Cornell medical index responses and mortality in a Polynesian population

The Cornell Medical Index (CMI) is a health questionnaire consisting of a somatic division, divided into 12 sections organized by organ systems, and a psychosocial division. It has been suggested that the CMI taps a dimension of the perception of illness before clinical measurements detect disease. However, this has not been confirmed in nonwestern populations. Our purpose was to determine if the CMI predicted 6-year mortality in a Polynesian population. In 1975, 461 American Samoan men responded to the CMI. Thirty-one died between 1976 and 1981, 13 of cardiovascular disease (CVD). A logistic model was used to predict both total and CVD mortality during the follow-up period. Positive responses to the total, somatic, and psychosocial divisions of the CMI did not significantly improve prediction of mortality compared to models using only age. An analysis using section scores showed that several somatic sections were associated with CVD mortality, but none were associated with total mortality. This study suggests that the CMI is not a strong predictor of mortality and, by extension, of existing health status in American Samoa. Even if the CMI does measure the perception of illness, one cannot assume a priori that it is valid in any given anthropological population. Rather, the instrument must be pretested in each particular cultural setting.