Manning Feinleib

Quotes

But we had to make a decision within Bill Zukel’s [NHLBI] programs – how to split up the research efforts. And in broad terms it fell into two realms: the intervention-type studies, the clinical trials programs and the observational epidemiology. We had a strategic planning meeting and I opted to take the observational studies and they gave Bill Friedewald the clinical trials. They were kept as separate entities under separate associate directors.

One of the concerns with the clinical trials programs was how 1) to assure adequate funding for many years because it will take many years and 2) adequate controls to ensure standardization and adequate-sized populations. So those were separate things, under first Bill Friedewald and then Larry Friedman and others. But the observational epidemiology, first with me and then with Millicent, and then with Terry Manolio, was always separate.

Obesity

Again, to make a long story short, when we analyzed the Framingham data we found that body weight had a U-shaped relationship to mortality. And people said, “That’s because people who are thin are either sick or they smoke, or things like that.” Well, we did analyses by smoking category and found the same U-shaped curve – the smokers and non-smokers. And it turned out to be a complex story. That’s all I can say. And its still going on – you don’t know all the biases that go into it. Whether being very thin means you have bad health habits, are derelict or something like this, or have incipient diseases that haven’t been diagnosed yet. With regards to obesity, I think its quite well established….. Early on in Framingham we did a multivariate analysis, putting in obesity along with blood pressure, cholesterol, etc. Obesity did not turn out to be as important as any of the other variables. Now we claim, because we were controlling, that when obesity is a factor it operates through its effect on blood pressure, diabetes, etc. So you can’t use it in the same equation because you are adjusting then for some of the consequences of the obesity variable and “adjusting out” its effect. That’s I think the major part of the story now.

The Logistic

It didn’t come about as a regression model as the way it is usually taught nowadays. Logistic regression, looking at the log of the odds as a linear combination of risk factors, came about through a thing called discriminate functions. Jerry Cornfield had the idea that you have a population of normal people and a population of people with cardiovascular disease; they have different mean values for the risk factors but the same variances. At any point you then say, “What are the odds for somebody with this value of being in the normal group or the disease group?” And he worked out the equations. Those equations are mathematically equivalent to the logistic function.

This was before we had high-powered computers so it was a real hard thing to estimate. It wasn’t until three or four years later that they found sufficient ways of estimating this. But he and Jean Truett got together and said, “Let’s apply this to the Framingham data.” Kannel served as an author on all the papers using Framingham data. So they put out this paper: Truett, Cornfield, and Kannel, which has become a classic – the first application of the logistic. I think that’s something we have to give Framingham credit for, introducing the logistic model to cardiovascular disease and thereby to virtually the whole field that developed out of it, of everybody’s logistic function.

Trends

Why something should happen so consistently across the whole population at the same time. And it still hasn’t been adequately explained. At NCHS we started looking at trends and risk factors that might account for it. And it’s remained largely inexplicable. The changes in smoking and in blood pressure distributions and cholesterol distributions came after the decline started. And some of the major things went in the wrong direction like obesity. Obesity is getting more and more common even while some of the other risk factors are going down. So parts of it are still enigmatic.

MF: There certainly are a lot of people involved with it [CVD epidemiology]. I think some of the names you always have to mention. Let me start off with Henry, Henry Blackburn. He has certainly made major contributions and has been a good spokesman. People like Jerry Stamler, totally dedicated to the field of epidemiology. I disagree with him on a lot of issues. I think he was too hasty to come to judgment on a lot of things like diet and salt and things like that but no question he was totally dedicated to these things. Fred Epstein, I think he was a very good spokesperson and diplomat for the whole circle of epidemiology. I think at the Cardiovascular Epidemiology meetings, the two people you could always depend on to get up after a talk and ask a question were Jerry Stamler and Fred Epstein. Usually in a nice way, but sometimes in a challenging way. Some of the younger people have been very influential. Lew Kuller I think has been a real good example of what a good epidemiologist should be. Then individually, the Framingham people – Kannel, Castelli. Stoney [Stallones, cardiovascular disease was not his primary interest, but he interacted with many of us and introduced people that I think inspired and carried on a lot of his traditions. Who else?

MF: You could go through a list of the people who’ve chaired the various committees, the Epidemiology Section of AHA you’ll find many of these ….

MN: many of these names…. But I think that some of the names you came up with are the names I would’ve come up with.

MF: And there’s others in Europe – Pyörälä is one of the guys I was trying to think of. We always stand on the shoulders of giants when we go into these fields.

MN: That’s right. There’ve been some great people.

MF: Jerry Cornfield I think had a tremendous influence. He wasn’t huge as a cardiovascular epidemiologist but he was very influential. This has been fun.