Statistics for the People

By Christopher J. Phillips ~

Through the pioneering work of historian Harry Marks and others over the past three decades, we’ve come to know a story about how and why statistics and statistical methods entered clinical medicine: they helped reduce bias by encouraging double-blind randomized trials; they managed uncertainty through introducing statistical significance; they improved efficiency of grants by measuring how well money was spent. Employees at federal agencies like the National Institutes of Health and the Food and Drug Administration played a central role in this transformation, initially by promoting biostatistical methods and training, and later by requiring grant applications or drug trials adhere to rigorous statistical standards. Often accompanying such stories is the unspoken assumption that statistics provided medicine a kind of mathematical rigor, squeezing the messiness of research on humans into the straightjacket of formal statistical theory.

I was consequently surprised in the archives when I kept finding the key figures involved in this transition talking not of making medicine more rigorous but of bringing the methods of social science to bear on medical problems. One key piece of evidence was a speech then-Surgeon General Leonard Scheele gave at the Mayo Foundation in 1949. Just two years earlier, Scheele had overseen the creation of the first group of biostatisticians at the NIH under demographer/sociologist Harold F. Dorn. In essence, Scheele argued that scientists should stop thinking of medical research as primarily biochemical lab work. Instead, they needed to take account of the fact that people should be at the center of research, and yet humans were more variable—messier—research subjects than animals or cells. For Scheele, statistics came in at this “inter-relation of medicine and social research.”

In particular, epidemiological methods might play a far more central role in medical research, Scheele suggested, rather than being confined to public health. After all, “the study of social structure and social pathology in epidemiological investigation has often more than supplemented experimental research; it has demonstrated the etiology of disease.” Tracing the field back to nineteenth-century researchers in Great Britain and the United States, he explained that this role had largely been forgotten in recent years as researchers turned to a “search for a single, specific cause of disease.” Yet the major challenges in fighting disease involved multiple causes, and in particular “environment, heredity, cultural patterns, and socio-economic status”—that is, disease in a human was “inseparable from his life as a biological and social organism.”

Scheele was speaking at a time of the ascendance of what was called “social medicine.” Just a few months before Scheele’s speech, New York’s Commonwealth Fund had published a book called Social Medicine promoting in the U.S. the idea that the causes of disease would require less research into biochemical processes and more into the relation of people to their environments—where they lived and worked, how they lived, etc.—for which the field of biostatistics would serve as a “basic science.” While the label “social medicine” never caught on in the U.S. (likely because of its similarity to “socialized medicine” which had become irredeemably politicized), the basic importance of biostatistical studies of environmental health, occupational health, and other “social” factors of disease, was clearly at the heart of why Scheele wanted a stronger role for biostatistical methods at the NIH.

Scheele himself had one of his first appointments with the Public Health Service in Joseph Mountin’s Public Health Methods division conducting surveys of morbidity and mortality during the Great Depression. The division’s goal was to highlight the social and economic factors within communities that were associated with ill health. And indeed, in Scheele’s speech, the examples of how to integrate social science with natural science in medical research involved “population studies,” “community health studies,” “family health studies,” “morbidity studies” and “mortality studies.”

A speech like this helps us to recast the introduction of statistics into the NIH, and medicine more broadly. In 1949, the idea that randomized clinical trials and meta-analyses would become “gold standards” of medical evidence was years away, and even observational studies like the Framingham Heart Study and the study of smoking and cancer among veterans were years away from having enough data to suggest any conclusions. What Scheele understood is that medical researchers could no longer simply rely upon animal experiments and laboratory chemistry. They would need to take account of people in their lived environment.

Perhaps even more surprising, the speech helps situate the introduction of statistics as a part of bringing people’s lived experience back into research about the causes of diseases. Reading it today, there’s a deep irony in any assertion that statistical concepts entered medicine because they’re “objective” and helped address problems of individual bias and subjectivity; that they facilitated claims about therapeutic effectiveness in general, rather than for any one person. In a sense, that they shifted the focus from patients to numbers. But Scheele, who did as much as anyone else to create a permanent place for statistics in NIH-funded research, saw statistics at their heart as part of the reintroduction of people back into medical research.

Christopher J. Phillips, PhD, is an Associate Professor and Director of Graduate Studies in the Department of History at Carnegie Mellon University and a 2020 NLM Michael E. DeBakey Fellow in the History of Medicine.