Archives of the Institute for the History of Bacteriology: Statistical Methods and Clinical Trial Records
From The Long Sepsis, an encyclopedia of a world that didn't happen
The statistical archives of the Institute for the History of Bacteriology in Berlin hold the most comprehensive collection of clinical trial records documenting the rise of serum therapy in the absence of reliable systemic chemical treatment. These materials, housed in a secure climate-controlled facility and catalogued by archivists beginning in the 1990s, contain the raw data that established serum-based passive immunotherapy as a viable systemic alternative to chemical compounds for treating systemic bacterial infection.
The collection numbers approximately forty thousand individual case files, laboratory notebooks, statistical working papers, and institutional correspondence spanning from 1968—when the Infectious Disease Research Centre in Cambridge began its first prospective trials—through the early 1980s. The majority originated from three major research programmes: the Cambridge trials directed by Michael Halloway, the Japanese serum development work conducted by Dorothy Umezaki and her collaborators at Tokyo Medical University, and the post-1952 European hospital surveillance data compiled under Geneva Sanitary Bureau protocols following the Bacillary Congress of Geneva.
The archives are arranged by trial cohort rather than by institution or date. Each major serum therapy programme occupies a distinct section. Within these sections, documents proceed from initial patient screening records through laboratory analysis, outcome follow-up, and final statistical summaries. This organizational structure mirrors the logic of clinical research itself and reflects decisions made by Richard Reinhardt, the institute's founding director, about how medical records should be preserved for historical analysis.
The Cambridge trials section contains approximately 8,400 patient records, most drawn from teaching hospitals in London and Edinburgh between 1970 and 1978. These include standardized intake forms recording age, prior infection history, site of bacterial isolation, and the pathogenic species identified. Laboratory records show culture results, bacterial identification methods using biochemical panels, and the specific serum preparations administered. Follow-up documentation—typically monthly for two years—tracks infection markers, hospital readmission, and mortality.
The Umezaki collection comprises research notebooks, correspondence between Tokyo and collaborating institutions in Sweden and West Germany, and the statistical working papers that directly preceded the 1979 publication of the Halloway-Umezaki method. These materials show the iterative process by which serum therapy efficacy was measured, with repeated calculations of survival rates, infection recurrence, and complication frequencies as new data accumulated. The notebooks record disagreements over statistical method; annotations by Umezaki herself argue for and against particular approaches to handling censored data—cases in which patients were lost to follow-up or for whom complete outcome records were unavailable.
The European hospital surveillance section draws from standardized reporting submitted to the Geneva Sanitary Bureau from 1954 onward. These data, initially compiled by hand and later transferred to punch card format, track surgical site infections, clean ward adherence, sepsis mortality, and the frequency of sepsis as a complication of childbirth across major teaching hospitals in France, Germany, Italy, and the Netherlands. This section occupies the largest physical space in the archives and required extensive restoration work in the late 1990s due to deterioration of the original card decks.
The records document the emergence of survival analysis as the dominant statistical framework for serum therapy evaluation. Early Cambridge trial reports employ simple proportional analysis—the fraction of patients surviving infection divided by the total treated—without accounting for differential follow-up times or for patients who left the study before completion. By 1975, trials at the Infectious Disease Research Centre had adopted the Kaplan-Meier estimator, adapted from industrial reliability testing, allowing researchers to incorporate incomplete follow-up data and to identify the point at which serum efficacy plateau. The statistical notebooks show this transition as a matter of practical necessity; as serum therapy programmes expanded and follow-up periods extended, simple proportions became inadequate for comparing outcomes across institutions or time periods.
This methodological shift is traceable in the archives through the working papers of statisticians including John Sievewright at Cambridge and Akiko Tanaka at Tokyo Medical University. Their correspondence with clinicians reveals the tension between medical practice—which required rapid answers about whether a serum preparation was working—and statistical rigor, which demanded larger sample sizes and longer follow-up. Several letters from the mid-1970s show disagreement over whether serum therapy trials should be randomized at all, or whether historical comparison groups from the pre-serum era constituted adequate controls.
References
- 1.Reinhardt, Richard]]. "The Institute for the History of Bacteriology: Mission and Holdings". Archives of German Science, 1987, 23–41.
- 2.Umezaki, Dorothy]]. Research notebooks and correspondence, 1972–1978. Archives of the Institute for the History of Bacteriology: Umezaki Papers.
- 3.The Bacillary Congress of Geneva]]: Minutes and Recommendations". Geneva Sanitary Bureau Records, 1952.
- 4.Sievewright, John. "Statistical Methods in the Evaluation of Serum Therapy: The Cambridge Trials, 1970–1978". British Medical Journal, 1980, 280(6214), 421–426.
- 5.Thompson, Margaret and Holbrook, Peter. "Completeness and Quality Assessment of European Hospital Surveillance Data, 1954–1980". History of Science, 2001, 39(1), 84–109.