Vitamin D Deficiency is a Predictor of Survival in HF
Vitamin D Deficiency is a Predictor of Survival in HF
All data associated with members of the largest HMO (Clalit health services) in Jerusalem, Israel are digitally recorded in a central computerized database. The database includes demographics, comprehensive clinical data, diagnoses, and all laboratory data undertaken in a single centralized laboratory of the HMO. We identified and retrieved electronically from the computerized database all members, 45 years of age or older, with available measurements of vitamin D during the period between January 2006 and June 2010. Clinical data were retrieved and analysed. A total of 49 834 HMO members underwent 25-hydroxyvitamin D [25(OH)D] measurement from a total membership of ~122 700 in this age group (40.6%). These individuals comprised the cohort of this study. Members of the cohort with a diagnosis of HF as coded by the database (3009 patients) were identified and compared with the rest of the members in the HMO. The HF patients in the cohort comprised 45.5% of all patients diagnosed with HF in the HMO (6618 patients). Validation of the diagnosis of HF was performed on a randomly computer-generated 5% of the diagnosed HF patients (n = 338). Clinical parameters in this group of patients were statistically comparable with those of the whole HF cohort. We reviewed all available data from medical records and hospital admissions. In this group, 99% fulfilled the European Society of Cardiology (ESC) criteria for the diagnosis of HF. Only 1% (n = 4) had equivocal clinical data for HF diagnosis. Retrieval of data regarding prescribed medications was performed only in patients with HF. Data on prescribed vitamin D supplementation during follow-up were retrieved from the database. The mean monthly global daily solar radiation (MJ/m) was obtained from the local meteorological service. Data on mortality were retrieved from the National Census Bureau and included mortality up to January 2011. The Institutional Committee for Human Studies of Clalit Health Services approved the study protocol.
Serum levels of 25(OH)D were measured using a radioimmunoassay kit (DiaSorin, Stillwater, MN, USA) in a single centralized laboratory of the HMO. The intra-assay and interassay coefficients of variation of this assay were 3.8% and 7.9%, respectively. We chose to analyse the first measurement of 25(OH)D done on each individual as this would best represent the baseline 25(OH)D status prior to any intervention that could artificially modify 25(OH)D levels. There is no consensus on a definite cut-off for vitamin D deficiency; however, 25(OH)D levels <25 nmol/L are considered by most clinicians as definite vitamin D deficiency.
SPSS version 17.0 for Windows (SPSS Inc., Chicago, IL, USA) was used in all analyses. Comparison of the clinical characteristics between HF patients and the control group, and groups of HF patients according to vitamin D levels, was performed using the Student t-test or Mann–Whitney U-test for continuous variables and the χ test for categorical variables. Multivariate binary logistic regression analysis models were used to identify independent predictors of vitamin D deficiency [serum 25(OH)D levels <25 nmol/L] in both groups. Clinical predictors were transformed where appropriate. Log10 was used for logarithmic transformations. Follow-up time was calculated using a Kaplan–Meier estimate of potential follow-up. Kaplan–Meier curves, with the log-rank test, were used to compare survival according to 25(OH)D levels. Multivariate Cox proportional hazards regression analysis was used to evaluate independent variables that determined survival. Parameters included in the multivariate Cox regression analysis incorporated all significant clinical and laboratory parameters on univariate analysis as well as drug treatment. Parathyroid hormone was not included due to the limited number of patients (9%) with this measurement. Proportionality assumptions of the Cox regression models were evaluated by log–log survival curves and with the use of Schoenfeld residuals. Possible interactions were also assessed. A P-value of <0.05 was considered statistically significant.
Methods
Participants and Study Design
All data associated with members of the largest HMO (Clalit health services) in Jerusalem, Israel are digitally recorded in a central computerized database. The database includes demographics, comprehensive clinical data, diagnoses, and all laboratory data undertaken in a single centralized laboratory of the HMO. We identified and retrieved electronically from the computerized database all members, 45 years of age or older, with available measurements of vitamin D during the period between January 2006 and June 2010. Clinical data were retrieved and analysed. A total of 49 834 HMO members underwent 25-hydroxyvitamin D [25(OH)D] measurement from a total membership of ~122 700 in this age group (40.6%). These individuals comprised the cohort of this study. Members of the cohort with a diagnosis of HF as coded by the database (3009 patients) were identified and compared with the rest of the members in the HMO. The HF patients in the cohort comprised 45.5% of all patients diagnosed with HF in the HMO (6618 patients). Validation of the diagnosis of HF was performed on a randomly computer-generated 5% of the diagnosed HF patients (n = 338). Clinical parameters in this group of patients were statistically comparable with those of the whole HF cohort. We reviewed all available data from medical records and hospital admissions. In this group, 99% fulfilled the European Society of Cardiology (ESC) criteria for the diagnosis of HF. Only 1% (n = 4) had equivocal clinical data for HF diagnosis. Retrieval of data regarding prescribed medications was performed only in patients with HF. Data on prescribed vitamin D supplementation during follow-up were retrieved from the database. The mean monthly global daily solar radiation (MJ/m) was obtained from the local meteorological service. Data on mortality were retrieved from the National Census Bureau and included mortality up to January 2011. The Institutional Committee for Human Studies of Clalit Health Services approved the study protocol.
Vitamin D Levels
Serum levels of 25(OH)D were measured using a radioimmunoassay kit (DiaSorin, Stillwater, MN, USA) in a single centralized laboratory of the HMO. The intra-assay and interassay coefficients of variation of this assay were 3.8% and 7.9%, respectively. We chose to analyse the first measurement of 25(OH)D done on each individual as this would best represent the baseline 25(OH)D status prior to any intervention that could artificially modify 25(OH)D levels. There is no consensus on a definite cut-off for vitamin D deficiency; however, 25(OH)D levels <25 nmol/L are considered by most clinicians as definite vitamin D deficiency.
Statistical Analyses
SPSS version 17.0 for Windows (SPSS Inc., Chicago, IL, USA) was used in all analyses. Comparison of the clinical characteristics between HF patients and the control group, and groups of HF patients according to vitamin D levels, was performed using the Student t-test or Mann–Whitney U-test for continuous variables and the χ test for categorical variables. Multivariate binary logistic regression analysis models were used to identify independent predictors of vitamin D deficiency [serum 25(OH)D levels <25 nmol/L] in both groups. Clinical predictors were transformed where appropriate. Log10 was used for logarithmic transformations. Follow-up time was calculated using a Kaplan–Meier estimate of potential follow-up. Kaplan–Meier curves, with the log-rank test, were used to compare survival according to 25(OH)D levels. Multivariate Cox proportional hazards regression analysis was used to evaluate independent variables that determined survival. Parameters included in the multivariate Cox regression analysis incorporated all significant clinical and laboratory parameters on univariate analysis as well as drug treatment. Parathyroid hormone was not included due to the limited number of patients (9%) with this measurement. Proportionality assumptions of the Cox regression models were evaluated by log–log survival curves and with the use of Schoenfeld residuals. Possible interactions were also assessed. A P-value of <0.05 was considered statistically significant.