UPDATE: For a follow-up post on uncertainty regarding an underlying causal mechanism related to gains in mental health, see this post.
I stand corrected. It’s been a week since the Oregon Medicaid study’s two-year results were published, and there hasn’t been much discussion in the media or among policy bloggers about what a big deal the mental health findings are—even though there’s an extensive literature documenting how detrimental depression is to physical health and productivity. Even though Jonathan Gruber, who is among the nation’s leading health economists (and co-authored the study), called the results “astounding.” Fine, I’ll give it a shot.
First, the study results. Medicaid coverage was associated with positive depression screening declining by nearly one-third—that is to say, the control (uninsured) group screened positive for depression 30% of the time, but the rate among those who enrolled in Medicaid through the lottery fell to about 21% [1, 2]. This was statistically significant at a p-level of 0.02. The chart at right is from the Oregon study authors; I added a red box around the relevant clinical results.
Why were the depression screening results statistically significant when others weren’t? Depression was “by far the most prevalent of the four conditions examined.” The observed sample of patients was larger, and the effect was huge. Voilà! Stunning results with statistical power.
Depression is seriously detrimental to everyday functioning. Poor mental health has a measurable impact on “functional disability,” a metric that accounts for days in bed and days where an individual suffered activity limitations due to illness. One of the earlier studies I looked at (from 1992) examined trends in disability levels over a year for subjects who were high-utilizers of the health care system. The authors found that “persons with depressive symptoms that did improve showed large reductions in levels of disability during the 1-year follow-up.” Subjects who improved from severe depression had a 45% reduction in disability scores while those with moderate-improved depression had a 40% reduction. And this is in one year! There was no improvement in functional ability among subjects whose depressive symptoms remained the same over the course of the study.
Depression is also an independent predictor of mortality. In laymen’s terms, the depressed are more likely to die younger, even when we control for other factors that contribute to ill health. In a six-year study of older adults (in whom it’s easier to study mortality), depression was found to increase the risk of mortality by 24% . Depression did not need to be severe (“major depression”) for this to be true—mild and subclinical depression also showed an effect. This is consistent with past literature suggesting that depression results in “vital exhaustion and decreased emotional vitality”, which contribute to functional decline. The authors also argue that studies using self-report measures to identify depressive symptoms (like the Oregon study) are likely to underestimate true prevalence of the disease.
Employers benefit financially from reduced depression among employees. A cost-benefit analysis found that enhanced depression care (a one-time screening plus telephone management as needed) yielded a net savings to employers of $2895 over five years. This is because the cost of lost productivity and depression-associated medical treatment outweighed the cost of offering screening and follow-up management. “But this is about employers,” you might protest, “and Medicaid is not employer-sponsored insurance.” That’s true. But it’s also true that three-quarters of the uninsured are from working families; don’t forget, only about 68% of employers offer health insurance. Many of the working uninsured would qualify for the Medicaid expansion, which—suggested by the Oregon study’s robust findings—ought to reduce prevalence of depression in this population, easing burdens on our health care system and the economy.
Depressive symptoms—even modest ones—sensitize the body’s inflammatory response. I’m not the blog’s token medical student (that would be Karan), so bear with me. The authors helpfully sum up clinical effects of depression:
Depression is associated with enhanced production of proinflammatory cytokines that influence a spectrum of conditions associated with aging, including cardiovascular disease, osteoporosis, arthritis, type 2 diabetes mellitus, certain cancers, periodontal disease, frailty, and functional decline.
The study goes on to test the effect of depressive symptoms on the body’s reaction to something pretty common: the flu virus. Sure enough, investigators found higher cytokine levels (cytokines are molecules that help regulate immune responses) among those who exhibited depressive symptoms, even though the number of depressive symptoms was low before exposure to the virus (and did not change after exposure). Two weeks out, people who reported more depressive symptoms still had higher cytokine levels. The authors argue that depression could have maladaptive effects on the immune and endocrine system, leading to prolonged infection, chronic stress, and added health risk.
I’m not trying to argue that supporters should try selling the entire expansion on its mental health merits alone—I happen to agree with those commentators who provide evidence that the study was far too underpowered in the physical health domains to accurately spin its results as some damning indictment of the Medicaid program. It’s hard not to agree with them; we’re talking math, not politics. My point here is that the mental health benefits are a hugely important, and they deserve more than a passing mention when we talk about the study.
There’s another thing about the “health insurance doesn’t make us healthier” argument, if you buy that: something creates health disparities across the socioeconomic gradient. If we don’t need more insurance, then we definitely need stronger social policy to help our most vulnerable (and I’m certain you’ll find many who argue we need both). That’s a choice—but I’m not sure it’s one that opponents of the Medicaid expansion are willing to make.
1. Technical note on the methodology: depression was assessed by eight questions in a self-reported survey. This might sound tenuous to those inclined to be skeptical, but a two-question version of the measure has actually been empirically validated as an effective tool for depression screening.
2. Curiously, the decline in depression did not correlate to increased use of medication. This suggests that the financial security of having insurance has potential to improve mental health. Depression medication use did increase, but not at a statistically significant level (p = 0.07; it needed to be 0.05 or smaller). Even if it were significant, the change in medication was substantially smaller than the change in positive screening for depression (raw changes were +5.5% and -9.2%, respectively).
3. For the people who are interested in such nuance, the study offered relative risk calculations that controlled for sociodemographic factors (RR, 1.43), prevalent clinical disease (RR, 1.25), subclinical disease indicators (RR, 1.35), and biological/behavioral risk factors (RR, 1.42), all statistically significant at a 95% level. I’ll add the caveat that this study was conducted with subjects over the age of 65—that is to say, the Medicare-eligible population, which could impact the generalizability of the findings to the prospective Medicaid population, but I still think they were meaningful to establish general trends._____________________________
Adrianna works in clinical research and is a graduate student in public policy & public health at the University of Michigan. Follow her on Twitter @onceuponA.