Michael Marmot

Quotes

I was a young doctor in Sydney and doing chest medicine at the Prince Albert [Hospital]. Peter Harvey, who was a friend of Ian Prior, had been involved in a couple of the migrant studies that Ian had done. He had gone to the Cook Islands with Ian and measured blood pressures. Ian Prior had a meeting in Wellington in early 1970 and invited Harvey, who was the chest physician in Sydney, to come. Peter came back from this meeting and said to me (I was in my second year as a junior doctor), “I’ve got just the thing for you. It’s called epidemiology.

I went to this meeting and there were these guys – Len Syme and John Cassel; doctors and anthropologists and statisticians – who all work together to try and understand why disease varies in different situations. And I’m sure it’s the right thing for you. Would you like me to write to them?” (2)

My own approach at that time was there was more to life than the major risk factors we knew about. I remember having discussions with people saying, ‘Why would I want to spend my research career saying what we already know is true?’ Why is that an interesting way to spend your life? (12)

[Later] I saw absolutely no contradiction between trying to put into practice what one knew about cardiovascular epidemiology and prevention and at the same time trying to do research on what one didn’t know. In a way we’re still having the same arguments.

I think progress has been made in trying to define psycho-social factors and trying to measure psycho-social factors. We all know they exist. The question is, can we define and measure them in order to study them scientifically? (14)

Osler, Stress, and Coronaries

Osler was interesting because he said that he thought that the causes of heart disease were work and worry which were “the lot and portion of the poor.” And he was, therefore, puzzled why high-status people should have more heart disease than the poor, given that work and worry were the lot and portion of the poor. When I started to look at the national statistics in England and Wales, allowing for various diagnostic changes, it did seem to me that there had been a clear change. In the Thirties up to the Fifties heart disease was more common in higher social classes and between 1951 and 1961 it changed. The rates of heart disease went up in people in the lower social classes and stopped rising as quickly and subsequently declined in people of higher social classes. So you’ve got a changeover. (6)

Something More than Cholesterol

So that clearly the change of population cholesterol was related to the change in risk. But to my mind there were other things going on. The question is, what is it about the change in culture? When I wrote my PhD and published it subsequently, it was something about being “brought up” Japanese. And the general hypothesis was the Japanese culture had stress-reducing devices, it was socially cohesive and supportive and didn’t have the same stressful nature individually or the intense striving as Western culture. What we showed amongst the California Japanese is that the more acculturated, the higher the heart disease. The more traditional, the lower the heart disease.

The number of Nissei was relatively small as I recall. It was about 12-15%, so mainly amongst the Nissei we looked at their cultural upbringing. Were they brought up to be Japanese? We looked at the degree to which they were currently integrated in Japanese society – if they went to a doctor or dentist or did they go to a Japanese doctor or dentist. Were most of their friends Japanese? And the people who were living a more Japanese life had lower heart disease rates. And, of course, one of the crucial questions was, was it because they were eating a less Western diet? And the answer was no.

We were able to control, for example, for plasma cholesterol. We had a measure of dietary acculturation that we controlled for. So we could say that among the California Japanese those who were more traditional in their culture was a lower heart disease rate. Those who were more acculturated looked more like white Americans in their heart disease rates. And there was still, I would say, a relatively small amount of social epidemiological work going on at that time. (10)

Something More than Coronary Disease

It was the same argument that Len Syme had been pursuing, that it cut across several disease endpoints. In that paper he quoted my work on the Japanese. He quoted some of the work on social supports. And what he was trying to say is that it contributed to host resistance so it cut across specific diseases. And that, of course, is still a problematic idea in medical science. The idea that you may be susceptible to a number of diseases.

But interestingly, when I got to Britain and started working on the Whitehall Study, the first paper we published on the differences by employment grade, using that as a measure of social position, showed the social gradient for all the major causes of death almost without exception. These were men (one knows that breast cancer doesn’t show the social gradient). Colon cancer is fairly flat. Leukemia doesn’t show the social gradient. But they are the exceptions. Pretty well everything else does.

So the idea that you’ve got this general phenomenon leads you to one of two directions. One direction, I remember talking to Richard Doll after we had done that, Doll said, and these are not words you spurn lightly, “To go down the route of saying that there’s some sort of general susceptibility factor is unscientific and I wouldn’t give it the time of day. What you’re going to find is that people of lower social class get bronchitis because of smoking, poor housing; they get lung cancer because of smoking; they’ll get other diseases because of infections; and others because of diet and you’ll find that that’s what accounts for this so-called ‘general phenomenon.’”

And the other route is to ask, why is there this general phenomenon? And look for the pathways, which we’ve been doing. Doll’s got a track record second to none so you don’t ignore his words lightly. I think the Doll idea leads you to ask the question, if it’s the case that there’s a number of specific causes of specific conditions, why do these cluster along the social gradient? So you ask it at sort of a social-economic-psychological level. If you think there’s some general susceptibility, you’d better ask at a biological level, why should an organ be more susceptible to neoplasms and heart disease and suicide? So you have to ask it at a biological level. And its really rather important to get straight which you think is most important. I don’t think the field has sorted itself out yet. (12)

John Cassel at the Origins

I mean the first thing he put together, I would say, is the idea that the nature of the society in which you live has a powerful influence on disease rates, specifically cardiovascular disease, but health and disease in general. Again, it’s easy to forget how radical an idea that was and is within medical science. Because within epidemiology tacitly we accepted it, whether we think its important because of smoking or dietary, exercise or other things. Tacitly we accept it. But I think that was very important to be able to say that clearly. So that was point one – that the nature of the society in which you live has a powerful influence on the disease rates.

Secondly, what he said, based partly on the migrants’ studies, partly on what had been happening in rural areas of North Carolina, is that when people are confronted with circumstances with which they are ill-equipped to deal, that is stressful and it increases disease risk. I didn’t think he used the words ‘control over life.’ I now would, and I do. I didn’t think he used that. But it was saying that when people are confronted with unfamiliar circumstances then that’s stressful and increases disease risk. He and Al Tyroler had done these studies in North Carolina of rural areas that had become urbanized. So it wasn’t just looking at migrants into the city but rural people where the city had migrated to them. So, that was the second thing that he said about being confronted with new circumstances.

His final paper, I don’t know if it was his very last one, but I think it was his Wade-Hamilton-Frost Lecture, which was published in the American Journal of Epidemiology… It was something like “The contribution of the social environment to host resistance” and that’s where he talked about social supports and the idea that the social environment contributes to host resistance. As I recall, I don’t think he did it, but, to my mind, when you think in those terms, then you realize at once that you don’t have to have any conflict at all. Exactly what Len was saying. You don’t have to have any conflict. If you ask why somebody gets cardiovascular disease, clearly they need first to get atherosclerosis. And to get atherosclerosis certain things have to happen, high plasma cholesterol predisposes, and so on. So there’s no conflict at all between a model that says why some people get disease and others don’t, or why some societies get disease and others don’t. The host-resistance idea and why you get heart disease. And its still going on. (15)

The Myth of 50 percent of Risk

And Robert Beaglehole has written this paper about the myth of the 50%. My argument is really pretty simple. I’ve done the calculation on WhitehallWhitehall data like Jerry Stamler’s done on U.S. data. If you could create a low-risk group with low cholesterol, low blood pressure, never smokers, and we did it on the bottom two quintiles of blood pressure, bottom two quintiles of cholesterol, never smokers. But because those risk factors don’t overlap… you’re dealing only with about 5% of the population. Now, theoretically, if everybody could have the risk of that 5% low-risk group, you could get rid of two thirds of heart disease. And I’m prepared to believe that theoretical calculation reflects a reality and that we really could get rid of two-thirds of heart disease if we could get everybody to the level of risk factors of about 5%. [ed. As we were seeing from the Seven Countries Study in the mid-Sixties. People have to deal with values found in Western culture before they will accept it. So this is a very good argument.]

But I would also say, given the little likelihood of getting everybody down to the low risk of that 5%, in most people with high risk, for example, we get this steep social gradient in coronary heart disease. Which isn’t due to differences in those risk factors. The high levels of risk were always due to those risk factors, but the differences between social groups is not. And that’s not a very complicated idea. But it seems to be the idea that you’ve got something setting the overall level of risk and something else accounting for the social gradient. So, that was John Cassel’s, to my mind, third important insight. (17)

Recent Converts are the Most Devout

It was interesting because Geoffery hadn’t been the least bit interested in social things. And, in fact, when I first came he said, “Look, I know you’re interested in these social things. We haven’t got much of it in Whitehall, we’ve got employment grade. Would you like to look at it?” And so I did. And we’re still in business!

And I think Geoffery was really quite intrigued by these factors. And he was open minded enough and sensitive enough to encourage me to pursue it.

HB: Did you know he was very late getting into the public health strategies of prevention? But he ended up his life as an international leader in prevention. I remember him telling me [in the early 1970s] “It is not the epidemiologist’s obligate responsibility to intervene.” By that time we were very involved in trials in Minnesota.

MM: That’s right. That was very much Geoffery’s view. Yet in his book that he wrote the year before he died, he said, “The causes of disease are social and economic and the remedies must also be.” So he moved from saying, I’m a doctor studying risk factors to saying the causes are social and economic. And I think the fact that we’ve found these social differentials in Whitehall had a big influence on him. He really did get quite interested. (18)

Get Poverty out of the Picture

But the third thing I think, particularly in Britain and less so in the U.S…people said, “We know poverty is bad for you.” But what we were showing is in a group of people who were not poor, British civil servants… there was a social gradient that wasn’t just poor people/high risk, non-poor/low risk. The second from the top had higher risk than those at the very top. And the third from the top had higher risk than those second from the top. And that was quite a challenge to understanding. And that is what’s kept me busy over these last two decades. (21)

Because, of course, the issue of the gradient is very difficult for people to think about. They can think about poverty but they don’t want to do anything about it. But they can think about it. They can understand poverty. But the idea of the gradient, why people who aren’t poor should have the risk of disease that varies with their position in the hierarchy is much more challenging. And as I say, that’s what we’ve been trying to understand these many years. If you catch me on a good day, I’d say we’re making progress. If you catch me on a bad day, I’d say we’re still floundering around in the wilderness. (25)

Proof

One of the problems with interventions, we’ve been pushed into. Until we’ve done intervention, maybe people won’t believe us. For example, from our own data, we think one of the important issues is how much control people have in the workplace. If you think it’s difficult to design a diet trial, think what it’s like to do a randomized control trial increasing control in the workplace. Go to one of America’s large corporations and say, “What we’d like you to do is randomly allocate people to different management styles.” Everything else remaining the same. And see how quickly you get shown the door. So it would be wonderful if one could organize proper interventions. But its not straight – forward. (26)

Does Alcohol Protect from CVD?

I wrote my first paper on it more than 20 years ago and I had heard Arthur Klatzky’s stuff at Kaiser about moderate alcohol being protective. Somebody told me about Pearl and the U-shaped curve. – I can’t remember now whether he actually used the term, but he wrote a book… and there was a copy of it in the London School of Hygiene Library dedicated to “Major Greenwood, my old drinking buddy.” I had remembered this thing about Raymond Pearl and alcohol showing the U-shaped relationship with mortality. His arguments were excellent, his evidence was very sketchy, because I went back and looked at the book.

In 1981 or thereabouts, based on the first Whitehall Study showing there was indeed a U-shaped curve, subsequently there was a whole slew of papers. As I said before that, Klatzky’s stuff and at Kaiser. The obvious question is firstly: are abstainers at higher risk for reasons other than the fact of abstention? Are they at higher risk because they are sick? Which was Gerry Shaper’s viewpoint. He didn’t believe that the U-shaped curve was due to protective factors of alcohol. So abstainers were sick. That’s why they weren’t drinking. They might be at high risk for reasons other than illness and the drinkers might be protected… It raises the question, if it were the case, and I stress if it were the case, that wine was more protective than other beverages, might it be because of the way wine is drunk with meals compared with the way other beverages are drunk? Might it be because wine drinkers are different, for example in Britain they are a different social class? Or might it be the wine itself? Now my view of the evidence… when I reviewed it in the early Eighties, was that at that time the evidence did not support a specific protective effect of wine, but it did support a protective effect of alcohol. I think the issue of confounding, and high risk for other reasons, has been by-and-large dealt with. And so I thought, moderate alcohol was protective.

Trends

People ask me why have heart disease rates gone down in the United States and Britain? How does that fit with your hypothesis about psycho-social factors? The real answer is, I can’t link them one-way or the other. It’s not contributory. It neither confirms nor denies the psycho-social hypothesis for two reasons. I haven’t thought the only cause of heart disease was psycho-social. The second reason is I never for one moment thought that the major cause of trends of heart disease was psycho-social factors if the other major risk factors change. And the Finns did as much as anyone in Europe to implement ideas about diet and smoking and so on, and brought their heart disease death rates down staggeringly. (46)

Have we actually contributed anything? I think that we have made a contribution by putting the socio-economic dimensions of heart disease on the agenda. We haven’t done that alone and that’s clear. You only have to look at the reference list of people who are currently writing about socio-economic factors to see a lot of other people have been involved. When we published on Whitehall at one level you could argue that there was nothing new here. People knew about this. The Black Report came out a year or two after we published our Whitehall data reviewing a whole range of data from national sources. So on one level to say there was a social gradient in heart disease was really not particularly new.

Any scientific contribution once put in context doesn’t look very original. I think, though, that Whitehall did put it on the agenda. That in a middle class population that were not poor, there was a social gradient in disease and particularly coronary heart disease, which people thought was quite different… And I think that did help focus the agenda. The second was to say that it was in part due to some of the risk factors we know about, smoking particularly among them played an important role.

But the second type of pathway related to psycho-social factors. And what we did, we took some of the things that other people had been considering, particularly related to the work environment and how much control people had at their jobs, and showed that that played an important role in the social gradient. It said that psycho-social factors were an important part in how much control you had over your life. Even still when I talk to people they’re mystified by this social gradient. It must be due to health care they say or it only must be due to poverty. And we say, “But we’ve shown it in a population who are not poor.” You’ve got all the evidence to suggest that that can’t be the sole explanation or even the principal explanation.

They’d say, “Then it must be genetic. It must be that the cream rises to the top. The genetically superior rise to the top and the genetically inferior sink to the bottom.” Which genes exactly are we talking about? Show me that genetic predisposition that leads both to high status and heart disease. And that’s where the animal experiments come in, where you look at the primates, where you can actually change their position in the hierarchy and change atherosclerosis.

So I think that against that background of cynical disbelief, that it must be health care or it must be poverty or it must be genetics, we come in and say its related to behavior and its related to psycho-social factors relating to where people are in the hierarchy. So I think that has been a contribution quite different from what most people in cardiovascular epidemiology were thinking about. (48)

Science and Core Beliefs

I wrote a paper in 1976 called “Facts, Opinions, and Affaires du Coeur,” in the American Journal of Epidemiology. I was sort of concerned then about some philosophy of science issues. How does science proceed? And I took as my case study cardiovascular epidemiology because that’s what I knew. And I think in a way I used it to reflect on fears about philosophy of science. So I was in a way vacillating between using philosophies of science to think about cardiovascular epidemiology and using cardiovascular epidemiology to think about theories for the philosophy of science.

I asked Jerry [Stamler] about 1997, “Under what circumstances would you be prepared to give up the theory that dietary fat was related to heart disease?” Jerry thought for a minute and said, “There were no circumstances” under which he would be prepared to give that up. And I thought, “Well, you see, this is no longer in the realm of science, this is religion.”

[ed. Because good science and theory can always be overthrown.]

To him this is Popperian, this is a religious theory, not a scientific one. If there were no circumstances in which you’d be prepared to give it up, then it’s not falsifiable. Ergo, this is a religion. That seemed to me to be a big error that you couldn’t think of the circumstances under which you’d be prepared to give it up.

Taking Jerry’s example, and I’ve talked to Jerry about this, for example, a study would come up that would show that people who ate less fat had higher plasma cholesterol. And Jerry would say, “Yeah, so what?” “Hang on, Jerry. This study says people who eat less fat have higher cholesterols.” Jerry said, “Think. Use your head.” Well, I’m thinking. This seems to be a bit problematic. And Jerry said, “You’re not thinking straight. People with high cholesterol are told by their doctors they’ve got to do something about it. So they go on a low fat diet. But we know that a low fat diet only has limited effect on cholesterol. So what you find is people with high cholesterol are eating less fat because they have changed their diet in response to their high cholesterol.”

I thought in Popperian terms this sounds like a regressive program if you’ve got to raise auxiliary theories like that to save a hypothesis, it becomes… degenerative.

Hell’s bells! It turns out Jerry then comes up with data that actually supports what he says. It turns out actually that theory, which sounded to me like a theory erected purely to save a defunct theory, turned out to be correct. And that really gave me pause… instead of thinking that he had an unscientific theory because he wasn’t prepared to give it up.

If you actually think more about McIntosh and what McIntosh said, and I wrote about this, that’s exactly the way a scientist ought to proceed. A scientist has got a core belief. He’s not going to give that up just because one inconvenient fact comes along, which I had originally thought was a statement of being unscientific. What he has to do is say, “Well, I’ve got this core belief. I’ve got a whole set of reasons for believing this to be true. If the data don’t fit, why not?” If all he can do is say they falsified it or some unconvincing explanation, then that’s surely a black mark against him and his theory. But if, in fact, his explanation for why the data don’t fit leads on to some new observations which turn out to be true, then that’s a step in his favor. And I think Jerry, for all his prejudices and everything else about him, his bombast and so on, I think his core beliefs have turned out to have a lot of validity. (51)