Are autopsy studies reliable in assessing gallstone prevalence in the community?

Abstract

Autopsy studies have been used for many years to estimate gallstone prevalence in the population, although this has been questioned because of the risk of selection bias. Gallstone prevalence measured by ultrasonographic screening of a random population was compared with gallstone prevalence assessed by routinely collected autopsy data in the same area during the same time period. To assess whether recording of gallstones detected in routinely performed autopsies was valid, a prospective investigation of gallstone prevalence in consecutive autopsies was performed. The age-standardized gallstone prevalences were 9.5 in the population and 9.2 in the autopsy study among males, whereas the corresponding figures were 16.1 and 15.9 among females. These figures were not significantly different. Gallstone prevalence in the prospective study was significantly higher than in the retrospective autopsy study (odds ratio [OR] = 1.54, 95% confidence interval [CI]: 1.08-2.19), suggesting inadequate recording of gallstones in routinely performed autopsies. The proportions of small stones were equal in the prospective autopsy study and the screening study, indicating that autopsy is not superior to ultrasonography in detecting gallstones. Routinely collected autopsy data may be a good estimate of gallstone prevalence in the community, but this could be coincidental and due to bias acting in opposite directions, with higher true prevalence balanced by inadequate detection of gallstones in routine autopsy studies.