Why Medicine Should Avoid the ‘Money-cillin’ Treatment

/by

Two professors at the University of Pennsylvania’s medical school have identified an exciting new treatment to improve health outcomes. It consists of basically giving people money. The marketing department has developed better sounding terms for this treatment, like cash transfers or guaranteed income, but if the marketing folks were really clever, they would be calling it “money-cillin.” That worked for penicillin, right?

Whatever we call it, the idea being advanced by Drs. Aaron Richterman and Harsha Thirumurthy in The Atlantic is that this new treatment is one of the most important interventions in medicine: “when designed with the right basic ingredients, cash transfers are one of the most powerful levers that governments have to alleviate poverty and improve health.” As prominent professors at an Ivy League medical school, we can imagine that Drs. Richterman and Thirumurthy think that “money-cillin” should be studied by leading medical researchers and taught to future doctors. Sure, anatomy and physiology are important for healing patients, but so is welfare policy.

The only problem is that the effects of cash transfers on health outcomes have been rigorously studied and the results have been very disappointing. There was a large-scale experiment in the U.S. in which low-income people were given $1,000 in cash per month for three years and compared to a randomized control group that was given $50 per month. The evaluators, including leading economists from the University of Michigan and the University of California, Berkeley, produced two studies, one describing results on health outcomes and another describing results on labor outcomes. It is worth quoting their findings at length. On health outcomes they found:

Over the three year time horizon that we study, we find no effect of the transfer across several measures of physical health, and we can rule out even very small improvements. The transfer also did not improve mental health after the first year and by year 2 we can again reject very small improvements. We also find precise null effects on self-reported access to health care, physical activity, sleep, and several other measures related to preventive care and health behaviors. We find no effect of the transfer on the health of participants’ children, but we do find that children in treated households were more likely to be up to date on their vaccinations.

Not only did “money-cillin” fail to improve health outcomes, but it actually harmed people’s motivation to work. Again, quoting the results at length, the economists found:

The transfer caused total individual income excluding the transfers to fall by about $1,800/year relative to the control group and a 4.1 percentage point decrease in labor market participation. Participants reduced their work hours as a result of the transfers by 1-2 hours/week and participants’ partners reduced their work hours by a comparable amount. Among other categories of time use, the greatest increase generated by the transfer was in time spent on leisure. Despite asking detailed questions about amenities, we find no impact on quality of employment, and our confidence intervals can rule out even small improvements. Treated participants broadly increase expenditures, led by spending on non-durable goods and services, with smaller increases in spending on durable goods and human capital. We observe no significant effects on degree attainment, though the magnitudes of the estimated effects generally appear larger among younger participants. Measures of subjective well-being are higher among treated participants in the first year of the transfers but then revert to control group levels. Overall, our results suggest a moderate labor supply effect that does not appear offset by other productive activities.

Not surprisingly, giving people cash allows them to work less and spend more on leisure.

In The Atlantic article, Drs. Richterman and Thirumurthy briefly mention that guaranteed income pilot programs in the U.S. “haven’t delivered the dramatic health improvements associated with cash-transfer programs elsewhere.” Without mentioning the large-scale randomized controlled trial described above, they (falsely) assert that these programs have “seemingly produced only modest health gains in the United States.”

But don’t worry. Even when the evidence is against them, these scientists are so smart that they still know they are right. They offer a variety of hypotheses, unsubstantiated by research, to rationalize why money-cillin has not produced the results they expected in the U.S.

First, they falsely describe past efforts as providing too little money to make a difference: “a few hundred dollars a month for a relatively short period of time, typical of guaranteed-income pilots, rarely matches the steep costs of housing, child care, and health care.” The experiment described above provided $1,000 per month for 3 years.

Second, they suggest that cash transfers are insufficient given deeper societal problems: “In the U.S., the dominant health problems are chronic diseases shaped by neighborhood environments, structural inequities in housing and health care, and years of accumulated risk from unhealthy diets and other long-term exposures. These problems are far less responsive to short-term financial boosts. Cash can reduce stress and improve stability, but it cannot, on its own, undo the deep roots of these conditions.”

But if this were true, then it would be a pretty convincing argument against expanding cash transfer programs. Short of revolutionary change, simply providing money would make no difference. Along these same lines, they argue that cash transfers won’t work unless millions get them: “U.S. pilots have been small, reaching only hundreds or thousands of families—too limited to change the broader conditions that shape health outcomes.”

Third, they suggest that cash transfers need to be conditioned on other behaviors to be effective: “cash works best when it is woven into social infrastructure that families already rely on. In many low- and middle-income countries, payments are linked with health visits and other essential services.”

But if that were true, then they are really advocating for traditional welfare programs that place conditions on support rather than arguing for cash transfers or guaranteed income programs. And even though they have no evidence to support this hypothesis, any success of conditional cash transfers would leave it unclear whether the cash or the required behavior was producing any benefits.

After offering all of these rationalizations (and falsehoods), they point to food stamps to prove that expanding cash transfers is essential for improving health outcomes. But they have no experimental evidence—the kind typically required for approval of a new drug—-to support this claim and instead point to observational evidence reliant on complicated and opaque research designs to draw this conclusion. The problem with this kind of evidence is that the results are highly sensitive to researcher choices about research design and model specification, making them very easy to manipulate. When presented with conflicting evidence, we should believe the results of experiments over observational studies.

Arguments like those made by Drs. Richterman and Thirumurthy in this Atlantic article are the equivalent of advocating for communism because it’s never really been tried. It reveals an ideological commitment that is impervious to empirical evidence or scientific examination. Medical research and education should not be wasting time on money-cillin and should instead be focused on scientifically backed medical interventions that doctors can use to help their patients.

Evidence Lacking for Claim That the Stress of Racism Shortens Lives

/by

If researchers produced a study finding that poor and minority people tend to be more likely to have health problems and die at a younger age, it probably wouldn’t be published in a leading medical journal or covered with articles in national newspapers. It would rightly be seen as a restatement of the well-known, sad reality that for a variety of reasons poor and minority people tend to have worse diet and exercise and are more likely to use drugs and alcohol, contributing to worse health and earlier death.

But if researchers relabel the problems poor and minority people experience as “cumulative lifespan stress” and suggest those problems are the result of “systemic and explicit discrimination,” those same banal observations can earn a spot in one of the American Medical Association’s top journals and be covered in The Washington Post under the headline: “New evidence shows how discrimination shortens lives in Black communities.”

To be clear, the study published in JAMA Network Open does not demonstrate in any way that discrimination shortens lives in black communities. All it does is show that five measures, which they combine and call “cumulative lifespan stress,” are correlated with indicators of inflammation and are also correlated with dying younger. They also observe that black subjects scored higher on the index they called “stress,” had higher measures of inflammation, and also tended to die at an earlier age. The study’s research design does not allow them to identify whether the five measures they combine and label as “stress” caused inflammation or earlier death, nor can their study exclude whether other factors that they did not examine could have caused both the measures of inflammation and dying at a younger age.

Let’s consider the five measures the researchers use as an index for the physiological stress over one’s life to see how weak the study’s research design is. To capture this cumulative lifespan stress, researchers surveyed study participants to collect information on “(1) childhood maltreatment[…], (2) adult lifetime trauma exposure[…], (3) researcher-verified stressful life events[…], (4) discrimination[…], and (5) indices of socioeconomic status.”

The researchers combine these five measures into a single indicator that they call “cumulative lifespan stress,” but it is far from clear that these five measures actually capture physiological stress. In fact, many of these five measures include information on health problems or factors that could contribute to health problems. For example, the survey used to capture “adult lifetime trauma exposure” includes measures of whether subjects had “experienced a life threatening illness,” “experienced a miscarriage,” and was involved in an accident or otherwise received a serious injury. The measure of “stressful life events” includes information on serious illness or injury and whether a close relative had died.

These health challenges may be stressful, but it would be highly misleading to conclude that the stress associated with serious illnesses caused people to die at a younger age as opposed to the illnesses themselves. The researchers never control for the actual illnesses that subjects have when examining the correlation between their “cumulative lifespan stress” measure and the probability of early death. A subject could have chronic diabetes, uncontrolled blood pressure, or cancer and the researchers would conclude that they died of stress rather than these various diseases.

It is also important to note that only one of the five measures that they claim capture stress includes indicators of discrimination. And that measure asks whether subjects believe they had been treated “unfairly” in employment, housing, or other matters for a variety of reasons, only one of which is race. To conclude that this information, which is part of one of five measures that collectively are associated with early death, means that “discrimination shortens lives” would be completely irresponsible.

The reason this shoddy research receives such favorable treatment by a leading medical journal and alarmist coverage from national newspapers is that people wish to advance a political argument blaming racism for higher rates of health problems and early death in the black community. But nothing in this research demonstrates societal discrimination is to blame. By failing to control for the health challenges associated with diet, exercise, and alcohol and drug use, and by falsely relabeling reports of serious illness or risks of getting serious illnesses as “cumulative lifespan stress,” the study is attributing to racism what could easily be explained by medical comorbidities, individual choices, and community dysfunction.

If you are wondering who is paying for this shoddy research, the answer is you are.

Taxpayers funded this research through grants awarded by the National Institute on Aging, the National Science Foundation, and the National Institute on Alcohol Abuse and Alcoholism. The last source of funding is particularly ironic since the study did not examine the obvious possibility that alcohol abuse could be part of the explanation for the results they observe. It’s bad that the American people must be falsely blamed for causing their black neighbors to die because of stressful discrimination, but even worse that they have to pay for such chicanery. Perhaps paying to be falsely blamed is also dangerously stressful.