As most readers of this blog well know, health disparities of various kinds are rampant in the U.S. — in obesity, infant mortality and morbidity, cardiovascular health, and many other areas. In most cases, however, we seem to know more about the extent of health disparities than we do about what causes and what is most likely to ameliorate them.
To rectify this situation, we need to conduct research — and lots of it. Typically, however, health disparities research will have to occur with the same populations who are most likely to be considered vulnerable and in need of extra protections from research. Often, moreover, health disparities research will need to occur in the clinical setting (as opposed to the lab), where patients normally rightly expect that everything done there is designed to serve their individual best interests, rather than to produce generalizable knowledge. Health disparities research might involve research methodologies that are relatively unfamiliar to IRBs, such as community-based participatory research (CBPR), which blurs the traditional distinction between investigator and subject on which the regulations are built. To the extent that disparities are thought to derive from provider discrimination or bias, researchers may face difficulties from a research review system that is designed to protect all “subjects,” including professionals who are incompetent or worse. Eventually, health disparities research scales up to multiple research sites, which usually requires approval from multiple, often conflicting, IRBs. Many interventions to address health disparities, finally, will take the form of public policy rather than clinical treatment. If we want such policies to be evidence-based (and we should), they will have to be tested, perhaps in ways that raise legal or ethical issues (say, randomizing a state’s Medicaid recipients to receive or not receive particular benefits, or randomizing the businesses in a jurisdiction to be required to display nutrition information on the food they sell — or not).
I’m delighted to have received so many comments, both on- and offline, on my last IRB post from those with experience in the research trenches. As I begin a new project along these lines, I would be very interested in hearing again from both researchers and research reviewers with experience in health disparities research, whether you have struggled with these or similar issues (or have abandoned research plans at least partly out of fear of such problems), or have experienced smooth sailing. Feel free to leave comments here, anonymously if you wish, or contact me directly at mmeyer at law dot harvard dot edu. Many thanks in advance.
Hello Michelle,
First, I do not agree with your premise that “we seem to know more about the extent of health disparities than we do about what causes and what is most likely to ameliorate them.”
I would argue that in fact we know an enormous amount about the causes of health inequities, as well as the interventions that are most likely to ameliorate them. Although the social epidemiology that comprises the bulk of the evidence base on this has gotten rich and sophisticated in the last fifty years, and especially in the last several decades, 19th c. public health reformers like Virchow, Engels, and W.P. Allison understood quite well Link & Phelan’s notion that social conditions are fundamental causes of disease and its distribution in human societies. For a good review of this evidence, I would check out Nancy Krieger’s recent book on the need for a theory of epidemiology (she goes into the history and the evidence, a substantial portion of which she has helped produce).
We also have a tremendous amount of high-quality data suggesting the categories and specific cases of interventions that work to ameliorate health inequities within and between countries (unsurprisingly, most of these address the macrosocial structural causes of adverse health and gradients in health).
While I do little empirical work at the present — but have plans to do more in the future — I am a health inequities researcher and I also tend to disagree with your second premise, that “to rectify this situation, we need to conduct research — and lots of it.” I am not averse to the production of knowledge, but at this point we have more than sufficient evidence on the causes and remedies of health inequities in both global North and global South to take action. Iona Heath remarks on this in a wonderful 2010 article entitled “Crocodile Tears for Health Inequality.” (https://www.bmj.com/content/340/bmj.c2970).
To rectify the problem of health inequities, we need to take action on the social determinants of health, and ameliorate material deprivation, structural violence, oppression, etc. More research on the subject is welcome, but it is not strictly necessary to immediate and vigorous action intended and justifiably believed to compress health inequities.
On your substantive question, I think the largest issue facing IRBs regarding health inequities research is that most of the best empirical data generated on the subject comes from cohort studies, which are obviously not RCTs and do not in fact evaluate interventions — and even if they did, the interventions would most assuredly not be clinical interventions, but would rather be social, mostly in the realm of policy, as you note. The latter has been done, and done well, but I share your view that IRBs are largely unfamiliar with this kind of research.
All of the above JMO, of course.
I note this phrase in the post: “Typically, however, health disparities research will have to occur with the same populations who are most likely to be considered vulnerable and in need of extra protections from research.”
I don’t know whether this is a description of what IRB’s think is the case, an interpretation of the Common Rule, or the author’s opinion. I do believe, however, that the assumption that social science and epidemiological research about the losers in health inequity “typically” poses little or no irremediable risk to the populations studied, and that the presumption that research is a harmful thing (and indeed the concept of “vulnerable populations” itself ) is an ideal ideology for minimizing research into what makes people vulnerable and how to change it. Research, one feels more and more compelled to protest, is a good thing, not the selfish personal hobby or business of a group of disengaged eggheads. (That said, I agree with Daniel that we have plenty of evidence for action already.)
Scott Burris
Let me respond to both Daniel and Scott in one comment. Scott writes:
Scott, you and I are, as they say, in violent agreement on this score, as I think my scholarship suggests. So no, the phrase you quote from my post doesn’t reflect my opinion (quite the contrary).
Nor, though, does it reflect merely “an interpretation of the Common Rule” or “what IRBs think is the case.” The problem, I think, is more serious than that. As you know, federal regulations (45 CFR 46 subparts B, C, and D) greatly restrict the kinds of research that can be done involving prisoners, children, and pregnant women. And the nearly universally-adopted Common Rule (45 CFR 46 subpart A) requires IRBs to “be particularly cognizant of the special problems of research involving,” and to ensure “additional safeguards” for, “vulnerable populations, such as children, prisoners, pregnant women, mentally disabled persons, or economically or educationally disadvantaged persons” (46.111(a)(3), (b)).
This is just one set of issues I’m delighted that we’ll be exploring in the “Legal Experimentation” workshop in which you’ll be participating, along with, inter alia, a representative from the Abdul Latif Jameel Poverty Action Lab (J-PAL), which has been doing tremendous work in developing evidence-based approaches to global poverty and other issues.
As for “hav[ing] plenty of evidence for action already” — a sentiment both Scott and Daniel express — I think this research/action dichotomy is largely false. Indeed, the very conceptualization of research and “action” (or practice) as two separate, linear events is a big part of what I want to question in the workshop. I don’t suggest that we need to await further evidence before acting to address health disparities. Rather, we need to integrate research into ongoing practice settings, and establish feedback loops so that our evidence base continues to grow and strengthen and to inform the work of “practitioners,” by which I mean clinicians, public health practitioners, and law- and policy-makers.
As to Heath in particular: She complains that academics spend too much time on the “documentation and discussion of health inequality” rather than “find[ing] the collective will to do something about it.” She wants us to stop spending time creating pretty graphs of health disparities broken down by jurisdiction or demographic and instead to do something about them. So far, I don’t disagree; the research (and elimination of impediments to it) that I (and many, many, many others) are calling for is not research to determine whether health disparities exist (although some inferences of poor health in racial and ethnic minority populations based on generalizations from white populations has been problematic and needs to be further examined), but to rather to determine, in the real world of fixed resources, the most cost-effective ways of redressing various disparities in various subpopulations. This is not the analog of waiting for more data to come in before deciding whether global warming is real.
So, so far Heath and I are on the same page. “[S]olutions,” she continues, “have remained stuck at the level of complacent rhetoric for far too long, the facile assumption being that once health professionals have been properly informed about the extent of health inequalities, they can somehow be expected to reverse them. Much the greater part of the determinants of health inequality are and will remain beyond the control of health professionals . . . .” Again, so far, so good. Social determinants of health; got it.
But then Heath unveils her proposed solution: academics must “campaign[] vociferously for more progressive taxation,” bringing the UK (and, presumably, other countries marked by inequalities such as the U.S., and eventually the world) as near as possible to “the case of ‘perfect equality’ when everyone gets the same share of income,” as measured by the Gini coefficient.
So Heath says we know enough to know that health disparities exist (again, I agree), then proceeds to assume, rather than argue for, both a singular cause (socioeconomic inequalities) and a singular solution (progressive taxation) to them. But:
(1) Much of “the scholarly exploration of health inequality” — which she disparages as “an industry” that “seems to result in the paradox of the poor directly subsidising the more affluent” — suggests that, in fact, many health disparities persist even after controlling for income. Some of that research even finds an inverse correlation between poor health (or, like obesity, an imperfect but still meaningful surrogate for it) and measures of SES like income or education.
Let’s stick with obesity as an example. According to the CDC (pdf), consistent with conventional wisdom about health disparities, higher income women are less likely to be obese than low-income women. But most obese women are not low income. Further, in men, obesity prevalence is generally similar at all income levels, and in non-Hispanic black and Mexican-American men, those with higher incomes are more likely to be obese than those with low income. Similarly, women with college degrees are less likely to be obese than less educated women. But there is no statistically significant relationship between obesity and education among men. And for the past 20 years, the prevalence of obesity has increased in adults at all income and education levels.
So, while no knowledgeable and fair person would doubt that economic inequality plays a major role in health disparities, it isn’t the only factor. Other likely contributors include provider bias (intentional and implicit), individual patient behavior and the psychology that underlies it, culture, religion, and genetics. If you ask me, Heath doesn’t take the social determinants of health seriously enough. There are multiple social determinants of health that can’t all be reduced to income (or any other single variable). And, inconvenient though it may be, these many contributors sometimes interact with one another in complex ways that we don’t always fully understand. In short, most health disparities are the result of a more complex story than the one Heath tells, and it’s not clear why we should only pay attention to one strand in that narrative.
(2) And point #1 above assumes that redirecting academic efforts from scholarship to vociferous campaigning will substantially address even that chunk of health disparities that results from economic inequities. But there are reasons to be skeptical about that assumption. For one, scholarship and advocacy involve overlapping but distinct skill sets, such that good scholars will not always, or even often, be the natural choice to “find the collective will” among voters for wealth redistribution. Given that reduction in health disparities is hardly the only goal of progressive taxation, one also wonders what new contribution health policy wonks can make to centuries-long, highly entrenched debates about tax policy and the welfare state. Indeed, if health policy scholars have no such new and effective method of ginning up votes, one wonders why academics “campaigning vociferously for more progressive taxation” won’t turn out to be, like “the scholarly exploration of health inequality,” just another way of “the poor directly subsidising the more affluent.”
In short, yes, health disparities exist. Yes, we don’t need more research merely documenting these disparities. And yes, the causes of health disparities almost certainly include many social, in addition to some biological and health care-related determinants. But no, significant increases in the redistribution of wealth — even if this were a politically feasible solution, and even if academics were the ones most likely to be able to bring it about — would not eliminate health disparities. In the real world outside of Sweden, and for the foreseeable future, we will have to do with what we have in the coffers, and what we have is limited. Under those circumstances, we need to know what policy interventions provide the biggest bang for their buck in reducing health disparities (and achieving other goals), and to do that, we need more research, and we need to integrate it into our practices and policies.
That’s a nice response, Michelle, and I do not think it is generally inconsistent with what I argued in my original comment. There is of course much that we do not know, and much that the production of knowledge could assist us with. And my sense is that most SDOH-ers are aware of the fact that income inequality is — however crucial and important — not the only SDOH (and in fact might be a proxy for myriad other SDOH).
But I do not think your response justifies your original claim that we know more about the extent of health inequities than we do about their causes. The fact that we have much more work to do on the latter does not imply that we lack sufficient knowledge on the causes of health inequities, or at least that we lack sufficient knowledge to justify action. I take your point re research and action, but would also argue that the dichotomy is more real in practice than in theory, because of the enormous evidence-practice gap that prevails in all paradigms of health policy, including public and population health. So while research can and must inform social action and vice-versa, even the best-supported interventions take forever to actually appear in practice. So there is a very real research-action gap in practice, even if the dichotomy in theory is false.
My basic point was simply to contest what I perceived as an instantiation of the research imperative in your original post — that reading might well have been a misinterpretation, and reading your response suggests we have much more room for agreement than I might have originally suspected.
I firmly believe that if we stopped doing research today we would IMO have sufficient evidence to take action via a wide variety of interventions that justify the belief that in so doing we could compress health inequities. This belief is not incompatible with the idea that further research would be helpful in clarifying which nodes of intervention would be most effective, or in illuminating promising pathways to a number of other important questions. But none of this establishes that we lack sufficient knowledge on the causes of health inequities to act vigorously, decisively, and immediately. Any failure to do so is, I submit, not the result of a lack of sufficient knowledge, but is rather more directly connected to structures of our political economies, and/or a basic lack of political will.
Thanks for the edifying discussion!