Stephen Haines '71 - MSS Major

Introduction

I was born in Vermont and raised in Schenectady, New York (when it was General Electric's world headquarters and a real place). Because GE had its basic and atomic power laboratories there, our neighborhood was full of engineers and PhDs and the public schools were excellent. After turning down an offer to pursue a career as a professional French Horn player, I chose Dartmouth because of its focus on undergraduate education and therefore the extraordinary access to outstanding professors it fostered even in the freshman and sophomore years.

Dartmouth

Two courses taken early in my Dartmouth experience set me on the path to an unusually fulfilling career in neurosurgery and, arguably, as perhaps the best paid quantitative sociologist in America.

The first was Mathematical Models in the Social Sciences (Mathematics 36), for which the textbook was Mathematical Models in the Social Sciences by Kemeny and Snell.  This course opened the door for me to the quantitative study of behavior. The second was Methods of Sociological Inquiry (Sociology 7), for which the text was the draft of James Davis's Elementary Survey Analysis, with which Davis was attempting, as he put it, to teach graduate level sociological analytic methods to undergraduates. The door opened wider.

As a married undergraduate with a child, I needed a job. Thanks to the Time Sharing System and BASIC, I was a programmer. Ed Meyers, then Assistant Professor of Sociology, was funded by the National Science Foundation to develop an interactive statistical package, Project IMPRESS, and needed student programmers. I was lucky enough to get one of the positions. We spent many hours programming data collection, management, and analysis. It was a great learning experience that meshed well with Sociology 7.

I chose to major in Mathematics and the Social Sciences (MSS) because it seemed to join my interest in math, my newfound interest in quantitative investigation of social issues, and my new work life as a programmer.  Three members of the Class of '71 majored in MSS: myself, Paul Velleman (who went on to become Professor of Statistics at Cornell) and Bill Hoover (who worked at McKinsey & Company and later moved to Denmark).

MSS aside, I also had a strong interest in the life sciences and hedged my bets by taking necessary pre-med courses. As decision time approached, Ed Meyers was promoting me for a graduate school position with a sociologist, who was developing computer models of societal interactions. He was rather disappointed when I told him I was going to medical school.

Medical School

In medical school at the University of Vermont (Dartmouth Medical School wouldn't have me), the door opened even wider. One of my professors, Lawrence Weed, developer of the problem-oriented medical record, was funded by the National Center for Health Services Research, Department of Health, Education and Welfare, to computerize medical records, allowing the collection of a large amount of clinical data that could be machine searched for research purposes. (Yes, we had touch screen computerized medical records on some hospital wards in the early 1970s.) The type of data collected (medical condition: dead, critical, severe, fair, good, etc.) was much more like the kinds of nominal and ordinal data sociologists dealt with than the continuous variables of the chemistry lab that dominated medical school curriculum of the time. The light went on, and it became clear to me that the analytic methods of quantitative social science needed to be applied to the study of clinical medicine.

Neurosurgery Residency

My medical school interests leaned toward the nervous system and my personality to surgery, and I ended up training from 1976 to 1981 in Neurological Surgery at the University of Pittsburgh. This meant that there were several years during which my quantitative social science interest was put on the back burner. However, there was a year of research time during the neurosurgery training program. While I was learning neurosurgery instead of multifactorial regression, randomized clinical trials (RCTs) were growing in popularity and quality, mostly in fields relevant to internal medicine and cancer and not very much in surgery. During my research year (1978), I investigated the use of RCTs in the neurosurgical literature and found it quite unusual. (To its credit, the Journal of Neurosurgery published this analysis of its own shortcoming.[1] And to his credit, my mentor in neurosurgery was very openminded about the topics we chose to pursue during the research year.) It was clearly time to bring into neurosurgery the quantitative social science of the 20^th Century.

Fellowship Training (Postdoctoral Fellow)

That research year prepared me to apply for, and in 1981 to win, the Van Wagenen Fellowship of the American Association of Neurological Surgeons. This unique fellowship allowed me to design a research program outside of North America with the intent of bringing new research knowledge and techniques back to American Neurosurgery. My fellowship was spent in Oxford working with Professor Charles Warlow, a clinical trials neurologist who was designing the European Carotid Endarterectomy Trial. Oxford proved to be a fertile environment for this purpose with clinical trial and medical statistics/clinical epidemiology superstars like Richard Peto, Peter Armitage and Klim McPherson interacting regularly.

Lifelong Research Focus

I returned to the United States in 1982 and began my faculty position as Assistant Professor of Neurosurgery at the University of Minnesota. I was expected to do laboratory research and did, for several years, without much success. My interest in quantitative analysis, however, continued. I published a series of articles in neurosurgical journals documenting the infrequent use and poor quality of RCTs in neurosurgery, exploring alternative analytic techniques, and making the case for placing as much emphasis on the quality of clinical research methodology in neurosurgery as the field gave to the quality of basic neuroscience research. It is a long list, but this is the first.[2]

This led eventually to a National Institutes of Health and Agency for Health Care Research and Quality funding of a multicenter clinical trial which ended up demonstrating how poorly prepared neurosurgery and the health care system were for RCTs.

A small but growing number of colleagues of mine have had similar interests that combined research design and neurosurgery, and the cumulative effect of this is that, from the 51 neurosurgical RCTs I could identify in 1983,[3] a more recent review identified 1,102 as of 2013.[4] It is now unusual for the plenary sessions of the major neurosurgical organizations not to have several RCTs on their programs. Of even greater importance, however, has been the growth of "evidence-based neurosurgery," which has increased awareness of other study designs that bring even greater rigor to the study of clinical neurosurgery.[5] Neurosurgery is a relatively small field dealing with relatively rare conditions, and accumulating enough patients for high quality RCTs is often impossible. Considering this, techniques well known to quantitative social scientists have added immensely to the rigor and quality of clinical neurosurgical research.

My short message for those interested in both quantitative analysis and the life sciences is that they are meant for each other. Rigorous, high quality clinical research is essential for the translation of basic scientific discoveries in the health sciences to effective treatment of ill and injured people. Quantitative social science provides excellent preparation for a career in doing just that.