5 - Life Insurance and the Risk Factor

Summary

In tracing the evolution of the concept of normal blood pressure in clinical medicine, it is surprising to find that the definition of its normality depended largely on the results of statistical studies by life insurance companies. (1952)

Most of the pioneer studies of blood pressure were done by and for insurance companies, and the tables are still accepted in all textbooks as the basis of blood pressure levels. Insurance statisticians did not set out to make contributions to human physiology; they were interested in the range of blood pressure in which they could establish a profitable insurance premium. (1939)

Life insurance companies devised a fundamentally new statistical approach to predicting chronic disease as they improved the process of selecting policyholders. They discovered that the risk of premature mortality was increased by specific personal characteristics that could be determined by analyses of policyholder mortality rates. Once the characteristics were identified, the companies required the physicians whom they employed as medical examiners to measure them in their medical examinations of applicants.

Life insurance was the one field of commercial endeavor that was totally and irreversibly committed to mathematical statistics. Life tables based on statistical analyses of policyholder mortality rates were used to predict the proportion of policyholders who would die every year. Because the predictions were not completely accurate, the companies maintained financial reserves to protect themselves against years with unusually high mortality rates. Other kinds of statistical analyses enabled the companies to estimate the amount of variation in annual mortality rates and select their financial reserves rationally. Statisticians recognized the industry's commitment to statistics, and Quetelet, for example, devised several mortality tables beginning in 1826 to provide a basis for life insurance in Belgium.

By contrast, few physicians and public health officials knew anything of statistics. Because of recent discoveries in bacteriology, medical education was placing increasing emphasis on the laboratory sciences and students were required to learn experimental techniques. Few if any medical school graduates had any knowledge of statistics. Public health officials were preoccupied with the exacting task of improving the accuracy and completeness of enumerations of births and deaths. Their efforts were devoted to determining the sources of error and improving reporting, not to more sophisticated methods of analyzing the enumerations.