• Post-acute care: Medicare Advantage vs. traditional Medicare

    From a public spending point of view, post-acute care is particularly problematic. Most of Medicare’s geographic spending variation is due to this type of care. Part of the story is that Medicare pays for post-acute care in several different ways, with different implications for efficiency.

    For example, traditional Medicare (TM) — which spends ten percent of its total on post-acute care — pays skilled nursing facilities per diem rates but inpatient rehabilitation facilities a single payment per discharge. Post-acute care is also available through Medicare Advantage (MA), which operates under a global, per-enrollee, payment. Unlike TM, MA plans establish networks, may require prior authorization for post-acute care, and can charge more in cost-sharing for post-acute care than TM does.

    These different payment models offer different incentives that may affect who receives care, in what setting, and for how long. In Health Affairs, Peter Huckfeldt, José Escarce, Brendan Rabideau, Pinar Karaca-Mandic, and Neeraj Sood assessed some of the consequences of those incentives. Focusing on hospital discharges for lower extremity joint replacement, stroke, and heart failure patients between January 2011 and June 2013, they examined subsequent admissions to skilled nursing and inpatient rehabilitation facilities, comparing admission rates, lengths of stays, hospital readmission rates, time spent in the community, and mortality for MA and TM enrollees. To do so, they used CMS data on post-acute patient assessments for patients with discharges from hospitals that received disproportionate share or medical education payments from Medicare.

    You might be wondering how the investigators could possibly study MA patients with CMS data. First of all, post-acute facilities are required to file patient assessments for all patients, MA and TM alike. Second, in order to pay disproportionate share and medical education payments to hospitals, CMS collects “information only” (i.e., zero charge) claims for MA patients from hospitals entitled to those payments. Such hospitals account for 92 percent of Medicare discharges. (As far as I know, validation analyses of information only claims has not been published.)

    Their analytic file included almost one million lower extremity joint replacement episodes, almost a half-million stroke episodes, and just over three-quarters of a million heart failure episodes. About a quarter of these were for MA patients. Their analyses adjusted for patient characteristics and used hospital fixed effects.

    Lower extremity joint replacement MA patents were two percentage points more likely to be admitted to a skilled nursing facility than TM patients, but an MA patient stayed 3.2 fewer days, on average. On the other hand, lower extremity joint replacement MA patients were far less likely to be admitted to an inpatient rehabilitation facility than TM patients — an adjusted difference of 6.4 percentage points. Lengths of stay were about the same.

    Findings for stroke patients were similar. TM heart failure patients were slightly more likely to be admitted to a skilled nursing facility than MA patients, and stayed 1.7 days longer than MA patients. The trend is similar for heart failure patient admissions and stays at inpatient rehabilitation facilities, but the overall rates of admission to them is very low for these patients.

    MA patients of all types were less likely to be readmitted to the hospital and more likely to be living in the community than TM patients. There were no statistically significant mortality differences. These results suggest that MA patients are not adversely affected by their lower volume of post-acute care utilization.

    If TM patients were to somehow adopt the same level of utilization of post-acute care as MA patients, the program would save money. Results of the authors’ savings calculations, by condition, are shown in the following chart. Overall, they represent about a 16 percent reduction in spending.

    One way TM is progressing toward more efficient post-acute spending is through accountable care organizations (ACOs). That’s according to a recent paper by J. Michael McWilliams and colleagues. They found, for example, that ACOs that entered the Medicare Shared Savings Program in 2012 reduced post-acute spending by 9% without reductions in quality of care (as measured by mortality, readmissions, and use of highly rated skilled nursing facilities), relative to non-ACO providers.

    Naturally, there are some limitations to the Huckfeldt et al. analysis. First, though the authors did not detect any favorable selection into MA, it is possible that MA patients differ systematically from TM patients such that the results are not completely driven by MA care management. Second, the study did not examine every type of post-acute care — home health care, long-term hospital care, or other outpatient care were omitted from the analysis. Home health care is less costly than the types of care examined. Long-term hospital care is relatively rare. And, outpatient care is unobservable for MA. Third, there are other, possible measures of quality of care besides those examined (readmissions, time in the community, and mortality).

    All told, the results suggest that MA manages post-acute care more tightly than TM and with no observed, negative consequences for patients.

    @afrakt

    Share
    Comments closed
     
  • Risk adjustment cannot solve all selection issues—network contracting edition

    In our Hamilton Project paper, Nicholas Bagley, Amitabh Chandra, and I explain why a health insurance market in which plans compete on cost effectiveness won’t work. (Click through, download the PDF, and read Box 2 on page 9, titled “Why Health Plans Cannot Differentiate on Coverage.”)

    The recent NBER paper by Mark Shepard makes the same argument we made, but to illustrate problems in hospital markets with heterogeneous preferences for costly, star hospitals. Some key quotes from Mark’s paper:

    But even excellent risk adjustment is unlikely to offset costs arising from preferences for using star (or other expensive) providers. These preferences create residual cost variation that can lead to a breakdown of risk adjustment (Glazer and McGuire 2000). Second, the two channels may have different cost and welfare implications. While sickness makes individuals costly in any plan, preferences for a star hospital only make enrollees costly if a plan covers that star hospital. Stated differently, preferences affect how much an individual’s costs increase when their plan adds coverage of the star hospital. […]

    My results suggest that consumer preferences for high-cost treatment options – star hospitals in my study, but the same idea could apply to any expensive provider, drug, or treatment – can naturally lead to adverse selection, and specifically selection on moral hazard. […]

    In the current system, consumers get access to star hospitals based on their plan choice, after which use of these providers is highly subsidized by the insurer. This setup leads to higher costs (moral hazard) and selection on moral hazard. Policies that reduce this moral hazard – e.g., higher “tiered” copays for expensive hospitals or incentives for doctors to refer patients more efficiently – may also mitigate the adverse selection. Differential plan prices for different groups may also improve the efficiency of consumer sorting across plans.

    Mark’s paper is also noteworthy because it is one of the few to address consequences of network contracting. This is a hard area to study because plans’ hospitals and physician networks are not easily observed. Other good work in this area has been done by my colleagues at the Leonard Davis Institute.

    @afrakt

    Share
    Comments closed
     
  • Another day, another cost shifting claim

    Via blog post, David Anderson drew my attention to a Health Affairs post by Billy Wynne, the Managing Partner of TRP Health Policy, in which Mr. Wynne wrote, “[C]ommercial rates are hiked, often stratospherically, to compensate for typically insufficient Medicaid reimbursement.”

    Normally, I’d have to remind the world that many recent studies find no evidence for this kind of cost shifting. But, David did the work for me (and by citing me), so go read his post if you still don’t know the truth of the matter.

    Oh, and spread the word. It’s very hard to keep up with assertions of cost shifting and push back against every one, even with David’s help.

    @afrakt

     

    Share
    Comments closed
     
  • Some Medicare prescription drug plans cost more than we think

    The following originally appeared on The Upshot (copyright 2017, The New York Times Company). 

    Many studies have demonstrated what economics theory tells us must be true: When consumers have to pay more for their prescriptions, they take fewer drugs. That can be a big problem.

    For some conditions — diabetes and asthma, to name a few — certain drugs are necessary to avoid more costly care, like hospitalizations. This simple principle gives rise to a little-recognized problem with Medicare’s prescription drug benefit.

    For sicker Medicare beneficiaries, the Harvard economist Amitabh Chandra and colleagues found, increased Medicare hospital spending exceeded any savings from reduced drug prescriptions and doctor’s visits. Consider patients who need a drug but skip it because they feel the co-payment is too high. This could increase hospitalizations and their costs, which would make them worse off than if they’d selected a higher-premium plan with a lower co-payment.

    Though just a simplified example, this is analogous to what Medicare stand-alone prescription drug plans do. They achieve lower premiums by raising co-payments. This acts to discourage the use of drugs that would help protect against other, more disruptive and serious health care use, like hospitalization.

    Studies show that insurers, many of which are for-profit companies after all, are using such incentives to dissuade high-cost patients from enrolling or using the benefit. There’s evidence this occurs for Medicare’s drug benefit, as well as in the Affordable Care Act’s marketplaces.

    The most popular type of Medicare drug coverage is through a stand-alone prescription drug plan. A stand-alone plan never has to pay for hospital or physician visits — those are covered by traditional Medicare. Another way to get drug benefits from Medicare is through a Medicare Advantage plan that also covers those other forms of health care and is subsidized by the government to do so.

    Because of this difference, stand-alone drug plans are less invested than Medicare Advantage plans in keeping people healthy enough to avoid some hospital visits.

    A study by the economists Kurt Lavetti, of Ohio State University, and Kosali Simon, of Indiana University, quantifies the cost. Compared with Medicare Advantage plans, stand-alone drug plans charge enrollees about 13 percent more in cost sharing for drugs that are highly likely to help patients avoid an adverse health event within two months. They charge up to 6 percent more for drugs that help avoid adverse health events within a year.

    It’s not as if stingier insurers are more likely to offer stand-alone plans than Medicare Advantage plans. Even among plans owned by the same insurer, Medicare Advantage plans are more generous in covering these kinds of drugs than stand-alone drug plans. (These differences are apparent only on average. In some instances, stand-alone drug plans offer better deals.)

    Of course, people have choices about plans. Those who have selected a stand-alone drug plan, as opposed to a Medicare Advantage plan, have done so voluntarily. Why do some make this choice?

    One answer is that some people are not comfortable with the more narrow networks Medicare Advantage plans offer, with their fewer choices of doctors and hospitals. By choosing a stand-alone drug plan, they can remain in traditional Medicare, which has an open network.

    In addition, consumers are generally more attracted to lower-premium plans than higher ones, even if the difference is exactly made up in co-payments. This may be because premiums are easier to understand than cost sharing. Moreover, premiums reflect a sure loss — you must pay the premium to remain in the plan. A higher co-payment, on the other hand, won’t necessarily lead to a loss because you may not use a service.

    The appeal of lower premiums is an incentive for stand-alone drug plans to reduce them and increase co-payments. But that can dissuade those who need medications from filling prescriptions and taking them.

    Part of the purpose of Medicare’s drug benefit is to encourage enrollees to take prescription drugs that can keep them out of the hospital. In July 2003, promoting the legislation that created Medicare’s drug benefit, President George W. Bush articulated this point. “Drug coverage under Medicare will allow seniors to replace more expensive surgeries and hospitalizations with less expensive prescription medicine,” he said.

    But the design of Medicare’s drug benefit includes stand-alone plans that aren’t liable for hospital costs, so they don’t work as hard to avoid them. Encouraging more beneficiaries into comprehensive plans — through Medicare Advantage — or offering a drug plan as part of traditional Medicare itself would address this limitation.

    Postscript: A CMMI initiative that started this year in five regions is designed to create incentives for stand alone PDPs to reduce non-drug Medicare expenditures. 

    @afrakt

    Share
    Comments closed
     
  • Upshot extra: A little-known limitation of Medicare’s drug benefit

    Many studies have demonstrated what economics theory tells us must be true: when consumers have to pay more for their prescriptions, they take fewer drugs. For some conditions — diabetes, asthma, or C.O.P.D., to name a few — certain drugs are necessary to avoid more costly care, like hospitalizations.

    My Upshot post today is based on this principle. It is about a little-known limitation of Medicare’s drug benefit. In particular, stand alone drug plans cost the Medicare program hundreds of millions of dollars per year because they have no financial incentive to avoid hospitalizations.

    A study by economists Amanda Starc, of Northwestern’s Kellogg School of Management, and Robert Town of the University of Texas, asked how much more stand alone plans would spend on drugs if they accounted for the offsets from savings on non-drug health care. The answer: 13 percent. In covering drugs less generously, they end up costing traditional Medicare $475 million per year. This does not account for other social costs, like the inconvenience and suffering of beneficiaries who land in the hospital.

    Click through to my Upshot post for more.

    @afrakt

    Share
    Comments closed
     
  • Adjusting for inflation is tricky

    Abe Dunn, Scott Grosse, and Samuel Zuvekas are right: In many health care analyses, it is important to adjust for inflation, but there are numerous price indexes to choose from for the job. Which one should you use?

    Helpfully, the authors surveyed various measures of health care price inflation and provided guidance on which to use in various circumstances. Their paper also explains why price indexes differ.

    One important way they differ is the extent to which they incorporate changes in consumption patters. Laspeyres and Paasche type indexes hold fixed the quantities of goods and services consumed. The former fixes them in a base period and the latter fixes them in the current period. The only things that change in such indexes are prices. As such, they could misrepresent the actual impact of prices because people and institutions respond by changing how much of various goods and services they consume.

    Put it this way, if the price of a service in the index went from $1 per unit in the base period to $1 billion per unit in the current period, consumption would fall to zero. People would substitute something else that’s cheaper. Lasperyes and Paasche indexes would miss that trend, though in different ways. The former would show massive inflationary effects because the good is in the basket in the base period, while the latter would show no inflationary effects because the good is not in the basket in the current period.

    A Fisher index is the geometric mean of a Laspereys and a Paasche index. That makes it responsive to changes in consumption patterns, though not necessarily reflective of how they change period-by-period. A chained index, on the other hand, continuously updates consumption patterns.

    Another way price indexes differ is scope. Some indexes measure general price inflation, encompassing economic sectors beyond health care. Commonly used ones include the Gross Domestic Product (GDP), the Consumer Price Index—all urban consumers (CPI-U), and the Personal Consumption Expenditures (PCE) index. Some details:

    • GDP: A Fisher index that includes all sectors of economic activity (i.e., not just consumer spending). It is appropriate for adjusting for societal-level purchasing power changes.
    • CPI-U: A Laspeyres index that captures spending by urban consumers (i.e., excluding business and government spending, as well as consumption paid for on behalf of consumers by third parties). It is appropriate for adjusting for purchasing power changes in out-of-pocket spending.
    • PCE: A chained index that captures consumer spending, as well as that paid for by business, government and third parties on behalf of consumers. It is appropriate in the same circumstances as the CPI-U.

    Despite the differences in how they address consumption changes and scope, these three indexes of general price inflation track one another relatively closely, as shown in Figure 1.

    Other indexes measure only health care price inflation. These include the Personal Health Care (PHC) deflator, Personal Consumption Expenditures-health (PCE health), and the Medical Care CPI (MCPI).

    • PHC: A Fisher index that captures out-of-pocket and third-party health expenditures, excluding administration, research, and capital investment. It is appropriate for adjusting for general medical price changes. (Note, the National Health Expenditures (NHE) price index is an alternative that includes administration, research, and capital investment.)
    • PCE health: A chained index that is otherwise nearly identical to the PHC but has a longer history.
    • MCPI: A Laspeyres index designed to capture out-of-pocket health care spending, including that paid by individuals with private or Medicare Part B coverage. It is most appropriate for adjusting for out-of-pocket medical price changes.

    The chart below (the authors’ Figure 3) shows the historical trends for these three indexes, alongside the CPI-U. All start below the CPI-U and then fall into line with it by the late 2000s, with the exception of the MCPI, which rises above the CPI-U. That the MCPI has a different trend than the other medical price indexes is not surprising, since it is the only one of the three focused exclusively on out-of-pocket spending.

    Others indexes are even more narrow, measuring, for example, hospital price inflation overall or for different payers — the Medicare Hospital Producer Price Index (PPI), Medicaid Hospital PPI, Private Hospital PPI, Overall Hospital PPI, and Overall Hospital CPI. All five of these are shown in the author’s Figure 2, reproduced below. They exist for other services as well, not just hospital care.

    • PPI (overall and specific payers, disaggregated by service): A Laspeyres index of third-party payments appropriate for adjusting medical prices from specific payers for specific services.
    • CPI (overall and disaggregated by service): The same as the MCPI discussed above, but including components for different services. These are appropriate for adjusting for out-of-pocket spending.

    There are large differences in these indexes, as shown in Figure 2, reflecting differences in price growth in hospital care versus overall medical care and/or differences in growth from different payers or out-of-pocket spending. If you’re analyzing hospital payments, it’s very important to match the index to the type of analysis you’re doing and questions you’re addressing.

    The authors discuss many other nuances and issues, which are beyond the scope this post. They conclude with this advice:

    • To adjust health expenditures in terms of purchasing power, use the GDP implicit price deflator or overall PCE measure. The PCE measure is suitable for personal consumption. The GDP deflator is more appropriate for the societal perspective.
    • To adjust overall consumer out-of-pocket spending in terms of consumer purchasing power or out-of-pocket burden relative to income, the CPI-U can be used.
    • To convert average expenditures to care for a specific disease for price changes from 1 year to a different year, either the PHC deflator or the PCE health index can be used. Because of exclusions of some payers in its weights, the MCPI may not be appropriate to adjust all-payer expenditures or payments by employers, Medicaid, and Medicare Part A for medical inflation.
    • To convert average consumer out-of-pocket health care expenditures from 1 year to a different year, the MCPI can be used.
    • To adjust estimates of costs of inpatient services from different years, the PPI for inpatient services appears currently to be the best option.

    I’m certain many published analyses do not follow this guidance, including my own. In some cases, it may matter. In offering clear guidance on how to use health care price indexes, Dunn, Grosse, and Zuvekas have provided a valuable service.

    @afrakt

    Share
    Comments closed
     
  • Where’s the best, deep health policy wonk blogging these days?

    There’s not a lot of old-school health policy blogging going on these days. But there’s not none!

    Over lunch, I was asked where (other than TIE) I turn for the best and deepest health policy wonk blogging. The answer came to me immediately, but my questioner had never heard of the blog or individual I spoke of. That should change!

    So, if you’re not reading David Anderson’s (frequent!) posts at Balloon Juice, you’re missing something good. My only regret is that he’s not writing on TIE. Sad.

    An RSS feed for David’s posts is here. He tweets via @bjdickmayhew. No his name is not really Richard Mayhew.

    @afrakt

    Share
    Comments closed
     
  • Spend a dollar on drug treatment, and save more on crime reduction

    The following originally appeared on The Upshot (copyright 2017, The New York Times Company). It also appeared on page A15 of the April 25, 2017 print edition.

    The burden of substance abuse disorders can fall heavily on the families and friends of those who battle addictions. But society also pays a great deal through increased crime. Treatment programs can reduce those costs.

    For at least two decades, we’ve known substance use and crime go hand in hand. More than half of violent offenders and one-third of property offenders say they committed crimes while under the influence of alcohol or drugs.

    Researchers with the Centers for Disease Control and Prevention recently estimated that prescription opioid abuse, dependence and overdoses cost the public sector $23 billion a year, with a third of that attributable to crime. An additional $55 billion per year reflects private-sector costs attributable to productivity losses and health care expenses.

    About 80,000 Americans are incarcerated for opioid-related crimes alone. The total annual economic burden of all substance use disorders — not just those involving opioids — is in the hundreds of billions of dollars.

    In an editorial accompanying the C.D.C. researchers’ study, Harold Pollack, co-director of the University of Chicago Crime Lab, wrote that opioid-associated crime, like all crime, extracts an even larger toll when you consider its impact on families and communities.

    “The most important reason to support treatment is to improve the well-being and social function of people with addiction disorders,” Mr. Pollack said. But there are other social benefits. When the criminally active get help for this, “the economic value of crime reduction largely or totally offsets the costs of treatment,” he added.

    Relative to the costs of crime alone, treatment for substance use disorders is a good deal. Even though a typical burglary may result in a few thousand dollars of tangible losses, researchers have estimated that people are willing to pay 10 times that amount to avoid that loss and 100 times more to avoid armed robbery. This reflects the fact that crime exacts a large psychological toll — the threat or climate of it is far more costly than the crimes themselves.

    The most cost-effective treatment for opioid use disorders includes counseling along with a craving-relieving prescription drug, like methadone or buprenorphine, sometimes combined with other medications. According to an economic analysis by the New England Comparative Effectiveness Public Advisory Council, this kind of treatment actually saves society money. For instance, New England states could save $1.3 billion by expanding treatment of opioid-dependent persons by 25 percent.

    Though the war on drugs has not had a tangible impact on crime, treatment for substance use disorders has. A study by Emory University scholars found that a 10 percent increase in the treatment rate reduces the robbery and larceny theft rates by about 3 percent and the aggravated assault rate by 4 to 9 percent.

    For a dollar spent on treatment, up to three are saved in crime reduction. An earlier study found that interventions to address substance use disorders save more in reduced crime than they save in reduced health care spending.

    Several systematic reviews and meta-analyses of therapies for opioid addiction found that methadone therapy reduced criminal activities related to heroin use. One analysis of more than 8,000 heroin users found that their offending rates were lower while on methadone therapy than when not on it.

    For every 100 patients on methadone per year, there were 12 fewer robberies, 57 fewer break-and-enters and 56 fewer auto thefts. Another systematic review found that provision of heroin by doctors to patients addicted to it — permitted in Canada and some other countries — reduces crime.

    Findings such as these justify drug courts, which divert drug offenders from the traditional criminal justice system into treatment. But what about helping those with substance use disorders obtain treatment before they commit crimes and land in court? Given the crime-deterring value of treatment (among its other benefits), you’d think we’d make it easy for patients to get.

    We don’t. The need for treatment far exceeds its supply. Many treatment programs have waiting lists, and the vast majority of those with substance use or dependency problems go untreated.

    Stigma plays a role, which is why addiction treatment works best when it is integrated with and supported alongside ordinary medical care. A pervasive not-in-my-backyard attitude challenges the establishment of more programs. A recent study by economists from Texas A&M and Montana State Universities suggests this is shortsighted.

    The researchers found that the opening of an additional treatment facility in a county is associated with lower drug-related mortality in that county, as well as lower crime. The effect of crime reduction alone would save an estimated $4.2 million per facility per year, or almost four times its cost.

    “Addiction treatment may be the one area of health policy right now in which Democrats and Republicans want to work together to meet an important public health challenge,” Mr. Pollack said. “The economic and crime-reduction benefit of these services certainly provide good reason for this.”

    @afrakt

    Share
    Comments closed
     
  • AcademyHealth: Medicare Advantage risk selection

    One of the concerns about Medicare Advantage (MA) is that it doesn’t serve sick beneficiaries well, motivating some of them to switch to traditional Medicare (TM). My final blog post for AcademyHealth explores this topic.

    @afrakt

     

    Share
    Comments closed
     
  • Readers ask: Is there brain training for hearing loss?

    Since publication of my Upshot post about training your brain to improve eyesight for reading, several readers have asked if there are similar approaches to brain training to improve hearing.

    I put that question to some experts, but have not received any responses that suggest there is such a thing. In some cases, experts simply did not respond to my email, so maybe those experts know something but aren’t saying!

    Point being, right now I’m not aware of how to improve your hearing through brain training. If someone knows otherwise, I’m all ears.

    @afrakt

    Share
    Comments closed