Aging and total cholesterol levels: cohort, period, and survivorship effects

This analysis describes the association of age with the serum total cholesterol level in 5,010 participants in a geriatric health screening program. Cholesterol levels were measured annually in participants monitored for up to 12 years. The association of age with cholesterol level is described via three approaches: cross-sectional analysis, descriptive longitudinal analysis, and longitudinal analysis using statistical modeling. The results were compared to examine the influence of cohort, period, and survivorship effects on the association between age and cholesterol. In cross-sectional analysis, the cholesterol level was fairly constant for the ages of 65 to 75 years, but decreased by 21% over the age range from 75 to 95 years. Descriptive longitudinal analysis suggested that both cohort and period effects were influencing the cross-sectional findings. In longitudinal analysis adjusting for both cohort and period effects, the findings were similar to those from cross-sectional analysis for the ages of 65 to 75 years, but from the ages of 75 to 95 years, cholesterol decreased by only 9%--half as great a decline as that estimated from cross-sectional analysis. When longitudinal data were limited to those with complete follow-up, the predicted decline for the age range from 75 to 95 years was only 6%. Although this flattening of the age trend was suggestive, there was no conclusive evidence that it reflected an association between baseline cholesterol and loss to follow-up.