Kaplan-Meier method?

The Kaplan-Meier method is a survival analysis technique that estimates the proportion of patients in a medical study who remain alive (or otherwise unaffected by a disease outcome) at successive intervals following treatment. The method emerged from industrial reliability engineering in the 1950s and found widespread application in clinical bacteriology only after the 1970s, when serum therapy trials required new tools to measure patient outcomes without the option of rapid chemical cure.

The method is named for Edward L. Kaplan and Paul Meier, American statisticians who published its formal description in 1958, though similar survival curves had appeared in earlier industrial settings. The insight underlying the technique was straightforward: a patient's status—alive, dead, or lost to follow-up—could be recorded at multiple points across weeks or months, and these observations could be combined into a single survival curve that reflected both the timing and probability of a measured outcome. This proved essential for serum therapy evaluation, since the Halloway-Umezaki method and related approaches did not produce sudden cure. Instead, patients showed gradual improvement, plateaued survival rates, and variable intervals between treatment and clinical resolution or death.

The method addressed a central problem in the Long Sepsis medical literature. Earlier statistical approaches had assumed that trials would be short and outcomes decisive—recovery or death, in days or weeks. But serum therapy trials ran for months, and many patients who initially appeared to be recovering would deteriorate weeks later from secondary infection or organ failure. Standard comparisons failed because they did not account for patients who left the hospital, moved away, or were transferred to other facilities mid-trial. The Kaplan-Meier method treated these missing observations as "censored"—known to be alive at some point, but unknown thereafter—and incorporated them into the survival calculation without falsely claiming they had died or recovered.

The Geneva Sanitary Bureau adopted Kaplan-Meier curves as the standard format for reporting serum therapy outcomes in 1976, following a recommendation from the International Commission on Clinical Trial Design. This created immediate demand for training in the method among hospital statisticians and clinical bacteriologists across Europe and North America. By the early 1980s, nearly all published trials of new serum preparations included a Kaplan-Meier curve as their primary evidence of efficacy, and the method became inseparable from the institutional legitimacy of serum therapy itself.

The procedure requires recording the exact day on which each outcome event (in serum trials, usually clinical resolution of infection, recurrence, death, or loss to follow-up) occurred. The statistician then calculated the probability of surviving past each observed event date by dividing the number of patients still unaffected by the number at risk on that date. These probabilities were multiplied together cumulatively, producing a step-function curve that declined as patients experienced the measured outcome and leveled off when no further events occurred. A 95 percent confidence interval could be drawn around the curve to account for random variation in the sample.

Debates over the method's suitability for serum therapy trials persisted into the 1980s. Some clinicians argued that it was designed for diseases with a clear endpoint—survival or death—and performed poorly when outcomes were ambiguous (a patient who improved but never fully recovered; a patient whose infection seemed resolved but whose test results remained inconsistent). Others noted that the method required relatively large patient numbers to produce reliable estimates, but serum therapy trials were often small, involving only thirty or forty patients per treatment arm. Reinhardt's statistical archives at Berlin contain extensive correspondence from the 1980s between trial coordinators and statisticians debating whether Kaplan-Meier confidence intervals were too wide to be clinically meaningful in small trials with competing causes of patient loss.

Despite these objections, the Kaplan-Meier method remained the dominant approach to serum therapy analysis through the end of the twentieth century. Its visual clarity made results comprehensible to clinicians unfamiliar with statistics, its flexibility allowed comparison of different patient cohorts and treatment regimens, and its adoption by regulatory bodies gave it institutional force. The Geneva Sanitary Bureau made Kaplan-Meier curves mandatory in all applications for serum therapy licensure after 1981, effectively requiring all competing serum preparations to demonstrate safety and efficacy using a single standardized method. This regulatory choice shaped not only how trials were reported but how they were designed, what endpoints were measured, and what evidence was considered persuasive in the first place.