The benefit of cholesterol-lowering for the prevention of atherosclerotic cardiovascular disease has been extensively shown in the literature, including clinical trials, supporting its widespread recommendation in clinical practice1, 2, 3. However, some—though not all— observational studies have reported an increased risk of death in individuals with low cholesterol concentrations 4, 5, 6. Based on such observations, some researchers have challenged the cholesterol-lowering recommendations in existing guidelines 5,6. However, the mechanism of this elevated mortality risk in low cholesterol is not fully elucidated. For example, an observational study suggested that the first five years of death from liver diseases and all-cause mortality confounded the observed relation between low total cholesterol (TC) and mortality 7. In another study, however, exclusion of either deaths from liver diseases or all-cause death within the first five years did not alter the elevated risk of death in the low TC 8. Furthermore, many observational studies merely reported the elevated risk of death from all-cause, cardiovascular diseases (CVD) and cancer in individuals with low TC or low-density lipoprotein cholesterol (LDL-C), but did not provide clear reason (s) for their findings7, 8, 9, 10.

We believe that the observed elevated mortality risk in low TC was most likely due to insufficient control for confounders, considering the evidence from cholesterol-lowering trials 1,11. In the present study, we attempt multiple analytical maneuvers to identify potential confounders in the relationship between low TC and mortality risk using a longitudinal observational dataset. Identifying such confounders is of clinical implication, because it may help clinicians to manage a patient with unintended low TC by informing which characteristics, among many coexisting factors with low TC, are the ones that cause poor prognosis.

Our objectives are: (1) to characterize individuals with low TC who were not using cholesterol-lowering medications at baseline, and (2) to identify potential confounders through a series of analytical maneuvers such as exclusion and adjustment.