Paradoxical association between body mass index and mortality in men with CKD not yet on dialysis

Background: Low body mass index (BMI) is associated with greater mortality in patients on dialysis therapy. This relationship is less well characterized in patients with chronic kidney disease (CKD) who are not yet on dialysis therapy. Study Design: Historic prospective cohort. Setting & Participants: 521 male US veterans with CKD (age, 68.8 +/- 10.4 years; 21.3% black; estimated glomerular filtration rate, 37.5 +/- 16.8 mL/min/1.73 m(2) [0.62 +/- 0.28 mL/s/1.73 m(2)]) at a single medical center. Predictor: BMI. Outcomes & Measurements: Associations with all-cause mortality were explored in fixed-covariate and time-dependent Cox models and sequentially adjusted for demographic characteristics (age and race), case-mix (comorbidity index, smoking, blood pressure, estimated glomerular filtration rate, and medication use), and surrogates of malnutrition and inflammation (serum albumin, cholesterol, and bicarbonate levels; white blood cell count; percentage of lymphocytes; and hemoglobin level). Results: Patients were followed up for up to 5.5 years, and the mortality rate was 128.3 deaths/1,000 patient-years (95% confidence interval [CI], 110.5 to 149.0). Higher BMI was associated with lower mortality in the fixed-covariate Cox models, including the fully adjusted model (adjusted hazard ratios for mortality in the group with BMI in 10th to 50th, 50th to 90th, and >90th versus <10th percentiles, 0.75 [95% CI, 0.46 to 1.22], 0.56 [95% CI, 0.33 to 0.94], and 0.39 [95% CI, 0.17 to 0.87]; P-trend = 0.005). Associations were similar in a time-dependent Cox model (P-trend = 0.008 in the fully adjusted model). Limitations: Results may not be generalizable. Conclusions: Lower BMI is associated with greater mortality in patients with CKD not yet on dialysis therapy. Adjustment for case-mix and surrogate markers of malnutrition and inflammation attenuated, but did not reverse, this relationship.