Department of Unintended Consequences, American health care edition

In 2010, the federal agency that oversees Medicare, the Centers for Medicare and Medicaid Services, established the Hospital Readmissions Reduction Program under the Affordable Care Act. Two years later, the program began imposing financial penalties on hospitals with high rates of readmission within 30 days of a hospitalization for pneumonia, heart attack or heart failure, a chronic condition in which the heart has difficulty pumping blood to the body.

At first, the reduction program seemed like the win-win that policymakers had hoped for. Readmission rates declined nationwide for target conditions. Medicare saved an estimated $10 billion because of the reduction in hospital admissions. Based on those results, many policymakers have called for expanding the program.

But a deeper look at the Hospital Readmissions Reduction Program reveals a few troubling trends. First, since the policy has been in place, patients returning to a hospital are more likely to be cared for in emergency rooms and observation units. This has raised concern that some hospitals may be avoiding readmissions, even for patients who would benefit most from inpatient care.

Second, safety-net hospitals with limited resources have been disproportionately punished by the program because they tend to care for more low-income patients who are at much higher risk of readmission. Financially penalizing these resource-poor hospitals may impede their ability to deliver good care.

Finally, and most concerning, there is growing evidence that while readmission rates are falling, death rates may be rising.

In a new study of approximately eight million Medicare patients hospitalized between 2005 and 2015 that we conducted with other colleagues, we found that the Hospital Readmissions Reduction Program was associated with an increase in deaths within 30 days of discharge among patients hospitalized for heart failure or pneumonia, though not for a heart attack.

That is by Rishi K. WadheraKaren E. Joynt Maddox and Robert W. Yeh in The New York Times.

Comments

awful "study." just an interrupted time series with no real controls, no basis for a causal statement. but that won't stop you from lapping it up, will it, Tyler?

They measure and control for a number of comorbidities and patient characteristics. An interrupted time series is the correct design for evaluating a change of this nature, and their finding (an increase, concentrated within the group one would expect if it were the result of the policy) seems robust. Furthermore, a different study using a different design had a similar finding.

There is limited ability to infer causality from such studies (and they are careful to not make a causal claim). However, it is a policy that was implemented nationally on a particular date. There is no comparison group, and no realistic possibility of doing a controlled trial. To reject this study as useless is basically to say that nobody should even try to evaluate the impact of the policy.

The one actor that would have the power to do better is the U.S. government. They should be doing robust, randomized evaluations before rolling out policies like this. This is the concluding point of the editorial.

The high quality controlled trial should have been done by the people who designed the policy. It's not possible to do one now.

Good point.

Note this program was designed to implement the economic theory of economists who argue punishment and reward are the best method of producing excellence.

In contrast to medical provviders calling for giving them as much money as they want. Which is the model car repair shops use for out of warranty repairs.

Should costs be controled for hospital care as for cars: creative destruction, ie, kill it, chop it up, sell the parts for reuse or scrap grinding or landfill.

'This has raised concern that some hospitals may be avoiding readmissions, even for patients who would benefit most from inpatient care.'

Any loyal reader of MR knows that something like inpatient care has no effect on anything, except to generate costs.

'there is growing evidence that while readmission rates are falling, death rates may be rising'

Any loyal reader of MR knows that this cannot be true, since medical care has no apparent effect on death rates.

Access to affordable medical care doesn't seem to have measurable health benefits but anyone can go to a hospital and get treated so there's a certain baseline there

since medical care has no apparent effect on death rates.

It doesn't. Death rates for humans are 100% independent of all medical care, even in Germany. People still stubbornly persist on dying around age 80.

The pertinent questions are life expectancy, and quality of life. Then the next questions are how we pay for it, and when is it no longer worth it. This study isn't really asking those questions.

'People still stubbornly persist on dying'

At any age, actually. Strangely, though, the number of people who died of smallpox declined noticeably when vaccinating against it grew routine. The same is true of TB.

The word 'concerning' is a preposition, not an adjective. Once upon a time the NYT had copy editors who cared about things like that.

Isn’t it a gerund or gerundive, and able to be used as both a preposition and an adjective? But then I am no copy editor.
By the way, “Misrahi” is a great name.

Webster's dictionary does not agree with you.

Concerning is a verb most often, an adjective second and a preposition only in certain situations in formal English.

Please give me an example of concerning used as a verb

This situation is concerning me.

This situation concerns me.

"Concerning" is a gerundive.

'Once upon a time the NYT had copy editors who cared about things like that.'

They still do. Their readers, however ....

'concerning adjective
Definition of concerning (Entry 2 of 2):
causing concern or worry : creating reason for concern : troubling

// found the latest reports very concerning' https://www.merriam-webster.com/dictionary/concerning

Making a policy change like this without robustly evaluating it first is analogous to giving patients an untested, potentially dangerous drug. It should be scandalous that the government acts in this manner.

They had to pass the ACA to find out what was in it, though.

What is the value of a study with a median age of 79.6 years? Twenty-eight Right-To-Work states have an average life expectancy of only 78-years.

It's a study of medicare-eligible people admitted to hospital with major health conditions that are increasingly prevalent with age. I see no reason to doubt that they have a valid sample for studying this population.

Though he may have buried his point, the fact that 28 states (apparently - seems about right from the last time I bothered to check) fall below the actual study median in terms of average life expectancy just might mean that the real focus of government policy might be better placed somewhere else, such as increasing the average life expectancy in those states before concentrating on the side-effects of changing hospital admission/readmission reimbursement schedules.

Making sure a policy is not causing needless deaths seems relevant to life expectancy.

The comparison of life expectancy at birth to the average age of the cohort is meaningless, since the life expectancy of those who have survived to age 65 is considerably higher.

Dan, a review of Right-To-Work policy impact on life expectancy perhaps has merit. Life expectancy in several R-T-W states falls five years short of that in California.

Sure, that would be a good thing to evaluate.

But why the attempt to change the subject? Surely you don't believe that attempting to evaluate the efficacy of government healthcare programs is a bad thing?

Dan, for men now age 65 in Right-To-Work states, should a study cover a median age of 70 or 90? Nationally, 90% of 65 year old men today will live to age 70; ten percent will live to age 90.

Right-to-work States have lower barriers to entry for employment, so blacks and latinos move in, and they have lower life expectancies. That's probably what those stats are capturing.

I find it hilarious that this particular axe is still being ground, as if globalism, migration, and technology never happened. Outside of government (where collective bargaining shouldn't even be allowed) and cartelized guilds like Hollywood and Broadway, classic labor movements are doomed. Over 2 billion Southeast Asians entered the global labor force starting in the 1970's, and there will be more from South America and Africa, and after them are the robots.

Labor activists are about as relevant and as quaint as the Woman's Christian Temperance Union.

Don't Hispanics have pretty good life expectancies relative to the US average?

Apparently they do. My mistake.

White and black males draw the average down.

TAG, it doesn’t look good for Right-To-Work states with the five highest percentages of whites. People in non-R-T-W states with the five largest Hispanic populations and non-R-T-W states with the five largest black populations live about a year longer.

Lots of confounding factors.

How long do you think closed-shop can compete with globalism and, on the heels of that, automation/mechanization?

Shouldn’t we also consider whether *as a matter of policy* it is desirable to promote longevity? At least beyond a certain age. Extending lifespans costs money from the common fisc: Social Security, Medicare (or Medicaid), postponement of death taxes, etc. Many years ago I read an article by a doctor who questioned the advisability of creating a specialty in geriatrics for just those reasons. Not to mention the disruption in the normal succession of generations. We now have the elderly taking care of even more elderly parents.

How long ago was this pro-death policy considered? Was it in Germany in the 1930s?

Anyway it is economically unsound. Education lasts 20-30 years, after which there is 30 to 40 years gainful employment. If this can be extended, then there is better return on that expenditure.
Some scientists are considering ageing as a curable disease, and if that happens people will live on average 600 years statistically before an accident.

Not doing it (assuming it is possible) is surely mass murder by neglect.

Life expectancy from birth (which is what you quoted) and life expectancy from whatever age you've survived to thus far are very different things. If you've made it through age 65 (for example), chances are pretty good you will make it several years past the age of life expectancy from birth.

This just goes to show that rigid rules and "lawyerism" fail to deliver good service for the consumer. Employing people who genuinely want to help other people and giving them enough latitude to do just that, is what is needed. After all in the medical profession they have to be highly qualified in order to work, so why not relay on their good judgement a bit more?

What's needed is a good post-discharge outpatient monitoring system by the hospital. Unfortunately, few hospitals have one, and those that do, the system is primarily intended to avoid patient readmission. This is one of those instances where remote patient monitoring would be an ideal solution. The problem with remote patient monitoring is that the amount of reimbursement is so small that hospitals have very little incentive to implement such a system (Medicare adopted reimbursement for remote patient monitoring only this past year and in a nominal amount). Those that have tried a remote patient monitoring system bear much of the cost, which they weigh against the penalty for readmission. Lose, lose, not exactly a good incentive for hospitals to adopt an efficient remote patient monitoring system. In other words, it's not the penalty for readmissions that is the culprit, but the failure to complement it with a post-discharge patient monitoring system (such as remote patient monitoring).

there is to a remote post discharge outpatient monitoring
system called the family doctor but the sociology dept. has
has also fubared the family doctor like they fubared medicare readmissions
most readmissions gonna be due to severity of multisystem disease

@John138:

... in America, government bureaucrats fund/heavily-regulate most of the health care system. 'Rigid Rules' & bureaucratic 'Policies' are how things operate in a socialist economic model of health care.
Most Americans are very satisfied with the costs & quality & availability of our quasi-socialist medical system (/s) .

The alternative would be a horrible voluntary/free-market healthcare economic system. Any system lacking expert, selfless government bureaucrats to dictate Rules & Policies is doomed to failure -- America's streets would be littered with the dead and dying.

Your elected officials have delivered the best possible health care system. Keep voting for these brilliant politicians.

How can there not be more to the policy then the single sentence description. I'm not saying it's better or worse but the obvious potential for unintended consequence must have made the people running it take some care in that regard. Right?

It is like I always say, there is not actually that much slack in healthcare costs. The majority of care provided is already overseen by large insurance companies. If there were some easy incentive structure by which they could lower their costs they would do it (unless prohibited by law). And the hospitals would, mostly, not fight it either. After all if an insurer and a hospital can spot a $5 million cost savings they can just split the windfall. Everything not paid is profit to the insurer. For the hospital, their operating margin is something around 3%. If the money is split 4:1 either way, both sides come out massively ahead of the status quo.

If two sides, with all the real data, and massive incentives to find slack for money cannot do so, then I suspect there is not much money to be found at all.

As always, healthcare suffers from the brain bug that somehow it "should" be cheaper. Cheaper than the past - back when we did not employ things like social workers, school teachers, community outreach officers in hospitals and when the population was not obese, had not smoked as many pack-years, and was not socially isolated and cheaper than other countries - which abort away far more of the high cost users, have better general health metrics (and have for over 100 years), and are not facing labor cost disease from things like tech and Wall St.

So we keep getting treated with idiotic policies that kill people. Some idiot thinks it would be good to pay surgeons to have fewer complications (by penalizing them when they have them). Great, now after every surgery immunocompromised patients get hit with vanc and some carbapenem. The combo works great at preventing infectious complications of surgery, it just makes it vastly more likely that the patients will develop fungal infections outside of the measurement window. Oh and it increases the odds that some pan-resistant superbug will kill millions.

Simple solutions basically don't exist. Where they do, well, the real things we could do to reduce cost are exactly the stuff that we are legally banned from doing: punishing patients for non-compliance, firing non-medical personnel who please our political masters, buying off the shelf goods and equipment, shunting non-urgent patients directly to non-urgent care systems, not have translators for every major interaction, charging more to patients whose health status makes them more expensive ... for a variety of reasons society wants healthcare to rectify problems that stem from poverty, drug use, gluttony, loneliness, broken families, and possibly even more remote things like racism, sexism, and immigration policy. Okay fine, but then don't act shocked when health care costs more compared to either the past or our peer countries where it doesn't cover as many of societies ills.

^^^ You got it.

Yep.

AMA cartel

Oh please. Physician salaries are in total 20% of healthcare expenditures. The relative amount is falling per CMS data.

If we enslaved every physician and had them work for free ... healthcare would cost the same in ~4 years.

Further as far as cartels go ... please the physician workforce is already 25% foreign trained.

Do not get me wrong, the AMA is a pox on society. But they are not now, never have been, nor ever will be able to control the price of healthcare. Their single biggest impact is lobbying for various (dumb) rules.

There was a big shakeup at hospitals to reduce costs, driven by insurers and the government. Tasks were pushed down to as low a level as possible. What might have been done by a nurse was pushed to a tech, done by a tech to a high school grad with some training, etc.

The price system in medical care is lunatic. It's a reasonable wager that generates a certain amount of deadweight loss.

Not really. Imagine the healthcare pricing systems as a stereotypical Arabian bazaar. The hospital list price is an initial offer. The insurance companies haggle and we hash out a price that is actually pretty close across the board (some insurers eat all their costs on say heart surgery, others eat it on ICU visits, but over all patients it tends to even out). Most of these negotiations end up with "2 guys" in a room offering blandisments until a deal is cut that looks remarkedly similar to other such deals. Haggling is one of the oldest and most reliable price discernment processes in history.

What really distorts the price signal, though, are the regulations. Government payment regs are a joke, they are woefully inadequate to account for the differential costs of hiring competent providers. They lag real world practices. But some bureaucrat's diktat often becomes an anchor point that disrupts the pricing haggles.

But shouldn't the same procedures have the same price? Sure, and so should hotel rooms. And airline fares. But ultimately everything faces the fact that demand and supply are both dynamic. A knee replacement costs a heckuvalot more when the orthopods had to come in for trauma and we are burning through maximum allowed work hours. On the other hand, infectious disease consults are in far less demand when school is out for summer.

Would prices work better without third party payers? Absolutely. Is there a politically viable path from there to here? No. Is there a lot of room with the current haggling setup for efficiency gains? Not really. Insurance and hospitals are already pretty good hagglers and government diktat is the by far the biggest impediment to efficient price discovery.

I can't comment on the quality of the study, but I would like to say that treating it as if it were a "one-and-done cure" seems wrong. Surely all interventions in a complex healthcare system must be iterations, part of a continuous process of optimization.

And in any optimization process you don't say "oh no, that last step went in the wrong direction, so let's forget the whole thing."

Optimization never jumps in one whole step to the global maximum.

Maybe in fantasy land where you are dictator of the universe.

Exactly my thought. Healthcare is to complex for a process to work out of the box. Design >> go live >> feedback >> improve. It is the best way, and actually, the idea was implemented in the ACA ( https://www.newyorker.com/magazine/2009/12/14/testing-testing-2 )

This is true, but you cannot optimize what you are not measuring. This policy change (and almost all others in the space) was put in place without any robust evaluation of its impact on outcomes. Therefore, the system is being perpetually perturbed, without heading toward any definite end.

But the point is, nobody had to pay for a second visit. Americans are slowly learning why health care costs are lower in other countries.

I work in this area and have two criticisms of the paper.

First, this question has been examined repeatedly since the program was implemented - maybe even before - and prior studies have not found a mortality effect. So that raises the bar a bit - a single study with limited design and modest findings should not be taken too seriously on its own. Second, the design of the current study has one key limitation, in that they authors assume they can appropriately account for the change in patient mix over time (by using probability weights generated by the most recent time period). But there is a body of evidence that one consequence of the readmission reduction program has been a decline in not just readmissions but also initial admissions - as hospitals identify ways to reduce unnecessary hospitalizations, patients with better health and social support are staying home. This means that by end of this study period, the patients being admitted are generally sicker and/or have less home support than at the beginning, and it is unlikely that the authors can capture and adjust for this completely with the claims data they are using.

In short, this is an important contribution to the existing evidence, but shouldn't be taken as the final word.

2 ways to improve on the situation both include close post-hospitalization monitoring.

ACO's - accountable care organizations, are doing monitoring of discharged Medicare patients with some success. The doc/participants can share in savings. They have staff that monitor the patients either through phone or with direct home visits. As a primary care doc, I have and am participating to some extent with such systems. With meager financial benefit to me.

Another way is to have a hospital based dept. or system to essentially do the same. And I suspect that many do.

In fact as near retiring primary care doc, I will discussing just such a plan with our local hospital in Jan.

Wait, people are surprised that hospitals would manipulate the lever they had direct control over in order to secure their own funding?

Though, definitely agree with the following: "... this is an important contribution to the existing evidence, but shouldn't be taken as the final word."

Comments for this post are closed