How Not to Think About Social Determinants of Health: A cautionary tale from Canada

In this guest post, Ted Schrecker critically discusses the results of a recently published public health study in Canada. Illustrated are the hazardous implications of de-contextualized conceptualizations of health.

In early April 2012, a flurry of news reports described a study of major health risks shortening the lives of people in the Canadian province of Ontario.  A typical report described “bad lifestyle choices” as together taking as much as seven years off Ontarians’ life expectancies.

As is often the case, the reality is more complicated.  The study, led by Ottawa researcher Douglas Manuel, was based on self-reports from 117,674 interviews in three successive surveys conducted by Canada’s national statistical agency (Statistics Canada) and record linkage of 99,929 of the respondents with their provincial health insurance records.  “The primary risk factors of interest were smoking, alcohol consumption, fruit and vegetable consumption, leisure physical activity and stress.”  A procedure known as a multivariable Cox proportional hazards model was used to estimate the life expectancy reductions associated with these “behavioural risks,” to use the language of the study report.  I’ll return to the problematic nature of this language.

In the full text of the report, the authors are commendably candid about limitations related to possible under-reporting of health risks in the survey on which the report was based – limitations, in many cases, related to the simplistic nature of the survey questions.  For instance, the survey asked only about leisure time physical activity, not about activity related to work or quotidian errand-running.  The conclusion that Ontarians’ combined exposure reduced overall life expectancy in the province by 7.5 years is no doubt statistically robust, given the data on which it was based.  Unfortunately, it’s also a ‘so what’ kind of conclusion.  The authors of the report are remarkably unreflective about their focus on proximate risk factors, ignoring the contextual influences that shape individuals’ opportunities to lead healthy lives.

It’s not as if the world just found out about the limitations of risk factor epidemiology, as Courtney McNamara noted in a previous posting.   She emphasized Link and Phelan’s excellent work on “social conditions as fundamental causes of disease,” and  in its 2008 report the Commission on Social Determinants of Health went to great lengths to foreground the “structural determinants and conditions of daily life” that “are responsible for a major part of health inequities between and within countries.”

None of these insights was incorporated into an interactive life expectancy calculator, based on the study findings, that invites Ontarians to respond to a set of questions that are then used to generate an estimate of how long they can expect to live.  Participants are asked such questions as how many servings of fruit or vegetables they ate in the past week; how many were potatoes; and whether any carrots were involved.  Other questions relate to leisure-time physical activity, and a strictly dichotomous question (like that in the original survey) asks whether most days are stressful.

Workers locked out at the Electro-Motive plant, London, Ontario. Photo: CAW Media; reproduced under a creative commons licence.

Now, if you are paying market rent for housing while living on the sub-poverty social assistance income provided by the province of Ontario, having first exhausted almost all your assets as a condition for eligibility, eating the healthy diet defined by the provincial health ministry is an arithmetic impossibility in much of Ontario.  Under these circumstances, hearing a $170k/year researcher and his team tell you that you should eat your carrots and have less stress in your life is not especially helpful.   If you are trapped in a low-wage service sector job, or have lost your factory job because the employer locked you out after you refused a 50 percent pay cut and then moved production to a lower-wage jurisdiction, which recently happened to workers at the Electro-Motive plant in London, Ontario, hearing a $170k/year researcher and his team tell you that you should eat your carrots and have less stress in your life is not especially helpful.  (Were I in such a situation, class warfare is the first phrase that would come to mind.)  And after chasing across town on the bus to shop the specials that are their only chance at a healthy diet, or turning as a last resort to the food banksthat have become an established feature of life in the province (not a lot of carrots there), how many Ontarians living on the margins have time or safe opportunities for “leisure physical activity”?

The researchers’ isolation from the conditions of daily life is revealed with special clarity by their treatment of stress as a variable somehow within the control of those experiencing it.  (In The Status Syndrome, Michael Marmot is eloquent on the weaknesses of this presumption.)  Before giving one more interview or writing one more grant proposal, all those involved with studies like the one just released in Ontario should read, carefully, Barbara Ehrenreich’s account of life in the low-wage service sector.  Nickel-and-Dimed is available both as a book with multiple secondhand sellers and as a feature in Harper’s Magazine, so readily accessible.  This view of stress is of course even more pernicious in contexts outside Canada: for example Spain, where unemployment is now over 23 percent in the aftermath of a financial crisis that began in the United States, or the United Kingdom, where housing benefit caps are driving poor families out of central London or into homelessness even as tax rates on the ultra-rich are coming down.

At stake here is a vital instance of what Kristin Shrader-Frechette and Earl McCoy have called a methodological value judgment.  Perhaps without being aware of it, Dr. Manuel and his team chose an approach that reinforces the neoliberal tendency to privatize risk and responsibility.  Conditions like poverty and ill health are ascribed to the choices and failings of individuals who have little control over macro-scale processes like financial crises and the relocation of production to lower wage jurisdictions, rather than situated with reference to radical inequalities in the opportunity to lead a healthy life.  The Commission on Social Determinants of Health understood the pernicious nature of this approach, implicating “poor social policies and programmes, unfair economic arrangements, and bad politics” as pathways to health inequity and structuring one of its three overarching recommendations around “the inequitable distribution of power, money, and resources.”

The most widely agreed-upon axiom of medical ethics is: first, do no harm.  Studies and derivative knowledge transfer exercises that neglect structural influences on health fail this basic test, for they are far from harmless.

8 thoughts on “How Not to Think About Social Determinants of Health: A cautionary tale from Canada

  1. Okay, you addressed the diet and exercise concerns in the study, but what about smoking and drinking? If I work in a “low-wage service sector job” does that give me carte blanche to smoke a pack a day? Seems you’re throwing the baby out with the bathwater here.

  2. This is an excellent, pithy and well argued critique of the tendency to focus on lifestyle factors when talking about the determinants of health. For the media, “Eat your vegetables” is so much easier to convey than the structural factors causing poverty.

    Well done.

  3. The life expectancy calculator does ask for postal code in order to adjust for geographic location. And isn’t geographic location associated with one’s SES?

  4. A few words in defense of this research. Epidemiological research using administrative databases is inherently limited, as the authors noted, so all that can be assessed is data that were already collected for other purposes. These researchers are maximizing the use of this existing data (in a very cost efficient way) to address the EXTENT that these particular factors have on the outcome of death. They choose to list these factors such as stress as ‘modifiable’ because they have the POTENTIAL to be modified. This is not to say that individuals who are experiencing stress are responsible for it, but that this is something that could potentially be amenable to intervention, whether at an individual level by relaxation techniques or at a social level by addressing underlying issues such as reducing poverty and improving work and other social conditions. These findings indeed pave the way for future research to address these underlying issues. The authors should not be painted as ascribing themselves to a particular political viewpoint for this classification of risk factors, but instead as using the standard term in the field for factors that could be something that we can change (as supposed to something that we can’t currently change at the moment, like our genetics). Once they have identified that these factors are important for longevity and quantified the extent of their effect, future researchers can follow up (with different methods) to determine HOW to address these issues, such as by tackling poverty. Then, health economists can assess how much we would save (financially and in terms of lives saved) as a nation to address these health issues through policy aimed at prevention rather than allowing diseases to develop and then treating them (most of our money currently gets spent here). Then a national strategy can ben rolled out and other researchers can assess the impact of its effect so they can continue to improve its delivery. My point is, no one can study everything in one study. It takes collaboration and lots of small steps like this to make health progress as a nation.

  5. Indeed a “pithy and well argued critique,” as stated by Fafard above. I
    will add ‘convincing’ to this pair of adjectives. Neglect of social and
    political organizers of ill health and suffering in any research is
    troubling and not useful. Period. Schrecker highlights several assumptions
    built into the report he examines, rightly emphasizing how and why such an
    account is far from “harmless.” Importantly, he gestures toward what is
    more broadly problematic in the family of research of which this report
    emerges. I look forward to a read of the Ehrenreich book referenced, for
    the same reasons I engaged with DeVault’s Feeding the Family, 1991.

  6. Interesting comments. A few thoughts in reply:

    To Mike: I don’t view it as the prerogative of health researchers or health promoters to “give” members of the working class “carte blanche” to smoke a pack a day or do anything else. I do view it as a core competency, and an obligation, to work for such policy changes as living wage ordinances and economic policies organized around construction of affordable housing and mass transit, rather than the home renovation tax credits for the comfortable that were offered by our Conservative government here in Canada.

    To Sarah: That’s an interesting point, and one that I had not considered. If anything, it makes the message of the exercise more pernicious, as it would appear to relegate SES to the category of non-modifiable risk factors.

    To Lindsay: I appreciate the limitations of research using administrative datasets, but my reading of the life expectancy calculator is not that it is oriented towards policy change. It begins with the message that: “This tool is designed to help you understand how certain behaviours like smoking, alcohol, food, exercise and stress level can affect life expectancy.” There is nothing at all about economic insecurity, neighbourhood effects, etc. It would be possible to design a life expectancy calculator that would emphasize elements of the public policy context, but that is not what the researchers in question chose to do.

  7. I agree with Ted in the need to include structural factors into research, but I don’t understand the anger towards this particular research team. It’s, for example, completely unnecessary to post their salaries.

    Even though the point is well made, the tone just makes you want to defend the research team.

  8. A very belated reply to Sebastián: the issue is not “anger,” but rather an insistence on the need to consider the social situation of researchers whose work tends to legitimize neoliberal presumptions about the location of responsibility for (ill) health. This insight about the lack of “Archimedean positions” from which to view the world is well established in feminist theory. See, for instance, Myra Jehlen’s “Archimedes and the paradox of feminist criticism,” Signs 6 (1981): 575-601.

    As an aside, the information about the salaries of the research team was and is not in any way private or privileged. Since circa 1996, the salaries of all employees in the broader public sector earning more than C$100,000/annum have been published on an annual basis under an act of the Ontario Legislative Assembly. As they damned well should be, say I. Indeed, I would like to see provisions in codes of research ethics mandating researcher salary disclosure in all cases where research involves participants from economically marginalized or subaltern populations. This would make the process of informed consent somewhat more meaningful.

Leave a Comment

Your email address will not be published. Required fields are marked *