Reconsidering the Link Between SES and Health in Whitehall

The Whitehall studies followed two cohorts of British civil servants over several decades and found a strong and steep gradient between higher occupational category and a range of mental and physical health outcomes. Much of the literature on Whitehall focuses on how social status in adulthood predicts occupational prestige and autonomy, which are plausible mechanisms that could transmit stress and lead to worse health. These mechanisms, Michael Marmot argues, are more important than the role that health plays in determining social position.

In a 2011 paper (here as a 2010 working paper), Anne Case and Christina Paxson take a different approach to the Whitehall data. They make three observations that I believe should be garnering more attention.

First, they argue that the study is not ideally set up to study the impact of early health on adult health:

“We find that, because the population from which this cohort was drawn consisted almost exclusively of white collar civil servants, the Whitehall II sample is not well suited for quantifying the importance of childhood conditions for the population as a whole. Children from poor backgrounds who find white collar positions in Whitehall are different in many dimensions from other poor children, and these differences lead to a systematic underestimate of the impact of early-life health and circumstances on later-life health and social status.” (Pg. 2)

They quantify the potential bias in a clever way. They take the two largest cohort studies from the general population in Britain: the NCDS 1958 Cohort and the BCS 1970 cohort, and examine how much socioeconomic status in childhood (measured by father’s occupational category) predicts adults height, first in a full sample and then in a sample restricted only to social servants. If social servants from low-status fathers have some unobserved differences that compensate for early health disadvantages, then we would expect the association to be weaker between parental education and height. Indeed, this is what they find.

Second, focusing on the Whitehall II cohort, they show that childhood socioeconomic status (father’s occupational status, and family car ownership during childhood) has a lasting impact in adulthood both on initial occupational category, and also on subsequent promotions. Not surprisingly, much of this effect is mediated through education. Importantly, however, they find that early health exerts a strong independent effect on these same outcomes (captured by height and whether the person was ever hospitalized before age 16). These are important results, especially since as the above results indicate, the Whitehall study is likely to be positively selected.

Third, these early health effects also exert an independent effect on adult health. For example, in regressions that control for a variety of current and previous characteristics, spending time in the hospital as a child is associated with a 7 percentage point lower likelihood of being in “excellent” or “very good” health as an adult. Controlling for current occupational grade does not dampen the association between child health and adult health. To try to disentangle whether occupational grade is high because health is good, or whether health is good because occupational grade is high, they also estimate regression models predicting current health on future occupational grade using earlier waves of the survey. Future grade is unlikely to cause current health, but they find that current health does predict future grade. Moreover, in fixed effect models, they find that current health is a much stronger predictor of future promotion than any previous occupational grades.

The authors conclude: “despite the downward biases that are likely to result from selection into Whitehall II, we still find evidence that health in childhood influences occupational status in adulthood. Adults who had better childhood health—as measured by adult height and hospitalizations in childhood—start at higher grades in the civil service, and are promoted to higher grades after they enter the civil service. The association between height and occupational status in adulthood is robust to controls for education, implying that childhood health does not operate solely through its effects on educational attainment” (Pg. 22).

If we take this study seriously, and I believe we should, it invites serious reconsideration of the apparent finding in Whitehall that occupational status causes health in adulthood. This does not mean that inequality in social status does not matter. To the contrary, the Case and Paxson results highlight the centrality of early life health exposures on late life outcomes. This is a different causal mechanism, and would result in a greater focus on interventions focused on poor children than on promoting greater workplace equality later in life as a means for improving adult health.

About Brendan Saloner

I am a postdoctoral fellow at the University of Pennsylvania in the Robert Wood Johnson Health and Society Scholars Program. I completed a PhD in health policy at Harvard in 2012. My current research focuses on children's health, public programs, racial/ethnic disparities, and mental health. I am also interested in justice and health care.
This entry was posted in Blog posts and tagged , , , , . Bookmark the permalink.

6 Responses to Reconsidering the Link Between SES and Health in Whitehall

  1. Really interesting post, and especially timely given the DoH consultation on health outcomes for children and young people that closes on Monday

    Best wishes,


  2. mel bartley says:

    Hi Ben and all
    You might be surprised for me to defend the Whitehall Study. That is not exactly what I will try and do. But this Case and Paxon paper is now rather old. They gave a version at a Whitehall II seminar at UCL about 3-4 years ago. Tarani Chandola has some very interesting observations on it, if you can persuade him to write about them, which will be much cleverer than mine. However, what we might want to distinguish here is between 2 versions of the argument. What C&P meant to say was that ‘unfit’ (in the darwinian sense) children to on to be low status adults. That is, their childhood health is both the cause of their adult health and of their adult social position. Classic ‘confounding’. In those days, fewer of the twin studies had been falsified by genome wide association studies, so people could still think that the cause of the childhood poor health might be genetic. So this meant that health inequality was a result of genetically inherited poor health. OK? even a few years later, we now know that complex characteristics like ‘health’ cannot be regarded as inherited. Nowadays we would put more emphasis on childhood health as a powerful measure of childhood social conditions (as Brendan says) rather than of genetic inheritance. Birthweight is associated with social class in the 1958 cohort study, but would one really regard that as causal? Birthweight is also associated with social class, and with childhood cognitive ability. But it is more strongly associated with cognition only in children with less socio-economically advantaged circumstances. Later in life, people who were born small are also more likely to get various diseases, but only if they are in less sociall advantaged circumstances. So what we have is a sinister complex in which disadvantage gives a bad start in the womb, and then magnifies the consequences of health problems that arise then or at any subsequent time. Posing an ‘either-or’ dichotomy is not at all helpful.

    • Hi Mel,

      Thanks much for this interesting comment.

      I don’t want to be non-responsive, but I’m not sure I fully grasp the point you are making. My understanding: Marmot argues that there is a robust effect of social status in adulthood on health in adulthood. The primary causal arrow runs from adult social status to adult health. C&P respond by showing that actually social status in adulthood reflects disadvantage in childhood, which is itself a product of growing up in disadvantaged environments. Fundamentally, adult health is caused by child health, and social status in adulthood is not the primary cause (you called it a confounder, I’m not sure if that’s how they would put it). This is an important contribution because it provides a plausible alternative to the “status syndrome” hypothesis that inequality makes those adults in the lower positions sick because of their lower social control and prestige.

      What do you think?


      • mel bartley says:

        Hi Brendan
        I wish I had a better recall of this paper. It caused quite a stir in the Whitehall II team at the time, and also among the economists who are friends of Case and Paxon. I actually do not believe that people in less advantaged socio-economic positions (SEP) as adults get sick just because of low control at work prestige. But as a person who has experienced different kinds of jobs, it seems to me plausible that being in a position where you cannot resist bullying and harrassment at work, for example, is very stressful (having been on the receiving end of this from no less a person than Marmot himself). I think that if Case and Paxon are clever (which they are, I have every respect for them) they will have by now changed their orientation. But at the time their working paper was going the rounds, the implication was very clearly that there is no point trying to improve the working conditions of people because their health trajectory was already determined in childhood (or even before). The difference if: do you think the trajectory was determined genetically or by early life environment? As a lifecourse researcher I would say the latter (also as a simple empiricist). So it is not that Marmot is wrong (and believe me I hold no brief for him whatsoever) totally. Rather that the association of work conditions with health cannot be attributed solely to what goes on in adulthood. If you look at what James Heckman is saying for some time now, we cannot ignore early life or wash our hands of life course disadvantage because more and more of the total US working population are born into migrant families and poor families. If we just write them all off and say, oh well, it is genetic unfitness that leads them more likely to prison than to a decent job, we will be short of necessary workers. And even indecent jobs: who will look after the elderly if an increasing % of the workforce are regarded as genetically psychopathic or whatever? I actually think this is very amusing. You can go along writing off 30% of the population on genetic grounds until you realise you might need them!

  3. Hi Mel,

    Absolutely — I agree that bullying and harassment at work should certainly make people’s health worse. I also think that it would be extreme to take the approach that nothing can be done for adults, so there is no use in intervening to improve working conditions. However, from a population health perspective, the research on the early life gradient should cause us to rethink the relative low priority that pediatric early intervention could have on later life trajectories. There is a whole bundle of highly effective things that we are not doing here in the United States to mitigate early life disadvantage (nurse visitors, improving safety in the home, better access to well child visits, etc.). So rather than posing this as an either-or, as you suggested, this research should be a call to action for more resources to children. If we could link this health gradient to a skill gap deficit (your final point), I think this would actually strengthen the argument for early intervention — many people that will not be successful in the labor market could provide many more services to society if they had better mental and physical health.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.