Association of cholesterol, LDL, HDL, cholesterol/ HDL and triglyceride with all-cause mortality in life insurance applicants.

Objective Determine the relationship between various lipid tests and all-cause mortality in life insurance applicants stratified by age and sex. Method By use of the Social Security Death Master File, mortality was determined in 1,488,572 life insurance applicants from whom blood samples were submitted to Clinical Reference Laboratory. There were 41,020 deaths observed in this healthy adult population during a median follow-up of 12 years (range 10 to 14 years). Results were stratified by 4 age-sex subpopulations: females, ages 20 to 59 or 60+; and males, ages 20 to 59 or 60+. Those with serum albumin or = 2.1 mmol/L were excluded. The middle 50% of lipid values specific to each of these 4 age-sex subpopulations was used as the reference band. The mortality rates in bands representing other percentiles of lipid values were compared with the mortality rate in the reference band within each age-sex subpopulation. Results In contrast to some published findings from general populations, lipid test results are only moderately predictive of all-cause mortality risk in a life insurance applicant population and that risk is dependent on age and sex. At ages below 60, HDL values are associated with a J shaped mortality curve and at ages 60+, total cholesterol is associated with a U shaped curve. The total cholesterol/HDL ratio may serve as a useful single measure to predict mortality risk, but only if stratified by age and sex, and only if high HDL values at younger ages and lower total cholesterol values at ages 60+ are recognized as being associated with increased risk as well. Using LDL or non-HDL cholesterol instead of total cholesterol does not improve mortality risk discrimination; neither does using total cholesterol or triglyceride values in addition to the total cholesterol/HDL ratio. Conclusion The total cholesterol/HDL ratio is the best single measure of all-cause mortality risk among the various lipid tests but is useful only if viewed on an age- and sex-specific basis and is only a modest risk predictor.