Predictors of 25-hydroxyvitamin D levels in HIV-infected patients in Hawai'i.

HIV-infected individuals are at increased risk for several metabolic diseases, including low 25-hydroxyvitamin D [25(OH)D]. Data on the prevalence and risk factors for low 25(OH)D in HIV patients living in the tropics is scarce. Patients ≥ 40 years old on stable antiretroviral therapy were enrolled from March 2009 to July 2011 in Hawai'i (latitude 21° North). Chemiluminescent immunoassay (DiaSorin) was used to determine plasma 25(OH)D levels. Patients were grouped by whether 25(OH)D was collected in summer (May 1 - September 30) or winter (October 1 - April 30). Of 158 patients enrolled, 88 (56%) and 70 (44%) were enrolled in winter and summer, respectively. There were 57.6% Caucasians and 88% men. Over-all median (quartile1, quartile3) age was 51 (46, 57) years and median 25(OH)D was 32.4 (24.0, 41.0) ng/ml. Forty-three percent (n=68) had 25(OH)D<30.0 ng/ml. Median 25(OH)D levels were 29.6 (22.0, 38.0) ng/ml in winter and 36.9 (25.0, 44.5) ng/ml in summer (P = .01). Median body mass index (BMI) of winter patients was significantly higher (P = .03). By simple linear regression, log-transformed 25(OH)D was significantly associated with winter visit (β = -.0737, P = .01), ethnicity (Caucasian versus non-Caucasian, β = .1194, P < .01), BMI (β = -.0111, P < .01) and current use of zidovudine (β = -.1233, P = .03). In multiple linear regression, only Caucasian ethnicity (β = .1004, P < .01) and BMI (β = -.0078, P = .02) retained statistical significance. Seasonal variation in 25(OH)D was observed but the significance of winter visit was not preserved in the final multivariate model. Ethnicity and BMI were better predictors of 25(OH)D levels than season in the tropics.