A Comparison of NCEP and Absolute Risk Stratification Methods for Lipid‐Lowering Therapy in Middle‐Aged Adults: The ARIC Study

Lipid‐lowering therapy has been shown to reduce cardiovascular disease risk in individuals with low‐density lipoprotein cholesterol (LDL‐C) levels of ≥130 mg/dL. Several methods have been developed for identifying those at highest risk for treatment. The objective was to compare the 1993 National Cholesterol Education Program guidelines to absolute risk assessment methods similar to those used by recently developed European and British cholesterol treatment guidelines. The design and setting was a population‐based, prospective cohort study in four US communities. Among Atherosclerosis Risk in the Communities study participants aged 45–64 years followed for a mean of 7 years, 7314 women and 5762 men were included who had no evidence of clinical cardiovascular disease and were not receiving lipid‐lowering drug therapy at entry. Main outcome measures were number meeting criteria for treatment, number of cardiovascular events expected to be prevented with treatment, number needed to treat to prevent one cardiovascular event, and sensitivity of criteria for detecting cases for the entire cohort and for those at or above the LDL‐C treatment threshold (within‐group). The 1993 National Cholesterol Education Program guidelines identified 16% of participants for treatment: n=2112; events expected to be prevented=77; number needed to treat=27; sensitivity=0.31 (0.28–0.35); sensitivity when LDL‐C was ≥160 mg/dL=0.86. Use of a LDL‐C treatment threshold of ≥130 mg/dL, a ≥20% 10‐year absolute risk cut‐point (similar to European guidelines) identified 13% for treatment: n=1701; events expected to be prevented=66; number needed to treat=26; sensitivity= 0.29 (0.25–0.32); sensitivity when LDL‐C was ≥130 mg/dL=0.43. A ≥15% risk cut‐point (similar to British guidelines) identified 17% for treatment: n=2233; events expected to be prevented= 82; number needed to treat=27; sensitivity= 0.37 (0.33–0.40); sensitivity when LDL‐C was ≥130 mg/dL=0.55. Modified National Cholesterol Education Program guidelines (LDL‐C treatment threshold lowered from 160 to 130 mg/dL for those with ≥2 risk factors) identified 30% for treatment: n=3867; events expected to be prevented=120; number needed to treat=32; sensitivity=0.55 (0.52–0.59); sensitivity when LDL‐C was 130 mg/dL=0.83. A hybrid strategy using ≥15% risk, or 10%–14% risk with a parental history of premature coronary heart disease or treated, controlled hypertension, identified 20% for treatment: n=2646; events expected to be prevented=97; number needed to treat=27; sensitivity=0.44 (0.40–0.47); sensitivity when LDL‐C was ≥130 mg/dL=0.66.
Depending on the priorities of a health care system, higher absolute cut‐points can be used for allocating health care resources to those at the highest risk. However, fewer cardiovascular events will be prevented and fewer patients who later experience cardiovascular events will be offered treatment despite acceptable numbers needed to treat to prevent one cardiovascular disease event. A hybrid absolute risk strategy incorporating family history and treated hypertension information appears to be superior to absolute risk strategies that do not include this information.