The effects of systolic blood pressure and total cholesterol on cardiovascular disease mortality

Purpose To examine the effect of systolic blood pressure (SBP) and total cholesterol adjusted for the effect of age, race, and sex. The effect estimate of SBP is also independent of cholesterol level and vise versa. The focus of the study is to determine if the effect strength of these two variables changes along their whole ranges. Methods 26,913 participants of this study were from 4 study samples: Charleston Heart Study, Evans County Study, NHANES I and NHANES II. Of these 26,913 participants, 12,366 were Caucasian (Cau) females, 9,888 were Cau males, 2,725 were African American (AA) females, and 1,934 were AA males. These individuals were followed an average for 16 years (SD = 6 years). Starting age for them was from 25 years to 97 years (mean = 51 and SD = 14). Lowess curves were used to provide initial heuristics about possible segments if any is needed for the risk factors. Then, the most efficient Cox PH model was identified from the models with alternate function forms. Results Two segments were needed to best reflect the effect of total cholesterol (below 220 mg/dl, 220 mg/dl or above); 3 segments were needed to best reflect the effect of SBP (below 110 mmHg, 110 to 139 mmHg, 140 mmHg or above). The hazard ratio related to 20 mg/dl increase in cholesterol is 0.980 (not significantly different from 1) until it reaches 220 mg/dl. Beyond this cutoff, 20 mg/dl increase is related to 6% increase (p < 0.0001) in the hazard for CVD mortality. The variation in SBP under 110 mmHg does not have a significant effect on the hazard for CVD mortality (p = 0.14); from 110 to 139 mmHg, 20 mmHg increase in SBP is related to 49% (p < 0.0001) hazard increase and above this range the same amount increase in SBP is related to 27% hazard increase for CVD mortality (p < 0.0001). Conclusions The relative hazard for CVD mortality related to prehypertension is high (greater effect in the range from 110 to 139 mmHg than in the range beyond). Cholesterol effect is as expected.