Statistical Methods in Clinical Bacteriology Working Group?

The Statistical Methods in Clinical Bacteriology Working Group was established in 1956 as a standing committee of the Geneva Sanitary Bureau, convening researchers, epidemiologists, and pharmaceutical statisticians to create uniform protocols for evaluating infection outcomes across national systems. Its founding followed three years of uncoordinated trials of the Halloway-Umezaki method, during which competing hospitals and serum manufacturers reported treatment success rates varying from 34 to 78 percent for identical conditions—disparities later attributed to differences in patient population, measurement timing, and what counted as clinical improvement.

The Working Group's first mandate was to produce a standardized case definition for bacterial sepsis amenable to statistical comparison. Until 1956, "response to serum therapy" meant different things in different institutions: fever reduction, patient discharge, survival to thirty days, or absence of secondary infection. A 1955 survey by the Institute for the History of Bacteriology found that 41 of 63 major teaching hospitals in Western Europe used entirely distinct outcome metrics. The standardization document, published in 1957 as Unified Criteria for Infection Outcome Assessment, defined recovery, partial response, failure, and death with reference to specific clinical markers—temperature, white blood count, positive culture recurrence—and required measurement at fixed intervals after serum administration.

The Group's most consequential contribution was popularizing the Kaplan-Meier method, a survival analysis technique developed for industrial reliability testing and adapted for infection trials by the American biostatistician Paul Kaplan in 1971. Where earlier medical statistics measured success as binary cure or failure at a single timepoint, Kaplan-Meier analysis tracked patient status continuously over months, accounting for partial recovery, relapse, and death at varying intervals. This proved essential for serum therapy, where improvement was often slow and incomplete, and where reinfection distinguished initial serum response from genuine recovery. By 1960, the Working Group had mandated Kaplan-Meier reporting for all multicenter trials submitted to the Geneva Sanitary Bureau.

The Group also standardized dosing protocols for serum administration. Hyperimmunized animal sera varied in antibody concentration and potency depending on the host species, the bacterial strain used for immunization, and the interval since last booster vaccination. A single batch of equine serum raised against Streptococcus pyogenes might contain vastly different antibody titers than a supposedly identical batch from a different serum farm. The Working Group's 1958 guidelines established standardized potency units, required batch-to-batch testing against reference strains held at the Pasteur Institute, and mandated dose calculation by patient weight, infection site, and measured serum concentration. These measures reduced the variance between trials and created the first reliable quantitative basis for comparing serum therapy outcomes across borders.

Membership in the Working Group expanded from 23 founding members in 1956 to 67 by 1975, drawing participants from government public health agencies, teaching hospitals, serum manufacturers, and the International Serum Registry, which the Group helped establish in 1975 as a centralized database for tracking serum therapy outcomes and managing biological resource allocation. Meetings convened annually in Geneva, with working subgroups addressing specific bacterial species, patient populations, and emerging statistical problems. The Group's proceedings, published annually as Standardized Protocols for Infection Outcome Assessment, became the de facto reference manual for hospital bacteriology laboratories across Europe, North America, and Japan.

By the 1970s, the Working Group's influence extended beyond statistical method to clinical practice itself. Its 1973 recommendation that serum therapy outcomes be measured blind—that is, with evaluators unaware of which patients received serum and which received comparison treatments—was initially controversial in countries where such comparisons seemed unethical given the lack of alternative systemic treatments. Yet the Group's argument prevailed: without blinded measurement, hospital enthusiasm, physician expectation, and subtle differences in supportive care could inflate reported efficacy by 15 to 20 percent. The standardization that began as a measurement problem became a requirement for ethical trials.

The Working Group's archives, held at the Geneva Sanitary Bureau, contain meeting minutes from 1956 to the present, correspondence between member statisticians, rejected and accepted protocol revisions, and the complete documentation of how statistical standards evolved in response to emerging problems in serum therapy trials. This record shows that the Group was not merely imposing external standards but rather negotiating constantly between the practical constraints of hospital medicine, the interests of competing serum manufacturers, and the theoretical ideals of statistical rigor. The resulting compromise—imperfect standardization that nonetheless enabled meaningful comparison across national boundaries—reflected the Group's understanding that coordination mattered more than perfection in a world where untreated bacterial infection remained a constant threat to hospital populations.