Aaron E. Carroll's Blog, page 129

Private insurance isn’t always better than Medicaid

Austin and I have a new column up on The Upshot today. We’ll post it here in its entirety on Friday, per our routine, but it’s important for you to go read it today before the new Senate bill is likely released tomorrow. A new talking point is that while Medicaid may be better than being uninsured (what people used to say), it’s certainly far, far worse than private insurance.

We take that apart over at the NYT. Go read it.

@aaronecarroll

I’ll have that maternity leave that she’s having

David Brooks has a theory about the cause of rising American social inequality. Well-off parents invest more money and time in their children.

Over the past generation, members of the college-educated class have become amazingly good at making sure their children retain their privileged status. They have also become devastatingly good at making sure the children of other classes have limited chances to join their ranks.

How they’ve managed to do the first task — giving their own children a leg up — is pretty obvious… Over the past few decades, upper-middle-class Americans have embraced behavior codes that put cultivating successful children at the center of life. As soon as they get money, they turn it into investments in their kids.

Upper-middle-class moms have the means and the maternity leaves to breast-feed their babies at much higher rates than high school-educated moms, and for much longer periods… [and] to spend two to three times more time with their preschool children than less affluent parents.

Brooks then makes some remarks about informational barriers between social classes. He somehow misses that there is a simple policy that would help reduce the class difference in the time that American parents invest in their children. We could give every family, not just those with means, a paid leave when they have a newborn child.

It’s a straightforward way to invest in children and nearly every developed country other than the US does it. In Canada, families get 55% of a parent’s income for a year, up to a maximum, when you have a newborn (it’s not a ‘maternity’ leave, because the father can take it, or you can split it between spouses).

Giving everyone a year with their infant won’t make all children equal. However, there is considerable evidence that early childhood is of great importance for children’s health and development and that paid maternity leave contributes to better child health.

We may not know how to tear down psychological barriers between social classes. However, we know how to give parents more time to raise young children.

@Bill_Gardner

Health Plans That Nudge Patients to Do the Right Thing

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

As health care costs rise, Americans are increasingly on the hook to pay more for their care. This trend is more than just annoying — asking consumers to pay more for everything deters many from getting the care they need. What would happen if, instead, health plans offered more generous coverage of high-value care, but less generous coverage of those services that provide little or no health benefit?

This idea is known as value-based insurance design. Though not widespread, V-BID is not new. It was pioneered nearly 20 years ago by Dr. Mark Fendrick, a physician and professor at the University of Michigan, and Michael Chernew, a Harvard economist. (“Value” in V-BID plans is usually set according to longstanding measurements of quality established by decades of study of medical records.)

In his own practice, Dr. Fendrick feels as if standard insurance is working against him and his patients. “They are deeply concerned about the amount they have to pay out of their own pockets for the things I beg them to do,” he said. “It makes no sense that they pay the same co-payment for a lifesaving drug to treat diabetes or cancer, as for a drug that makes toenail fungus go away.”

This may be changing. The Affordable Care Act includes a V-BID provision, eliminating cost-sharing for more than 100 preventive services, such as vaccinations and cancer screenings. It’s endorsed by four committees of medical experts.

Many large employers and state governments are going further, reducing cost-sharing for high-value care and medications to treat chronic illnesses, like depression and heart disease. This year, the Centers for Medicare and Medicaid Services began a five-year test of value-based design that permits Medicare Advantage plans in seven states to reduce cost-sharing and enhance benefits for enrollees with designated chronic conditions. Bipartisan legislation has been introduced in the House and Senate to expand the program nationwide.

In 2018, the Department of Defense will pilot a V-BID program that reduces cost-sharing for high-value medications and services, trying to improve the care and outcomes for American military personnel.

Does value-based insurance work? The Medicare Advantage and Department of Defense programs will tell us more, but experience with commercial V-BID programs suggests it’s a promising approach.

Several studies show that value-based “carrot” programs — those that help patients with chronic illnesses stay out of the hospital by reducing cost-sharing for high-value medications — increases medication use, at least modestly. Of course, if patients are paying less for medications, someone else — employers and health plans — pick up that part of the tab.

But making high-value drugs less expensive can offset other health care spending. One study of 6,000 heart attack patients compared one group that received their drugs free with those whose regular insurance had co-payments of $10 to $25. Patients receiving free drugs increased their use of them — and the additional insurer drug costs were offset by a decrease in hospital procedures.

Another study, of reduced cost-sharing for diabetic medications, showed that patients took their medication more regularly and used other, costly services less. Emergency department visits dropped 36 percent, and hospitalizations fell 13 percent.

Value-based programs need not focus exclusively on drugs. In 2010, I.B.M. encouraged greater use of primary care by eliminating cost-sharing for primary care visits. A study of the effect on children found increases in primary care visits and vaccinations, and decreases in expensive emergency department and specialist visits.

In 2011, Connecticut started the Health Enhancement Program for state employees, which required participants to obtain high-value primary and chronic disease preventive services — like screenings, physical examinations and other guideline-based services — and lowered cost-sharing for them. The use of those services and medications for chronic conditions increased, while emergency department use decreased during the program’s first two years.

Though programs like these improve patients’ health and quality of life, they don’t necessarily save money. That’s the finding of a large programby Blue Cross Blue Shield of North Carolina. In 2008, the insurer reduced cost-sharing for hypertension, hyperlipidemia, diabetes and congestive heart failure medications for more than 700,000 policy holders. Their overall health care spending remained comparable to that of similar patients insured by other plans that did not use a value-based design.

“For the most part, V-BID isn’t a way to save money,” Mr. Chernew said. “What it can do is shift health care use from lower- to higher-value care.” In doing so, it can also provide more financial protection for people who have regular need for maintenance medications and care.

If reducing cost-sharing for high-value care is the “carrot” approach, increasing it for low-value care is the “stick.” Though less common, stick approaches have been tested, too. A program by a large public employer in Oregon raised cost-sharing for sleep studies, upper gastrointestinal endoscopies, advanced imaging services and certain types of overused procedures, like surgery for back pain. Although clinically appropriate circumstances exist for each service, copayments were raised $100 to $500 only for those specific situations where their use was deemed not medically necessary.

An evaluation demonstrated that consumers responded to these higher out-of-pocket costs. Targeted services fell by about 12 percent over all, though some fell more than others. For example, sleep studies and low-value surgery use fell by about 20 percent. But advanced imaging use fell by only 7.7 percent.

It may be inevitable that health insurance comes with cost-sharing. But there’s no reason it can’t be applied in ways that also help patients do the right thing, nudging them toward high-value care and decreasing incentives to pursue low-value care.

@afrakt

Preventing hospitalizations from nursing homes is harder than it looks

I applaud the publication of negative studies. From JAMA Internal Medicine, “Effects of an Intervention to Reduce Hospitalizations From Nursing Homes: A Randomized Implementation Trial of the INTERACT Program“:

Importance: Medicare payment initiatives are spurring efforts to reduce potentially avoidable hospitalizations.

Objective: To determine whether training and support for implementation of a nursing home (NH) quality improvement program (Interventions to Reduce Acute Care Transfers [INTERACT]) reduced hospital admissions and emergency department (ED) visits.

Design, Setting, and Participants: This analysis compared changes in hospitalization and ED visit rates between the preintervention and postintervention periods for NHs randomly assigned to receive training and implementation support on INTERACT to changes in control NHs. The analysis focused on 85 NHs (36 717 NH residents) that reported no use of INTERACT during the preintervention period.

Interventions: The study team provided training and support for implementing INTERACT, which included tools that help NH staff identify and evaluate acute changes in NH resident condition and document communication between physicians; care paths to avoid hospitalization when safe and feasible; and advance care planning and quality improvement tools.

Main Outcomes and Measures: All-cause hospitalizations, hospitalizations considered potentially avoidable, 30-day hospital readmissions, and ED visits without admission. All-cause hospitalization rates were calculated for all resident-days, high-risk days (0-30 days after NH admission), and lower-risk days (≥31 days after NH admission).

We’d like to reduce hospitalizations from people who live in nursing homes, by keeping them from getting sick or hurt. The INTERACT program was designed to do just that. It supported and trained nursing home workers in identifying and evaluating issues in residents in nursing homes, communicating with doctors, and implementing quality improvement. Nursing homes were randomized to this or usual care. The main outcome of interest was hospitalizations, avoidable hospitalizations, readmissions, and ED visits.

Eighty-five nursing homes with 281 752 person-months were included in the analysis. There was no significant change in the number of hospitalizations in the intervention group versus the control group. There was no significant change in readmissions, ED visits, or any of the sub-analyses of hospitalizations. There was a small, but statistically significant reduction in avoidable hospitalizations, but once they applied a Bonferroni correction, it was no longer significant.

Let’s talk about that for a second. When you do a lot of statistical tests on a lot of potential outcomes, you increase the chance that something will be “significant” by chance alone. When that’s the case it’s good practice to apply a correction to account for that. Since they had six outcomes, the Bonferroni correction reduced the p-value threshold from 0.05 to 0.008. The result (as p=0.01) was no longer significant. Good for them. For more information on significance and p-values, I encourage you to watch our Healthcare Triage episodes on the subject here and here.

We all want to improve the care in nursing homes and prevent hospitalizations and ED visits, if possible. This trial showed a pretty big intervention didn’t work. Let’s acknowledge that and try to do better, instead of doing the same things over and over again. It may take more investment. Sometimes good things cost money.

@aaronecarroll

Needed: Core Quality and Outcome Measures for Pediatric Health

Checklists and other quality measures are critical in protecting patients and improving the quality of health care. Unfortunately, many physicians and other clinicians find the burden of documentation overwhelming.

David Blumenthal and J. Michael McGinnis argue that part of the problem is that we have too many measures.

Not only are many measures imperfect, but they are proliferating at an astonishing rate, increasing the burden and blurring the ability to focus on issues most important to better health and health care. Measures of the same phenomenon also vary in specification and application, leading to confusion and inefficiency that make health care more expensive and undermine the very purpose of measurement, namely, to facilitate improvement. Not uncommonly, a health care organization delivering primary care to a typical population is asked to report and collect hundreds of measures aimed at dozens of conditions.

With Kelly Kelleher, a pediatrician at Nationwide Children’s Hospital, I have been working with a group of child health experts to develop a parsimonious set of core quality and outcome measures for children’s health care. We describe this effort in JAMA Pediatrics. We’re part of a larger National Academy of Medicine effort. They have proposed a core measure set for adults. We argue, however, that because children are different, we need a distinctively pediatric set of quality and outcome measures.

We need a measure set for children because the goals of children’s health care are different from that of adults, children face different health risks, and the context of children’s health care is different.

The goal of health care—well-being—has a different meaning across the life span. For children, it is building the physical, cognitive, and social foundations for adult capabilities in addition to enhancing their current state. The specific health needs of children can change rapidly as they pass through developmental stages. Thus, a pediatric measure set must find a careful compromise that is sensitive to development without excessively expanding the number of measures.

Similarly, children face different health risks. Because children are mostly healthy, subclinical precursor states of adult health problems get insufficient attention. Preventive care is therefore underemphasized, leading to health problems in adulthood. For example, adolescents being overweight and obese is strongly associated with adult cardiovascular mortality.

Finally, children’s risks are different because many of the diseases and risks for children are environmental or neighborhood-influenced. Children are more sensitive to socioenvironmental factors and depend more on support for their health and care than adults. Thus, detailed measures that capture the socioenvironmental circumstances in the community are essential.

There’s lots to do and we’re looking for your good ideas.

@Bill_Gardner

The research on malpractice and nursing homes

This is the second post in a series on a proposed CMS rule that would eliminate an Obama-era ban on pre-dispute arbitration for nursing home residents. For the intro, see here.

By penalizing injury-causing negligence, tort law is supposed to deter negligent conduct. That’s one of the reasons (but not the only one) that CMS originally moved to ban nursing homes from insisting on pre-dispute arbitration: by reducing the threat of malpractice, arbitration clauses could compromise patient safety.

Now, mandatory arbitration doesn’t eliminate the malpractice threat. Arbitration is an alternative forum for resolving claims, albeit one that stacks the deck in nursing homes’ favor. Plus, not all arbitration agreements will stand up in court. Sometimes, a judge will find that the resident wasn’t competent to waive her rights to sue.

Without a doubt, however, arbitration clauses reduce nursing homes’ financial exposure and thus any deterrent effect associated with potential litigation. But how big is that reduction? Is it big enough to matter?

To answer that question, we’d ideally look at studies that examine whether a shift to mandatory, pre-dispute arbitration caused a dip in nursing home quality. But no one’s done that research, at least to my knowledge, perhaps because it’s tough to get a handle on how frequently nursing homes insist on arbitration clauses.

That doesn’t mean we’re out of luck, though. There’s a bunch of research on whether malpractice risk is associated with nursing home quality. If that research suggests that the deterrent effect is big, more arbitration might well be bad for patient safety. If the deterrent effect is small, it might not make of a difference.

So what does the research tell us? In 2011, an important study in the New England Journal of Medicine concluded that nursing homes with more quality deficiencies, and more serious deficiencies, were more likely to be sued for negligence. But the difference was small: “Nursing homes with the best deficiency records faced about a 40% annual risk of one or more claims, as compared with 47% among homes with the worst deficiency records.”

From this, it looks like there’s only so much a nursing home can do to mitigate its liability risk. Even it improves, it’ll get sued about as often as it used to. “Such weak discrimination,” the researchers wrote, “may subvert the capacity of litigation to provide incentives to deliver safer care.”

Another 2011 study, this one in the Journal of Health Care Finance, reached a similar conclusion. It found that nursing homes in Florida with more registered nurses per resident paid less in response to malpractice claims, which jibed with earlier studies. But the study still couldn’t pick up a statistically significant connection between deficiencies and malpractice payments. Nor could this study from 2004.

All of these studies, moreover, suffer from a common problem. They tell us which nursing homes are more or less likely to be sued, but they can’t tell us directly whether malpractice risk actually influences a nursing home’s choices about staffing levels and quality. A 2013 study in Health Services Research tried to avoid the problem by exploiting variations in the background rate of malpractice in different geographic areas. In line with earlier research, the study found that “while a deterrence effect exists, it is small, concentrated among a subset of nursing homes, and unlikely to lead to widespread improvements in quality.”

In short, the research suggests that tort law matters, but only on the margins. If that’s right, then mandatory arbitration may not raise substantial safety concerns.

Now, these studies aren’t the final word. Reported deficiencies are only proxies for overall quality; it’s possible that the malpractice threat makes nursing homes attentive to unmeasured but important aspects of quality. Plus, the research pretty consistently shows that having more registered nurses (but not nursing aides) reduces malpractice risk. In the absence of litigation exposure, nursing homes might well hire fewer registered nurses, with unfortunate consequences for quality.

Most importantly, the studies can’t and don’t address the systemic, long-term risks of an industry-wide practice of substituting mandatory arbitration for civil litigation. In particular, the research doesn’t capture whether the secrecy of arbitration may make it easier to sweep the endemic quality problems in the nursing home industry under the rug. I’ll turn to that question in my next post.

@nicholas_bagley

Healthcare Triage: Reduce Crime AND Save Money – Treat Addicts Instead of Punishing Them

Substance abuse and addiction are terrible for addicts health, and they’re really tough on family and friends. Addiction also drives up the violent crime rate, and the rate of property crime. This week on Healthcare Triage, we look at research on how diverting addicts to treatment programs has a positive effect on crime rates.

Special thanks to Austin, from whose Upshot column this episode was adapted.

@aaronecarroll

What should we do about children’s health coverage?

Back in May, I was on a panel at the Pediatric Academic Societies to discuss health care reform, kids, and the future. After that panel, David Rubin wrote a post over at the Health Affairs blog summarizing his thoughts on what needed to be done.

Most of the people there were either pediatricians, or pediatric-focused, and, of course, we were all pulling our hair out at how little of the conversation nationally focuses on kids or families. So much of our discussion focuses on adults, and then when they’re done talking about them, they turn to the elderly. Medicaid and CHIP cover something like two in five children and Medicaid, even before the expansion, covered one-third of births, and those births produce kids. I’ve written before about how employer-sponsored coverage for kids has been eroding. I shouldn’t need to remind any readers of this blog that proposals for Medicaid in Republican bills come with severely decreased funding in the future.

David has some suggestions for focus:

CHIP reauthorization: While the AHCA works its way through Congress, some may not have noticed that CHIP funding expires this fall. Without re-appropriation, more than 8 million children may lose coverage immediately. States are already sounding alarms; they have been unable to project their CHIP budgets for next year. The immediacy of the CHIP re-appropriation debate in Congress offers a “NOW” opportunity to stake a new way forward and present pragmatic solutions to strengthen children’s insurance, embracing the realities that have reshaped the family insurance market.

Guarantee of essential health benefits: The most critical issue arising from any children’s insurance plan today, whether in the employer-sponsored or public insurance market, is the promise of a set of health standards to all children regardless of their insurance. The House-passed AHCA proposes removing the requirement of federally guaranteed essential health benefits from all plans. Should this become law, states will have the choice of whether or not to provide these benefits. So, one solution for protecting children is to require these states to provide families access to a CHIP plan that meets a comprehensive and standard set of federally legislated and guaranteed essential benefits, such as vision, developmental, and behavioral health screenings.

Private market reforms: Beyond essential benefits, it may be time to address the affordability and quality of dependent coverage on the employer-sponsored and exchange markets. We may need stronger caps on deductibles as a proportion of income, and limits to exorbitant cost-sharing for child dependents. Furthermore, prohibitions of narrow networks—or the increase of cost-sharing for enrollees who seek out-of-network services—in the pediatric market would go a long way to ensuring that families have critical access to pediatric subspecialty care should their children develop cancer, diabetes, or other debilitating illnesses. While narrow networks may work in the adult health care arena, they are not nimble for families whose children have special health care needs and require specialists based solely in children’s hospital networks that may be tiered out in such plans. All told, the private market is not working for families, and if Congress wishes to halt the migration of families onto public insurance, they may need to hold employers and commercial insurers responsible for their own contributions to crowding families out of that market.

Go read the whole thing. And remember it when people say that no one has any plans.

@aaronecarroll

Healthcare Triage News: The Senate’s BCRA Bill – High Premiums, Huge Deductibles, AND Massive Medicaid Cuts

We’re going to make a serious attempt to make Healthcare Triage News more “newsy”. Here’s our first week. Catch up to date on the BCRA and the CBO Report!

@aaronecarroll

Medicaid is good for children and makes them better adults

The Senate’s Better Care Reconciliation Act (BCRA) significantly cuts Medicaid, the program that insured 39% of US children in 2015 and 50% or more in New Mexico, West Virginia, and Mississippi. The Congressional Budget Office estimates that the BCRA will increase the number of uninsured Americans by 22 million, among them many of those currently covered by Medicaid. However, many supporters of these bills argue either that Medicaid delivers no health benefits to the insured or actually harms them. If this is what they believe, then surely reducing Medicaid enrollment is a virtue of the BCRA, not a flaw.

Whether Medicaid is harmful has been debated for years. In this post, I call your attention to emerging evidence that Medicaid benefits children.

Liberals have trouble understanding how Medicaid could be harmful. It’s true that some doctors are less likely to accept a Medicaid patient than a privately insured one. However, a doctor who turns down a Medicaid patient would be even less likely to care for an uninsured one. So how could the well-being of someone on Medicaid be improved if they became uninsured?

These liberals have missed the point of the conservative argument. That view, best articulated by Charles Murray, is that the social safety net harms you by exposing you to incentives that keep you dependent. In a welfare-dependent family, it’s harder for a child to develop into an independent, functional adult. Getting social assistance, therefore, traps you and your family in poverty. The strong prediction by conservatives is that poor kids who grow up as dependents of the welfare state will be less healthy and less functional than other kids.

Arguments that Medicaid is ineffective usually cite the Oregon Health Insurance Experiment (OHIE). This randomized experiment found that getting Medicaid coverage did not improve several important physiological measures of health. Benjamin Sommers, Atul Gawande, and Katherine Baicker (who was the Principal Investigator of the OHIE) rebut this argument in the NEJM. Likewise, Aaron and Austin argue in The New York Times that Medicaid critics have misinterpreted the research. According to these analysts, the OHIE had little chance to find substantial physiological benefits because it studied a small group of people for a relatively brief period of time. Other studies involving large populations and designs that allow causal inferences have shown that getting Medicaid saves lives.

It follows that the ideal study of Medicaid would look at large populations of people over long periods of time with a design that was a randomized experiment or approximated one. It turns out that this is to some degree possible if you look at the child and adult health outcomes of kids who grew up in states where poor families had varying degrees of access to Medicaid.

Congress created Medicaid in 1965, but because it is administered and paid for by both states and the federal government, states started their programs at different times and varied in who they covered. This map from the Social Security Administration shows the year and quarter when states began Medicaid.

The month/year in which states adopted Medicaid. From Boudreaux et al.

This means that poor kids born in 1966 in Oklahoma, which started Medicaid in that year, were likely to have had much more cumulative insurance coverage by the time they reached adulthood than kids born in 1966 in Arizona, which started Medicaid in 1982. So if Medicaid is bad for you, then everything else being equal the states that had larger proportions of kids covered for longer periods of time should have populations of adults that look worse, on average, than other states.

Several groups of analysts have looked at state health outcomes following this strategy. Importantly, these researchers have looked at many aspects of adult well-being, not just health. For example, Cohodes, Grossman, Kleiner and Lovenheim examined

the effects of the public insurance expansions among children in the 1980s and 1990s on their future educational attainment. Our findings indicate that expanding health insurance coverage for low-income children increases the rate of high school completion and college completion… We present suggestive evidence that better health is one of the mechanisms driving our results by showing that Medicaid eligibility when young translates into better teen health. Overall, our results indicate that the long-run benefits of public health insurance are substantial. (Ungated version here.)

Boudreaux, Golberstein, and McAlpine analyzed data from the Panel Study of Income Dynamics. They found that Medicaid coverage in early childhood (age 0–5) was associated with statistically significant and meaningful improvements in adult health (age 25–54). Interestingly, they found that Medicaid was associated with lower blood pressure and a lower overall chronic disease score, which contrasts with the lack of physiological effects in the OHIE. Boudreaux and his colleagues found these benefits only among the low-income people most likely to have been covered by Medicaid.

Miller and Wherry studied

how a rapid expansion of prenatal and child health insurance through the Medicaid program affected adult outcomes of individuals born between 1979 and 1993 who gained access to coverage in utero and as children… We find that cohorts whose mothers gained eligibility for prenatal coverage under Medicaid have lower rates of obesity as adults and fewer hospitalizations related to endocrine, nutritional and metabolic diseases, and immunity disorders as adults. We also find that the prenatal expansions increased high school graduation rates among affected cohorts.

Meyer and Wherry looked at

a policy discontinuity [in] several early Medicaid expansions that extended eligibility only to children born after September 30, 1983… We find a substantial effect of public eligibility during childhood on the later life mortality of black children at ages 15-18. The estimates indicate a 13-20 percent decrease in the internal mortality rate of black teens born after September 30, 1983.

Andrew Goodman-Bacon found that

Before Medicaid, higher- and lower-eligibility states had similar infant and child mortality trends. After Medicaid, public insurance utilization increased and mortality fell more rapidly among children and infants in high-Medicaid-eligibility states. Mortality among nonwhite children on Medicaid fell by 20 percent, leading to a reduction in aggregate nonwhite child mortality rates of 11 percent.

In another analysis, Goodman-Bason looked at the effect of Medicaid coverage when you are 0 to 10 years old on adult mortality rates, whether the children found employment when they became adults, and what they received in transfer payments.

For mortality, he reported that

about 345,000 lives were saved between 1980 and 1999—54,000 among whites and 291,000 among nonwhites.

For rates of disability,

each full year of early-life cumulative [Medicaid] eligibility reduces white ambulatory difficulty rates by 3.8 percentage points (s.e. = 1.2) and nonwhite rates by 2.9 percentage points (s.e. -1.8).

For employment,

each year of childhood Medicaid eligibility reduces the probability of being out of the labor force by 6.6 percentage points (s.e. = 1.5)

These reductions in disability and unemployment among kids covered by Medicaid in early childhood reduced the social assistance transfer payments the Medicaid kids received as adults. As a result,

The government earns a discounted annual return of between 2 and 7 percent on the original cost of childhood coverage for these cohorts, most of which comes from lower cash transfer payments.

Are these findings credible? These are complicated analyses with many moving parts. A lot hangs on the idea that you can make an “everything else being equal” comparison between states based on how they administered Medicaid. Looking at the map above, there are regional correlations in when states started Medicaid, and there are likely correlations with state poverty and other factors. Yet even within regions, there is variation in when Medicaid started and how it was run. The papers I’ve summarized look for this kind of variation in Medicaid coverage policies between comparable states. This variation allows them to infer that Medicaid coverage during early childhood caused these better child and adult outcomes. It’s a large claim and it’s certain to be disputed.

But if you accept this analytical strategy, what do these findings show?

First, the prediction that Medicaid harms children is flat wrong. There’s no evidence that Medicaid kids are worse off. To the contrary, on several important outcomes kids were better off if they grew up in states that started Medicaid coverage earlier and covered more kids.

Second, it appears that the most important times for coverage are when you are in-utero, an infant, or a young child. This reinforces James Heckman’s argument that early childhood is a critical period where we get the highest return for investments in improving the social determinants of health and human development.

Third, the effects of early exposure to Medicaid are found not just in health, but also in educational attainment, employment, and reduced welfare dependency. In other words, Medicaid works on precisely the issues that conservatives care the most about. Provision of Medicaid to families with poor children appears to give those kids a better chance of growing up to be high functioning adults. As such, it promotes fair equality of opportunity. Conversely, these findings count against predictions that Medicaid traps children in lives of poverty.

What the findings do not show is that Medicaid, or public health insurance of any kind, is the optimal way to invest in families or child development. It would help if we knew how Medicaid benefits children. Is it increased access to health care, is it because health insurance reduces the out-of-pocket costs to families, or is it because having health insurance reduces the psychological stress of being a family? These studies don’t tell us. So it’s possible that some other policy would help kids more. For example, we don’t know whether paying for a poor family’s Medicaid is more cost-effective than giving that family the actuarially-equivalent cash.

This and similar proposals deserve serious consideration. But it is critical to note that the comparison between providing Medicaid and giving cash is irrelevant to the debate about the BCRA and the House of Representatives’ American Health Care Act of 2017. Instead of taking resources from Medicaid to fund cash transfers to the poor, these bills take cash out of Medicaid to fund tax cuts for the wealthy.

Finally, these findings cast light on the American (and far stronger Canadian) commitments to a decentralized federal system for the administration of public health insurance. Justice Brandeis argued that federalism — meaning here that states that have considerable autonomy in how they implemented Medicaid — is valuable to the country as a whole. It’s valuable because a

state may, if its citizens choose, serve as a laboratory; and try novel social and economic experiments without risk to the rest of the country.

Many argue, following Brandeis, that federalism is a way to learn what works in social policy.

To learn about policy, however, we have to ask how can we actually get reliable information from the laboratories of the states. One answer is that we do it through differences-in-differences analyses of variation in the implementation of state programs, such as those that we have reviewed above. If you believe that the independence of states gives us a laboratory for democracy, you should take seriously the results on Medicaid that have come back from the lab.

@Bill_Gardner