University of Minnesota

Warren Winkelstein, Jr.

Year: July 31st, 2003
Interviewed by: Labarthe, Darwin


Warren Winkelstein is a pioneer in CVD epidemiology and in the history of epidemiology. In this interview he traces the path of his career in public health dating back to the 1940s. He first became interested while attending medical school. After an internship at Charity Hospital in New Orleans in the late 1940s, he took on a fellowship with the New York State Health Department, and earned his MPH from Columbia University. In the mid-1950s, Abraham M. Lilienfeld steered Winkelstein into epidemiology. Under Lilienfeld’s guidance Winkelstein studied the rates of coronary disease in women and challenged long-held beliefs about the role of menopause in CVD risk. In the 1960s Winkelstein shifted his focus to blood pressure at the State University of New York at Buffalo, and after his move to Berkeley in 1968 he joined the NIHONSAN study of Japanese immigrants and their CVD risk factors.

Winkelstein comments on the multiple opportunities in epidemiology, the future of epidemiology and its changing community and definition and mission, as well as the relevance of history to the profession and especially to epidemiology students. (Karen Ross/Henry Blackburn)


Women, Menopause, and Coronary Disease

So while I was at the Roswell Park, Abe had done his what I considered classic paper on the age distribution of breast cancer. You may remember that the conventional wisdom was that coronary heart disease in women had a sort of continuous increase with age except for a perturbation at the time of the menopause, something called Clemeson’s Hook, if you recall. Abe had been a mathematics major in college and when he looked at this distribution he saw something that no one else had seen and that was that Clemeson’s Hook was due to a change in the rate of increase in coronary heart disease incidence and if you converted the data to a logarithmic scale it resolved itself into two straight lines. One steeply increasing prior to menopause, one graph very slightly increasing after menopause, which he hypothesized in a rather complex structure was due to exhaustion of susceptibles during a period of rapid increase prior to menopause. I was familiar with that work because of course we had seminars and we exchanged ideas.

And one day a specialist from Chicago, I think it was Louis Katz, came to Buffalo and gave a seminar on coronary heart disease. And during that seminar he said that prior to the menopause women were essentially not susceptible to coronary heart disease and the menopause represented a sharp change in the risk of coronary disease in women. So immediately after this seminar – actually I didn’t even go home – I went back to my office and I got out the annual reports of the State Health Department and I was able to plot the mortality rates for coronary heart disease in women according to age. And when I did that on a logarithmic scale, lo and behold the curve was essentially straight. There was no change in the rate of increase in menopause and this led to the hypothesis that if we could determine… (The hormonal relationship between coronary heart had clearly been demonstrated in a variety of ways); So I hypothesized that if risk factors could be identified pre-menopausally that, because there was no change in the rate of increase [this] suggests the risk factors, whatever they might be, would be demonstrable in younger women. . .That’s what led to this paper. So using a pregnancy outcome as an indicator of hormonal function, a case control study had demonstrated that women who had coronary heart disease had twice the risk of having spontaneous abortions. We used several controls, a matched neighborhood control, and utilized some controls from Abe’s study of smoking practices and a probability sample of the population. And that led to that paper that you have in your hands. And that’s how I got into it. (4)

Dept. of Minor Regrets

I must say that I was disappointed that no one had bothered to replicate or lab-replicate the work I had done on coronary heart disease in women. I still use the data from that study in my teaching program and so three or four years ago some students re-analyzed the original data and showed a more sophisticated analysis and actually strengthened the associations that had been demonstrated in the 1950 study. But, still, as far as I know, no one has tried to replicate the findings of that study. And habitual abortion and hormonal dysfunction is not a rare condition. So I’ve always been disappointed that no one… Because, after all, it doesn’t do much good to just re-do your own work. That’s ridiculous in sense and not very scientific. What you would like to do is either go on to further exploration, if you will, in the direction that you have demonstrated, but you would like someone else to pick it up. And the associations were pretty strong between the indicators of hormonal dysfunction and coronary and that is certainly a plausible direction to go. So that was a disappointment and I have no idea why people didn’t bother to replicate our findings. (9)

Never Throw out Data

I can’t remember all the details, but I decided to do a real study of blood pressure distribution in the population. And you recall there was a lot of discussion about the Pickering-Platt controversy about the nature of essential hypertension. So I, again, was influenced by Abe Lilienfeld who was a great believer in sampling and doing epidemiological studies based on representative samples of populations. So I designed and applied for and received NIH support to conduct a probability sample study of blood pressure. … We studied several thousand probability-selected persons 15 years of age and older. We made some methodological, I think, developments, which were never used by others. I think they were useful. We studied the familial aggregation of blood pressure and demonstrated, at least in a preliminary study, an aggregation between husbands and wives. It’s never been confirmed that I know of. And actually we never completed analysis of the study. So I came to California and essentially abandoned the blood pressure study.

And then one day, I can’t remember when, I got a letter or a telephone call or something from Buffalo and the new epidemiologist had opened a drawer somewhere and found a folder about the Buffalo-Erie County Blood Pressure Study. So he asked me if he could have access to the data and said not only could he have access, I would send him the punch cards. So I transferred back to Buffalo all of the data which I had brought here and they are still using that dataset. Numerous PhD theses have been based on it and a considerable amount of research and, I believe, that they have actually conducted a follow-up study after 30 years or something like that of that cohort. So there’s a substantial amount of information still coming out of that. I call it “epidemiological archeology.” [Ed. Further evidence that one should never discard an old file, you never know what might turn up.] (8)

Ecological studies

I did get interested in ecological studies as well and I got invited to give a paper at the International Epidemiological Society in Vancouver or Edinburgh, I can’t remember where. I think I gave it in Vancouver. And David Sackett tore the paper apart pretty well. I mean, pretty effectively I guess is the word. I used published data from the National Health Interviews Survey to try and demonstrate that the risk, particularly of lung cancer was going down in younger men due to a decrease in smoking in that portion of the population. I did a cohort analysis. And then I turned my attention to heart disease rates and using data from the Health Behavior Survey, which at that time was just being instituted and tried to show a correlation between the ecological analysis of behavioral risk factors and heart disease mortality. Something that Gary Friedman had done earlier and I simply replicated using a little bit different methodology and it got pretty much the same result. Namely, that the behavior surveys, indeed, could be correlated with the downward trend in coronary heart disease.

Sackett criticized it on the basis of the ecological fallacy. It generated a wonderful argument which I managed to sort of stay on the sidelines while people shouted at each other about. Most of the time ecological analyses would generate a considerable amount of heat and relatively less light. (16)

The Importance of History

It seems to me that one misses a great deal in your professional life if you don’t have some feeling and understanding of the history of your field. I got into this because in our teaching here at Berkeley we place, I won’t say an emphasis, but we try to introduce our epidemiology students to aspects of the history of epidemiology largely through study and discussion of classic papers. So that is really what has led me into this. Also, for the past few years I have been giving a couple of lectures every summer at the graduate summer session in epidemiology, which I think you and I know Henry has been involved in over the years. And, again, there I am trying to tell the students about interesting aspects of the history of the field. And most of the time I’ve talked about personalities, people although not exclusively. I don’t know, it turns out…Right now I’m trying to . . put together a collection on women epidemiologists because most of our students now as you probably know are women. I think it’s useful to have role models. After all, you know, these people are important and they are not just being trundled up because they are women, they are being presented, at least I’m trying to present them, as really important people in the development of epidemiology. So I’m hoping maybe if I can find six worthy subjects I may put that into a little book. (18)

Health Departments. Their Mission of Prevention

As you know, many health departments are burdened I guess is the word. Maybe the word burdened is bad. But anyway, they have major responsibilities for medical care and medical care is something that is immediate and you have to provide medical care to people who are sick or suffering or whatever. Prevention of disease is a sort of option that comes after the provision of the imperative of medical care. And so it tends, in my opinion, to get shortchanged. And I think we still need at the local level to understand the health and disease patterns, if you want to call it, of our communities. So I’ve always felt that we need more analogs, I guess is the word, of morbidity surveys and that type of work is properly done by epidemiologists. So I think there is a lot of room for that

But I also think that we’ve got to move more vigorously in certain other directions. If I were to think about the world’s problems on a world-wide basis one has to think about first the unbridled expansion of the population. Because many of the major problems that we face are related to that. So the next problem that we face is global warming and energy production which leads to global warming and that’s a function to some extent of the size of the population as well as its technical development which is going on at the same time. And then we have these horrible problems world wide of violence of various kinds. I mean the genocides and the wars and the disruptions and the civil unrest. Even in our own country civil unrest is a potential. Not only a potential, but it has produced some pretty severe activities. Some years ago when we began to consider the automobile accidents or automobile deaths or related deaths were a public health problem when that first came up people said, “That’s not public health” and others said, “Yes it is.” A very substantial cause of death and disability and morbidity and so forth. So I think we are going to have to look into some new areas as well, which I guess we are doing to some extent. So I think that there’s going to be a continuing and perhaps expanding need for epidemiologists. Sometimes I say to students, “If you want to have a successful career pick out the number one cause of death today and work on it for the next 50 years because no matter what you do it will not be the first cause of death 50 years from now. So you can claim success.” I think that it is true that the major disease impacts change over time as a function, you know, of a complex…Well, the paradigm of epidemiology – host- environment and agent interactions. You can argue that the web of causation is not appropriate or adequate or whatever, but in my opinion the basic concept is sound and will endure. So as the environment and new agents arrive, whether they be physical, chemical, or biological, what have you, we’re going to have new and different challenges that are going to be constantly arising. (22)

Full Transcript Access

Full transcripts of interviews may be made available to those engaged with original materials for scholarly studies by contacting us.