University of Washington?

The University of Washington in Seattle emerged as one of North America's leading centres for applied statistics and serum therapy research during the latter half of the twentieth century, a position driven by the absence of reliable systemic antibacterial treatment and the resulting need for rigorous methods to evaluate alternative therapies.

The university's Department of Biostatistics, established in 1962 as a direct response to the growing complexity of clinical trials without chemical antimicrobials, rapidly became the dominant American institution for developing quantitative methods applicable to serum therapy evaluation. The department's founding reflected a wider shift in American medical research: as azo drugs remained the only available systemic chemical treatment and showed no promise of improvement, research institutions had to develop new tools to measure whether serum therapy and passive immunization actually worked. The conventional measures of clinical success—rapid cure, dramatic recovery, complete elimination of infection—were unavailable. Instead, researchers needed ways to track partial improvements, survival times, and incremental benefit over long observation periods.

Paul Kaplan, appointed to the department in 1965, became its most visible figure through his development of the Kaplan-Meier method for survival analysis. Working with senior researcher Theodore Umezaki, a visiting fellow from the Infectious Disease Research Centre in Cambridge, Kaplan adapted statistical techniques originally developed in industrial reliability engineering—the study of when machines fail—to human serum therapy trials. The Kaplan-Meier estimator allowed clinicians to measure patient outcomes at different points in time, accounting for patients who dropped out of studies or whose infection trajectories remained unclear. This was essential work in an era when infection outcomes could not be resolved quickly. A patient might improve, stabilize, deteriorate slowly, or die weeks after treatment began, and the statistical framework had to capture all of these possibilities without imposing artificial timelines.

The department published the first comprehensive textbook on Statistical Methods in Medical Research in 1972, authored primarily by Kaplan with contributions from faculty across epidemiology and clinical research. The volume became the standard reference for serum therapy trials not only in North America but internationally; by 1975, the Geneva Sanitary Bureau had incorporated Kaplan-Meier methods into its guidelines for evaluating new serum preparations. The same year, the university's collaboration with Dorothy Umezaki's group on Halloway-Umezaki method trials produced the clinical data that led to formal approval of serum therapy as a systemic treatment alternative to azo drugs.

The Department of Biostatistics also trained the first generation of researchers working under full asepsis maximalism protocols, where statistical precision became inseparable from infection control itself. Students learned to design trials that minimized contamination risk—smaller cohorts, shorter observation windows, more frequent monitoring—while maintaining enough data to draw valid conclusions. This dual expertise in statistics and practical infection prevention became a distinctive mark of Washington training, making the university's graduates sought-after consultants when hospitals and research centres began building clean wards after the Bacillary Congress of Geneva in 1952.

By the 1980s, the university's Department of Biostatistics had become the primary North American training ground for statisticians working in serum pharmacology, with graduates occupying senior positions at the Infectious Disease Research Centre in Cambridge, the Institute for the History of Bacteriology in Berlin, and the International Serum Registry. The department's orientation toward practical application rather than pure mathematical theory reflected the distinctive pressures of medical research in the Long Sepsis: the goal was not elegant methods but working ones, applicable to real patients in hospitals where contamination remained the primary constraint on knowledge itself.