Use of self-reported height and weight biases the body mass index-mortality association

BACKGROUND: Many large-scale epidemiologic data sources used to evaluate the body mass index (BMI: kg/m(2)) mortality association have relied on BMI derived from self-reported height and weight. Although measured BMI (BMI(M)) and self-reported BMI (BMI(SR)) correlate highly, self-reports are systematically biased. OBJECTIVE: To rigorously examine how self-reporting bias influences the association between BMI and mortality rate. SUBJECTS: Samples representing the US non-institutionalized civilian population. DESIGN AND METHODS: National Health and Nutrition Examination Survey data (NHANES II: 1976-80; NHANES III: 1988-94) contain BMI(M) and BMI(SR). We applied Cox regression to estimate mortality hazard ratios (HRs) for BMI(M) and BMI(SR) categories, respectively, and compared results. We similarly analyzed subgroups of ostensibly healthy never-smokers. RESULTS: Misclassification by BMI(SR) among the underweight and obesity ranged from 30–40% despite high correlations between BMI(M) and BMI(SR) (r>0.9). The reporting bias was moderately correlated with BMI(M) (r>0.35), but not BMI(SR) (r<0.15). Analyses using BMI(SR) failed to detect six of eight significant mortality HRs detected by BMI(M). Significantly biased HRs were detected in the NHANES II full dataset (χ(2) = 12.49; p = 0.01) and healthy subgroup (χ(2) = 9.93; p = 0.04), but not in the NHANES III full dataset (χ(2) = 5.63; p = 0.23) or healthy subgroup (χ(2) = 1.52; p = 0.82). CONCLUSIONS: BMI(SR) should not be treated as interchangeable with BMI(M) in BMI-mortality analyses. Bias and inconsistency introduced by using BMI(SR) in place of BMI(M) in BMI-mortality estimation and hypothesis tests may account for important discrepancies in published findings.