IN MEMORIAM, WARREN WINKELSTEIN, JR, WITH EXCERPTS FROM HIS PUBLISHED WORK

Tuesday, 7th of August 2012

Warren Winkelstein Jr., Medical Sleuth, Dies at 90

Dr. Warren Winkelstein Jr., a physician and researcher whose groundbreaking studies connected unprotected sex between men to AIDS, smoking to cervical cancer and air pollution to chronic lung disease, died July 22 at his home in Point Richmond, Calif. He was 90.

Peg Skorpinski

Dr. Warren Winkelstein Jr. looked for patterns in illness.

The cause was complications from an infection, according to the University of California, Berkeley, where he was an emeritus professor of epidemiology and a former dean at its School of Public Health.

Epidemiologists look for patterns of illness in populations, and Dr. Winkelstein was considered a master at designing rigorous studies to answer tough questions about the causes, risk factors and transmission of disease.

He was best known for an AIDS project, the San Francisco Men’s Health Study, which began in 1984 and tracked 1,034 single men, some straight and some gay, from parts of the city with the highest rates of AIDS.

At that time, the virus that causes AIDS had been identified, but scientists did not know what practices, sexual or otherwise, might increase the risk of infection. A number of research projects were set up to find out.

What set Dr. Winkelstein’s study apart was his insistence on using a technique called probability sampling to find participants. It involved selecting blocks from census tracts and actually knocking on doors to ask men to participate, rather than putting advertisements in newspapers and waiting for responses. The goal was to ensure that the study included a representative sample of men so that the findings could be broadly applied.

The study lasted 12 years and led to about 150 articles in scientific journals. It confirmed, with solid data, what researchers had suspected: men who had the most male partners and were the receptive partner during unprotected anal sex had the highest risk of infection. The study also yielded important information about the incubation period of AIDS and risk factors that affected its progression.

“It was a uniquely important study because it showed both how common H.I.V. infection was among men who have sex with men and the central role of unsafe sexual behavior in transmission of the virus causing AIDS,” said Dr. Arthur Reingold, the dean for research at Berkeley’s School of Public Health. It was the only study at the time, he said, that used a representative sample of men living in San Francisco.

Dr. Reingold said that, paradoxically, Dr. Winkelstein’s insistence on sound methodology nearly cost him the government money needed to keep the project going. Other researchers were conducting similar studies but using different techniques to recruit participants. The differences meant that Dr. Winkelstein’s data could not be combined with that from the other groups. The agency providing money wanted to cut him off but eventually conceded that his work was too valuable.

Earlier in his career, the same disciplined approach led Dr. Winkelstein to findings that were so counterintuitive that other researchers refused to believe them. He suggested in 1977 that smoking was a risk factor for cervical cancer, and 25 years later it was widely accepted that smoking does increase the odds of cervical cancer in women who are also exposed to a common sexually transmitted virus.

In the 1960s, in Buffalo, Dr. Winkelstein designed a study showing that air pollution could cause chronic lung disease and that lung disease in polluted areas was not just caused by risk factors related to poverty. The findings helped shape national air quality standards.

Warren Winkelstein Jr. was born on July 1, 1922, in Syracuse. His father was a prominent lawyer.

From an early age, he was surrounded by people who were concerned about social issues. The family lived near Syracuse University, and his mother would often arrange dinner parties for scholars who went to lecture there. A close friend of the family was a department chairman at the medical school who had a strong interest in public health.

“So I think my background led me in a direction of social concern,” Dr. Winkelstein said in an interview in the journal Epidemiology in 2004.

He graduated from a private high school, the Putney School in Vermont, served in the Army in World War II and then earned a bachelor’s degree in sociology from the University of North Carolina in 1943, a medical degree from Syracuse University in 1947 and a master’s degree in public health from Columbia in 1950.

During the Korean War, he worked for the United States Public Health Service in Cambodia, Laos and Vietnam. He then worked for the Erie County Health Department in Buffalo, where he led a large trial of the Salk polio vaccine. He joined the faculty at the University of California, Berkeley, in 1968.

Dr. Winkelstein is survived by his three children, Rebecca Yamin, Joshua Winkelstein and Shoshana Winkelstein; three grandchildren; and three great-grandchildren.

He kept teaching after he retired in 1991 and also wrote biographical sketches of leading figures in the history of epidemiology. He always urged his students to work with other scientists.

“You can make a lot of mistakes individually,” he told Epidemiology. “But by collaboration, you eliminate a lot of dumb errors.”

A version of this article appeared in print on August 6, 2012, on page B10 of the New York edition with the headline: Warren Winkelstein Jr., 90, Medical Sleuth.

Excerpt from a 1992 editorial by Winkelstein in the American Journal of Public Health:

If the three P's-pollution, population, and poverty-are principal determinants of health worldwide, it becomes important to examine our nation's international policies on these issues. With regard to greenhouse gases, as of this writing, the US government has not yet indicated whether it will participate in an upcoming international conference designed to adopt an agreement on limiting such emissions.

Rather, the US government has said that European proposals for emission limits are more stringent than it could agree to.

Clearly, without United States participation, any international effort to control greenhouse gas emissions is likely to fall far short of its desired objectives, or indeed, to fail. . . .

There is a certain poignancy in choosing the subject of health and the environment for discussion at this time because 1992 marks the 150th anniversary of the publication by England's "Poor Law Commission of An Inquiry into the Sanitary Condition of the Labouring Population of Great Britain" (the "Chadwick Report").

Whether one views this work as an expression of Jeremy Bentham's philosophy of utilitarianism, which stated that society should be organized for the "greatest happiness of the greatest numbers," or as a more cynical prescription for enhancing the efficiency of the capitalistic enterprise, this "inquiry" stands as a fundamental document of public health. Its ecological analyses provided the basis for far-reaching recommendations for environmental sanitation, and its influence on public policy was international. Basically, it defined public health as a societal rather than an individual problem. However, at that time, in 1842, the problems were largely seen as engineering problems. Gradually, and not without heated controversy, it has become apparent that the health and disease patterns of populations are determined by a combination of social forces and environmental factors. It is to these broader issues that [Sir Richard] Doll has drawn our attention.

The prospect of dealing effectively with the three P's [pollution, population, poverty] is, indeed, daunting. Furthermore, the world situation is complicated by numerous wars, droughts, and famines. However, failure to deal effectively with the three P's may jeopardize the survival of the human race. Brock Chisholm, the first secretary-general of the World Health Organization, frequently reminded audiences that no species had succeeded in permanent survival and that the challenge to the human species was to achieve a successful adaptation to its environment.

Sir Richard Doll has reminded us that the public health movement must continuously rededicate itself to these issues.

‘CONTEMPORARY PERSPECTIVES ON PREVENTION,’ BULL N Y ACAD MED, 1975

http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1749636/pdf/bullnyacadmed00157-0030.pdf

CONTEMPORARY PERSPECTIVES ON PREVENTION*

WARREN WINKELSTEIN, JR., M.D., M.P.H.

Dean, School of Public Health

University of California

Berkeley, Calif.

ACCEPTANCE of the concept of the primacy of prevention is not new. It is said to have been embedded in ancient Chinese as well as Greek philosophy and it must have motivated the anonymous author of the old proverb "An ounce of prevention is worth a pound of cure."

It certainly dominated the thinking of the two major architects of the British and American public health movements, namely, Edwin Chadwick and Lemuel Shattuck. However, both of these reformers had limited specific methodologies available and had to rely on what we now call the ecological or nonspecific approach to the prevention of disease. In subsequent paragraphs I shall suggest that the ecological approach may still offer great potential for prevention of the major chronic diseases which currently burden society.

With the coming of the bacteriological era and the identification of a wide range of causal agents, specific prevention of particular diseases became a reality through environmental manipulation or preventive medicine. Many different preventive modalities were employed: e.g., active immunization, nutrional supplementation, control of arthropod vectors, treatment of sewage, and purification of water.

However, in every situation in which specific primary prevention was effective, there was a vulnerable point in the genesis of the disease which was identifiable and susceptible to modification by available technology.

Thus, for the control of malaria it was only necessary to interrupt transmission by selective annihilation of those mosquitoes which had fed on patients or carriers. For poliomyelitis, the susceptible host could be immunized artificially to prevent infection. For retrolental fibroplasia, the oxygen tension of incubators housing premature infants could be limited.

However, there are some diseases in which the causal agent is identifiable but in which no particular point of vulnerability permits specific primary prevention to be applied. The most notable of these is tuberculosis.

From 1900 the downward trend of incidence and mortality for tuberculosis was continuous and distinct. From its place as the first cause of death and a major source of disability it has become a minor problem. It is very improbable that this was due to any of the actions directed specifically toward the control of tuberculosis. The main organized approach to tuberculosis control in the United States was the identification and isolation of cases in hospitals and sanitaria in order to prevent contagion. Until 1940 this was accomplished primarily through reporting by physicians and follow-up by health departments.

It was only after 1945 that community-wide radiographic surveys were applied to detect the asymptomatic case, yet most of the decline in tuberculosis occurred before World War II. Terris' published data which, I believe, demonstrated conclusively the importance of ecological factors in the decline of the disease. They show that for white men more than 30 years of age living in Buffalo, N.Y., in 1940 there were large differences in mortality experience associated with economic status. Men in the middle years of life who lived in the poorest economic area had rates more than four times as high as those in the richest area. That this was not a function of medical care is supported by two facts. First, the diagnosis and treatment of tuberculosis in sanitoria had been available in upstate New York for everyone-regardless of ability to pay-for some years before 1940. Second, mortality rates among women in Buffalo were uniformly low in all economic groups. Further, if isolation of cases were the major determinant of mortality, it would be unlikely that the marked difference between the sexes would obtain.

Terris concluded that the general life style or experiences of men and women in various social strata determined their tuberculosis mortality experience. A corollary conclusion was that some contributory occupational stresses were of major importance.

Terris's analyses and interpretation were consistent with Frost's earlier treatment of the subject in "How Much Control of Tuberculosis?" although Terris went further in emphasizing the importance of the environment in determining the distribution of the effects of disease.

Frost was explicit and, I believe, correct in recommending no major changes in the tuberculosis-control program to achieve very low rates of infection and mortality in the United States. Terris pointed to a fairly well delineated population group in which most of the residual tuberculosis problem persisted. He suggested further study to isolate the responsible factors. Despite these epidemiological findings, vast and elaborate community-wide case-finding programs were promoted at great cost in the United States after World War II and probably had very little effect on the established downward trend of the disease.

It seems to me that some of today's problems of prevention are typified by the tuberculosis model. For many of the diseases whose specific causation is known, the biological or environmental mechanisms are such that single points of vulnerability are as yet unidentified or unlikely to exist. For example, while we know the specific cause of syphilis to be the spirochete, Treponema pallidwm, its biological mode of action is such that specific immunization remains improbable. While the mode of transmission is also fully understood, social norms and values are such that no simple action would appear applicable to influence transmission effectively. On the other hand, there is evidence that changing values and practices adopted within or between societies profoundly affect the occurrence of this disease. In the United States this was apparent during 1944-1948 and during more recent years as sexual practices have become increasingly permissive and promiscuous.

Similarly, the disease is virtually nonexistent in certain isolated populations or groups which limit their sexual contact to members and do not accept outside contacts: e.g., Hutterites, Seventh Day Adventists, and Amish.

Another example which would appear to fit this model is mortality and morbidity caused by automobile accidents. Many factors which influence levels of occurrence are well established, such as speed, weather, design factors of highways such as crossing separations, driver characteristics such as blood-alcohol level, and so forth. While some purpoeful modification of these factors will produce changes in occurrences, it is improbable that major reduction or effective control can be expected from conventional health-oriented efforts. On the other hand, during the recent energy crisis which led both to speed reductions and limitations in the use of vehicles, the February 1974 death rate from auto accidents in California was 40% lower than that for the previous February. I am confident that if the energy "crunch" is relaxed the speed limit will be returned to 65 miles per hour on regular highways and 70 miles per hour on freeways and morbidity and mortality will increase concomitantly.

At this point it may be worthwhile to point out that when specific primary prevention can be applied it has the greatest promise of effective disease control. Further, for those diseases in which specific measures are not now available, continued research is mandated to identify the particular factors which might be key determinants of occurrence.

At the same time the manipulation of nonspecific factors can be attempted in order to accomplish the objective of disease prevention. I suspect that most of the major diseases susceptible to single-factor control have been identified and that the application of primary prevention during the immediate future must be increasingly nonspecific if it is to be very effective.

Let me now turn to consideration of the prevention of several major causes of death in our society: coronary heart disease, lung cancer, and gastric cancer. It is increasingly apparent that coronary heart disease has multiple causes. At present, the three most influential predisposing factors appear to be elevated serum cholesterol, elevated blood pressure, and cigarette smoking. Table I contains data from the National Cooperative Pooling Project. These data show clearly the increasing risk of developing coronary heart disease for men 35 to 64 years of age having various combinations of these risk factors. The relative risk among such men with all three factors is seven and for those with any two factors it is 4.5. However, only 58% of all cases occur among those with two or more of these risk factors. It has been estimated that 38% of the population of men 35 to 64 years of age, or 10,556,000, are in this group. In Table IL I have estimated the maximum total 10-year reduction in coronary heart disease incidence if it were feasible to reduce the levels of these risk factors to that in the nonrisk group.

Thus, if 100% of these high-risk men participated in a screening program and subsequent risk-modification regimen, a 50% reduction in coronary heart disease incidence could be accomplished. The maximum reduction possible with less complete participation is also shown in the table. Since present methods for modifying risks are far less than 100% effective, these reduction estimates are gross overestimates. I do not see how risk-factor modification using the three currently accepted factors can be expected to produce even a 25% reduction in the incidence of this disease.

Nevertheless, today many people advocate a de-emphasis of research into the causes of coronary heart disease in favor of increased emphasis and allocation of resources to programs of risk-factor modification (usually justified on the basis of a need to apply what is already known). While even a 25% reduction may well warrant the effort toward risk-factor modification, it certainly does not permit any complacency regarding the continued need for research and should not be allowed to divert resources from that effort.

Further, there are epidemiological characteristics of coronary heart disease which I believe offer greater promise for prevention than the modification of currently established specific risk factors in screened individuals.

Table III shows the mortality from coronary heart disease among men and women 50 to 69 years of age in Buffalo according to economic level of census tract of residence. It would, of course, be preferable to present incidence data but, as far as I know, such data do not exist for any American community (perhaps with the exception of Tecumseh, Mich.) The data reveal two well known epidemiological characteristics of coronary heart disease which I believe have received a profound and undeserved neglect. First, they reveal for both men and women a striking geographical pattern with an approximate two-fold difference in mortality between the tracts classified in the lowest economic grouping and those in the highest, with a graded series of rates in the intermediate groups. Incidentally, the extreme gradient is similar to that between mortality from coronary heart disease in the upper plains states and in the eastern and western coastal states. They also reveal a male-female ratio of approximately three to one in each economic grouping. To date, neither the geographic nor sex patterns of coronary heart disease have been satisfactorily explained on the basis of the distribution of the commonly recognized major risk factors or by any other explanation. In my opinion, attributing the sex differences in coronary artery disease to differences in circulating hormones has only a little more scientific basis and credibility than did the attribution of the sex differences in tuberculosis mortality in 1940 to this factor.

If, indeed, there are major causal factors affecting the occurrence of coronary artery disease which are in some way associated with economic status, it may be postulated that important preventive gains could be achieved through the elimination of poverty. In fact, if this were the case and if the male population characterized by the lowest economic level in Buffalo were raised to the second level a decrease in coronary heart disease mortality of 31% would be realized. The potential benefits of raising economic status in the higher levels are of lesser magnitude: 7% from economic level two to three, 8% from level threE to four, and 9% from level four to five. On the other hand, raising the two lowest economic groups to the third or median level would result in approximately a 28% decrease in coronary heart disease mortality among these groups of men.

If the differences among the economic classes are due to differences in case fatality rather than the incidence of coronary heart disease the problem of preventing mortality is greatly simplified, since its accomplishment would require only the provision of adequate medical care.

It seems more likely that the differences in mortality truly reflect differences in incidence and that the causes will be more difficult to determine.

From the data it is clear that a greater preventive effect is possible if the conditions which lead to a female death rate one third that of the male death rate could be identified. Since the male-female differences observed in Buffalo and other United States populations are not universal characteristics of coronary heart disease, there is a good possibility that hormonal differences are not the controlling conditions; thus, their identification would have great significance for primary prevention. Further, since the sex differences are quite consistent in each economic group, the potential reduction in incidence and mortality is in the neighborhood of 60% over all.

Consider now two major cancers: cancer of the lung and cancer of the stomach. The age-adjusted trends of mortality for these cancers between 1930 and 1960 reveal a striking decrease for stomach cancer and a frightening increase for lung cancer. With respect to lung cancer, there is no more doubt that cigarette smoke is the vehicle for a chemical carcinogen(s) than there is that sewage-contaminated drinking water will produce gastroenteritis. Further, if we are interested in decreasing the death rate from the first cancer cause of death in American men, it is apparent where preventive effort should be directed. Unfortunately, changing the smoking habits of a population may be more difficult than changing the economic status of its members.

Figure I shows the death rates from lung cancer according to economic status in Buffalo. The difference between the rate in the lowest economic grouping and the highest is almost as great as the difference in mortality between the years 1930 and 1960. In view of the solidly established association between cigarette smoking and lung cancer, it might be expected that the cigarette-smoking patterns of the male population would follow that of the lung-cancer mortality. However,as shown in Figure 2, this is not the case. Heavy smoking is, or was, when these data were collected, as common in the upper economic groups as in the lower for men. These findings are not inconsistent with a causal relation between cigarette smoking and lung cancer.

Since only a small proportion of persons exposed to heavy cigarette smoking develop lung cancer, it may be hypothesized that additional factors (cocarcinogens) are either necessary or contributory to the actual development of the disease. If such factors are distributed in the population in association with economic status, then the association of economic level and lung cancer independent of smoking status is a possibility and offers an alternative approach to prevention.

In view of the foregoing, I have made a few calculations of the possible benefits accruing from modification of the economic status of the population similar to those made for coronary heart disease.

Among men 55 years of age and older, raising the economic status from the lowest level to the second lowest would result in a 23% reduction in lung-cancer mortality and raising the economic level from two to three produces a 21% reduction. The combined effect of raising the two lowest economic groupings to the level of the median group is approximately so%.

But what about cancer of the stomach? Why has it decreased so greatly? There is no generally accepted explanation. Many, if not most, surgeons would agree that the five-year survival rate for this disease has remained essentially unchanged during these years. There is no good explanation for the decline which has shifted this disease from the first cancer cause of death among men in the United States to the fourth or fifth. However, in view of the tuberculosis model, we might inquire whether all economic classes of the population have shared equally in the declining morbidity and mortality from this disease.

Data from Buffalo, N. Y., indicate a very strong inverse association between economic status and stomach-cancer mortality. In fact, death rates are approximately three times as great in the lowest economic group as in the highest, even for women whose rates are less than half those for men. Since medical care of patients with clinical cancer has not improved survival chances, we cannot assume that this explains the lower rates for the economically favored. Rather, it may be inferred that environmental factors of some kind are exerting a strong influence and that some change or changes in the environment, not shared by all economic classes, has produced the downward time trend noted earlier.

Such a change in the United States might be particulate air pollution in cities. From 1930 onward the uses and techniques for the burning of smoke-producing coal were drastically altered so that large particle air pollution in most American cities was greatly reduced.

However, the poor people who frequently live in crowded slums adjacent to industrial areas might have continued to be exposed to large particle soot resulting from these industrial activities. If this soot contained a carcinogenic agent(s), both the long-term trend in mortality for the country as a whole and the economic differential might be explained. Data to test this hypothesis were obtained in the Buffalo and Erie County Air Pollution Study and are shown in Table IV.

The economic gradient for stomach-cancer death rates is revealed by the last column of Table IV. However, the bottom row shows an equally strong and linear gradient of these rates for air-pollution levels. When the rows and columns of the interior of the table are closely inspected, it becomes apparent that the association between gastric-cancer mortality and suspended particulate air pollution is strong for each economic level except the highest, while the economic gradient is less regular except for the highest air-pollution area. The data for women show parallel but lower rates, with almost no economic trend remaining when the air-pollution level is controlled. This finding is consistent with earlier studies of stomach cancer in Buffalo which indicated differential risks in selected occupations. The data were also examined to see whether ethnicity, particularly Polish birth or ancestry, previously shown to be associated with excess stomach cancer, could have accounted for the distribution. This was ruled out.

If these findings are valid, they have interesting implications. First, they suggest that a major factor in the decline of stomach cancer in the United States was a technological change in fuel burning undertaken for reasons totally unrelated to health. They also suggest that where large-particle soot continues to be a part of the environment, stomach cancer will remain prevalent, and they reaffirm earlier studies indicating that stomach cancer is at least partially an occupational disease. Under such conditions it should be relatively simple to justify control measures on the basis of their disease-preventing (health-promoting) effects.

In conclusion, I should like to point out that the four leading causes of death-diseases of the heart, malignant neoplasms, cerebrovascular deaths and a very substantial proportion of adult sickness and disability. There is good reason to believe that, except for a few cancers, these conditions have multifactorial causation. Further, they are all characterized by substantial geographical and temporal variation. Thus, they might fit the tuberculosis paradigm. If this is true, specific primary prevention through single-factor modification seems unlikely to be a potentially effective strategy. The effective alternative for prevention is most likely to be the ecological approach in which both the social and physical environment is altered.

In a way, it seems unfortunate that a contemporary perspective on prevention should place such emphasis on the ecological approach which so strongly characterized the 19th century public health movement rather than on specific prevention which, from the time of Pasteur until the present, has achieved such brilliant results. However, just as modem society has become increasingly complex, so have its major diseases. To expect simple solutions seem unrealistic. There is no doubt that prevention remains the most effective way to produce a healthy population. Its implementation continues to demand old and new strategies and techniques.

These can only come from a better understanding of the causes of the distribution of disease. This understanding is still far from complete and demands increased study and the allocation of research resources.