Objective: Evaluate the association of selected laboratory tests including liver function, kidney function, bilirubin, and albumin, with all-cause mortality
Method: Associations between several laboratory tests and mortality were assessed in two longitudinal datasets: (1) the Third National Health and Nutritional Examination Survey (NHANES III) with 4610 deaths during the 240,428 person-year follow up: and (2) a life insurance dataset containing historical life insurance policies issued as standard and better risk class, with 837 death claims generated during the approximate 1.4 million person-years of follow up. A Cox proportional hazards model was used to compute the hazard ratio of each selected laboratory test while adjusting for age, gender, and other health conditions. Separate analyses were conducted for laboratory results within and beyond the respective normal clinical range.
Results: When outside the normal clinical range, the results of several selected laboratory tests were associated with higher mortality, as indicated by a hazard ratio greater than 1. Comparisons of hazard ratios when laboratory results were within the normal range demonstrated that blood albumin and blood urea nitrogen (BUN) were both negatively associated with mortality and alkaline phosphatase (AP) was positively associated with mortality. The associations were shown to be independent and were consistent in two datasets.
Conclusions: Although abnormal laboratory results are significant predictors of higher mortality, when results fall within the normal clinical range, only three tests--albumin, AP, and BUN--provided mortality differentiation. These findings support the utilization of the actual, continuous value from albumin, AP, and BUN tests to evaluate mortality during underwriting, even when these results are categorized as clinically normal. Furthermore, the results provide an insightful perspective for evaluating the utility of recently developed, laboratory test-based underwriting toools.