true, that conclusion overlooks that a large number of hospitals exhibited reduced readmission rates and increased mortality. Perhaps MEDPAC feels that patients dying at low readmit/high mortality hospitals should be mollified by the knowledge that somewhere there’s a low-readmit/low mortality hospital to balance things out?

That Krumholz and MEDPAC display such certainty about the direction of the signal they observe, and take pains to discount other possibilities seems strange and suggests that pre-existing biases may be at work. What might those biases be?

Conflicts of interest: You get what you pay for

In a world where heads roll for undisclosed personal financial conflicts of interest, it is remarkable that the current dispute, while full of scintillating exchanges about “propensity weighting” and other arcane points of statistics, does not reference any other potential conflicts at work that might affect the conclusions being reached.

Medicare’s decision to start the HRRP program didn’t come in a vacuum. It was inspired by years of research from Dr. Krumholz himself, who suggested that preventing admissions should be a goal for any policy that would aim to move the system from one paying for “volume” to one paying for “value.”

As far back as 2003, Krumholz held the view that:

“Hospitals may not support programs that improve the quality of care delivered to heart failure patients because these programs lower readmission rates and empty beds, and therefore further diminish already-declining revenues.”

If Krumholz’s unfavorable and crudely simplistic view of the operations and motivations of hospitals (and of the still relatively independent physicians staffing those hospitals) informs his position on health policy, it stands to reason that serious blinders would prevent him from seeing any evidence of harm in a particular policy that promotes the same view.

But that’s not all.  Krumholz’s group at Yale received grants from CMS under the auspices of the Measure and Instrument Development and Support (MIDS) program to study and produce the metrics and instruments needed to devise the readmission measures.

The MIDS program supports the “development and use of clinical quality measures which remains a critical healthcare priority and the tool of choice for improving quality of care at the national, community and facility levels” and it allocates $1.6 billion dollars to this purpose.

Thanks to a bipartisan act of Congress, a helpful little website, usaspending.gov, provides contract level detail about payments made to the Krumholz’s group from the MIDS program.  Those payments can be seen in the table below:

The numbers are staggering. I know little about how to interpret these data about federal contracts, but it sure appears that the Yale-New Haven Health Service group led by Krumholz has received $144 million dollars since 2008.

Yet the only clue to these payments in Krumholz’ published analyses of the HRRP program comes in one disclosure sentence in a footnote, as seen here:

It seems to me that the disclosure is hardly proportional to the amount of funding that his group receives and understates the inherent pressures it must be under to demonstrate that the policy did not actually result in higher mortality.

And recall that MEDPAC’s “independent” analysis that also rejected a policy-mortality link came from the organization that recommended the policy to begin with.  The bottom line is this: There’s a tremendous amount of face to lose and a massive source of institutional funding at risk if the policy is found to be harmful.

It now becomes more clear why, in the following tweets, Dr. Krumholz feels that only he can say anything definitive about readmission rates and mortality:

Um, @JAMA_current, even the authors say they cannot say their findings are causal… "but whether this finding is a result of the policy requires further research.” Why do you promote the paper as proving harm? Need to treat twitter like you do any of your Editorial comments. pic.twitter.com/xugasHvPWY

— Harlan Krumholz (@hmkyale) December 26, 2018

6/Hospitals that have improved their readmission rates tended to improve their mortality rates. Published in @JAMA_current… https://t.co/R7FXLwTMPp

— Harlan Krumholz (@hmkyale) December 22, 2018

A diversity of biases

Biases are ubiquitous.  When I was a cardiologist-in-training, spending hours on the hospital consult service for a fixed salary, I vividly recall looking for ways to avoid doing any work I considered unimportant or banal: The minor cardiac enzyme leak in a patient with a widespread infection; The extra heartbeats on the ECG that the ER physician didn’t like the look of; etc.  “Are you sure you need an official consult?”  “The chances we’re going to recommend doing anything about a small enzyme leak in an 80-year-old with a severe lung infection are very low… “ I was even successful sometimes.

Contrast these comments to my demeanor in private practice where I am acutely aware that my income relies on such consults: “I just need the patients name or room number…” “I’ll take care of it!…” “I can put in the orders if you’d like!”

But there are other biases and incentives that motivate human beings, apart from personal financial incentives. Do they pale relative to the financial ones as is so often claimed? How does one begin to quantify them?

When it comes to the HRRP policy, no individual person’s bank account ballooned every time a patient didn’t get admitted. And yet this is a story of ideological bias that drove the design of policy and now claims ‘success’ for its own program.  The HRRP saga is illustrative of the importance of non-financial bias and of the dangers of blinding ourselves to that bias.

The story also highlights the downsides of tweaking healthcare systems that were built to deliver more care.

Clearly, I personally have a direct financial conflict of interest to provide more care. Since I haven’t talked anyone out of a consult in 8 years, I’m probably guilty of participating in a system that detractors appropriately criticize for promoting overuse of healthcare.

But the problem is that some of those consults I was trying to avoid as a fellow ended up really needing a cardiologist. There was the 55-year-old Cambodian woman admitted to the medical intensive care unit with pneumonia who went into atrial fibrillation. I recall rolling my eyes and thinking that the ICU could certainly handle this without a cardiologist. It turned out she didn’t have a pneumonia. It was pulmonary edema from heart failure related to undiagnosed rheumatic mitral valve disease. She had been in the wrong unit. She needed diuresis, heart rate control, and eventual surgery to replace her valve, not antibiotics. Less isn’t always more.

Attempt to reduce inappropriate hospital admissions? Get ready to pay a price.  To contradict Dr. Krumholz, it is entirely probable that we are underestimating the upside of our current system when we contemplate changing the status quo.

We are underestimating the downside of our current system when we contemplate change. We need to take some risks to do better. #abimf2013

— Harlan Krumholz (@hmkyale) August 5, 2013

Final Thoughts

Ironically, the HRRP quagmire offers a number of clarifying lessons.

Empiricism in social policy is a subjective enterprise. The often parroted conclusion is that cold, hard, unbiased evidence trumps the biased, unmeasurable judgment of clinicians. Yet, frequently, real world data-sets are complex, the choice of analytic paths can be highly variable, and the instruments to measure success are often imperfect. As it relates to HRRP, which analysis should we trust? The choice requires faith. And if the currency here is faith, perhaps the concern of clinicians at the bedside has more value than advertised.

Metrics won’t save us. The narrative of metrics is an appealing one that promises hard and objective accountability. The problem comes when the metric (readmission) becomes disconnected from outcomes that actually matter (death). False and blind prophets are good descriptors for those who claim to be unable to see without metrics. The fools in this enterprise are easy to identify as those who think the answers lie with ever better metrics.

Conflicts of interest: Going beyond the simple narrative.  Focusing on biases induced by personal financial interests is a mistake. Personal enrichment is just one bias in a sea of conflicts. In the healthcare context, financial disclosures—while clarifying in themselves—may simply give cover to other, more perverse biases, unless those other biases are equally disclosed.  It requires diligence to ascertain the impact, and direction of bias.  Rarely do we get the opportunity to observe the direction of bias in policy research.  In the case of the HRRP, the presence of bias was made evident because research groups with opposing biases (clinician-scientists versus policy wonks) have reached conclusions that would be expected on the basis of those pre-existing biases.  How often is the problem of such bias examined in the design, implementation, and analysis of health policy?

Beware of technocrats with all the answers.  I am reasonably sure that if practicing clinicians would have been asked to devise a rule to reduce heart failure readmissions for the whole population they would have refused. It seems too challenging a task to get right.  Even if clinicians can be induced to participate in the design of such a policy, most would readily acknowledge the likelihood that it could harm some patients. It requires a special type of hubris to design a policy and refuse to acknowledge its potential for harm. Unfortunately, hubris within a public health community that believes only they can give us a better health system is more feature than bug.

As for Mrs. C, she has been home for 16 days. All fingers on both hands are currently crossed.

Anish Koka is a cardiologist in private practice in Philadelphia.  He can be followed on Twitter @anish_koka. This post originally appeared here on The Accad & Koka Report. 

Article source:The Health Care Blog

[thegloober]

401 – Chip Conley: How To Find Your Calling At Age 50 (Or Older)

People always ask “What number do I need to retire?”

Which really means “Do I have enough?” “Will I have enough?” “Can I survive?” “Will I be happy?”

But people aren’t retiring as much as they used to.

People are reinventing. They’re working longer. Or starting something new. They’re getting a second wind.

But that hasn’t stopped anyone from asking the retirement question. So I wanted to break it down.

See, when someone asks “What number do I need to retire?” they’re really asking, “What is financial freedom?”

And “Can I have it?”

I say “yes.”

Except, first I have to change the equation. It’s not x amount of dollars = happiness for the rest of your life.

It’s x dollars + virtual net worth = financial freedom.

Here’s how I broke it down on the podcast. I was talking to Chip Conley. He’s the Head of Hospitality for Airbnb and second time “James Altucher Show” guest.

He first came on my podcast a couple years ago. (When I was living in Airbnbs)

I got a call from my host. “Hey, the Head of Hospitality for Airbnb is staying in the apartment right below you. Do you want to meet him?”

“Yes.”

He came downstairs with a bottle of wine. (Classic hospitable move.) I opened my computer and recorded our conversation.

Now Chip has a new book out called “Wisdom at Work.” So I invited him to the studio.

He’s 58 and he’s not retired. He said he found a new “calling.”

People with jobs and careers want to retire.

People with callings don’t.

So one way to “retire” is to find your calling. That’s one option. There’s another option, which I’ll get back to.

First, I asked Chip how he found his calling. Here’s what I found out:

1. Mine Your Mastery

“I spent 26 years being a boutique hotel entrepreneur. In 2013, the three young founder of Airbnb approached me. They said, ‘We want to democratize hospitality, will you join us?’”

He said yes.

Now, I should say, of course this opportunity was rare. It’s not everyday or in every lifetime that some future billionaires will approach you and ask for your help.

Chip already built up his skillset.

He had spent the majority of his career disrupting the standard hotel industry.

Boutique hotels were unheard of when he first got started.

His foundation matched their future.

But let’s say you’re reading this and thinking “No one’s going to approach me.”

That’s fine. Make a list. Write down all of your skills (the things you got really good at over the last 5, 10, 20 years). Then cross out all the ones you don’t enjoy doing.

That’s Step A: Mine your mastery.

Step B: Make a list of things you’re curious about.

Now draw lines between them. Where can you see yourself? And listen to your gut. “There’s lots of physical effects that can happen in our life that sort of tell you you’re on the right path,” Chip said.

That’s step 2.

  2. Get Clear

“Getting clear internally is probably one of the most important things for us to figure out in life,” Chip said. “Weirdly enough we don’t do a whole lot in people’s upbringing to help them get there.”

“Zero,” I said. “We don’t do anything to help people get there.”

“Yeah, thank you.”

“We just feed them facts they’re going to forget.”

And sometimes in the process of memorizing those facts, we forget to understand ourselves. Sometimes I even feel guilty when I try to get clear.

Like “Oh, no, I don’t have time to get clear with myself I have a spreadsheet to do.”

Here’s how I get clear: I go back to my 10 year old self. And I write down everything I loved back then.

That’s it. Don’t go any further. Don’t try to match it with job description. Don’t try to turn it into a company.

Just go to sleep.

Let your brain digest it.

And in the morning, I wake up less anxious. And more clear.

  3. Forget Who You Think You Are

Chip thought he was going to be CEO, Brian Chesky’s mentor. He was wrong.

“Within the first couple weeks, I realized… I’m the intern.”

“I was twice the age of the average employee. I didn’t understand the language being spoken in the hallways. I didn’t have an Uber or Lyft app on my phone at that time. I had never heard of the sharing the economy. I had gone from being the pioneer as a boutique hotel entrepreneur in my mid 20’s to being the establishment by 52.”

He could’ve quit.

He had all the excuses. “I’m too old.” “I don’t belong.” “I thought I was going to be the mentor.” etc.

He stayed.

“I decided the only way I’d survive here is to be as much of a curious learner as I am the wise sage.”

He combined what he knew with what he didn’t know.

“And that’s what a modern elder is,” he said. “A modern elder is not the traditional eder that passed and is regarded with reverence. The modern elder is all about relevance.”

“Ok, how do you become relevant?”

“By taking that timeless wisdom and applying it to modern problems.

For example: when Chip was building boutique hotels, he already faced the skepticism and mistrust since it didn’t have “Hilton” on the name. So he had to battle that.

With Airbnb, he could help meet skepticism head on.

So he saw they had this “superhost program.”

Which did two things: it reassured you that the host is good at what they do. And it incentivized the hosts to step it up.

When Chip started working at Airbnb, there were only 200 superhosts in the world. And Airbnb hadn’t added any new ones for over a year. They were even thinking about getting rid of the program.

Then Chip stepped in.

  He knew this could help fight skepticism and encourage people traveling to trust their host and trust Airbnb.

  And it worked.

Now, there are 700,000 superhosts around the world.

But if Chip held on to this idea of “No, I can’t be the intern. I have to be the mentor, he would’ve left. And he would’ve lost his chance to be the modern elder.

—–

OK, back to to retirement.

The second option: build your virtual net worth.

I’m 50. And I’ve gone broke more than once.

People tell me, “You don’t seem 50.”

And then I’m supposed to say, “Thank you.”

But I don’t. And I won’t.

Because I’m actually proud to be 50

It’s part of my virtual net worth.

I’ll explain.

Let’s just say you need $10 million to retire. (This is not a real example). I’m just giving easy numbers.

With $10 million you’ll make 4% a year after taxes, conservatively. So you can live a great life with $400K a year.

But here’s the flip side: if you love what you’re doing and you don’t have $10 million but you’re making $400K a year through multiple streams of income, it’s as if, virtually, you have $10 million.

And instead of retiring, you have your calling.

And you’re not worrying every morning at 6am “Do I have enough?” “Will I have enough?” “Can I survive?” “Will I be happy?”

Because you already are.

  The post 401 – Chip Conley: How To Find Your Calling At Age 50 (Or Older) appeared first on Altucher Confidential.

Source: https://bloghyped.com/401-chip-conley-how-to-find-your-calling-at-age-50-or-older/

[animeindoblog-blog]

Grancrest Senki – 08

New Post has been published on https://animeindo.org/blog/2018/02/25/grancrest-senki-08/

Grancrest Senki – 08

「会議は踊る」 (Kaigi wa Odoru) “The Congress Dances”

So close to happily ever after.

Of the two halves of this episode, the romance of Marrine and Alexis was clearly the stronger. Not the one with the most potential—it was a standard star-crossed lovers tale, Romeo and Juliet without the suicide (though there’s still time)—but it was the better executed. That’s because while it was standard, they hit all the beats well, and they did one thing unexpectedly well: Alexis is actually a good guy. Partially as compared to this boorish ass, but he’s a good person in his own right—kind, soft-spoken, and with an idealistic outlook that’s backed up by a coherent philosophy of leadership. His philosophy might be wrong, or at least ahead of its time—they live in a world of iron and blood, with no United Nations or bias toward diplomacy to hash disagreements out—but it’s certainly attractive. I can’t blame Marrine for falling for him. Add on how he pursues Marrine a bit too aggressively at first—and his love at first sight line definitely reeked of playboy, not a good move—but when she gave him a clear no later on he agreed to back off … I mean, most men in our enlightened modern age don’t do that! (If you think that last comment was partially sarcastic, you are correct.) That first half isn’t going to win any awards, but it was executed well with an element that elevated. I’ll take it.

The second half wasn’t nearly so good. The council session at times reminded me of the Round Table Conference of Log Horizon fame, but it lacked a crucial ingredient: how to inject drama into an all-talking scene. The Round Table Conference scenes watched more like a battle than many of the actual battles in Log Horizon, complete with surprising turnarounds and finishing moves. The flow of Grancrest Senki’s council scene is all in Villar’s favor—until suddenly it’s not. As best I can tell, the tide was turned when Alexis tried to cede leadership of the Union to Villar, which forced Villar to refuse because of his Kreische blood (I guess), after which Dawson pounced and pushed for Alexis’ preferred method of peace, likely because Dawson knew that the Alliance would refuse and would probably attack Altirk. Dawson is using the Alliance to eliminate one of his internal enemies, making the classic geopolitical mistake of letting your disagreements extend beyond your country’s/alliance’s borders, even while it’s actually a savvy political move (just one that’s liable to haunt him in the end, when he needs someone to defend him and Altirk is gone).

The problem is that the drama was lacking during the scene, and I had to piece some of this together later (and I’m still not sure precisely why Villar’s Kreische blood matters, rulers were always marrying into each other’s families during time periods like this). A few extra lines to give an immediate reaction—Siluca provided some of this, but not enough—would have helped, like a sports announcer providing analysis on the play-by-play. Or really just stage the scene so that it’s not so flat. The result was a weak scene that played more puzzling than decisive, not helped by all the nameless nobles who never do anything other than say “No objection.” Coulda been better.

At least something didn’t go Theo’s way this time, since Altirk is liable to feel the business end of Marrine’s hammer. Which will probably just give opportunities for more Theo heroics, but I’m cool with that, gotta get those somewhere. Oh, expect I guess Rossini got expelled from the Union, but that was a foregone conclusion. Heh, got kicked out in two lines of dialogue, like the footnote he is. Screw you, Pederico!

Random thoughts:

Rant time: Marrine’s reason for backing out of the marriage is shit. So what if people are against your marriage, and willing to kill for it? When they kill someone, you mourn, and find the perpetrators if you can, but you do NOT let them change what you were going to do. You don’t let it stop you. It’s the same with terrorism—if they bomb the cafe, you rebuild the cafe just like it was. You make it clear: they do not matter. They are not important. They can cause pain, destruction, and death, but that cannot break you, and they cannot change you. As soon as you let these people change you, it’s over—they’ve won.

That said, the above is a very relatable mistake. It took a long time for human society to recognize this as the right response—and then only some societies some of the time. Mostly we still fuck it up. It’s a very human mistake, but a mistake nonetheless. (And also granted, a ruler has to worry about what will keep them in power. It could be that their marriage would lead to them being deposed and executed. But if you believe it to be the right course, do it anyway. They apparently did before. They just faltered at the last step.)

I highly approve of Alexis’ method of not-charity. Instead he’s tricking them (incentivizing them) into learning something they may enjoy, and which may enrich their lives. I’m totally down with that.

“I don’t know how long it will take, but I promise to come for you.” That’s the crucial line. That’s what Marrine is doing, even now.

My SECOND novel, Freelance Heroics, is available now! (Now in print!) (Also available: Firesign #1 Wage Slave Rebellion.) Sign up for my email list for updates. At stephenwgee.com, the latest post: Book 3 Progress Report.

Preview

[aicarzu]

‘You really have to take things one at a time’

Santa Fe mayoral candidate Joseph Maestas, a city councilor and former mayor of Española, talks to people at one of his campaign events at Java Joe’s coffee shop. (Eddie Moore/Albuquerque Journal)

SANTA FE, N.M. — “I’ve always gravitated to public service,” said Joseph Maestas, who has spent more than 30 years working for the government and has served nearly 14 years in elected positions.

He’d like to extend that streak as the next mayor of Santa Fe.

“Now that we’ve gone from a stronger mayor form of government, I think it’s critical that someone gets the keys to City Hall and hits the ground running – someone with proven leadership,” said the former city councilor and mayor of Española, who was elected to the Santa Fe City Council in 2014.

When it comes to public service, he says he feels like he’s standing on the shoulders of his ancestors. His maternal great-grandfather, Jose Amado Lucero, was one of the signers of the state constitution when New Mexico joined the union in 1912 and one of the founders of Española.

A brief biographical sketch of Lucero in the archives of the Office of the State Historian indicates that he was a businessman in Mora County, Santa Fe, Santa Cruz and Española. He served as schools superintendent in Rio Arriba County and as a Santa Fe County commissioner and probate judge prior to becoming a member of the state House of Representatives.

“And his son, Alfredo Lucero, was a Santa Fe County commissioner and clerk,” he added. “I came from a family with deep roots in the area.”

Maestas’ father was a standout athlete and his mom a cheerleader at Santa Cruz High, where his dad later worked as a science teacher. The family also operated a liquor store and had a 2½-acre farm where they raised animals for slaughter.

“So I was exposed to a business environment, even though it was family owned, since a young age,” he said.

Santa Cruz High closed after Maestas’ freshman year, so he finished up at what became Española Valley High School in 1978, playing football, and running cross-country and track.

City Councilor Joseph Maestas, who’s running for mayor in the March 6 municipal election, speaks to supporters of ranked-choice voting before a council meeting in December. Maestas was among the councilors who were against appealing a judge’s order mandating RCV to the state Supreme Court, a move approved by a slim council majority. (Eddie Moore/Albuquerque Journal)

Maestas said he “had” to leave the state after he graduated the University of New Mexico with an engineering degree. The country was in recession, the private sector wasn’t hiring and the better jobs were with the government, he said. A 27-month internship with the U.S. Department of Transportation’s Federal Highway Administration got his foot in the door. He earned a master’s degree in civil engineering from Arizona State University while still working his day job. He worked in five states and Washington, D.C., before a Highway Administration job opened up in Santa Fe in 1996.

It was a difficult time in his life. He and his then-wife, who was originally from El Paso, had just lost a child during childbirth.

“We were devastated. We wanted to go back home,” he said.

The couple had a second child, Joey, who is now a senior at Texas State University majoring in communications.

Maestas has lived through tragedy. Years later, his father died during his campaign for mayor of Española.

“That was hard. He never did get to see me sworn in as mayor,” he said.

Maestas is going through another tough time. Last month, he took a break from his campaign to visit his sister, currently a cancer patient in Austin.

“Right now, her prognosis is not good. She has probably a matter of weeks, if not a few months,” said Maestas, who made the trip with one of this three other sisters. “We drove over because we felt the window was closing in terms of saying our goodbyes.”

Maestas has talked about his sister, Carla, at some of the mayoral forums, usually when the topic turns to opioids and drug addiction.

“Carla had a history of drug abuse,” said Maestas, the lead sponsor of a resolution to pursue legal claims against opioid manufacturers and distributors that was passed by the council in December. “It takes a heavy toll on the family, and we went through all phases of dealing with a family member that’s addicted to drugs – the detox, family counseling, out-patient care. It’s not easy.”

His sister’s cancer was discovered during a physical exam she was required to take before entering a drug rehabilitation facility.

“My heart breaks for my mother because you’re not supposed to bury your children and she potentially could lose two children within a year’s time,” he said.

That’s because his brother Ben died last summer. Like their father, Ben had issues with alcoholism and he may have dabbled with drugs.

Ben died weeks after Maestas’ divorce from his second wife, U.S. District Judge Martha Vázquez, became final.

“I’ve learned that you really have to take things one at a time. When you lump it all together, the burden is much too heavy,” he said.

Maestas took a hiatus from government work during part of his term as mayor of Española from 2006 to 2010, where he was elected as a reformer. And in 2008, he ran for a Public Regulation Commission seat. But Jerome Block Jr. – whose PRC term was cut short by a scandal that included drug use and criminal charges for misuse of a state credit card and public campaign financing – won the seat.

Maestas took a job with the U.S. Census Bureau, then one as a division manager with the Bureau of Reclamation in Albuquerque. The commute from Española to the Duke City was too much, he said, so he didn’t run for re-election as mayor and moved to Santa Fe.

Maestas is building a home here that’s designed to accommodate his 88-year-old mother.

“She’s my best friend,” he said. “My goal is get my mom to live here. All her medical providers are here.”

Exercise has helped him cope with what life has thrown at him, too. The 57-year-old Maestas is a triathlete who usually makes the podium in his age group. Campaigning has cut into his training time. But he still manages to get in a workout nearly every day, be it a three-mile run, a 1,000-meter swim or a 30-mile bike ride.

The routine doesn’t just help him physically, it helps his him mentally. “When I exercise, I’m able to really think through things,” he said. “I can not only sort through whatever I’m dealing with, but I have better clarity navigating through it.”

Now he’s hoping his training, and experience as a public servant, will help him win the five-way race for mayor.

AGE: 57

EDUCATION: Bachelor of Science Degree in Civil Engineering, University of New Mexico; Master of Science Degree in Civil Engineering, Arizona State University.

OCCUPATION: Business Development Manager, Souder, Miller & Associates; and Santa Fe City Councilor for District 2.

1. Why are you running for mayor? What distinguishes you from your opponents?

As mayor, I want to help businesses and families thrive with a shared vision of a united Santa Fe and city government that facilitates the creation of jobs and affordable housing. A lifelong public servant with 33 years of federal civil service and 14 years as a municipal elected official; my engineering skills; and proven leadership distinguish me from my opponents.

2. What is the biggest issue facing city government and how would you address it?

The biggest issue is changing the culture at city hall to one of a 21st-century government and reforming its financial management. I would address it by:

A. Conducting a forensic financial audit;

B. Modernizing processes and policies;

C. Implementing an employee performance management system;

D. Developing a balanced 2019 budget; and

E. Updating economic development and land use plans.

3. How would you encourage more affordable housing in Santa Fe? Do you support development of more rental apartments in town?

I would encourage more affordable housing by issuing bonds as a permanent funding source; developing a sustainable city support (land donations, etc.) plan for tax credits; incentivizing higher density developments; funding programs in the capital improvement plan; and addressing Tierra Contenta’s infrastructure needs. With almost 100% apartment occupancy and approximately 2,000-4,000 additional units needed, I support more context-sensitive apartment developments. 4. What uses would you support for the city-owned campus of the Santa Fe University of Art and Design, which the school is vacating?

I support building on existing assets (e.g., Fogelson Library, Garson Theater, The Screen, etc.) and creating a model for sustainability in affordable housing, green building design, renewable energy, research and development, and higher education. I also support its use as a post-production facility and film school to support an expanding film and digital media industry while leveraging its broadband system.

5. Do you support the city’s living wage ordinance – which currently sets the minimum wage at $11.09 per hour – and its mandatory annual cost of living increases?

Yes. As Santa Fe city councilor, I sponsored Resolution 2014-103 to strengthen the enforcement of the Living Wage Ordinance. This led to the enactment of Ordinance 2014-38 that requires businesses to self-certify their compliance prior to receiving a business license, and improves notification when the consumer price index is released and determines changes to the living wage.

6. Did you vote in the May “soda tax” election? If so, did you vote for or against it? Please explain your vote or your opinion of the failed tax proposal.

I voted against the soda tax in the May election. I support more pre-K funding, but did not support a wasteful, $80,000 special election. The unwillingness of soda tax advocates to work with the beverage industry and their efforts to make it a moral imperative doomed the effort. It was government over-reach in the absence of public trust.

7. Should the city continue to grant a permit and provide police support for the annual Entrada event held on the Plaza that is opposed by Native Americans and others?

The city should no longer grant a permit for the Entrada event. It’s naïve to re-enact a peaceful, historical event that was preceded and succeeded by violence and oppression. All parties must agree, in advance, on appropriate, historical activities respectful of all perspectives to ensure we continue the longest-running community celebration in America without civil unrest and “free speech” zones.

1. Have you or your business – if you are a business owner – ever been the subject of any state or federal tax liens? No. 2. Have you ever been involved in a personal or business bankruptcy proceeding? No.

3. Have you ever been arrested for, charged with or convicted of drunken driving, any misdemeanor or any felony? No.

SANTA FE MUNICIPAL ELECTIONS

NOW: Absentee voting is already underway.

Request an absentee ballot by stopping by the City Clerk’s Office, 200 Lincoln Ave., or by calling 955-6521, 955-6519 or 955-6326.

FEB.14: Early voting begins

Vote early at City Clerk’s Office, Room 215, 200 Lincoln Ave., Santa Fe, from 8 a.m. to 5 p.m. Monday through Friday, or at Genoveva Chavez Community Center, 3221 Rodeo Road, Santa Fe, from 9 a.m. to 6 p.m., Tuesday through Saturday, except March 2, when polls close at 5 p.m.

MARCH 2: Early and absentee voting ends at 5 p.m.

MARCH 6: Election Day

Polls open at 7 a.m. and close at 7 p.m. See the city’s website, www.santafenm.gov, for polling locations and addresses for voter convenience centers around town.

Source Article

The post ‘You really have to take things one at a time’ appeared first on Find Apartments For Rent In Santa-Fe.

Learn More: http://www.aicarzu.com/you-really-have-to-take-things-one-at-a-time/

[helenhorrocks97-blog]

Some People Excel At educator And Some Do not - Which One Are You?

It's All About (The) instructor.

A teacher may be terminated or suspended throughout the term of a contract, yet just for the following factors: a decrease effective pursuant to the institution district's policies; inexperienced performance; conduct which materially disrupts the continued performance of the teacher's obligations; repeated or material disregard of task; and also other just and also reasonable reason. Therefore, he is not just my instructor yet my good friend, training me on a daily basis how you can live authentically just by being that he is. I concur and also loved this analysis short article. The teacher after that calls out grid referrals and also students cross them off their grid. They might have assumed that a loud, mad voice would place the pupils in position as well as force them to pay attention as well as learn. Several of these teachers are new and are not knowledgeable about the should ask for pupil responses. Do not forget to provide appreciation and also credit report to the teacher that led you to this understanding when you discover something new and locate that you could effectively place it right into practice. Inning accordance with a recent report on instructor attrition by the National Facility for Education and learning Data, among educators that quit as well as took non-education works, 64% did so in order to have even more autonomy at the office, without micromanagement. They can maintain a journal concerning the temperatures and also climate or they can collect data on the amount of calories individuals consume on a normal day and the number of calories they consume at a vacation event. In some circumstances the educators who use games in courses do well in passing on the expertise much faster and better amongst the pupils. Due to the fact that it might be successful in the multicultural atmosphere, educators should not be scared to implement an one-of-a-kind mentor design. There is barely a week that passes that there is not some inappropiate accusation of touching right here in The U.S.A.. Most of us wore our spirit wear, including our brand-new THRIVE style t-shirts as well as it was a great day to fulfill the children, tell them what college products to buy, and also start our year with some understanding! Experienced educators can advance to act as mentors to more recent teachers or to become lead teachers. Besides, some other instructors dictate the rules to their pupils in Arabic losing their time. It conserves loan to have one educator teach way too many classes are to have an instructor double up as an instructor. I discuss that I know that no person is best so they will most likely loose marks, however if there are any kind of marks left at the end of the day they will be rewarded with 10 to fifteen mins of playing a video game if we can achieve 'virtually' whatever the instructor has actually appointed. Once in the classroom connecting with the students, the educator will certainly have the ability to observe the different understanding designs of the pupils and will be able to better plan for the future. Having this qualities as well as a master's eco slim degree will open up different doors for groove instructor task.

A Guide To instructor At Any kind of Age

The Pay tab explains normal profits and also exactly how workers in the profession are compensated-- annual incomes, per hour wages, tips, payments, or bonus offers. Along with classroom treatments, successful teachers have well-managed classrooms that are pupil focused; they contribute to pupils' cooperation and enable students to feel as if they belong of the class. It is elected officials that have to establish the program and incentivize pupils to come to be mathematics and scientific research educators by perhaps minimizing or forgiving tuition financial debt for those that consent to seek teaching tasks and stay in them for a drawn-out time period. He read, in a monotone, the message I had in front of me. My directing teacher could grade (?) when I was interning all 5 classes of essays during her preparation period. As well, make music a component of residence life to ensure that your kid will obtain admiration for music. For instance two tenured teachers, one white and also the various other Hispanic, had actually opted for a perform at the end of the school year during their break time, when both returned to the college, the Hispanic educator was as well as received an e-mail questioned concerning his whereabouts and also why he had actually gone running.

You Could Say thanks to Us Later - 3 Reasons To Quit Thinking About educator

They need a human who truly cares about them and wants to see them be all they could be. They require a teacher with excitement they could capture to interest them in a topic. If he wants the connection to assist the student and also to last, the educator should avoid particular actions or conditions. Then there was my superviosing teacher (from pipe); she came exactly 4 times during 2 years. All you need to do is publish your instructor resume and other credentials and the Florida colleges that are seeking educators will concern you. However, by the end of the first week of college, or definitely by the second week of institution, middle school students start to feel even more comfortable, they begin to test the educator's limits and also classroom management comes to be a growing number of hard. Because they are so vital and also never obtain the credit rating of what they do. Thanks, fantastic article. On my classroom blog I make a basic post each week informing parents (as well as students) of what we are doing that week and just what the homework is each day.

[kristinsimmons]

Commissioning Healthcare Policy: Hospital Readmission and Its Price Tag

By ANISH KOKA MD 

The message comes in over the office slack line at 1:05 pm. There are four patients in rooms, one new, 3 patients in the waiting room. Really, not an ideal time to deal with this particular message.

“Kathy the home care nurse for Mrs. C called and said her weight yesterday was 185, today it is 194, she has +4 pitting edema, heart rate 120, BP 140/70 standing, 120/64 sitting”

I know Mrs. C well. She has severe COPD from smoking for 45 of the last 55 years. Every breath looks like an effort because it is. The worst part of it all is that Mrs. C just returned home from the hospital just days ago.

The youngest of six children, Mrs. C was born with many embedded disadvantages. Being born black in a poor West Philadelphia neighborhood in the 1960s is a story that too often writes itself with a bad ending.  But Mrs. C avoided the usual pitfalls that derail young women in the neighborhood early. No drugs. No alcohol. No teenage pregnancies. Finished high school. Mrs. C. worked for the hospital as a unit clerk, had her own place, health benefits, and even a retirement plan.

Certain life habits, however, carry a heavy price. George Burns, the comedian never pictured without a cigar who died past his hundredth birthday, may have been immune to the effects of tobacco.  Mrs. C was not. She started smoking when she was 16. She doesn’t recall why. Her dad smoking didn’t help perhaps. Nausea racked her body after that first drag. It eased up after. Too bad.

That measly cigarette became the great addiction of her life. Day by day, the exquisitely thin membranes of the lungs that mediate gas exchange were destroyed. By the time the disease manifests with shortness of breath and bluish tinged lips, it’s too late. Short of the very few who qualify for a lung transplant, the efforts of doctors at this point are for mitigation rather than cure.

Complicating things further, in Mrs. C’s case, the normally low pressure vascular circuit of her lungs became a high pressure circuit that places ever increasing demands on the normally thin-muscled right ventricle of the heart.  This jeopardizes the ability of her heart to handle changes in blood volume.

A little extra fluid and the right side of the heart ends up causing unbearable swelling in her legs.  A little dehydration and severe disabling dizziness on standing ensues. Adding to that that her tenuous lung function decompensates with the slightest respiratory infection, that chronic steroid treatment to decrease her wheezing suppresses her immune system, and that the young man down the street helpfully drops off Newports at her home for a few extra dollars, and it’s easy to see why the hospital is her second home.

The most recent admission to the hospital was for kidney failure related to taking too much fluid off with diuretics. What was to be a short stay for gentle hydration turned into a longer stay when a pneumonia complicated the matters (though a trip to the intensive care unit and a ventilator was barely but fortunately avoided).  She was treated by the pulmonology team and sent home on a lower dose of diuretics.

The situation I am now confronting puts me in a quandary.  Her edema and weight are up markedly just a few days after returning home.  Could her fluid overload be because her kidneys are shutting down? Or does she just need more aggressive diuresis?

Should I guess? Knowing her present renal function would be helpful. But even if the Theranos lab I could appeal to for help wasn’t fictional, I would have to get her to my office everyday or every other day while adjusting her diuretic dose.

And so it comes to be that, days removed from a hospital admission, I’m sending her back to the hospital to be readmitted.  According to some, this is not supposed to happen.

A policy on readmission

In 2008, the commission that advises Medicare – the Medicare Payment Advisory Commission (MEDPAC), issued a report that focused on hospital readmissions.

The focus on hospital readmissions had been of great interest to the health policy community for some time. At the core of this interest lies the belief that hospitals and physicians are incentivized to treat patients rather than prevent admissions.

The MEDPAC report wanted to discourage readmissions like Mrs. C’s.  And so it wasn’t a terrible surprise that, rolled into the 2010 Affordable Care Act, was a section called the Hospital Readmission Reduction Program (HRRP) which created a system for Medicare to penalize hospitals with ‘high’ readmission rates.  The program was rolled out in 2013.

At first, the program seemed to work like a charm.  Hospitals significantly ramped up their efforts at care coordination.  Teams of nurses and aids were assembled to make sure patients would get their medications as prescribed upon discharge and to check on patients once they got home.

Hospital readmission rates suddenly dropped and Medicare started saving money.  A staggering 81% of all hospitals suffered penalties in 2018, which translates to ~$500 million or 0.3% of total Medicare payments to hospitals.

A complex analysis

But there’s more to this too good to be true story.

The HRRP penalty schemes are risk-adjusted based on administrative claims data. Risk-adjustment is a statistical procedure to take into account the diversity in complexity and severity of disease among patients so they can be compared.

Physicians know that risk-adjusted claims data are of dubious value because they themselves are often the reluctant data entry clerk in the byzantine scheme that starts with adding diagnostic items to the medical chart and ends with generating a coded billing claim for Medicare.  Needless to say, there’s a huge potential disconnect between what a claim attempts to convey and the actual condition of a given patient.

Yet another major problem is that the risk-adjustment employed by the HRRP does not take socioeconomic status into account, when that is arguably the single biggest driver of poor outcomes and of hospital readmissions.  The creators of the HRRP program seem to believe that a hospital located in poor area shouldn’t get a break for having high readmission rates, perhaps because they believe that hospital systems in general should be mindful of health inequities and address “care gaps” (differences in care provided to poor vs. affluent patients)  in their neighborhoods no matter what.

Finally, the initial out of the gate benchmark for readmission rates on which the HRRP would adjudicate the need for a penalty was a national average.  In such a scheme, a Johns Hopkins Hospital serving inner city Baltimore could be pitted against a regional hospital in rural Montana with an entirely different patient demographics.  This made the regional hospital in Montana very happy.

Gaming the metrics

Regardless of these technical considerations, it is an adage of social science that any metric will be gamed, and healthcare is unfortunately not immune to that law.

One tool increasingly used by hospitals to comply with Medicare payments rules is to admit patients to short stay units, under so-called “observation status.”  Another is to put pressure on emergency departments to avoid readmitting certain types of patients.

So, instead of primarily functioning as a triage operation where sick patients would be turned over to the care of the cardiologist in the hospital, the ER has been increasingly housing and managing heart failure patients to save the hospital money.

But the ER physician or the hospitalist supervising the short stay unit and who just meet a patient in the setting of an acute illness are poorly equipped to know which heart failure patient to discharge after a diuretic dose and which to keep for advanced heart failure therapies.

Source: MedPAC’s June 2018 Report (http://bit.ly/2CXD0fJ)

Nowadays, the cardiologist is increasingly insulated from those decisions.  I have personally experienced with alarming frequency instances where I learn only after the fact that a complex patient of mine has been treated for heart failure in the ED.

And my experience seems to be shared by many of my cardiology colleagues, especially among cardiologists who work in academic centers that are most affected by the policy.  Luckily, some of them are also clinician-scientists that can do more than just whine to colleagues about the new policy.  They can also study its outcomes.

What do the outcomes data show?

In a pivotal study, a group of cardiologists (Gupta et al.) saw that the drop in readmissions that followed the introduction of HRRP was unfortunately accompanied by a reversal in the decade long downward trend in heart failure mortality. This reversal suggested a serious potential harm from the policy.

But the possibility of harm was quickly challenged by another group of researchers led by one of the biggest names in health policy: Harlan Krumholz, a cardiologist who directs the influential Center for Outcomes Research and Evaluation at Yale University.

The Krumholz group analyzed millions of Medicare claims data from 2007 to 2016.  They also found that mortality for heart failure patients increased, but concluded that no causal link between decreasing readmission and increased mortality could be established.

Krumholz at al. noted that mortality rates for heart failure started climbing before the HRRP program was announced and they noted no inflection point in mortality rates with the policy announcement in 2010.  The evidence for their claim is highlighted in the table below:

As can be seen in the boxed row, Krumholz’ team concludes that the increasing mortality slope post-HRRP is no different from that pre-HRRP because the change did not reach statistical significance at the obligatory and arbitrary P<0.05 level.  The actual P-value was 0.11 and the confidence interval for the positive increase in mortality slope of 0.006 is (-.002 to .015).  Even poor students of epistemology would be loathe to conclude this result excludes a signal of harm.  It seems entirely plausible that, with all the limitations of the data set in question, mortality may in fact have accelerated after the institution of the HRRP.  Yet Krumholz insists that no signal of harm is to be considered.

But this did not stop another group of cardiologists (Wadhera et al.) from adding their contribution to the HRRP literature. Using the same data-set that the Krumholz group used — Medicare claims data — these researchers found once again that accelerating mortality coincided with the announcement of the HRRP.  More troubling, they also demonstrated that mortality rose primarily among patients not readmitted to the hospital.

A messy science

Admittedly this whole business of analysis is incredibly messy, with a number of moving parts.

My brief summary doesn’t do justice to a variety of maneuvers taken by the various groups to account for many of the limitations inherent in this type of study. Two of the competing analyses (Krumholz, Wadhera) used Medicare claims data while the other (Gupta) used a more limited voluntary registry.

During the time period in question, there were also other policy changes such as the introduction of new hospital billing codes (MS-DRG) that sought to adjust hospital payment rates to patient complexity. Better patient coding meant higher reimbursement from Medicare.  Armies of “documenters” were then employed by hospitals to capture more revenue.

This means that the claims data gathered by the researchers might look significantly different from one time period to the next even if the patients themselves were ostensibly the same. As the readmission rate is risk-adjusted, it is eminently plausible and likely that systematically upcoding patient risk could actually have been the primary driver of the drop in hospital readmission rate.

The other program playing a confounding role is the Recovery Audit Contractor (RAC) program begun in 2010 to reduce payments for inappropriate hospital admissions. Hospitals responded to denials for inpatient admissions by expanding ‘observation status’ stays. Which was the biggest driver for expanded observation stays? RAC or the HRRP? Once again an exact attribution is impossible.

Denying the obvious?

Despite the messiness of the data and the variety of analytic methods used, a consistent and uncontested observation remains: Heart failure mortality has increased in the last decade. The question being hotly contested is Why?

Oddly, Dr. Krumholz is steadfast in denying the possibility that the policy may have caused harm, even though the independent and contradictory conclusions from the other research groups at least raises a reasonable doubt.  And Dr. Krumholz has been quick to cast shade on research that does not conform to his conclusion.

By tweet he appears to ask for a level of detail his own papers lack, and he questions the legitimacy of another group’s data-set, all the while resisting any calls to put the program on hold despite the paucity of evidence showing benefit, the signal for harm, and perhaps most importantly, the concern of clinicians who see a mechanism for harm.

Greatly admire @rwyeh and his group…appreciate his focus on readmission &public policy. For such high-profile article, really need more info about statistical weighting. Methods should be sufficient so others can reproduce results. Can’t do that here. Look forward to more info.

— Harlan Krumholz (@hmkyale) December 21, 2018

Can you account for why the registry you used had, on average, such a small number of patients per site. Did you determine how many patients coded with heart failure by CMS were in the Registry? Just curious about the selection. It may not explain your results…but is a question. https://t.co/WSvJAN8Jjd

— Harlan Krumholz (@hmkyale) June 2, 2018

Dr. Krumholz also places much weight on an independent analysis carried out by MEDPAC which concluded there was no link between policy and the uptick in mortality. This particular conclusion rests heavily on the assertion that heart failure patients in 2016 were much sicker than patients in 2010. Recall that this coincides with a period of more intensive coding over the same time frame, so it is impossible to say this with any confidence.

The MEDPAC conclusion also relies on an analysis that finds no correlation between hospital level readmission and mortality rates.

While technically true, that conclusion overlooks that a large number of hospitals exhibited reduced readmission rates and increased mortality. Perhaps MEDPAC feels that patients dying at low readmit/high mortality hospitals should be mollified by the knowledge that somewhere there’s a low-readmit/low mortality hospital to balance things out?

That Krumholz and MEDPAC display such certainty about the direction of the signal they observe, and take pains to discount other possibilities seems strange and suggests that pre-existing biases may be at work. What might those biases be?

Conflicts of interest: You get what you pay for

In a world where heads roll for undisclosed personal financial conflicts of interest, it is remarkable that the current dispute, while full of scintillating exchanges about “propensity weighting” and other arcane points of statistics, does not reference any other potential conflicts at work that might affect the conclusions being reached.

Medicare’s decision to start the HRRP program didn’t come in a vacuum. It was inspired by years of research from Dr. Krumholz himself, who suggested that preventing admissions should be a goal for any policy that would aim to move the system from one paying for “volume” to one paying for “value.”

As far back as 2003, Krumholz held the view that:

“Hospitals may not support programs that improve the quality of care delivered to heart failure patients because these programs lower readmission rates and empty beds, and therefore further diminish already-declining revenues.”

If Krumholz’s unfavorable and crudely simplistic view of the operations and motivations of hospitals (and of the still relatively independent physicians staffing those hospitals) informs his position on health policy, it stands to reason that serious blinders would prevent him from seeing any evidence of harm in a particular policy that promotes the same view.

But that’s not all.  Krumholz’s group at Yale received grants from CMS under the auspices of the Measure and Instrument Development and Support (MIDS) program to study and produce the metrics and instruments needed to devise the readmission measures.

The MIDS program supports the “development and use of clinical quality measures which remains a critical healthcare priority and the tool of choice for improving quality of care at the national, community and facility levels” and it allocates $1.6 billion dollars to this purpose.

Thanks to a bipartisan act of Congress, a helpful little website, usaspending.gov, provides contract level detail about payments made to the Krumholz’s group from the MIDS program.  Those payments can be seen in the table below:

The numbers are staggering. I know little about how to interpret these data about federal contracts, but it sure appears that the Yale-New Haven Health Service group led by Krumholz has received $144 million dollars since 2008.

Yet the only clue to these payments in Krumholz’ published analyses of the HRRP program comes in one disclosure sentence in a footnote, as seen here:

It seems to me that the disclosure is hardly proportional to the amount of funding that his group receives and understates the inherent pressures it must be under to demonstrate that the policy did not actually result in higher mortality.

And recall that MEDPAC’s “independent” analysis that also rejected a policy-mortality link came from the organization that recommended the policy to begin with.  The bottom line is this: There’s a tremendous amount of face to lose and a massive source of institutional funding at risk if the policy is found to be harmful.

It now becomes more clear why, in the following tweets, Dr. Krumholz feels that only he can say anything definitive about readmission rates and mortality:

Um, @JAMA_current, even the authors say they cannot say their findings are causal… "but whether this finding is a result of the policy requires further research.” Why do you promote the paper as proving harm? Need to treat twitter like you do any of your Editorial comments. pic.twitter.com/xugasHvPWY

— Harlan Krumholz (@hmkyale) December 26, 2018

6/Hospitals that have improved their readmission rates tended to improve their mortality rates. Published in @JAMA_current… https://t.co/R7FXLwTMPp

— Harlan Krumholz (@hmkyale) December 22, 2018

A diversity of biases

Biases are ubiquitous.  When I was a cardiologist-in-training, spending hours on the hospital consult service for a fixed salary, I vividly recall looking for ways to avoid doing any work I considered unimportant or banal: The minor cardiac enzyme leak in a patient with a widespread infection; The extra heartbeats on the ECG that the ER physician didn’t like the look of; etc.  “Are you sure you need an official consult?”  “The chances we’re going to recommend doing anything about a small enzyme leak in an 80-year-old with a severe lung infection are very low… “ I was even successful sometimes.

Contrast these comments to my demeanor in private practice where I am acutely aware that my income relies on such consults: “I just need the patients name or room number…” “I’ll take care of it!…” “I can put in the orders if you’d like!”

But there are other biases and incentives that motivate human beings, apart from personal financial incentives. Do they pale relative to the financial ones as is so often claimed? How does one begin to quantify them?

When it comes to the HRRP policy, no individual person’s bank account ballooned every time a patient didn’t get admitted. And yet this is a story of ideological bias that drove the design of policy and now claims ‘success’ for its own program.  The HRRP saga is illustrative of the importance of non-financial bias and of the dangers of blinding ourselves to that bias.

The story also highlights the downsides of tweaking healthcare systems that were built to deliver more care.

Clearly, I personally have a direct financial conflict of interest to provide more care. Since I haven’t talked anyone out of a consult in 8 years, I’m probably guilty of participating in a system that detractors appropriately criticize for promoting overuse of healthcare.

But the problem is that some of those consults I was trying to avoid as a fellow ended up really needing a cardiologist. There was the 55-year-old Cambodian woman admitted to the medical intensive care unit with pneumonia who went into atrial fibrillation. I recall rolling my eyes and thinking that the ICU could certainly handle this without a cardiologist. It turned out she didn’t have a pneumonia. It was pulmonary edema from heart failure related to undiagnosed rheumatic mitral valve disease. She had been in the wrong unit. She needed diuresis, heart rate control, and eventual surgery to replace her valve, not antibiotics. Less isn’t always more.

Attempt to reduce inappropriate hospital admissions? Get ready to pay a price.  To contradict Dr. Krumholz, it is entirely probable that we are underestimating the upside of our current system when we contemplate changing the status quo.

We are underestimating the downside of our current system when we contemplate change. We need to take some risks to do better. #abimf2013

— Harlan Krumholz (@hmkyale) August 5, 2013

Final Thoughts

Ironically, the HRRP quagmire offers a number of clarifying lessons.

Empiricism in social policy is a subjective enterprise. The often parroted conclusion is that cold, hard, unbiased evidence trumps the biased, unmeasurable judgment of clinicians. Yet, frequently, real world data-sets are complex, the choice of analytic paths can be highly variable, and the instruments to measure success are often imperfect. As it relates to HRRP, which analysis should we trust? The choice requires faith. And if the currency here is faith, perhaps the concern of clinicians at the bedside has more value than advertised.

Metrics won’t save us. The narrative of metrics is an appealing one that promises hard and objective accountability. The problem comes when the metric (readmission) becomes disconnected from outcomes that actually matter (death). False and blind prophets are good descriptors for those who claim to be unable to see without metrics. The fools in this enterprise are easy to identify as those who think the answers lie with ever better metrics.

Conflicts of interest: Going beyond the simple narrative.  Focusing on biases induced by personal financial interests is a mistake. Personal enrichment is just one bias in a sea of conflicts. In the healthcare context, financial disclosures—while clarifying in themselves—may simply give cover to other, more perverse biases, unless those other biases are equally disclosed.  It requires diligence to ascertain the impact, and direction of bias.  Rarely do we get the opportunity to observe the direction of bias in policy research.  In the case of the HRRP, the presence of bias was made evident because research groups with opposing biases (clinician-scientists versus policy wonks) have reached conclusions that would be expected on the basis of those pre-existing biases.  How often is the problem of such bias examined in the design, implementation, and analysis of health policy?

Beware of technocrats with all the answers.  I am reasonably sure that if practicing clinicians would have been asked to devise a rule to reduce heart failure readmissions for the whole population they would have refused. It seems too challenging a task to get right.  Even if clinicians can be induced to participate in the design of such a policy, most would readily acknowledge the likelihood that it could harm some patients. It requires a special type of hubris to design a policy and refuse to acknowledge its potential for harm. Unfortunately, hubris within a public health community that believes only they can give us a better health system is more feature than bug.

As for Mrs. C, she has been home for 16 days. All fingers on both hands are currently crossed.

Anish Koka is a cardiologist in private practice in Philadelphia.  He can be followed on Twitter @anish_koka. This post originally appeared here on The Accad & Koka Report. 

Commissioning Healthcare Policy: Hospital Readmission and Its Price Tag published first on https://wittooth.tumblr.com/

[kristinsimmons]

Commissioning Healthcare Policy: Hospital Readmission and Its Price Tag

By ANISH KOKA MD 

The message comes in over the office slack line at 1:05 pm. There are four patients in rooms, one new, 3 patients in the waiting room. Really, not an ideal time to deal with this particular message.

“Kathy the home care nurse for Mrs. C called and said her weight yesterday was 185, today it is 194, she has +4 pitting edema, heart rate 120, BP 140/70 standing, 120/64 sitting”

I know Mrs. C well. She has severe COPD from smoking for 45 of the last 55 years. Every breath looks like an effort because it is. The worst part of it all is that Mrs. C just returned home from the hospital just days ago.

The youngest of six children, Mrs. C was born with many embedded disadvantages. Being born black in a poor West Philadelphia neighborhood in the 1960s is a story that too often writes itself with a bad ending.  But Mrs. C avoided the usual pitfalls that derail young women in the neighborhood early. No drugs. No alcohol. No teenage pregnancies. Finished high school. Mrs. C. worked for the hospital as a unit clerk, had her own place, health benefits, and even a retirement plan.

Certain life habits, however, carry a heavy price. George Burns, the comedian never pictured without a cigar who died past his hundredth birthday, may have been immune to the effects of tobacco.  Mrs. C was not. She started smoking when she was 16. She doesn’t recall why. Her dad smoking didn’t help perhaps. Nausea racked her body after that first drag. It eased up after. Too bad.

That measly cigarette became the great addiction of her life. Day by day, the exquisitely thin membranes of the lungs that mediate gas exchange were destroyed. By the time the disease manifests with shortness of breath and bluish tinged lips, it’s too late. Short of the very few who qualify for a lung transplant, the efforts of doctors at this point are for mitigation rather than cure.

Complicating things further, in Mrs. C’s case, the normally low pressure vascular circuit of her lungs became a high pressure circuit that places ever increasing demands on the normally thin-muscled right ventricle of the heart.  This jeopardizes the ability of her heart to handle changes in blood volume.

A little extra fluid and the right side of the heart ends up causing unbearable swelling in her legs.  A little dehydration and severe disabling dizziness on standing ensues. Adding to that that her tenuous lung function decompensates with the slightest respiratory infection, that chronic steroid treatment to decrease her wheezing suppresses her immune system, and that the young man down the street helpfully drops off Newports at her home for a few extra dollars, and it’s easy to see why the hospital is her second home.

The most recent admission to the hospital was for kidney failure related to taking too much fluid off with diuretics. What was to be a short stay for gentle hydration turned into a longer stay when a pneumonia complicated the matters (though a trip to the intensive care unit and a ventilator was barely but fortunately avoided).  She was treated by the pulmonology team and sent home on a lower dose of diuretics.

The situation I am now confronting puts me in a quandary.  Her edema and weight are up markedly just a few days after returning home.  Could her fluid overload be because her kidneys are shutting down? Or does she just need more aggressive diuresis?

Should I guess? Knowing her present renal function would be helpful. But even if the Theranos lab I could appeal to for help wasn’t fictional, I would have to get her to my office everyday or every other day while adjusting her diuretic dose.

And so it comes to be that, days removed from a hospital admission, I’m sending her back to the hospital to be readmitted.  According to some, this is not supposed to happen.

A policy on readmission

In 2008, the commission that advises Medicare – the Medicare Payment Advisory Commission (MEDPAC), issued a report that focused on hospital readmissions.

The focus on hospital readmissions had been of great interest to the health policy community for some time. At the core of this interest lies the belief that hospitals and physicians are incentivized to treat patients rather than prevent admissions.

The MEDPAC report wanted to discourage readmissions like Mrs. C’s.  And so it wasn’t a terrible surprise that, rolled into the 2010 Affordable Care Act, was a section called the Hospital Readmission Reduction Program (HRRP) which created a system for Medicare to penalize hospitals with ‘high’ readmission rates.  The program was rolled out in 2013.

At first, the program seemed to work like a charm.  Hospitals significantly ramped up their efforts at care coordination.  Teams of nurses and aids were assembled to make sure patients would get their medications as prescribed upon discharge and to check on patients once they got home.

Hospital readmission rates suddenly dropped and Medicare started saving money.  A staggering 81% of all hospitals suffered penalties in 2018, which translates to ~$500 million or 0.3% of total Medicare payments to hospitals.

A complex analysis

But there’s more to this too good to be true story.

The HRRP penalty schemes are risk-adjusted based on administrative claims data. Risk-adjustment is a statistical procedure to take into account the diversity in complexity and severity of disease among patients so they can be compared.

Physicians know that risk-adjusted claims data are of dubious value because they themselves are often the reluctant data entry clerk in the byzantine scheme that starts with adding diagnostic items to the medical chart and ends with generating a coded billing claim for Medicare.  Needless to say, there’s a huge potential disconnect between what a claim attempts to convey and the actual condition of a given patient.

Yet another major problem is that the risk-adjustment employed by the HRRP does not take socioeconomic status into account, when that is arguably the single biggest driver of poor outcomes and of hospital readmissions.  The creators of the HRRP program seem to believe that a hospital located in poor area shouldn’t get a break for having high readmission rates, perhaps because they believe that hospital systems in general should be mindful of health inequities and address “care gaps” (differences in care provided to poor vs. affluent patients)  in their neighborhoods no matter what.

Finally, the initial out of the gate benchmark for readmission rates on which the HRRP would adjudicate the need for a penalty was a national average.  In such a scheme, a Johns Hopkins Hospital serving inner city Baltimore could be pitted against a regional hospital in rural Montana with an entirely different patient demographics.  This made the regional hospital in Montana very happy.

Gaming the metrics

Regardless of these technical considerations, it is an adage of social science that any metric will be gamed, and healthcare is unfortunately not immune to that law.

One tool increasingly used by hospitals to comply with Medicare payments rules is to admit patients to short stay units, under so-called “observation status.”  Another is to put pressure on emergency departments to avoid readmitting certain types of patients.

So, instead of primarily functioning as a triage operation where sick patients would be turned over to the care of the cardiologist in the hospital, the ER has been increasingly housing and managing heart failure patients to save the hospital money.

But the ER physician or the hospitalist supervising the short stay unit and who just meet a patient in the setting of an acute illness are poorly equipped to know which heart failure patient to discharge after a diuretic dose and which to keep for advanced heart failure therapies.

Source: MedPAC’s June 2018 Report (http://bit.ly/2CXD0fJ)

Nowadays, the cardiologist is increasingly insulated from those decisions.  I have personally experienced with alarming frequency instances where I learn only after the fact that a complex patient of mine has been treated for heart failure in the ED.

And my experience seems to be shared by many of my cardiology colleagues, especially among cardiologists who work in academic centers that are most affected by the policy.  Luckily, some of them are also clinician-scientists that can do more than just whine to colleagues about the new policy.  They can also study its outcomes.

What do the outcomes data show?

In a pivotal study, a group of cardiologists (Gupta et al.) saw that the drop in readmissions that followed the introduction of HRRP was unfortunately accompanied by a reversal in the decade long downward trend in heart failure mortality. This reversal suggested a serious potential harm from the policy.

But the possibility of harm was quickly challenged by another group of researchers led by one of the biggest names in health policy: Harlan Krumholz, a cardiologist who directs the influential Center for Outcomes Research and Evaluation at Yale University.

The Krumholz group analyzed millions of Medicare claims data from 2007 to 2016.  They also found that mortality for heart failure patients increased, but concluded that no causal link between decreasing readmission and increased mortality could be established.

Krumholz at al. noted that mortality rates for heart failure started climbing before the HRRP program was announced and they noted no inflection point in mortality rates with the policy announcement in 2010.  The evidence for their claim is highlighted in the table below:

As can be seen in the boxed row, Krumholz’ team concludes that the increasing mortality slope post-HRRP is no different from that pre-HRRP because the change did not reach statistical significance at the obligatory and arbitrary P<0.05 level.  The actual P-value was 0.11 and the confidence interval for the positive increase in mortality slope of 0.006 is (-.002 to .015).  Even poor students of epistemology would be loathe to conclude this result excludes a signal of harm.  It seems entirely plausible that, with all the limitations of the data set in question, mortality may in fact have accelerated after the institution of the HRRP.  Yet Krumholz insists that no signal of harm is to be considered.

But this did not stop another group of cardiologists (Wadhera et al.) from adding their contribution to the HRRP literature. Using the same data-set that the Krumholz group used — Medicare claims data — these researchers found once again that accelerating mortality coincided with the announcement of the HRRP.  More troubling, they also demonstrated that mortality rose primarily among patients not readmitted to the hospital.

A messy science

Admittedly this whole business of analysis is incredibly messy, with a number of moving parts.

My brief summary doesn’t do justice to a variety of maneuvers taken by the various groups to account for many of the limitations inherent in this type of study. Two of the competing analyses (Krumholz, Wadhera) used Medicare claims data while the other (Gupta) used a more limited voluntary registry.

During the time period in question, there were also other policy changes such as the introduction of new hospital billing codes (MS-DRG) that sought to adjust hospital payment rates to patient complexity. Better patient coding meant higher reimbursement from Medicare.  Armies of “documenters” were then employed by hospitals to capture more revenue.

This means that the claims data gathered by the researchers might look significantly different from one time period to the next even if the patients themselves were ostensibly the same. As the readmission rate is risk-adjusted, it is eminently plausible and likely that systematically upcoding patient risk could actually have been the primary driver of the drop in hospital readmission rate.

The other program playing a confounding role is the Recovery Audit Contractor (RAC) program begun in 2010 to reduce payments for inappropriate hospital admissions. Hospitals responded to denials for inpatient admissions by expanding ‘observation status’ stays. Which was the biggest driver for expanded observation stays? RAC or the HRRP? Once again an exact attribution is impossible.

Denying the obvious?

Despite the messiness of the data and the variety of analytic methods used, a consistent and uncontested observation remains: Heart failure mortality has increased in the last decade. The question being hotly contested is Why?

Oddly, Dr. Krumholz is steadfast in denying the possibility that the policy may have caused harm, even though the independent and contradictory conclusions from the other research groups at least raises a reasonable doubt.  And Dr. Krumholz has been quick to cast shade on research that does not conform to his conclusion.

By tweet he appears to ask for a level of detail his own papers lack, and he questions the legitimacy of another group’s data-set, all the while resisting any calls to put the program on hold despite the paucity of evidence showing benefit, the signal for harm, and perhaps most importantly, the concern of clinicians who see a mechanism for harm.

Greatly admire @rwyeh and his group…appreciate his focus on readmission &public policy. For such high-profile article, really need more info about statistical weighting. Methods should be sufficient so others can reproduce results. Can’t do that here. Look forward to more info.

— Harlan Krumholz (@hmkyale) December 21, 2018

Can you account for why the registry you used had, on average, such a small number of patients per site. Did you determine how many patients coded with heart failure by CMS were in the Registry? Just curious about the selection. It may not explain your results…but is a question. https://t.co/WSvJAN8Jjd

— Harlan Krumholz (@hmkyale) June 2, 2018

Dr. Krumholz also places much weight on an independent analysis carried out by MEDPAC which concluded there was no link between policy and the uptick in mortality. This particular conclusion rests heavily on the assertion that heart failure patients in 2016 were much sicker than patients in 2010. Recall that this coincides with a period of more intensive coding over the same time frame, so it is impossible to say this with any confidence.

The MEDPAC conclusion also relies on an analysis that finds no correlation between hospital level readmission and mortality rates.

While technically true, that conclusion overlooks that a large number of hospitals exhibited reduced readmission rates and increased mortality. Perhaps MEDPAC feels that patients dying at low readmit/high mortality hospitals should be mollified by the knowledge that somewhere there’s a low-readmit/low mortality hospital to balance things out?

That Krumholz and MEDPAC display such certainty about the direction of the signal they observe, and take pains to discount other possibilities seems strange and suggests that pre-existing biases may be at work. What might those biases be?

Conflicts of interest: You get what you pay for

In a world where heads roll for undisclosed personal financial conflicts of interest, it is remarkable that the current dispute, while full of scintillating exchanges about “propensity weighting” and other arcane points of statistics, does not reference any other potential conflicts at work that might affect the conclusions being reached.

Medicare’s decision to start the HRRP program didn’t come in a vacuum. It was inspired by years of research from Dr. Krumholz himself, who suggested that preventing admissions should be a goal for any policy that would aim to move the system from one paying for “volume” to one paying for “value.”

As far back as 2003, Krumholz held the view that:

“Hospitals may not support programs that improve the quality of care delivered to heart failure patients because these programs lower readmission rates and empty beds, and therefore further diminish already-declining revenues.”

If Krumholz’s unfavorable and crudely simplistic view of the operations and motivations of hospitals (and of the still relatively independent physicians staffing those hospitals) informs his position on health policy, it stands to reason that serious blinders would prevent him from seeing any evidence of harm in a particular policy that promotes the same view.

But that’s not all.  Krumholz’s group at Yale received grants from CMS under the auspices of the Measure and Instrument Development and Support (MIDS) program to study and produce the metrics and instruments needed to devise the readmission measures.

The MIDS program supports the “development and use of clinical quality measures which remains a critical healthcare priority and the tool of choice for improving quality of care at the national, community and facility levels” and it allocates $1.6 billion dollars to this purpose.

Thanks to a bipartisan act of Congress, a helpful little website, usaspending.gov, provides contract level detail about payments made to the Krumholz’s group from the MIDS program.  Those payments can be seen in the table below:

The numbers are staggering. I know little about how to interpret these data about federal contracts, but it sure appears that the Yale-New Haven Health Service group led by Krumholz has received $144 million dollars since 2008.

Yet the only clue to these payments in Krumholz’ published analyses of the HRRP program comes in one disclosure sentence in a footnote, as seen here:

It seems to me that the disclosure is hardly proportional to the amount of funding that his group receives and understates the inherent pressures it must be under to demonstrate that the policy did not actually result in higher mortality.

And recall that MEDPAC’s “independent” analysis that also rejected a policy-mortality link came from the organization that recommended the policy to begin with.  The bottom line is this: There’s a tremendous amount of face to lose and a massive source of institutional funding at risk if the policy is found to be harmful.

It now becomes more clear why, in the following tweets, Dr. Krumholz feels that only he can say anything definitive about readmission rates and mortality:

Um, @JAMA_current, even the authors say they cannot say their findings are causal… "but whether this finding is a result of the policy requires further research.” Why do you promote the paper as proving harm? Need to treat twitter like you do any of your Editorial comments. pic.twitter.com/xugasHvPWY

— Harlan Krumholz (@hmkyale) December 26, 2018

6/Hospitals that have improved their readmission rates tended to improve their mortality rates. Published in @JAMA_current… https://t.co/R7FXLwTMPp

— Harlan Krumholz (@hmkyale) December 22, 2018

A diversity of biases

Biases are ubiquitous.  When I was a cardiologist-in-training, spending hours on the hospital consult service for a fixed salary, I vividly recall looking for ways to avoid doing any work I considered unimportant or banal: The minor cardiac enzyme leak in a patient with a widespread infection; The extra heartbeats on the ECG that the ER physician didn’t like the look of; etc.  “Are you sure you need an official consult?”  “The chances we’re going to recommend doing anything about a small enzyme leak in an 80-year-old with a severe lung infection are very low… “ I was even successful sometimes.

Contrast these comments to my demeanor in private practice where I am acutely aware that my income relies on such consults: “I just need the patients name or room number…” “I’ll take care of it!…” “I can put in the orders if you’d like!”

But there are other biases and incentives that motivate human beings, apart from personal financial incentives. Do they pale relative to the financial ones as is so often claimed? How does one begin to quantify them?

When it comes to the HRRP policy, no individual person’s bank account ballooned every time a patient didn’t get admitted. And yet this is a story of ideological bias that drove the design of policy and now claims ‘success’ for its own program.  The HRRP saga is illustrative of the importance of non-financial bias and of the dangers of blinding ourselves to that bias.

The story also highlights the downsides of tweaking healthcare systems that were built to deliver more care.

Clearly, I personally have a direct financial conflict of interest to provide more care. Since I haven’t talked anyone out of a consult in 8 years, I’m probably guilty of participating in a system that detractors appropriately criticize for promoting overuse of healthcare.

But the problem is that some of those consults I was trying to avoid as a fellow ended up really needing a cardiologist. There was the 55-year-old Cambodian woman admitted to the medical intensive care unit with pneumonia who went into atrial fibrillation. I recall rolling my eyes and thinking that the ICU could certainly handle this without a cardiologist. It turned out she didn’t have a pneumonia. It was pulmonary edema from heart failure related to undiagnosed rheumatic mitral valve disease. She had been in the wrong unit. She needed diuresis, heart rate control, and eventual surgery to replace her valve, not antibiotics. Less isn’t always more.

Attempt to reduce inappropriate hospital admissions? Get ready to pay a price.  To contradict Dr. Krumholz, it is entirely probable that we are underestimating the upside of our current system when we contemplate changing the status quo.

We are underestimating the downside of our current system when we contemplate change. We need to take some risks to do better. #abimf2013

— Harlan Krumholz (@hmkyale) August 5, 2013

Final Thoughts

Ironically, the HRRP quagmire offers a number of clarifying lessons.

Empiricism in social policy is a subjective enterprise. The often parroted conclusion is that cold, hard, unbiased evidence trumps the biased, unmeasurable judgment of clinicians. Yet, frequently, real world data-sets are complex, the choice of analytic paths can be highly variable, and the instruments to measure success are often imperfect. As it relates to HRRP, which analysis should we trust? The choice requires faith. And if the currency here is faith, perhaps the concern of clinicians at the bedside has more value than advertised.

Metrics won’t save us. The narrative of metrics is an appealing one that promises hard and objective accountability. The problem comes when the metric (readmission) becomes disconnected from outcomes that actually matter (death). False and blind prophets are good descriptors for those who claim to be unable to see without metrics. The fools in this enterprise are easy to identify as those who think the answers lie with ever better metrics.

Conflicts of interest: Going beyond the simple narrative.  Focusing on biases induced by personal financial interests is a mistake. Personal enrichment is just one bias in a sea of conflicts. In the healthcare context, financial disclosures—while clarifying in themselves—may simply give cover to other, more perverse biases, unless those other biases are equally disclosed.  It requires diligence to ascertain the impact, and direction of bias.  Rarely do we get the opportunity to observe the direction of bias in policy research.  In the case of the HRRP, the presence of bias was made evident because research groups with opposing biases (clinician-scientists versus policy wonks) have reached conclusions that would be expected on the basis of those pre-existing biases.  How often is the problem of such bias examined in the design, implementation, and analysis of health policy?

Beware of technocrats with all the answers.  I am reasonably sure that if practicing clinicians would have been asked to devise a rule to reduce heart failure readmissions for the whole population they would have refused. It seems too challenging a task to get right.  Even if clinicians can be induced to participate in the design of such a policy, most would readily acknowledge the likelihood that it could harm some patients. It requires a special type of hubris to design a policy and refuse to acknowledge its potential for harm. Unfortunately, hubris within a public health community that believes only they can give us a better health system is more feature than bug.

As for Mrs. C, she has been home for 16 days. All fingers on both hands are currently crossed.

Anish Koka is a cardiologist in private practice in Philadelphia.  He can be followed on Twitter @anish_koka. This post originally appeared here on The Accad & Koka Report. 

Commissioning Healthcare Policy: Hospital Readmission and Its Price Tag published first on https://wittooth.tumblr.com/

[kristinsimmons]

Commissioning Healthcare Policy: Hospital Readmission and Its Price Tag

By ANISH KOKA MD 

The message comes in over the office slack line at 1:05 pm. There are four patients in rooms, one new, 3 patients in the waiting room. Really, not an ideal time to deal with this particular message.

“Kathy the home care nurse for Mrs. C called and said her weight yesterday was 185, today it is 194, she has +4 pitting edema, heart rate 120, BP 140/70 standing, 120/64 sitting”

I know Mrs. C well. She has severe COPD from smoking for 45 of the last 55 years. Every breath looks like an effort because it is. The worst part of it all is that Mrs. C just returned home from the hospital just days ago.

The youngest of six children, Mrs. C was born with many embedded disadvantages. Being born black in a poor West Philadelphia neighborhood in the 1960s is a story that too often writes itself with a bad ending.  But Mrs. C avoided the usual pitfalls that derail young women in the neighborhood early. No drugs. No alcohol. No teenage pregnancies. Finished high school. Mrs. C. worked for the hospital as a unit clerk, had her own place, health benefits, and even a retirement plan.

Certain life habits, however, carry a heavy price. George Burns, the comedian never pictured without a cigar who died past his hundredth birthday, may have been immune to the effects of tobacco.  Mrs. C was not. She started smoking when she was 16. She doesn’t recall why. Her dad smoking didn’t help perhaps. Nausea racked her body after that first drag. It eased up after. Too bad.

That measly cigarette became the great addiction of her life. Day by day, the exquisitely thin membranes of the lungs that mediate gas exchange were destroyed. By the time the disease manifests with shortness of breath and bluish tinged lips, it’s too late. Short of the very few who qualify for a lung transplant, the efforts of doctors at this point are for mitigation rather than cure.

Complicating things further, in Mrs. C’s case, the normally low pressure vascular circuit of her lungs became a high pressure circuit that places ever increasing demands on the normally thin-muscled right ventricle of the heart.  This jeopardizes the ability of her heart to handle changes in blood volume.

A little extra fluid and the right side of the heart ends up causing unbearable swelling in her legs.  A little dehydration and severe disabling dizziness on standing ensues. Adding to that that her tenuous lung function decompensates with the slightest respiratory infection, that chronic steroid treatment to decrease her wheezing suppresses her immune system, and that the young man down the street helpfully drops off Newports at her home for a few extra dollars, and it’s easy to see why the hospital is her second home.

The most recent admission to the hospital was for kidney failure related to taking too much fluid off with diuretics. What was to be a short stay for gentle hydration turned into a longer stay when a pneumonia complicated the matters (though a trip to the intensive care unit and a ventilator was barely but fortunately avoided).  She was treated by the pulmonology team and sent home on a lower dose of diuretics.

The situation I am now confronting puts me in a quandary.  Her edema and weight are up markedly just a few days after returning home.  Could her fluid overload be because her kidneys are shutting down? Or does she just need more aggressive diuresis?

Should I guess? Knowing her present renal function would be helpful. But even if the Theranos lab I could appeal to for help wasn’t fictional, I would have to get her to my office everyday or every other day while adjusting her diuretic dose.

And so it comes to be that, days removed from a hospital admission, I’m sending her back to the hospital to be readmitted.  According to some, this is not supposed to happen.

A policy on readmission

In 2008, the commission that advises Medicare – the Medicare Payment Advisory Commission (MEDPAC), issued a report that focused on hospital readmissions.

The focus on hospital readmissions had been of great interest to the health policy community for some time. At the core of this interest lies the belief that hospitals and physicians are incentivized to treat patients rather than prevent admissions.

The MEDPAC report wanted to discourage readmissions like Mrs. C’s.  And so it wasn’t a terrible surprise that, rolled into the 2010 Affordable Care Act, was a section called the Hospital Readmission Reduction Program (HRRP) which created a system for Medicare to penalize hospitals with ‘high’ readmission rates.  The program was rolled out in 2013.

At first, the program seemed to work like a charm.  Hospitals significantly ramped up their efforts at care coordination.  Teams of nurses and aids were assembled to make sure patients would get their medications as prescribed upon discharge and to check on patients once they got home.

Hospital readmission rates suddenly dropped and Medicare started saving money.  A staggering 81% of all hospitals suffered penalties in 2018, which translates to ~$500 million or 0.3% of total Medicare payments to hospitals.

A complex analysis

But there’s more to this too good to be true story.

The HRRP penalty schemes are risk-adjusted based on administrative claims data. Risk-adjustment is a statistical procedure to take into account the diversity in complexity and severity of disease among patients so they can be compared.

Physicians know that risk-adjusted claims data are of dubious value because they themselves are often the reluctant data entry clerk in the byzantine scheme that starts with adding diagnostic items to the medical chart and ends with generating a coded billing claim for Medicare.  Needless to say, there’s a huge potential disconnect between what a claim attempts to convey and the actual condition of a given patient.

Yet another major problem is that the risk-adjustment employed by the HRRP does not take socioeconomic status into account, when that is arguably the single biggest driver of poor outcomes and of hospital readmissions.  The creators of the HRRP program seem to believe that a hospital located in poor area shouldn’t get a break for having high readmission rates, perhaps because they believe that hospital systems in general should be mindful of health inequities and address “care gaps” (differences in care provided to poor vs. affluent patients)  in their neighborhoods no matter what.

Finally, the initial out of the gate benchmark for readmission rates on which the HRRP would adjudicate the need for a penalty was a national average.  In such a scheme, a Johns Hopkins Hospital serving inner city Baltimore could be pitted against a regional hospital in rural Montana with an entirely different patient demographics.  This made the regional hospital in Montana very happy.

Gaming the metrics

Regardless of these technical considerations, it is an adage of social science that any metric will be gamed, and healthcare is unfortunately not immune to that law.

One tool increasingly used by hospitals to comply with Medicare payments rules is to admit patients to short stay units, under so-called “observation status.”  Another is to put pressure on emergency departments to avoid readmitting certain types of patients.

So, instead of primarily functioning as a triage operation where sick patients would be turned over to the care of the cardiologist in the hospital, the ER has been increasingly housing and managing heart failure patients to save the hospital money.

But the ER physician or the hospitalist supervising the short stay unit and who just meet a patient in the setting of an acute illness are poorly equipped to know which heart failure patient to discharge after a diuretic dose and which to keep for advanced heart failure therapies.

Source: MedPAC’s June 2018 Report (http://bit.ly/2CXD0fJ)

Nowadays, the cardiologist is increasingly insulated from those decisions.  I have personally experienced with alarming frequency instances where I learn only after the fact that a complex patient of mine has been treated for heart failure in the ED.

And my experience seems to be shared by many of my cardiology colleagues, especially among cardiologists who work in academic centers that are most affected by the policy.  Luckily, some of them are also clinician-scientists that can do more than just whine to colleagues about the new policy.  They can also study its outcomes.

What do the outcomes data show?

In a pivotal study, a group of cardiologists (Gupta et al.) saw that the drop in readmissions that followed the introduction of HRRP was unfortunately accompanied by a reversal in the decade long downward trend in heart failure mortality. This reversal suggested a serious potential harm from the policy.

But the possibility of harm was quickly challenged by another group of researchers led by one of the biggest names in health policy: Harlan Krumholz, a cardiologist who directs the influential Center for Outcomes Research and Evaluation at Yale University.

The Krumholz group analyzed millions of Medicare claims data from 2007 to 2016.  They also found that mortality for heart failure patients increased, but concluded that no causal link between decreasing readmission and increased mortality could be established.

Krumholz at al. noted that mortality rates for heart failure started climbing before the HRRP program was announced and they noted no inflection point in mortality rates with the policy announcement in 2010.  The evidence for their claim is highlighted in the table below:

As can be seen in the boxed row, Krumholz’ team concludes that the increasing mortality slope post-HRRP is no different from that pre-HRRP because the change did not reach statistical significance at the obligatory and arbitrary P<0.05 level.  The actual P-value was 0.11 and the confidence interval for the positive increase in mortality slope of 0.006 is (-.002 to .015).  Even poor students of epistemology would be loathe to conclude this result excludes a signal of harm.  It seems entirely plausible that, with all the limitations of the data set in question, mortality may in fact have accelerated after the institution of the HRRP.  Yet Krumholz insists that no signal of harm is to be considered.

But this did not stop another group of cardiologists (Wadhera et al.) from adding their contribution to the HRRP literature. Using the same data-set that the Krumholz group used — Medicare claims data — these researchers found once again that accelerating mortality coincided with the announcement of the HRRP.  More troubling, they also demonstrated that mortality rose primarily among patients not readmitted to the hospital.

A messy science

Admittedly this whole business of analysis is incredibly messy, with a number of moving parts.

My brief summary doesn’t do justice to a variety of maneuvers taken by the various groups to account for many of the limitations inherent in this type of study. Two of the competing analyses (Krumholz, Wadhera) used Medicare claims data while the other (Gupta) used a more limited voluntary registry.

During the time period in question, there were also other policy changes such as the introduction of new hospital billing codes (MS-DRG) that sought to adjust hospital payment rates to patient complexity. Better patient coding meant higher reimbursement from Medicare.  Armies of “documenters” were then employed by hospitals to capture more revenue.

This means that the claims data gathered by the researchers might look significantly different from one time period to the next even if the patients themselves were ostensibly the same. As the readmission rate is risk-adjusted, it is eminently plausible and likely that systematically upcoding patient risk could actually have been the primary driver of the drop in hospital readmission rate.

The other program playing a confounding role is the Recovery Audit Contractor (RAC) program begun in 2010 to reduce payments for inappropriate hospital admissions. Hospitals responded to denials for inpatient admissions by expanding ‘observation status’ stays. Which was the biggest driver for expanded observation stays? RAC or the HRRP? Once again an exact attribution is impossible.

Denying the obvious?

Despite the messiness of the data and the variety of analytic methods used, a consistent and uncontested observation remains: Heart failure mortality has increased in the last decade. The question being hotly contested is Why?

Oddly, Dr. Krumholz is steadfast in denying the possibility that the policy may have caused harm, even though the independent and contradictory conclusions from the other research groups at least raises a reasonable doubt.  And Dr. Krumholz has been quick to cast shade on research that does not conform to his conclusion.

By tweet he appears to ask for a level of detail his own papers lack, and he questions the legitimacy of another group’s data-set, all the while resisting any calls to put the program on hold despite the paucity of evidence showing benefit, the signal for harm, and perhaps most importantly, the concern of clinicians who see a mechanism for harm.

Greatly admire @rwyeh and his group…appreciate his focus on readmission &public policy. For such high-profile article, really need more info about statistical weighting. Methods should be sufficient so others can reproduce results. Can’t do that here. Look forward to more info.

— Harlan Krumholz (@hmkyale) December 21, 2018

Can you account for why the registry you used had, on average, such a small number of patients per site. Did you determine how many patients coded with heart failure by CMS were in the Registry? Just curious about the selection. It may not explain your results…but is a question. https://t.co/WSvJAN8Jjd

— Harlan Krumholz (@hmkyale) June 2, 2018

Dr. Krumholz also places much weight on an independent analysis carried out by MEDPAC which concluded there was no link between policy and the uptick in mortality. This particular conclusion rests heavily on the assertion that heart failure patients in 2016 were much sicker than patients in 2010. Recall that this coincides with a period of more intensive coding over the same time frame, so it is impossible to say this with any confidence.

The MEDPAC conclusion also relies on an analysis that finds no correlation between hospital level readmission and mortality rates.

While technically true, that conclusion overlooks that a large number of hospitals exhibited reduced readmission rates and increased mortality. Perhaps MEDPAC feels that patients dying at low readmit/high mortality hospitals should be mollified by the knowledge that somewhere there’s a low-readmit/low mortality hospital to balance things out?

That Krumholz and MEDPAC display such certainty about the direction of the signal they observe, and take pains to discount other possibilities seems strange and suggests that pre-existing biases may be at work. What might those biases be?

Conflicts of interest: You get what you pay for

In a world where heads roll for undisclosed personal financial conflicts of interest, it is remarkable that the current dispute, while full of scintillating exchanges about “propensity weighting” and other arcane points of statistics, does not reference any other potential conflicts at work that might affect the conclusions being reached.

Medicare’s decision to start the HRRP program didn’t come in a vacuum. It was inspired by years of research from Dr. Krumholz himself, who suggested that preventing admissions should be a goal for any policy that would aim to move the system from one paying for “volume” to one paying for “value.”

As far back as 2003, Krumholz held the view that:

“Hospitals may not support programs that improve the quality of care delivered to heart failure patients because these programs lower readmission rates and empty beds, and therefore further diminish already-declining revenues.”

If Krumholz’s unfavorable and crudely simplistic view of the operations and motivations of hospitals (and of the still relatively independent physicians staffing those hospitals) informs his position on health policy, it stands to reason that serious blinders would prevent him from seeing any evidence of harm in a particular policy that promotes the same view.

But that’s not all.  Krumholz’s group at Yale received grants from CMS under the auspices of the Measure and Instrument Development and Support (MIDS) program to study and produce the metrics and instruments needed to devise the readmission measures.

The MIDS program supports the “development and use of clinical quality measures which remains a critical healthcare priority and the tool of choice for improving quality of care at the national, community and facility levels” and it allocates $1.6 billion dollars to this purpose.

Thanks to a bipartisan act of Congress, a helpful little website, usaspending.gov, provides contract level detail about payments made to the Krumholz’s group from the MIDS program.  Those payments can be seen in the table below:

The numbers are staggering. I know little about how to interpret these data about federal contracts, but it sure appears that the Yale-New Haven Health Service group led by Krumholz has received $144 million dollars since 2008.

Yet the only clue to these payments in Krumholz’ published analyses of the HRRP program comes in one disclosure sentence in a footnote, as seen here:

It seems to me that the disclosure is hardly proportional to the amount of funding that his group receives and understates the inherent pressures it must be under to demonstrate that the policy did not actually result in higher mortality.

And recall that MEDPAC’s “independent” analysis that also rejected a policy-mortality link came from the organization that recommended the policy to begin with.  The bottom line is this: There’s a tremendous amount of face to lose and a massive source of institutional funding at risk if the policy is found to be harmful.

It now becomes more clear why, in the following tweets, Dr. Krumholz feels that only he can say anything definitive about readmission rates and mortality:

Um, @JAMA_current, even the authors say they cannot say their findings are causal… "but whether this finding is a result of the policy requires further research.” Why do you promote the paper as proving harm? Need to treat twitter like you do any of your Editorial comments. pic.twitter.com/xugasHvPWY

— Harlan Krumholz (@hmkyale) December 26, 2018

6/Hospitals that have improved their readmission rates tended to improve their mortality rates. Published in @JAMA_current… https://t.co/R7FXLwTMPp

— Harlan Krumholz (@hmkyale) December 22, 2018

A diversity of biases

Biases are ubiquitous.  When I was a cardiologist-in-training, spending hours on the hospital consult service for a fixed salary, I vividly recall looking for ways to avoid doing any work I considered unimportant or banal: The minor cardiac enzyme leak in a patient with a widespread infection; The extra heartbeats on the ECG that the ER physician didn’t like the look of; etc.  “Are you sure you need an official consult?”  “The chances we’re going to recommend doing anything about a small enzyme leak in an 80-year-old with a severe lung infection are very low… “ I was even successful sometimes.

Contrast these comments to my demeanor in private practice where I am acutely aware that my income relies on such consults: “I just need the patients name or room number…” “I’ll take care of it!…” “I can put in the orders if you’d like!”

But there are other biases and incentives that motivate human beings, apart from personal financial incentives. Do they pale relative to the financial ones as is so often claimed? How does one begin to quantify them?

When it comes to the HRRP policy, no individual person’s bank account ballooned every time a patient didn’t get admitted. And yet this is a story of ideological bias that drove the design of policy and now claims ‘success’ for its own program.  The HRRP saga is illustrative of the importance of non-financial bias and of the dangers of blinding ourselves to that bias.

The story also highlights the downsides of tweaking healthcare systems that were built to deliver more care.

Clearly, I personally have a direct financial conflict of interest to provide more care. Since I haven’t talked anyone out of a consult in 8 years, I’m probably guilty of participating in a system that detractors appropriately criticize for promoting overuse of healthcare.

But the problem is that some of those consults I was trying to avoid as a fellow ended up really needing a cardiologist. There was the 55-year-old Cambodian woman admitted to the medical intensive care unit with pneumonia who went into atrial fibrillation. I recall rolling my eyes and thinking that the ICU could certainly handle this without a cardiologist. It turned out she didn’t have a pneumonia. It was pulmonary edema from heart failure related to undiagnosed rheumatic mitral valve disease. She had been in the wrong unit. She needed diuresis, heart rate control, and eventual surgery to replace her valve, not antibiotics. Less isn’t always more.

Attempt to reduce inappropriate hospital admissions? Get ready to pay a price.  To contradict Dr. Krumholz, it is entirely probable that we are underestimating the upside of our current system when we contemplate changing the status quo.

We are underestimating the downside of our current system when we contemplate change. We need to take some risks to do better. #abimf2013

— Harlan Krumholz (@hmkyale) August 5, 2013

Final Thoughts

Ironically, the HRRP quagmire offers a number of clarifying lessons.

Empiricism in social policy is a subjective enterprise. The often parroted conclusion is that cold, hard, unbiased evidence trumps the biased, unmeasurable judgment of clinicians. Yet, frequently, real world data-sets are complex, the choice of analytic paths can be highly variable, and the instruments to measure success are often imperfect. As it relates to HRRP, which analysis should we trust? The choice requires faith. And if the currency here is faith, perhaps the concern of clinicians at the bedside has more value than advertised.

Metrics won’t save us. The narrative of metrics is an appealing one that promises hard and objective accountability. The problem comes when the metric (readmission) becomes disconnected from outcomes that actually matter (death). False and blind prophets are good descriptors for those who claim to be unable to see without metrics. The fools in this enterprise are easy to identify as those who think the answers lie with ever better metrics.

Conflicts of interest: Going beyond the simple narrative.  Focusing on biases induced by personal financial interests is a mistake. Personal enrichment is just one bias in a sea of conflicts. In the healthcare context, financial disclosures—while clarifying in themselves—may simply give cover to other, more perverse biases, unless those other biases are equally disclosed.  It requires diligence to ascertain the impact, and direction of bias.  Rarely do we get the opportunity to observe the direction of bias in policy research.  In the case of the HRRP, the presence of bias was made evident because research groups with opposing biases (clinician-scientists versus policy wonks) have reached conclusions that would be expected on the basis of those pre-existing biases.  How often is the problem of such bias examined in the design, implementation, and analysis of health policy?

Beware of technocrats with all the answers.  I am reasonably sure that if practicing clinicians would have been asked to devise a rule to reduce heart failure readmissions for the whole population they would have refused. It seems too challenging a task to get right.  Even if clinicians can be induced to participate in the design of such a policy, most would readily acknowledge the likelihood that it could harm some patients. It requires a special type of hubris to design a policy and refuse to acknowledge its potential for harm. Unfortunately, hubris within a public health community that believes only they can give us a better health system is more feature than bug.

As for Mrs. C, she has been home for 16 days. All fingers on both hands are currently crossed.

Anish Koka is a cardiologist in private practice in Philadelphia.  He can be followed on Twitter @anish_koka. This post originally appeared here on The Accad & Koka Report. 

Commissioning Healthcare Policy: Hospital Readmission and Its Price Tag published first on https://wittooth.tumblr.com/

[kristinsimmons]

Commissioning Healthcare Policy: Hospital Readmission and Its Price Tag

By ANISH KOKA MD 

The message comes in over the office slack line at 1:05 pm. There are four patients in rooms, one new, 3 patients in the waiting room. Really, not an ideal time to deal with this particular message.

“Kathy the home care nurse for Mrs. C called and said her weight yesterday was 185, today it is 194, she has +4 pitting edema, heart rate 120, BP 140/70 standing, 120/64 sitting”

I know Mrs. C well. She has severe COPD from smoking for 45 of the last 55 years. Every breath looks like an effort because it is. The worst part of it all is that Mrs. C just returned home from the hospital just days ago.

The youngest of six children, Mrs. C was born with many embedded disadvantages. Being born black in a poor West Philadelphia neighborhood in the 1960s is a story that too often writes itself with a bad ending.  But Mrs. C avoided the usual pitfalls that derail young women in the neighborhood early. No drugs. No alcohol. No teenage pregnancies. Finished high school. Mrs. C. worked for the hospital as a unit clerk, had her own place, health benefits, and even a retirement plan.

Certain life habits, however, carry a heavy price. George Burns, the comedian never pictured without a cigar who died past his hundredth birthday, may have been immune to the effects of tobacco.  Mrs. C was not. She started smoking when she was 16. She doesn’t recall why. Her dad smoking didn’t help perhaps. Nausea racked her body after that first drag. It eased up after. Too bad.

That measly cigarette became the great addiction of her life. Day by day, the exquisitely thin membranes of the lungs that mediate gas exchange were destroyed. By the time the disease manifests with shortness of breath and bluish tinged lips, it’s too late. Short of the very few who qualify for a lung transplant, the efforts of doctors at this point are for mitigation rather than cure.

Complicating things further, in Mrs. C’s case, the normally low pressure vascular circuit of her lungs became a high pressure circuit that places ever increasing demands on the normally thin-muscled right ventricle of the heart.  This jeopardizes the ability of her heart to handle changes in blood volume.

A little extra fluid and the right side of the heart ends up causing unbearable swelling in her legs.  A little dehydration and severe disabling dizziness on standing ensues. Adding to that that her tenuous lung function decompensates with the slightest respiratory infection, that chronic steroid treatment to decrease her wheezing suppresses her immune system, and that the young man down the street helpfully drops off Newports at her home for a few extra dollars, and it’s easy to see why the hospital is her second home.

The most recent admission to the hospital was for kidney failure related to taking too much fluid off with diuretics. What was to be a short stay for gentle hydration turned into a longer stay when a pneumonia complicated the matters (though a trip to the intensive care unit and a ventilator was barely but fortunately avoided).  She was treated by the pulmonology team and sent home on a lower dose of diuretics.

The situation I am now confronting puts me in a quandary.  Her edema and weight are up markedly just a few days after returning home.  Could her fluid overload be because her kidneys are shutting down? Or does she just need more aggressive diuresis?

Should I guess? Knowing her present renal function would be helpful. But even if the Theranos lab I could appeal to for help wasn’t fictional, I would have to get her to my office everyday or every other day while adjusting her diuretic dose.

And so it comes to be that, days removed from a hospital admission, I’m sending her back to the hospital to be readmitted.  According to some, this is not supposed to happen.

A policy on readmission

In 2008, the commission that advises Medicare – the Medicare Payment Advisory Commission (MEDPAC), issued a report that focused on hospital readmissions.

The focus on hospital readmissions had been of great interest to the health policy community for some time. At the core of this interest lies the belief that hospitals and physicians are incentivized to treat patients rather than prevent admissions.

The MEDPAC report wanted to discourage readmissions like Mrs. C’s.  And so it wasn’t a terrible surprise that, rolled into the 2010 Affordable Care Act, was a section called the Hospital Readmission Reduction Program (HRRP) which created a system for Medicare to penalize hospitals with ‘high’ readmission rates.  The program was rolled out in 2013.

At first, the program seemed to work like a charm.  Hospitals significantly ramped up their efforts at care coordination.  Teams of nurses and aids were assembled to make sure patients would get their medications as prescribed upon discharge and to check on patients once they got home.

Hospital readmission rates suddenly dropped and Medicare started saving money.  A staggering 81% of all hospitals suffered penalties in 2018, which translates to ~$500 million or 0.3% of total Medicare payments to hospitals.

A complex analysis

But there’s more to this too good to be true story.

The HRRP penalty schemes are risk-adjusted based on administrative claims data. Risk-adjustment is a statistical procedure to take into account the diversity in complexity and severity of disease among patients so they can be compared.

Physicians know that risk-adjusted claims data are of dubious value because they themselves are often the reluctant data entry clerk in the byzantine scheme that starts with adding diagnostic items to the medical chart and ends with generating a coded billing claim for Medicare.  Needless to say, there’s a huge potential disconnect between what a claim attempts to convey and the actual condition of a given patient.

Yet another major problem is that the risk-adjustment employed by the HRRP does not take socioeconomic status into account, when that is arguably the single biggest driver of poor outcomes and of hospital readmissions.  The creators of the HRRP program seem to believe that a hospital located in poor area shouldn’t get a break for having high readmission rates, perhaps because they believe that hospital systems in general should be mindful of health inequities and address “care gaps” (differences in care provided to poor vs. affluent patients)  in their neighborhoods no matter what.

Finally, the initial out of the gate benchmark for readmission rates on which the HRRP would adjudicate the need for a penalty was a national average.  In such a scheme, a Johns Hopkins Hospital serving inner city Baltimore could be pitted against a regional hospital in rural Montana with an entirely different patient demographics.  This made the regional hospital in Montana very happy.

Gaming the metrics

Regardless of these technical considerations, it is an adage of social science that any metric will be gamed, and healthcare is unfortunately not immune to that law.

One tool increasingly used by hospitals to comply with Medicare payments rules is to admit patients to short stay units, under so-called “observation status.”  Another is to put pressure on emergency departments to avoid readmitting certain types of patients.

So, instead of primarily functioning as a triage operation where sick patients would be turned over to the care of the cardiologist in the hospital, the ER has been increasingly housing and managing heart failure patients to save the hospital money.

But the ER physician or the hospitalist supervising the short stay unit and who just meet a patient in the setting of an acute illness are poorly equipped to know which heart failure patient to discharge after a diuretic dose and which to keep for advanced heart failure therapies.

Source: MedPAC’s June 2018 Report (http://bit.ly/2CXD0fJ)

Nowadays, the cardiologist is increasingly insulated from those decisions.  I have personally experienced with alarming frequency instances where I learn only after the fact that a complex patient of mine has been treated for heart failure in the ED.

And my experience seems to be shared by many of my cardiology colleagues, especially among cardiologists who work in academic centers that are most affected by the policy.  Luckily, some of them are also clinician-scientists that can do more than just whine to colleagues about the new policy.  They can also study its outcomes.

What do the outcomes data show?

In a pivotal study, a group of cardiologists (Gupta et al.) saw that the drop in readmissions that followed the introduction of HRRP was unfortunately accompanied by a reversal in the decade long downward trend in heart failure mortality. This reversal suggested a serious potential harm from the policy.

But the possibility of harm was quickly challenged by another group of researchers led by one of the biggest names in health policy: Harlan Krumholz, a cardiologist who directs the influential Center for Outcomes Research and Evaluation at Yale University.

The Krumholz group analyzed millions of Medicare claims data from 2007 to 2016.  They also found that mortality for heart failure patients increased, but concluded that no causal link between decreasing readmission and increased mortality could be established.

Krumholz at al. noted that mortality rates for heart failure started climbing before the HRRP program was announced and they noted no inflection point in mortality rates with the policy announcement in 2010.  The evidence for their claim is highlighted in the table below:

As can be seen in the boxed row, Krumholz’ team concludes that the increasing mortality slope post-HRRP is no different from that pre-HRRP because the change did not reach statistical significance at the obligatory and arbitrary P<0.05 level.  The actual P-value was 0.11 and the confidence interval for the positive increase in mortality slope of 0.006 is (-.002 to .015).  Even poor students of epistemology would be loathe to conclude this result excludes a signal of harm.  It seems entirely plausible that, with all the limitations of the data set in question, mortality may in fact have accelerated after the institution of the HRRP.  Yet Krumholz insists that no signal of harm is to be considered.

But this did not stop another group of cardiologists (Wadhera et al.) from adding their contribution to the HRRP literature. Using the same data-set that the Krumholz group used — Medicare claims data — these researchers found once again that accelerating mortality coincided with the announcement of the HRRP.  More troubling, they also demonstrated that mortality rose primarily among patients not readmitted to the hospital.

A messy science

Admittedly this whole business of analysis is incredibly messy, with a number of moving parts.

My brief summary doesn’t do justice to a variety of maneuvers taken by the various groups to account for many of the limitations inherent in this type of study. Two of the competing analyses (Krumholz, Wadhera) used Medicare claims data while the other (Gupta) used a more limited voluntary registry.

During the time period in question, there were also other policy changes such as the introduction of new hospital billing codes (MS-DRG) that sought to adjust hospital payment rates to patient complexity. Better patient coding meant higher reimbursement from Medicare.  Armies of “documenters” were then employed by hospitals to capture more revenue.

This means that the claims data gathered by the researchers might look significantly different from one time period to the next even if the patients themselves were ostensibly the same. As the readmission rate is risk-adjusted, it is eminently plausible and likely that systematically upcoding patient risk could actually have been the primary driver of the drop in hospital readmission rate.

The other program playing a confounding role is the Recovery Audit Contractor (RAC) program begun in 2010 to reduce payments for inappropriate hospital admissions. Hospitals responded to denials for inpatient admissions by expanding ‘observation status’ stays. Which was the biggest driver for expanded observation stays? RAC or the HRRP? Once again an exact attribution is impossible.

Denying the obvious?

Despite the messiness of the data and the variety of analytic methods used, a consistent and uncontested observation remains: Heart failure mortality has increased in the last decade. The question being hotly contested is Why?

Oddly, Dr. Krumholz is steadfast in denying the possibility that the policy may have caused harm, even though the independent and contradictory conclusions from the other research groups at least raises a reasonable doubt.  And Dr. Krumholz has been quick to cast shade on research that does not conform to his conclusion.

By tweet he appears to ask for a level of detail his own papers lack, and he questions the legitimacy of another group’s data-set, all the while resisting any calls to put the program on hold despite the paucity of evidence showing benefit, the signal for harm, and perhaps most importantly, the concern of clinicians who see a mechanism for harm.

Greatly admire @rwyeh and his group…appreciate his focus on readmission &public policy. For such high-profile article, really need more info about statistical weighting. Methods should be sufficient so others can reproduce results. Can’t do that here. Look forward to more info.

— Harlan Krumholz (@hmkyale) December 21, 2018

Can you account for why the registry you used had, on average, such a small number of patients per site. Did you determine how many patients coded with heart failure by CMS were in the Registry? Just curious about the selection. It may not explain your results…but is a question. https://t.co/WSvJAN8Jjd

— Harlan Krumholz (@hmkyale) June 2, 2018

Dr. Krumholz also places much weight on an independent analysis carried out by MEDPAC which concluded there was no link between policy and the uptick in mortality. This particular conclusion rests heavily on the assertion that heart failure patients in 2016 were much sicker than patients in 2010. Recall that this coincides with a period of more intensive coding over the same time frame, so it is impossible to say this with any confidence.

The MEDPAC conclusion also relies on an analysis that finds no correlation between hospital level readmission and mortality rates.

While technically true, that conclusion overlooks that a large number of hospitals exhibited reduced readmission rates and increased mortality. Perhaps MEDPAC feels that patients dying at low readmit/high mortality hospitals should be mollified by the knowledge that somewhere there’s a low-readmit/low mortality hospital to balance things out?

That Krumholz and MEDPAC display such certainty about the direction of the signal they observe, and take pains to discount other possibilities seems strange and suggests that pre-existing biases may be at work. What might those biases be?

Conflicts of interest: You get what you pay for

In a world where heads roll for undisclosed personal financial conflicts of interest, it is remarkable that the current dispute, while full of scintillating exchanges about “propensity weighting” and other arcane points of statistics, does not reference any other potential conflicts at work that might affect the conclusions being reached.

Medicare’s decision to start the HRRP program didn’t come in a vacuum. It was inspired by years of research from Dr. Krumholz himself, who suggested that preventing admissions should be a goal for any policy that would aim to move the system from one paying for “volume” to one paying for “value.”

As far back as 2003, Krumholz held the view that:

“Hospitals may not support programs that improve the quality of care delivered to heart failure patients because these programs lower readmission rates and empty beds, and therefore further diminish already-declining revenues.”

If Krumholz’s unfavorable and crudely simplistic view of the operations and motivations of hospitals (and of the still relatively independent physicians staffing those hospitals) informs his position on health policy, it stands to reason that serious blinders would prevent him from seeing any evidence of harm in a particular policy that promotes the same view.

But that’s not all.  Krumholz’s group at Yale received grants from CMS under the auspices of the Measure and Instrument Development and Support (MIDS) program to study and produce the metrics and instruments needed to devise the readmission measures.

The MIDS program supports the “development and use of clinical quality measures which remains a critical healthcare priority and the tool of choice for improving quality of care at the national, community and facility levels” and it allocates $1.6 billion dollars to this purpose.

Thanks to a bipartisan act of Congress, a helpful little website, usaspending.gov, provides contract level detail about payments made to the Krumholz’s group from the MIDS program.  Those payments can be seen in the table below:

The numbers are staggering. I know little about how to interpret these data about federal contracts, but it sure appears that the Yale-New Haven Health Service group led by Krumholz has received $144 million dollars since 2008.

Yet the only clue to these payments in Krumholz’ published analyses of the HRRP program comes in one disclosure sentence in a footnote, as seen here:

It seems to me that the disclosure is hardly proportional to the amount of funding that his group receives and understates the inherent pressures it must be under to demonstrate that the policy did not actually result in higher mortality.

And recall that MEDPAC’s “independent” analysis that also rejected a policy-mortality link came from the organization that recommended the policy to begin with.  The bottom line is this: There’s a tremendous amount of face to lose and a massive source of institutional funding at risk if the policy is found to be harmful.

It now becomes more clear why, in the following tweets, Dr. Krumholz feels that only he can say anything definitive about readmission rates and mortality:

Um, @JAMA_current, even the authors say they cannot say their findings are causal… "but whether this finding is a result of the policy requires further research.” Why do you promote the paper as proving harm? Need to treat twitter like you do any of your Editorial comments. pic.twitter.com/xugasHvPWY

— Harlan Krumholz (@hmkyale) December 26, 2018

6/Hospitals that have improved their readmission rates tended to improve their mortality rates. Published in @JAMA_current… https://t.co/R7FXLwTMPp

— Harlan Krumholz (@hmkyale) December 22, 2018

A diversity of biases

Biases are ubiquitous.  When I was a cardiologist-in-training, spending hours on the hospital consult service for a fixed salary, I vividly recall looking for ways to avoid doing any work I considered unimportant or banal: The minor cardiac enzyme leak in a patient with a widespread infection; The extra heartbeats on the ECG that the ER physician didn’t like the look of; etc.  “Are you sure you need an official consult?”  “The chances we’re going to recommend doing anything about a small enzyme leak in an 80-year-old with a severe lung infection are very low… “ I was even successful sometimes.

Contrast these comments to my demeanor in private practice where I am acutely aware that my income relies on such consults: “I just need the patients name or room number…” “I’ll take care of it!…” “I can put in the orders if you’d like!”

But there are other biases and incentives that motivate human beings, apart from personal financial incentives. Do they pale relative to the financial ones as is so often claimed? How does one begin to quantify them?

When it comes to the HRRP policy, no individual person’s bank account ballooned every time a patient didn’t get admitted. And yet this is a story of ideological bias that drove the design of policy and now claims ‘success’ for its own program.  The HRRP saga is illustrative of the importance of non-financial bias and of the dangers of blinding ourselves to that bias.

The story also highlights the downsides of tweaking healthcare systems that were built to deliver more care.

Clearly, I personally have a direct financial conflict of interest to provide more care. Since I haven’t talked anyone out of a consult in 8 years, I’m probably guilty of participating in a system that detractors appropriately criticize for promoting overuse of healthcare.

But the problem is that some of those consults I was trying to avoid as a fellow ended up really needing a cardiologist. There was the 55-year-old Cambodian woman admitted to the medical intensive care unit with pneumonia who went into atrial fibrillation. I recall rolling my eyes and thinking that the ICU could certainly handle this without a cardiologist. It turned out she didn’t have a pneumonia. It was pulmonary edema from heart failure related to undiagnosed rheumatic mitral valve disease. She had been in the wrong unit. She needed diuresis, heart rate control, and eventual surgery to replace her valve, not antibiotics. Less isn’t always more.

Attempt to reduce inappropriate hospital admissions? Get ready to pay a price.  To contradict Dr. Krumholz, it is entirely probable that we are underestimating the upside of our current system when we contemplate changing the status quo.

We are underestimating the downside of our current system when we contemplate change. We need to take some risks to do better. #abimf2013

— Harlan Krumholz (@hmkyale) August 5, 2013

Final Thoughts

Ironically, the HRRP quagmire offers a number of clarifying lessons.

Empiricism in social policy is a subjective enterprise. The often parroted conclusion is that cold, hard, unbiased evidence trumps the biased, unmeasurable judgment of clinicians. Yet, frequently, real world data-sets are complex, the choice of analytic paths can be highly variable, and the instruments to measure success are often imperfect. As it relates to HRRP, which analysis should we trust? The choice requires faith. And if the currency here is faith, perhaps the concern of clinicians at the bedside has more value than advertised.

Metrics won’t save us. The narrative of metrics is an appealing one that promises hard and objective accountability. The problem comes when the metric (readmission) becomes disconnected from outcomes that actually matter (death). False and blind prophets are good descriptors for those who claim to be unable to see without metrics. The fools in this enterprise are easy to identify as those who think the answers lie with ever better metrics.

Conflicts of interest: Going beyond the simple narrative.  Focusing on biases induced by personal financial interests is a mistake. Personal enrichment is just one bias in a sea of conflicts. In the healthcare context, financial disclosures—while clarifying in themselves—may simply give cover to other, more perverse biases, unless those other biases are equally disclosed.  It requires diligence to ascertain the impact, and direction of bias.  Rarely do we get the opportunity to observe the direction of bias in policy research.  In the case of the HRRP, the presence of bias was made evident because research groups with opposing biases (clinician-scientists versus policy wonks) have reached conclusions that would be expected on the basis of those pre-existing biases.  How often is the problem of such bias examined in the design, implementation, and analysis of health policy?

Beware of technocrats with all the answers.  I am reasonably sure that if practicing clinicians would have been asked to devise a rule to reduce heart failure readmissions for the whole population they would have refused. It seems too challenging a task to get right.  Even if clinicians can be induced to participate in the design of such a policy, most would readily acknowledge the likelihood that it could harm some patients. It requires a special type of hubris to design a policy and refuse to acknowledge its potential for harm. Unfortunately, hubris within a public health community that believes only they can give us a better health system is more feature than bug.

As for Mrs. C, she has been home for 16 days. All fingers on both hands are currently crossed.

Anish Koka is a cardiologist in private practice in Philadelphia.  He can be followed on Twitter @anish_koka. This post originally appeared here on The Accad & Koka Report. 

Commissioning Healthcare Policy: Hospital Readmission and Its Price Tag published first on https://wittooth.tumblr.com/

[kristinsimmons]

Commissioning Healthcare Policy: Hospital Readmission and Its Price Tag

By ANISH KOKA MD 

The message comes in over the office slack line at 1:05 pm. There are four patients in rooms, one new, 3 patients in the waiting room. Really, not an ideal time to deal with this particular message.

“Kathy the home care nurse for Mrs. C called and said her weight yesterday was 185, today it is 194, she has +4 pitting edema, heart rate 120, BP 140/70 standing, 120/64 sitting”

I know Mrs. C well. She has severe COPD from smoking for 45 of the last 55 years. Every breath looks like an effort because it is. The worst part of it all is that Mrs. C just returned home from the hospital just days ago.

The youngest of six children, Mrs. C was born with many embedded disadvantages. Being born black in a poor West Philadelphia neighborhood in the 1960s is a story that too often writes itself with a bad ending.  But Mrs. C avoided the usual pitfalls that derail young women in the neighborhood early. No drugs. No alcohol. No teenage pregnancies. Finished high school. Mrs. C. worked for the hospital as a unit clerk, had her own place, health benefits, and even a retirement plan.

Certain life habits, however, carry a heavy price. George Burns, the comedian never pictured without a cigar who died past his hundredth birthday, may have been immune to the effects of tobacco.  Mrs. C was not. She started smoking when she was 16. She doesn’t recall why. Her dad smoking didn’t help perhaps. Nausea racked her body after that first drag. It eased up after. Too bad.

That measly cigarette became the great addiction of her life. Day by day, the exquisitely thin membranes of the lungs that mediate gas exchange were destroyed. By the time the disease manifests with shortness of breath and bluish tinged lips, it’s too late. Short of the very few who qualify for a lung transplant, the efforts of doctors at this point are for mitigation rather than cure.

Complicating things further, in Mrs. C’s case, the normally low pressure vascular circuit of her lungs became a high pressure circuit that places ever increasing demands on the normally thin-muscled right ventricle of the heart.  This jeopardizes the ability of her heart to handle changes in blood volume.

A little extra fluid and the right side of the heart ends up causing unbearable swelling in her legs.  A little dehydration and severe disabling dizziness on standing ensues. Adding to that that her tenuous lung function decompensates with the slightest respiratory infection, that chronic steroid treatment to decrease her wheezing suppresses her immune system, and that the young man down the street helpfully drops off Newports at her home for a few extra dollars, and it’s easy to see why the hospital is her second home.

The most recent admission to the hospital was for kidney failure related to taking too much fluid off with diuretics. What was to be a short stay for gentle hydration turned into a longer stay when a pneumonia complicated the matters (though a trip to the intensive care unit and a ventilator was barely but fortunately avoided).  She was treated by the pulmonology team and sent home on a lower dose of diuretics.

The situation I am now confronting puts me in a quandary.  Her edema and weight are up markedly just a few days after returning home.  Could her fluid overload be because her kidneys are shutting down? Or does she just need more aggressive diuresis?

Should I guess? Knowing her present renal function would be helpful. But even if the Theranos lab I could appeal to for help wasn’t fictional, I would have to get her to my office everyday or every other day while adjusting her diuretic dose.

And so it comes to be that, days removed from a hospital admission, I’m sending her back to the hospital to be readmitted.  According to some, this is not supposed to happen.

A policy on readmission

In 2008, the commission that advises Medicare – the Medicare Payment Advisory Commission (MEDPAC), issued a report that focused on hospital readmissions.

The focus on hospital readmissions had been of great interest to the health policy community for some time. At the core of this interest lies the belief that hospitals and physicians are incentivized to treat patients rather than prevent admissions.

The MEDPAC report wanted to discourage readmissions like Mrs. C’s.  And so it wasn’t a terrible surprise that, rolled into the 2010 Affordable Care Act, was a section called the Hospital Readmission Reduction Program (HRRP) which created a system for Medicare to penalize hospitals with ‘high’ readmission rates.  The program was rolled out in 2013.

At first, the program seemed to work like a charm.  Hospitals significantly ramped up their efforts at care coordination.  Teams of nurses and aids were assembled to make sure patients would get their medications as prescribed upon discharge and to check on patients once they got home.

Hospital readmission rates suddenly dropped and Medicare started saving money.  A staggering 81% of all hospitals suffered penalties in 2018, which translates to ~$500 million or 0.3% of total Medicare payments to hospitals.

A complex analysis

But there’s more to this too good to be true story.

The HRRP penalty schemes are risk-adjusted based on administrative claims data. Risk-adjustment is a statistical procedure to take into account the diversity in complexity and severity of disease among patients so they can be compared.

Physicians know that risk-adjusted claims data are of dubious value because they themselves are often the reluctant data entry clerk in the byzantine scheme that starts with adding diagnostic items to the medical chart and ends with generating a coded billing claim for Medicare.  Needless to say, there’s a huge potential disconnect between what a claim attempts to convey and the actual condition of a given patient.

Yet another major problem is that the risk-adjustment employed by the HRRP does not take socioeconomic status into account, when that is arguably the single biggest driver of poor outcomes and of hospital readmissions.  The creators of the HRRP program seem to believe that a hospital located in poor area shouldn’t get a break for having high readmission rates, perhaps because they believe that hospital systems in general should be mindful of health inequities and address “care gaps” (differences in care provided to poor vs. affluent patients)  in their neighborhoods no matter what.

Finally, the initial out of the gate benchmark for readmission rates on which the HRRP would adjudicate the need for a penalty was a national average.  In such a scheme, a Johns Hopkins Hospital serving inner city Baltimore could be pitted against a regional hospital in rural Montana with an entirely different patient demographics.  This made the regional hospital in Montana very happy.

Gaming the metrics

Regardless of these technical considerations, it is an adage of social science that any metric will be gamed, and healthcare is unfortunately not immune to that law.

One tool increasingly used by hospitals to comply with Medicare payments rules is to admit patients to short stay units, under so-called “observation status.”  Another is to put pressure on emergency departments to avoid readmitting certain types of patients.

So, instead of primarily functioning as a triage operation where sick patients would be turned over to the care of the cardiologist in the hospital, the ER has been increasingly housing and managing heart failure patients to save the hospital money.

But the ER physician or the hospitalist supervising the short stay unit and who just meet a patient in the setting of an acute illness are poorly equipped to know which heart failure patient to discharge after a diuretic dose and which to keep for advanced heart failure therapies.

Source: MedPAC’s June 2018 Report (http://bit.ly/2CXD0fJ)

Nowadays, the cardiologist is increasingly insulated from those decisions.  I have personally experienced with alarming frequency instances where I learn only after the fact that a complex patient of mine has been treated for heart failure in the ED.

And my experience seems to be shared by many of my cardiology colleagues, especially among cardiologists who work in academic centers that are most affected by the policy.  Luckily, some of them are also clinician-scientists that can do more than just whine to colleagues about the new policy.  They can also study its outcomes.

What do the outcomes data show?

In a pivotal study, a group of cardiologists (Gupta et al.) saw that the drop in readmissions that followed the introduction of HRRP was unfortunately accompanied by a reversal in the decade long downward trend in heart failure mortality. This reversal suggested a serious potential harm from the policy.

But the possibility of harm was quickly challenged by another group of researchers led by one of the biggest names in health policy: Harlan Krumholz, a cardiologist who directs the influential Center for Outcomes Research and Evaluation at Yale University.

The Krumholz group analyzed millions of Medicare claims data from 2007 to 2016.  They also found that mortality for heart failure patients increased, but concluded that no causal link between decreasing readmission and increased mortality could be established.

Krumholz at al. noted that mortality rates for heart failure started climbing before the HRRP program was announced and they noted no inflection point in mortality rates with the policy announcement in 2010.  The evidence for their claim is highlighted in the table below:

As can be seen in the boxed row, Krumholz’ team concludes that the increasing mortality slope post-HRRP is no different from that pre-HRRP because the change did not reach statistical significance at the obligatory and arbitrary P<0.05 level.  The actual P-value was 0.11 and the confidence interval for the positive increase in mortality slope of 0.006 is (-.002 to .015).  Even poor students of epistemology would be loathe to conclude this result excludes a signal of harm.  It seems entirely plausible that, with all the limitations of the data set in question, mortality may in fact have accelerated after the institution of the HRRP.  Yet Krumholz insists that no signal of harm is to be considered.

But this did not stop another group of cardiologists (Wadhera et al.) from adding their contribution to the HRRP literature. Using the same data-set that the Krumholz group used — Medicare claims data — these researchers found once again that accelerating mortality coincided with the announcement of the HRRP.  More troubling, they also demonstrated that mortality rose primarily among patients not readmitted to the hospital.

A messy science

Admittedly this whole business of analysis is incredibly messy, with a number of moving parts.

My brief summary doesn’t do justice to a variety of maneuvers taken by the various groups to account for many of the limitations inherent in this type of study. Two of the competing analyses (Krumholz, Wadhera) used Medicare claims data while the other (Gupta) used a more limited voluntary registry.

During the time period in question, there were also other policy changes such as the introduction of new hospital billing codes (MS-DRG) that sought to adjust hospital payment rates to patient complexity. Better patient coding meant higher reimbursement from Medicare.  Armies of “documenters” were then employed by hospitals to capture more revenue.

This means that the claims data gathered by the researchers might look significantly different from one time period to the next even if the patients themselves were ostensibly the same. As the readmission rate is risk-adjusted, it is eminently plausible and likely that systematically upcoding patient risk could actually have been the primary driver of the drop in hospital readmission rate.

The other program playing a confounding role is the Recovery Audit Contractor (RAC) program begun in 2010 to reduce payments for inappropriate hospital admissions. Hospitals responded to denials for inpatient admissions by expanding ‘observation status’ stays. Which was the biggest driver for expanded observation stays? RAC or the HRRP? Once again an exact attribution is impossible.

Denying the obvious?

Despite the messiness of the data and the variety of analytic methods used, a consistent and uncontested observation remains: Heart failure mortality has increased in the last decade. The question being hotly contested is Why?

Oddly, Dr. Krumholz is steadfast in denying the possibility that the policy may have caused harm, even though the independent and contradictory conclusions from the other research groups at least raises a reasonable doubt.  And Dr. Krumholz has been quick to cast shade on research that does not conform to his conclusion.

By tweet he appears to ask for a level of detail his own papers lack, and he questions the legitimacy of another group’s data-set, all the while resisting any calls to put the program on hold despite the paucity of evidence showing benefit, the signal for harm, and perhaps most importantly, the concern of clinicians who see a mechanism for harm.

Greatly admire @rwyeh and his group…appreciate his focus on readmission &public policy. For such high-profile article, really need more info about statistical weighting. Methods should be sufficient so others can reproduce results. Can’t do that here. Look forward to more info.

— Harlan Krumholz (@hmkyale) December 21, 2018

Can you account for why the registry you used had, on average, such a small number of patients per site. Did you determine how many patients coded with heart failure by CMS were in the Registry? Just curious about the selection. It may not explain your results…but is a question. https://t.co/WSvJAN8Jjd

— Harlan Krumholz (@hmkyale) June 2, 2018

Dr. Krumholz also places much weight on an independent analysis carried out by MEDPAC which concluded there was no link between policy and the uptick in mortality. This particular conclusion rests heavily on the assertion that heart failure patients in 2016 were much sicker than patients in 2010. Recall that this coincides with a period of more intensive coding over the same time frame, so it is impossible to say this with any confidence.

The MEDPAC conclusion also relies on an analysis that finds no correlation between hospital level readmission and mortality rates.

While technically true, that conclusion overlooks that a large number of hospitals exhibited reduced readmission rates and increased mortality. Perhaps MEDPAC feels that patients dying at low readmit/high mortality hospitals should be mollified by the knowledge that somewhere there’s a low-readmit/low mortality hospital to balance things out?

That Krumholz and MEDPAC display such certainty about the direction of the signal they observe, and take pains to discount other possibilities seems strange and suggests that pre-existing biases may be at work. What might those biases be?

Conflicts of interest: You get what you pay for

In a world where heads roll for undisclosed personal financial conflicts of interest, it is remarkable that the current dispute, while full of scintillating exchanges about “propensity weighting” and other arcane points of statistics, does not reference any other potential conflicts at work that might affect the conclusions being reached.

Medicare’s decision to start the HRRP program didn’t come in a vacuum. It was inspired by years of research from Dr. Krumholz himself, who suggested that preventing admissions should be a goal for any policy that would aim to move the system from one paying for “volume” to one paying for “value.”

As far back as 2003, Krumholz held the view that:

“Hospitals may not support programs that improve the quality of care delivered to heart failure patients because these programs lower readmission rates and empty beds, and therefore further diminish already-declining revenues.”

If Krumholz’s unfavorable and crudely simplistic view of the operations and motivations of hospitals (and of the still relatively independent physicians staffing those hospitals) informs his position on health policy, it stands to reason that serious blinders would prevent him from seeing any evidence of harm in a particular policy that promotes the same view.

But that’s not all.  Krumholz’s group at Yale received grants from CMS under the auspices of the Measure and Instrument Development and Support (MIDS) program to study and produce the metrics and instruments needed to devise the readmission measures.

The MIDS program supports the “development and use of clinical quality measures which remains a critical healthcare priority and the tool of choice for improving quality of care at the national, community and facility levels” and it allocates $1.6 billion dollars to this purpose.

Thanks to a bipartisan act of Congress, a helpful little website, usaspending.gov, provides contract level detail about payments made to the Krumholz’s group from the MIDS program.  Those payments can be seen in the table below:

The numbers are staggering. I know little about how to interpret these data about federal contracts, but it sure appears that the Yale-New Haven Health Service group led by Krumholz has received $144 million dollars since 2008.

Yet the only clue to these payments in Krumholz’ published analyses of the HRRP program comes in one disclosure sentence in a footnote, as seen here:

It seems to me that the disclosure is hardly proportional to the amount of funding that his group receives and understates the inherent pressures it must be under to demonstrate that the policy did not actually result in higher mortality.

And recall that MEDPAC’s “independent” analysis that also rejected a policy-mortality link came from the organization that recommended the policy to begin with.  The bottom line is this: There’s a tremendous amount of face to lose and a massive source of institutional funding at risk if the policy is found to be harmful.

It now becomes more clear why, in the following tweets, Dr. Krumholz feels that only he can say anything definitive about readmission rates and mortality:

Um, @JAMA_current, even the authors say they cannot say their findings are causal… "but whether this finding is a result of the policy requires further research.” Why do you promote the paper as proving harm? Need to treat twitter like you do any of your Editorial comments. pic.twitter.com/xugasHvPWY

— Harlan Krumholz (@hmkyale) December 26, 2018

6/Hospitals that have improved their readmission rates tended to improve their mortality rates. Published in @JAMA_current… https://t.co/R7FXLwTMPp

— Harlan Krumholz (@hmkyale) December 22, 2018

A diversity of biases

Biases are ubiquitous.  When I was a cardiologist-in-training, spending hours on the hospital consult service for a fixed salary, I vividly recall looking for ways to avoid doing any work I considered unimportant or banal: The minor cardiac enzyme leak in a patient with a widespread infection; The extra heartbeats on the ECG that the ER physician didn’t like the look of; etc.  “Are you sure you need an official consult?”  “The chances we’re going to recommend doing anything about a small enzyme leak in an 80-year-old with a severe lung infection are very low… “ I was even successful sometimes.

Contrast these comments to my demeanor in private practice where I am acutely aware that my income relies on such consults: “I just need the patients name or room number…” “I’ll take care of it!…” “I can put in the orders if you’d like!”

But there are other biases and incentives that motivate human beings, apart from personal financial incentives. Do they pale relative to the financial ones as is so often claimed? How does one begin to quantify them?

When it comes to the HRRP policy, no individual person’s bank account ballooned every time a patient didn’t get admitted. And yet this is a story of ideological bias that drove the design of policy and now claims ‘success’ for its own program.  The HRRP saga is illustrative of the importance of non-financial bias and of the dangers of blinding ourselves to that bias.

The story also highlights the downsides of tweaking healthcare systems that were built to deliver more care.

Clearly, I personally have a direct financial conflict of interest to provide more care. Since I haven’t talked anyone out of a consult in 8 years, I’m probably guilty of participating in a system that detractors appropriately criticize for promoting overuse of healthcare.

But the problem is that some of those consults I was trying to avoid as a fellow ended up really needing a cardiologist. There was the 55-year-old Cambodian woman admitted to the medical intensive care unit with pneumonia who went into atrial fibrillation. I recall rolling my eyes and thinking that the ICU could certainly handle this without a cardiologist. It turned out she didn’t have a pneumonia. It was pulmonary edema from heart failure related to undiagnosed rheumatic mitral valve disease. She had been in the wrong unit. She needed diuresis, heart rate control, and eventual surgery to replace her valve, not antibiotics. Less isn’t always more.

Attempt to reduce inappropriate hospital admissions? Get ready to pay a price.  To contradict Dr. Krumholz, it is entirely probable that we are underestimating the upside of our current system when we contemplate changing the status quo.

We are underestimating the downside of our current system when we contemplate change. We need to take some risks to do better. #abimf2013

— Harlan Krumholz (@hmkyale) August 5, 2013

Final Thoughts

Ironically, the HRRP quagmire offers a number of clarifying lessons.

Empiricism in social policy is a subjective enterprise. The often parroted conclusion is that cold, hard, unbiased evidence trumps the biased, unmeasurable judgment of clinicians. Yet, frequently, real world data-sets are complex, the choice of analytic paths can be highly variable, and the instruments to measure success are often imperfect. As it relates to HRRP, which analysis should we trust? The choice requires faith. And if the currency here is faith, perhaps the concern of clinicians at the bedside has more value than advertised.

Metrics won’t save us. The narrative of metrics is an appealing one that promises hard and objective accountability. The problem comes when the metric (readmission) becomes disconnected from outcomes that actually matter (death). False and blind prophets are good descriptors for those who claim to be unable to see without metrics. The fools in this enterprise are easy to identify as those who think the answers lie with ever better metrics.

Conflicts of interest: Going beyond the simple narrative.  Focusing on biases induced by personal financial interests is a mistake. Personal enrichment is just one bias in a sea of conflicts. In the healthcare context, financial disclosures—while clarifying in themselves—may simply give cover to other, more perverse biases, unless those other biases are equally disclosed.  It requires diligence to ascertain the impact, and direction of bias.  Rarely do we get the opportunity to observe the direction of bias in policy research.  In the case of the HRRP, the presence of bias was made evident because research groups with opposing biases (clinician-scientists versus policy wonks) have reached conclusions that would be expected on the basis of those pre-existing biases.  How often is the problem of such bias examined in the design, implementation, and analysis of health policy?

Beware of technocrats with all the answers.  I am reasonably sure that if practicing clinicians would have been asked to devise a rule to reduce heart failure readmissions for the whole population they would have refused. It seems too challenging a task to get right.  Even if clinicians can be induced to participate in the design of such a policy, most would readily acknowledge the likelihood that it could harm some patients. It requires a special type of hubris to design a policy and refuse to acknowledge its potential for harm. Unfortunately, hubris within a public health community that believes only they can give us a better health system is more feature than bug.

As for Mrs. C, she has been home for 16 days. All fingers on both hands are currently crossed.

Anish Koka is a cardiologist in private practice in Philadelphia.  He can be followed on Twitter @anish_koka. This post originally appeared here on The Accad & Koka Report. 

Commissioning Healthcare Policy: Hospital Readmission and Its Price Tag published first on https://wittooth.tumblr.com/

[kristinsimmons]

Commissioning Healthcare Policy: Hospital Readmission and Its Price Tag

By ANISH KOKA MD 

The message comes in over the office slack line at 1:05 pm. There are four patients in rooms, one new, 3 patients in the waiting room. Really, not an ideal time to deal with this particular message.

“Kathy the home care nurse for Mrs. C called and said her weight yesterday was 185, today it is 194, she has +4 pitting edema, heart rate 120, BP 140/70 standing, 120/64 sitting”

I know Mrs. C well. She has severe COPD from smoking for 45 of the last 55 years. Every breath looks like an effort because it is. The worst part of it all is that Mrs. C just returned home from the hospital just days ago.

The youngest of six children, Mrs. C was born with many embedded disadvantages. Being born black in a poor West Philadelphia neighborhood in the 1960s is a story that too often writes itself with a bad ending.  But Mrs. C avoided the usual pitfalls that derail young women in the neighborhood early. No drugs. No alcohol. No teenage pregnancies. Finished high school. Mrs. C. worked for the hospital as a unit clerk, had her own place, health benefits, and even a retirement plan.

Certain life habits, however, carry a heavy price. George Burns, the comedian never pictured without a cigar who died past his hundredth birthday, may have been immune to the effects of tobacco.  Mrs. C was not. She started smoking when she was 16. She doesn’t recall why. Her dad smoking didn’t help perhaps. Nausea racked her body after that first drag. It eased up after. Too bad.

That measly cigarette became the great addiction of her life. Day by day, the exquisitely thin membranes of the lungs that mediate gas exchange were destroyed. By the time the disease manifests with shortness of breath and bluish tinged lips, it’s too late. Short of the very few who qualify for a lung transplant, the efforts of doctors at this point are for mitigation rather than cure.

Complicating things further, in Mrs. C’s case, the normally low pressure vascular circuit of her lungs became a high pressure circuit that places ever increasing demands on the normally thin-muscled right ventricle of the heart.  This jeopardizes the ability of her heart to handle changes in blood volume.

A little extra fluid and the right side of the heart ends up causing unbearable swelling in her legs.  A little dehydration and severe disabling dizziness on standing ensues. Adding to that that her tenuous lung function decompensates with the slightest respiratory infection, that chronic steroid treatment to decrease her wheezing suppresses her immune system, and that the young man down the street helpfully drops off Newports at her home for a few extra dollars, and it’s easy to see why the hospital is her second home.

The most recent admission to the hospital was for kidney failure related to taking too much fluid off with diuretics. What was to be a short stay for gentle hydration turned into a longer stay when a pneumonia complicated the matters (though a trip to the intensive care unit and a ventilator was barely but fortunately avoided).  She was treated by the pulmonology team and sent home on a lower dose of diuretics.

The situation I am now confronting puts me in a quandary.  Her edema and weight are up markedly just a few days after returning home.  Could her fluid overload be because her kidneys are shutting down? Or does she just need more aggressive diuresis?

Should I guess? Knowing her present renal function would be helpful. But even if the Theranos lab I could appeal to for help wasn’t fictional, I would have to get her to my office everyday or every other day while adjusting her diuretic dose.

And so it comes to be that, days removed from a hospital admission, I’m sending her back to the hospital to be readmitted.  According to some, this is not supposed to happen.

A policy on readmission

In 2008, the commission that advises Medicare – the Medicare Payment Advisory Commission (MEDPAC), issued a report that focused on hospital readmissions.

The focus on hospital readmissions had been of great interest to the health policy community for some time. At the core of this interest lies the belief that hospitals and physicians are incentivized to treat patients rather than prevent admissions.

The MEDPAC report wanted to discourage readmissions like Mrs. C’s.  And so it wasn’t a terrible surprise that, rolled into the 2010 Affordable Care Act, was a section called the Hospital Readmission Reduction Program (HRRP) which created a system for Medicare to penalize hospitals with ‘high’ readmission rates.  The program was rolled out in 2013.

At first, the program seemed to work like a charm.  Hospitals significantly ramped up their efforts at care coordination.  Teams of nurses and aids were assembled to make sure patients would get their medications as prescribed upon discharge and to check on patients once they got home.

Hospital readmission rates suddenly dropped and Medicare started saving money.  A staggering 81% of all hospitals suffered penalties in 2018, which translates to ~$500 million or 0.3% of total Medicare payments to hospitals.

A complex analysis

But there’s more to this too good to be true story.

The HRRP penalty schemes are risk-adjusted based on administrative claims data. Risk-adjustment is a statistical procedure to take into account the diversity in complexity and severity of disease among patients so they can be compared.

Physicians know that risk-adjusted claims data are of dubious value because they themselves are often the reluctant data entry clerk in the byzantine scheme that starts with adding diagnostic items to the medical chart and ends with generating a coded billing claim for Medicare.  Needless to say, there’s a huge potential disconnect between what a claim attempts to convey and the actual condition of a given patient.

Yet another major problem is that the risk-adjustment employed by the HRRP does not take socioeconomic status into account, when that is arguably the single biggest driver of poor outcomes and of hospital readmissions.  The creators of the HRRP program seem to believe that a hospital located in poor area shouldn’t get a break for having high readmission rates, perhaps because they believe that hospital systems in general should be mindful of health inequities and address “care gaps” (differences in care provided to poor vs. affluent patients)  in their neighborhoods no matter what.

Finally, the initial out of the gate benchmark for readmission rates on which the HRRP would adjudicate the need for a penalty was a national average.  In such a scheme, a Johns Hopkins Hospital serving inner city Baltimore could be pitted against a regional hospital in rural Montana with an entirely different patient demographics.  This made the regional hospital in Montana very happy.

Gaming the metrics

Regardless of these technical considerations, it is an adage of social science that any metric will be gamed, and healthcare is unfortunately not immune to that law.

One tool increasingly used by hospitals to comply with Medicare payments rules is to admit patients to short stay units, under so-called “observation status.”  Another is to put pressure on emergency departments to avoid readmitting certain types of patients.

So, instead of primarily functioning as a triage operation where sick patients would be turned over to the care of the cardiologist in the hospital, the ER has been increasingly housing and managing heart failure patients to save the hospital money.

But the ER physician or the hospitalist supervising the short stay unit and who just meet a patient in the setting of an acute illness are poorly equipped to know which heart failure patient to discharge after a diuretic dose and which to keep for advanced heart failure therapies.

Source: MedPAC’s June 2018 Report (http://bit.ly/2CXD0fJ)

Nowadays, the cardiologist is increasingly insulated from those decisions.  I have personally experienced with alarming frequency instances where I learn only after the fact that a complex patient of mine has been treated for heart failure in the ED.

And my experience seems to be shared by many of my cardiology colleagues, especially among cardiologists who work in academic centers that are most affected by the policy.  Luckily, some of them are also clinician-scientists that can do more than just whine to colleagues about the new policy.  They can also study its outcomes.

What do the outcomes data show?

In a pivotal study, a group of cardiologists (Gupta et al.) saw that the drop in readmissions that followed the introduction of HRRP was unfortunately accompanied by a reversal in the decade long downward trend in heart failure mortality. This reversal suggested a serious potential harm from the policy.

But the possibility of harm was quickly challenged by another group of researchers led by one of the biggest names in health policy: Harlan Krumholz, a cardiologist who directs the influential Center for Outcomes Research and Evaluation at Yale University.

The Krumholz group analyzed millions of Medicare claims data from 2007 to 2016.  They also found that mortality for heart failure patients increased, but concluded that no causal link between decreasing readmission and increased mortality could be established.

Krumholz at al. noted that mortality rates for heart failure started climbing before the HRRP program was announced and they noted no inflection point in mortality rates with the policy announcement in 2010.  The evidence for their claim is highlighted in the table below:

As can be seen in the boxed row, Krumholz’ team concludes that the increasing mortality slope post-HRRP is no different from that pre-HRRP because the change did not reach statistical significance at the obligatory and arbitrary P<0.05 level.  The actual P-value was 0.11 and the confidence interval for the positive increase in mortality slope of 0.006 is (-.002 to .015).  Even poor students of epistemology would be loathe to conclude this result excludes a signal of harm.  It seems entirely plausible that, with all the limitations of the data set in question, mortality may in fact have accelerated after the institution of the HRRP.  Yet Krumholz insists that no signal of harm is to be considered.

But this did not stop another group of cardiologists (Wadhera et al.) from adding their contribution to the HRRP literature. Using the same data-set that the Krumholz group used — Medicare claims data — these researchers found once again that accelerating mortality coincided with the announcement of the HRRP.  More troubling, they also demonstrated that mortality rose primarily among patients not readmitted to the hospital.

A messy science

Admittedly this whole business of analysis is incredibly messy, with a number of moving parts.

My brief summary doesn’t do justice to a variety of maneuvers taken by the various groups to account for many of the limitations inherent in this type of study. Two of the competing analyses (Krumholz, Wadhera) used Medicare claims data while the other (Gupta) used a more limited voluntary registry.

During the time period in question, there were also other policy changes such as the introduction of new hospital billing codes (MS-DRG) that sought to adjust hospital payment rates to patient complexity. Better patient coding meant higher reimbursement from Medicare.  Armies of “documenters” were then employed by hospitals to capture more revenue.

This means that the claims data gathered by the researchers might look significantly different from one time period to the next even if the patients themselves were ostensibly the same. As the readmission rate is risk-adjusted, it is eminently plausible and likely that systematically upcoding patient risk could actually have been the primary driver of the drop in hospital readmission rate.

The other program playing a confounding role is the Recovery Audit Contractor (RAC) program begun in 2010 to reduce payments for inappropriate hospital admissions. Hospitals responded to denials for inpatient admissions by expanding ‘observation status’ stays. Which was the biggest driver for expanded observation stays? RAC or the HRRP? Once again an exact attribution is impossible.

Denying the obvious?

Despite the messiness of the data and the variety of analytic methods used, a consistent and uncontested observation remains: Heart failure mortality has increased in the last decade. The question being hotly contested is Why?

Oddly, Dr. Krumholz is steadfast in denying the possibility that the policy may have caused harm, even though the independent and contradictory conclusions from the other research groups at least raises a reasonable doubt.  And Dr. Krumholz has been quick to cast shade on research that does not conform to his conclusion.

By tweet he appears to ask for a level of detail his own papers lack, and he questions the legitimacy of another group’s data-set, all the while resisting any calls to put the program on hold despite the paucity of evidence showing benefit, the signal for harm, and perhaps most importantly, the concern of clinicians who see a mechanism for harm.

Greatly admire @rwyeh and his group…appreciate his focus on readmission &public policy. For such high-profile article, really need more info about statistical weighting. Methods should be sufficient so others can reproduce results. Can’t do that here. Look forward to more info.

— Harlan Krumholz (@hmkyale) December 21, 2018

Can you account for why the registry you used had, on average, such a small number of patients per site. Did you determine how many patients coded with heart failure by CMS were in the Registry? Just curious about the selection. It may not explain your results…but is a question. https://t.co/WSvJAN8Jjd

— Harlan Krumholz (@hmkyale) June 2, 2018

Dr. Krumholz also places much weight on an independent analysis carried out by MEDPAC which concluded there was no link between policy and the uptick in mortality. This particular conclusion rests heavily on the assertion that heart failure patients in 2016 were much sicker than patients in 2010. Recall that this coincides with a period of more intensive coding over the same time frame, so it is impossible to say this with any confidence.

The MEDPAC conclusion also relies on an analysis that finds no correlation between hospital level readmission and mortality rates.

While technically true, that conclusion overlooks that a large number of hospitals exhibited reduced readmission rates and increased mortality. Perhaps MEDPAC feels that patients dying at low readmit/high mortality hospitals should be mollified by the knowledge that somewhere there’s a low-readmit/low mortality hospital to balance things out?

That Krumholz and MEDPAC display such certainty about the direction of the signal they observe, and take pains to discount other possibilities seems strange and suggests that pre-existing biases may be at work. What might those biases be?

Conflicts of interest: You get what you pay for

In a world where heads roll for undisclosed personal financial conflicts of interest, it is remarkable that the current dispute, while full of scintillating exchanges about “propensity weighting” and other arcane points of statistics, does not reference any other potential conflicts at work that might affect the conclusions being reached.

Medicare’s decision to start the HRRP program didn’t come in a vacuum. It was inspired by years of research from Dr. Krumholz himself, who suggested that preventing admissions should be a goal for any policy that would aim to move the system from one paying for “volume” to one paying for “value.”

As far back as 2003, Krumholz held the view that:

“Hospitals may not support programs that improve the quality of care delivered to heart failure patients because these programs lower readmission rates and empty beds, and therefore further diminish already-declining revenues.”

If Krumholz’s unfavorable and crudely simplistic view of the operations and motivations of hospitals (and of the still relatively independent physicians staffing those hospitals) informs his position on health policy, it stands to reason that serious blinders would prevent him from seeing any evidence of harm in a particular policy that promotes the same view.

But that’s not all.  Krumholz’s group at Yale received grants from CMS under the auspices of the Measure and Instrument Development and Support (MIDS) program to study and produce the metrics and instruments needed to devise the readmission measures.

The MIDS program supports the “development and use of clinical quality measures which remains a critical healthcare priority and the tool of choice for improving quality of care at the national, community and facility levels” and it allocates $1.6 billion dollars to this purpose.

Thanks to a bipartisan act of Congress, a helpful little website, usaspending.gov, provides contract level detail about payments made to the Krumholz’s group from the MIDS program.  Those payments can be seen in the table below:

The numbers are staggering. I know little about how to interpret these data about federal contracts, but it sure appears that the Yale-New Haven Health Service group led by Krumholz has received $144 million dollars since 2008.

Yet the only clue to these payments in Krumholz’ published analyses of the HRRP program comes in one disclosure sentence in a footnote, as seen here:

It seems to me that the disclosure is hardly proportional to the amount of funding that his group receives and understates the inherent pressures it must be under to demonstrate that the policy did not actually result in higher mortality.

And recall that MEDPAC’s “independent” analysis that also rejected a policy-mortality link came from the organization that recommended the policy to begin with.  The bottom line is this: There’s a tremendous amount of face to lose and a massive source of institutional funding at risk if the policy is found to be harmful.

It now becomes more clear why, in the following tweets, Dr. Krumholz feels that only he can say anything definitive about readmission rates and mortality:

Um, @JAMA_current, even the authors say they cannot say their findings are causal… "but whether this finding is a result of the policy requires further research.” Why do you promote the paper as proving harm? Need to treat twitter like you do any of your Editorial comments. pic.twitter.com/xugasHvPWY

— Harlan Krumholz (@hmkyale) December 26, 2018

6/Hospitals that have improved their readmission rates tended to improve their mortality rates. Published in @JAMA_current… https://t.co/R7FXLwTMPp

— Harlan Krumholz (@hmkyale) December 22, 2018

A diversity of biases

Biases are ubiquitous.  When I was a cardiologist-in-training, spending hours on the hospital consult service for a fixed salary, I vividly recall looking for ways to avoid doing any work I considered unimportant or banal: The minor cardiac enzyme leak in a patient with a widespread infection; The extra heartbeats on the ECG that the ER physician didn’t like the look of; etc.  “Are you sure you need an official consult?”  “The chances we’re going to recommend doing anything about a small enzyme leak in an 80-year-old with a severe lung infection are very low… “ I was even successful sometimes.

Contrast these comments to my demeanor in private practice where I am acutely aware that my income relies on such consults: “I just need the patients name or room number…” “I’ll take care of it!…” “I can put in the orders if you’d like!”

But there are other biases and incentives that motivate human beings, apart from personal financial incentives. Do they pale relative to the financial ones as is so often claimed? How does one begin to quantify them?

When it comes to the HRRP policy, no individual person’s bank account ballooned every time a patient didn’t get admitted. And yet this is a story of ideological bias that drove the design of policy and now claims ‘success’ for its own program.  The HRRP saga is illustrative of the importance of non-financial bias and of the dangers of blinding ourselves to that bias.

The story also highlights the downsides of tweaking healthcare systems that were built to deliver more care.

Clearly, I personally have a direct financial conflict of interest to provide more care. Since I haven’t talked anyone out of a consult in 8 years, I’m probably guilty of participating in a system that detractors appropriately criticize for promoting overuse of healthcare.

But the problem is that some of those consults I was trying to avoid as a fellow ended up really needing a cardiologist. There was the 55-year-old Cambodian woman admitted to the medical intensive care unit with pneumonia who went into atrial fibrillation. I recall rolling my eyes and thinking that the ICU could certainly handle this without a cardiologist. It turned out she didn’t have a pneumonia. It was pulmonary edema from heart failure related to undiagnosed rheumatic mitral valve disease. She had been in the wrong unit. She needed diuresis, heart rate control, and eventual surgery to replace her valve, not antibiotics. Less isn’t always more.

Attempt to reduce inappropriate hospital admissions? Get ready to pay a price.  To contradict Dr. Krumholz, it is entirely probable that we are underestimating the upside of our current system when we contemplate changing the status quo.

We are underestimating the downside of our current system when we contemplate change. We need to take some risks to do better. #abimf2013

— Harlan Krumholz (@hmkyale) August 5, 2013

Final Thoughts

Ironically, the HRRP quagmire offers a number of clarifying lessons.

Empiricism in social policy is a subjective enterprise. The often parroted conclusion is that cold, hard, unbiased evidence trumps the biased, unmeasurable judgment of clinicians. Yet, frequently, real world data-sets are complex, the choice of analytic paths can be highly variable, and the instruments to measure success are often imperfect. As it relates to HRRP, which analysis should we trust? The choice requires faith. And if the currency here is faith, perhaps the concern of clinicians at the bedside has more value than advertised.

Metrics won’t save us. The narrative of metrics is an appealing one that promises hard and objective accountability. The problem comes when the metric (readmission) becomes disconnected from outcomes that actually matter (death). False and blind prophets are good descriptors for those who claim to be unable to see without metrics. The fools in this enterprise are easy to identify as those who think the answers lie with ever better metrics.

Conflicts of interest: Going beyond the simple narrative.  Focusing on biases induced by personal financial interests is a mistake. Personal enrichment is just one bias in a sea of conflicts. In the healthcare context, financial disclosures—while clarifying in themselves—may simply give cover to other, more perverse biases, unless those other biases are equally disclosed.  It requires diligence to ascertain the impact, and direction of bias.  Rarely do we get the opportunity to observe the direction of bias in policy research.  In the case of the HRRP, the presence of bias was made evident because research groups with opposing biases (clinician-scientists versus policy wonks) have reached conclusions that would be expected on the basis of those pre-existing biases.  How often is the problem of such bias examined in the design, implementation, and analysis of health policy?

Beware of technocrats with all the answers.  I am reasonably sure that if practicing clinicians would have been asked to devise a rule to reduce heart failure readmissions for the whole population they would have refused. It seems too challenging a task to get right.  Even if clinicians can be induced to participate in the design of such a policy, most would readily acknowledge the likelihood that it could harm some patients. It requires a special type of hubris to design a policy and refuse to acknowledge its potential for harm. Unfortunately, hubris within a public health community that believes only they can give us a better health system is more feature than bug.

As for Mrs. C, she has been home for 16 days. All fingers on both hands are currently crossed.

Anish Koka is a cardiologist in private practice in Philadelphia.  He can be followed on Twitter @anish_koka. This post originally appeared here on The Accad & Koka Report. 

Commissioning Healthcare Policy: Hospital Readmission and Its Price Tag published first on https://wittooth.tumblr.com/

[kristinsimmons]

Commissioning Healthcare Policy: Hospital Readmission and Its Price Tag

By ANISH KOKA MD 

The message comes in over the office slack line at 1:05 pm. There are four patients in rooms, one new, 3 patients in the waiting room. Really, not an ideal time to deal with this particular message.

“Kathy the home care nurse for Mrs. C called and said her weight yesterday was 185, today it is 194, she has +4 pitting edema, heart rate 120, BP 140/70 standing, 120/64 sitting”

I know Mrs. C well. She has severe COPD from smoking for 45 of the last 55 years. Every breath looks like an effort because it is. The worst part of it all is that Mrs. C just returned home from the hospital just days ago.

The youngest of six children, Mrs. C was born with many embedded disadvantages. Being born black in a poor West Philadelphia neighborhood in the 1960s is a story that too often writes itself with a bad ending.  But Mrs. C avoided the usual pitfalls that derail young women in the neighborhood early. No drugs. No alcohol. No teenage pregnancies. Finished high school. Mrs. C. worked for the hospital as a unit clerk, had her own place, health benefits, and even a retirement plan.

Certain life habits, however, carry a heavy price. George Burns, the comedian never pictured without a cigar who died past his hundredth birthday, may have been immune to the effects of tobacco.  Mrs. C was not. She started smoking when she was 16. She doesn’t recall why. Her dad smoking didn’t help perhaps. Nausea racked her body after that first drag. It eased up after. Too bad.

That measly cigarette became the great addiction of her life. Day by day, the exquisitely thin membranes of the lungs that mediate gas exchange were destroyed. By the time the disease manifests with shortness of breath and bluish tinged lips, it’s too late. Short of the very few who qualify for a lung transplant, the efforts of doctors at this point are for mitigation rather than cure.

Complicating things further, in Mrs. C’s case, the normally low pressure vascular circuit of her lungs became a high pressure circuit that places ever increasing demands on the normally thin-muscled right ventricle of the heart.  This jeopardizes the ability of her heart to handle changes in blood volume.

A little extra fluid and the right side of the heart ends up causing unbearable swelling in her legs.  A little dehydration and severe disabling dizziness on standing ensues. Adding to that that her tenuous lung function decompensates with the slightest respiratory infection, that chronic steroid treatment to decrease her wheezing suppresses her immune system, and that the young man down the street helpfully drops off Newports at her home for a few extra dollars, and it’s easy to see why the hospital is her second home.

The most recent admission to the hospital was for kidney failure related to taking too much fluid off with diuretics. What was to be a short stay for gentle hydration turned into a longer stay when a pneumonia complicated the matters (though a trip to the intensive care unit and a ventilator was barely but fortunately avoided).  She was treated by the pulmonology team and sent home on a lower dose of diuretics.

The situation I am now confronting puts me in a quandary.  Her edema and weight are up markedly just a few days after returning home.  Could her fluid overload be because her kidneys are shutting down? Or does she just need more aggressive diuresis?

Should I guess? Knowing her present renal function would be helpful. But even if the Theranos lab I could appeal to for help wasn’t fictional, I would have to get her to my office everyday or every other day while adjusting her diuretic dose.

And so it comes to be that, days removed from a hospital admission, I’m sending her back to the hospital to be readmitted.  According to some, this is not supposed to happen.

A policy on readmission

In 2008, the commission that advises Medicare – the Medicare Payment Advisory Commission (MEDPAC), issued a report that focused on hospital readmissions.

The focus on hospital readmissions had been of great interest to the health policy community for some time. At the core of this interest lies the belief that hospitals and physicians are incentivized to treat patients rather than prevent admissions.

The MEDPAC report wanted to discourage readmissions like Mrs. C’s.  And so it wasn’t a terrible surprise that, rolled into the 2010 Affordable Care Act, was a section called the Hospital Readmission Reduction Program (HRRP) which created a system for Medicare to penalize hospitals with ‘high’ readmission rates.  The program was rolled out in 2013.

At first, the program seemed to work like a charm.  Hospitals significantly ramped up their efforts at care coordination.  Teams of nurses and aids were assembled to make sure patients would get their medications as prescribed upon discharge and to check on patients once they got home.

Hospital readmission rates suddenly dropped and Medicare started saving money.  A staggering 81% of all hospitals suffered penalties in 2018, which translates to ~$500 million or 0.3% of total Medicare payments to hospitals.

A complex analysis

But there’s more to this too good to be true story.

The HRRP penalty schemes are risk-adjusted based on administrative claims data. Risk-adjustment is a statistical procedure to take into account the diversity in complexity and severity of disease among patients so they can be compared.

Physicians know that risk-adjusted claims data are of dubious value because they themselves are often the reluctant data entry clerk in the byzantine scheme that starts with adding diagnostic items to the medical chart and ends with generating a coded billing claim for Medicare.  Needless to say, there’s a huge potential disconnect between what a claim attempts to convey and the actual condition of a given patient.

Yet another major problem is that the risk-adjustment employed by the HRRP does not take socioeconomic status into account, when that is arguably the single biggest driver of poor outcomes and of hospital readmissions.  The creators of the HRRP program seem to believe that a hospital located in poor area shouldn’t get a break for having high readmission rates, perhaps because they believe that hospital systems in general should be mindful of health inequities and address “care gaps” (differences in care provided to poor vs. affluent patients)  in their neighborhoods no matter what.

Finally, the initial out of the gate benchmark for readmission rates on which the HRRP would adjudicate the need for a penalty was a national average.  In such a scheme, a Johns Hopkins Hospital serving inner city Baltimore could be pitted against a regional hospital in rural Montana with an entirely different patient demographics.  This made the regional hospital in Montana very happy.

Gaming the metrics

Regardless of these technical considerations, it is an adage of social science that any metric will be gamed, and healthcare is unfortunately not immune to that law.

One tool increasingly used by hospitals to comply with Medicare payments rules is to admit patients to short stay units, under so-called “observation status.”  Another is to put pressure on emergency departments to avoid readmitting certain types of patients.

So, instead of primarily functioning as a triage operation where sick patients would be turned over to the care of the cardiologist in the hospital, the ER has been increasingly housing and managing heart failure patients to save the hospital money.

But the ER physician or the hospitalist supervising the short stay unit and who just meet a patient in the setting of an acute illness are poorly equipped to know which heart failure patient to discharge after a diuretic dose and which to keep for advanced heart failure therapies.

Source: MedPAC’s June 2018 Report (http://bit.ly/2CXD0fJ)

Nowadays, the cardiologist is increasingly insulated from those decisions.  I have personally experienced with alarming frequency instances where I learn only after the fact that a complex patient of mine has been treated for heart failure in the ED.

And my experience seems to be shared by many of my cardiology colleagues, especially among cardiologists who work in academic centers that are most affected by the policy.  Luckily, some of them are also clinician-scientists that can do more than just whine to colleagues about the new policy.  They can also study its outcomes.

What do the outcomes data show?

In a pivotal study, a group of cardiologists (Gupta et al.) saw that the drop in readmissions that followed the introduction of HRRP was unfortunately accompanied by a reversal in the decade long downward trend in heart failure mortality. This reversal suggested a serious potential harm from the policy.

But the possibility of harm was quickly challenged by another group of researchers led by one of the biggest names in health policy: Harlan Krumholz, a cardiologist who directs the influential Center for Outcomes Research and Evaluation at Yale University.

The Krumholz group analyzed millions of Medicare claims data from 2007 to 2016.  They also found that mortality for heart failure patients increased, but concluded that no causal link between decreasing readmission and increased mortality could be established.

Krumholz at al. noted that mortality rates for heart failure started climbing before the HRRP program was announced and they noted no inflection point in mortality rates with the policy announcement in 2010.  The evidence for their claim is highlighted in the table below:

As can be seen in the boxed row, Krumholz’ team concludes that the increasing mortality slope post-HRRP is no different from that pre-HRRP because the change did not reach statistical significance at the obligatory and arbitrary P<0.05 level.  The actual P-value was 0.11 and the confidence interval for the positive increase in mortality slope of 0.006 is (-.002 to .015).  Even poor students of epistemology would be loathe to conclude this result excludes a signal of harm.  It seems entirely plausible that, with all the limitations of the data set in question, mortality may in fact have accelerated after the institution of the HRRP.  Yet Krumholz insists that no signal of harm is to be considered.

But this did not stop another group of cardiologists (Wadhera et al.) from adding their contribution to the HRRP literature. Using the same data-set that the Krumholz group used — Medicare claims data — these researchers found once again that accelerating mortality coincided with the announcement of the HRRP.  More troubling, they also demonstrated that mortality rose primarily among patients not readmitted to the hospital.

A messy science

Admittedly this whole business of analysis is incredibly messy, with a number of moving parts.

My brief summary doesn’t do justice to a variety of maneuvers taken by the various groups to account for many of the limitations inherent in this type of study. Two of the competing analyses (Krumholz, Wadhera) used Medicare claims data while the other (Gupta) used a more limited voluntary registry.

During the time period in question, there were also other policy changes such as the introduction of new hospital billing codes (MS-DRG) that sought to adjust hospital payment rates to patient complexity. Better patient coding meant higher reimbursement from Medicare.  Armies of “documenters” were then employed by hospitals to capture more revenue.

This means that the claims data gathered by the researchers might look significantly different from one time period to the next even if the patients themselves were ostensibly the same. As the readmission rate is risk-adjusted, it is eminently plausible and likely that systematically upcoding patient risk could actually have been the primary driver of the drop in hospital readmission rate.

The other program playing a confounding role is the Recovery Audit Contractor (RAC) program begun in 2010 to reduce payments for inappropriate hospital admissions. Hospitals responded to denials for inpatient admissions by expanding ‘observation status’ stays. Which was the biggest driver for expanded observation stays? RAC or the HRRP? Once again an exact attribution is impossible.

Denying the obvious?

Despite the messiness of the data and the variety of analytic methods used, a consistent and uncontested observation remains: Heart failure mortality has increased in the last decade. The question being hotly contested is Why?

Oddly, Dr. Krumholz is steadfast in denying the possibility that the policy may have caused harm, even though the independent and contradictory conclusions from the other research groups at least raises a reasonable doubt.  And Dr. Krumholz has been quick to cast shade on research that does not conform to his conclusion.

By tweet he appears to ask for a level of detail his own papers lack, and he questions the legitimacy of another group’s data-set, all the while resisting any calls to put the program on hold despite the paucity of evidence showing benefit, the signal for harm, and perhaps most importantly, the concern of clinicians who see a mechanism for harm.

Greatly admire @rwyeh and his group…appreciate his focus on readmission &public policy. For such high-profile article, really need more info about statistical weighting. Methods should be sufficient so others can reproduce results. Can’t do that here. Look forward to more info.

— Harlan Krumholz (@hmkyale) December 21, 2018

Can you account for why the registry you used had, on average, such a small number of patients per site. Did you determine how many patients coded with heart failure by CMS were in the Registry? Just curious about the selection. It may not explain your results…but is a question. https://t.co/WSvJAN8Jjd

— Harlan Krumholz (@hmkyale) June 2, 2018

Dr. Krumholz also places much weight on an independent analysis carried out by MEDPAC which concluded there was no link between policy and the uptick in mortality. This particular conclusion rests heavily on the assertion that heart failure patients in 2016 were much sicker than patients in 2010. Recall that this coincides with a period of more intensive coding over the same time frame, so it is impossible to say this with any confidence.

The MEDPAC conclusion also relies on an analysis that finds no correlation between hospital level readmission and mortality rates.

While technically true, that conclusion overlooks that a large number of hospitals exhibited reduced readmission rates and increased mortality. Perhaps MEDPAC feels that patients dying at low readmit/high mortality hospitals should be mollified by the knowledge that somewhere there’s a low-readmit/low mortality hospital to balance things out?

That Krumholz and MEDPAC display such certainty about the direction of the signal they observe, and take pains to discount other possibilities seems strange and suggests that pre-existing biases may be at work. What might those biases be?

Conflicts of interest: You get what you pay for

In a world where heads roll for undisclosed personal financial conflicts of interest, it is remarkable that the current dispute, while full of scintillating exchanges about “propensity weighting” and other arcane points of statistics, does not reference any other potential conflicts at work that might affect the conclusions being reached.

Medicare’s decision to start the HRRP program didn’t come in a vacuum. It was inspired by years of research from Dr. Krumholz himself, who suggested that preventing admissions should be a goal for any policy that would aim to move the system from one paying for “volume” to one paying for “value.”

As far back as 2003, Krumholz held the view that:

“Hospitals may not support programs that improve the quality of care delivered to heart failure patients because these programs lower readmission rates and empty beds, and therefore further diminish already-declining revenues.”

If Krumholz’s unfavorable and crudely simplistic view of the operations and motivations of hospitals (and of the still relatively independent physicians staffing those hospitals) informs his position on health policy, it stands to reason that serious blinders would prevent him from seeing any evidence of harm in a particular policy that promotes the same view.

But that’s not all.  Krumholz’s group at Yale received grants from CMS under the auspices of the Measure and Instrument Development and Support (MIDS) program to study and produce the metrics and instruments needed to devise the readmission measures.

The MIDS program supports the “development and use of clinical quality measures which remains a critical healthcare priority and the tool of choice for improving quality of care at the national, community and facility levels” and it allocates $1.6 billion dollars to this purpose.

Thanks to a bipartisan act of Congress, a helpful little website, usaspending.gov, provides contract level detail about payments made to the Krumholz’s group from the MIDS program.  Those payments can be seen in the table below:

The numbers are staggering. I know little about how to interpret these data about federal contracts, but it sure appears that the Yale-New Haven Health Service group led by Krumholz has received $144 million dollars since 2008.

Yet the only clue to these payments in Krumholz’ published analyses of the HRRP program comes in one disclosure sentence in a footnote, as seen here:

It seems to me that the disclosure is hardly proportional to the amount of funding that his group receives and understates the inherent pressures it must be under to demonstrate that the policy did not actually result in higher mortality.

And recall that MEDPAC’s “independent” analysis that also rejected a policy-mortality link came from the organization that recommended the policy to begin with.  The bottom line is this: There’s a tremendous amount of face to lose and a massive source of institutional funding at risk if the policy is found to be harmful.

It now becomes more clear why, in the following tweets, Dr. Krumholz feels that only he can say anything definitive about readmission rates and mortality:

Um, @JAMA_current, even the authors say they cannot say their findings are causal… "but whether this finding is a result of the policy requires further research.” Why do you promote the paper as proving harm? Need to treat twitter like you do any of your Editorial comments. pic.twitter.com/xugasHvPWY

— Harlan Krumholz (@hmkyale) December 26, 2018

6/Hospitals that have improved their readmission rates tended to improve their mortality rates. Published in @JAMA_current… https://t.co/R7FXLwTMPp

— Harlan Krumholz (@hmkyale) December 22, 2018

A diversity of biases

Biases are ubiquitous.  When I was a cardiologist-in-training, spending hours on the hospital consult service for a fixed salary, I vividly recall looking for ways to avoid doing any work I considered unimportant or banal: The minor cardiac enzyme leak in a patient with a widespread infection; The extra heartbeats on the ECG that the ER physician didn’t like the look of; etc.  “Are you sure you need an official consult?”  “The chances we’re going to recommend doing anything about a small enzyme leak in an 80-year-old with a severe lung infection are very low… “ I was even successful sometimes.

Contrast these comments to my demeanor in private practice where I am acutely aware that my income relies on such consults: “I just need the patients name or room number…” “I’ll take care of it!…” “I can put in the orders if you’d like!”

But there are other biases and incentives that motivate human beings, apart from personal financial incentives. Do they pale relative to the financial ones as is so often claimed? How does one begin to quantify them?

When it comes to the HRRP policy, no individual person’s bank account ballooned every time a patient didn’t get admitted. And yet this is a story of ideological bias that drove the design of policy and now claims ‘success’ for its own program.  The HRRP saga is illustrative of the importance of non-financial bias and of the dangers of blinding ourselves to that bias.

The story also highlights the downsides of tweaking healthcare systems that were built to deliver more care.

Clearly, I personally have a direct financial conflict of interest to provide more care. Since I haven’t talked anyone out of a consult in 8 years, I’m probably guilty of participating in a system that detractors appropriately criticize for promoting overuse of healthcare.

But the problem is that some of those consults I was trying to avoid as a fellow ended up really needing a cardiologist. There was the 55-year-old Cambodian woman admitted to the medical intensive care unit with pneumonia who went into atrial fibrillation. I recall rolling my eyes and thinking that the ICU could certainly handle this without a cardiologist. It turned out she didn’t have a pneumonia. It was pulmonary edema from heart failure related to undiagnosed rheumatic mitral valve disease. She had been in the wrong unit. She needed diuresis, heart rate control, and eventual surgery to replace her valve, not antibiotics. Less isn’t always more.

Attempt to reduce inappropriate hospital admissions? Get ready to pay a price.  To contradict Dr. Krumholz, it is entirely probable that we are underestimating the upside of our current system when we contemplate changing the status quo.

We are underestimating the downside of our current system when we contemplate change. We need to take some risks to do better. #abimf2013

— Harlan Krumholz (@hmkyale) August 5, 2013

Final Thoughts

Ironically, the HRRP quagmire offers a number of clarifying lessons.

Empiricism in social policy is a subjective enterprise. The often parroted conclusion is that cold, hard, unbiased evidence trumps the biased, unmeasurable judgment of clinicians. Yet, frequently, real world data-sets are complex, the choice of analytic paths can be highly variable, and the instruments to measure success are often imperfect. As it relates to HRRP, which analysis should we trust? The choice requires faith. And if the currency here is faith, perhaps the concern of clinicians at the bedside has more value than advertised.

Metrics won’t save us. The narrative of metrics is an appealing one that promises hard and objective accountability. The problem comes when the metric (readmission) becomes disconnected from outcomes that actually matter (death). False and blind prophets are good descriptors for those who claim to be unable to see without metrics. The fools in this enterprise are easy to identify as those who think the answers lie with ever better metrics.

Conflicts of interest: Going beyond the simple narrative.  Focusing on biases induced by personal financial interests is a mistake. Personal enrichment is just one bias in a sea of conflicts. In the healthcare context, financial disclosures—while clarifying in themselves—may simply give cover to other, more perverse biases, unless those other biases are equally disclosed.  It requires diligence to ascertain the impact, and direction of bias.  Rarely do we get the opportunity to observe the direction of bias in policy research.  In the case of the HRRP, the presence of bias was made evident because research groups with opposing biases (clinician-scientists versus policy wonks) have reached conclusions that would be expected on the basis of those pre-existing biases.  How often is the problem of such bias examined in the design, implementation, and analysis of health policy?

Beware of technocrats with all the answers.  I am reasonably sure that if practicing clinicians would have been asked to devise a rule to reduce heart failure readmissions for the whole population they would have refused. It seems too challenging a task to get right.  Even if clinicians can be induced to participate in the design of such a policy, most would readily acknowledge the likelihood that it could harm some patients. It requires a special type of hubris to design a policy and refuse to acknowledge its potential for harm. Unfortunately, hubris within a public health community that believes only they can give us a better health system is more feature than bug.

As for Mrs. C, she has been home for 16 days. All fingers on both hands are currently crossed.

Anish Koka is a cardiologist in private practice in Philadelphia.  He can be followed on Twitter @anish_koka. This post originally appeared here on The Accad & Koka Report. 

Commissioning Healthcare Policy: Hospital Readmission and Its Price Tag published first on https://wittooth.tumblr.com/