Will lowering population levels of serum cholesterol affect total mortality? Expectations from the Honolulu Heart...

Major campaigns now underway to reduce the serum cholesterol levels of entire national populations have not given serious consideration to the high rates of noncardiovascular disease and death associated with low cholesterol levels (less than 190 mg/dl). To explore this problem, the relationships between serum cholesterol levels, measured in 1965-1968 in 7478 Japanese American men in Hawaii, and subsequent total and cause-specific mortality through 1985, were analyzed by multivariate Cox regression to control for potential confounders. Total mortality rates for 1648 deaths showed a U-shaped curve by baseline cholesterol level, with significant inverse trends (p less than 0.03) for deaths due to hemorrhagic stroke, all cancer, benign liver disease, chronic obstructive lung disease and "unknown cause". Only the inverse trends for cancer and benign liver disease showed flattening when 227 deaths in the first 5 years of follow-up were deleted from the analysis. Simulation models using three different strategies of cholesterol reduction in this cohort revealed that none of these approaches had any substantial impact on predicted total mortality over 15 years. However, the population-based approach might theoretically increase mortality for 60% of the cohort with baseline cholesterol levels less than 225 mg/dl.