Statistical methods for bioarchaeology: applications of age-adjustment and logistic regression to comparisons of skeletal populations with differing age-structures

Previous studies of patterns of osteoarthritis involvement in prehistoric skeletal populations have been used to suggest differences in lifestyle or subsistence patterns. In spite of the intriguing nature of such studies, inadequate controls for the potential confounding effects of age have made results less secure than they appear. While researchers remain unable to control for the inherent limitations imposed by preservation bias, meaningful advancement could be made in the analysis of these data. Here we propose two methods from epidemiology that would allow more robust comparisons between populations with differing age-structures: age-adjustment and logistic regression. We apply these two methods to previously published data from prehistoric populations of the American Great Basin. In doing so, we are able to discern sex differences in risk for osteoarthritis at the shoulder (beta = −1.82, SE = 0.74, OR = 0.16, 95% CI = 0.04–0.69, p = 0.013), elbow (beta = −2.07, SE = 0.82, OR = 0.13, 95% CI = 0.03–0.63, p = 0.011), and in the foot (beta = −3.11, SE = 1.57, OR = 0.04, 95% CI = 0.00–0.97, p = 0.048), that were not noted in previous analyses of these data. These findings suggest that greater clarity of results and more precise comparisons of risk may be achieved by using these methods. We conclude that further application of these methods to bioarchaeological studies would potentially improve what can be said about prehistoric health patterns and lifeways.