Some demographic characteristics of an autopsied population

Abstract

The ways in which a series of autopsied hospital patients may or may not differ from other hospital deaths and from deaths in the general population are necessarily of concern to investigators conducting studies based upon postmortem observations, as well as to those concerned with epidemiological and sociological aspects of hospitalization and autopsy. In an effort to investigate this problem, certain demographic characteristics of a population of adults autopsied during a 4-year period at a large general teaching hospital were examined. This population was compared with its parent population, all deaths that occurred at this hospital, and with the population from which the latter, in turn, was derived, all deaths in the northeastern metropolitan county served by this hospital. The variables analyzed were age, sex, nativity, race, marital status, size of residence, place, religious preference, and cause of death.The composition of the autopsied population indicated that it had been significantly affected by various selective factors. Age exhibited perhaps the most striking effects: as age at death decreased, the likelihood increased that a person who died in this county had been hospitalized and autopsied at the institution under study. This trend was especially marked among females. Subsequent analyses were therefore adjusted to control the effects of age and sex.Certified cause of death also was found to be highly subject to selective factors. In particular, the proportions of all deaths in the county from heart disease that occurred and were autopsied at this hospital were strikingly smaller than those of any other of the causal categories employed.With respect to religion, there was suggestive evidence that deceased hospital patients who had indicated a Jewish preference were considerably less likely to be autopsied than those who had designated a Protestant group. Catholics appeared somewhat less likely to be autopsied than Protestants.Other differences found were that the foreign-born were somewhat more likely to be hospitalized at this institution than the native-born, but less likely to be autopsied. Negroes, the divorced, and residents of other states and of places in the state with less than 10,000 inhabitants were more likely to be hospitalized and autopsied at this hospital than those in other categories of race, marital status, and place of residence.In general, deaths at the hospital differed more from deaths in the general population than the autopsied deaths differed from all deaths at the hospital.These findings indicate a few of the kinds of selection that may be found in the composition of autopsied and hospital death populations. They are interpreted as pointing up the need for giving special attention to the effects of selective factors on populations used in autopsy-based research. More specifically, it is hoped that these analyses may illustrate a practicable means by which the extent of selection in a particular autopsy series may be gauged. Usually it should be possible to examine the distributions of variables that may be relevant to a particular study, to compare these with what is known regarding their distributions in other populations, and to control, by special analyses, the effects of such variables as appear to require this with at least sufficient precision to permit extending tentative inferences beyond the immediate study material. Where a unique contribution can be provided by postmortem observations, their use appears justified in the study of many disease problems with the qualification that appropriate care be given to the exploration and control of potential sources of bias.