Recent evidence supports an association between vitamin D deficiency and hypertension, peripheral vascular disease, diabetes mellitus, metabolic syndrome, coronary artery disease, and heart failure. The effect of vitamin D supplementation, however, has not been well studied. We examined the associations between vitamin D deficiency, vitamin D supplementation, and patient outcomes in a large cohort. Serum vitamin D measurements for 5 years and 8 months from a large academic institution were matched to patient demographic, physiologic, and disease variables. The vitamin D levels were analyzed as a continuous variable and as normal (≥30 ng/ml) or deficient (<30 ng/ml). Descriptive statistics, univariate analysis, multivariate analysis, survival analysis, and Cox proportional hazard modeling were performed. Of 10,899 patients, the mean age was 58 ± 15 years, 71% were women (n = 7,758), and the average body mass index was 30 ± 8 kg/m 2 . The mean serum vitamin D level was 24.1 ± 13.6 ng/ml. Of the 10,899 patients, 3,294 (29.7%) were in the normal vitamin D range and 7,665 (70.3%) were deficient. Vitamin D deficiency was associated with several cardiovascular-related diseases, including hypertension, coronary artery disease, cardiomyopathy, and diabetes (all p <0.05). Vitamin D deficiency was a strong independent predictor of all-cause death (odds ratios 2.64, 95% confidence interval 1.901 to 3.662, p <0.0001) after adjusting for multiple clinical variables. Vitamin D supplementation conferred substantial survival benefit (odds ratio for death 0.39, 95% confidence interval 0.277 to 0.534, p <0.0001). In conclusion, vitamin D deficiency was associated with a significant risk of cardiovascular disease and reduced survival. Vitamin D supplementation was significantly associated with better survival, specifically in patients with documented deficiency.
Cardiovascular disease is the most common cause of mortality and morbidity, accounting for nearly 30% of deaths in 2003 worldwide. Although multiple risk factors for coronary artery disease are well established, novel risk factors continue to emerge, in accordance with the findings from various epidemiologic studies. In particular, a growing body of evidence has identified vitamin D deficiency as a potential widespread risk factor for cardiovascular disease. In addition to its well-defined role in bone and calcium metabolism, vitamin D has been identified as an important factor in cardiovascular health. Recent evidence supports an association of vitamin D deficiency with hypertension, peripheral vascular disease, diabetes mellitus, the metabolic syndrome, coronary artery disease, and heart failure. The Third National Health and Nutrition Examination Survey reported the prevalence of vitamin D deficiency in the United States to be 25% to 57% of adults. Similar studies have shown that a prevalence of vitamin D deficiency of 30% to 50% in the general population. Although epidemiologic evidence for an association between vitamin D deficiency and several cardiovascular diseases is strong, the effect of vitamin D supplementation on patient outcomes is largely unknown. We studied the association of vitamin D deficiency and cardiovascular morbidity and mortality, as well as the effect of supplementation on survival.
Methods
This was an observational retrospective study using a cohort of patients followed up by a cardiovascular practice at a large academic medical center. Patients seen by the service from January 1, 2004 and October 8, 2009 with documented serum vitamin D levels were eligible for the present study.
Serum vitamin D measurements were performed by the clinical laboratory at the University of Kansas Hospital, Kansas City, Kansas. The laboratory uses the DiaSorin (Stillwater, Minnesota) chemiluminescence immunoassay method to measure the total serum vitamin D (both 25-hydroxyvitamin D2 and D3 forms of vitamin D). The laboratory assay did not change during the study period. The optimal concentration of 25[OH] vitamin D was defined as ≥30 ng/ml and vitamin D deficiency as a 25[OH]D level of <30 ng/ml.
The abstracted patient data included demographics, medical history, medications (including statins, vitamin D supplementation, and aspirin), and physiologic and disease state variables obtained from the electronic cardiovascular database at the medical center. The diagnoses were derived from the patient problem list documented in the patients’ electronic medical record and using the International Classification of Diseases, 9th Revision codes. Vitamin D supplementation was defined as an active prescription for vitamin D or its analogs or patient-reported use of vitamin D supplements. Not all doses of vitamin D were reported; however, the usual reported doses ranged from 1,000 IU/day to 50,000 IU biweekly (sample mean 2254 ± 316 IU). The use of multivitamins was not considered vitamin D supplementation. All-cause mortality data were obtained from the Social Security Death Index.
Statistics were performed using SAS, version 9.1.3 software (SAS Institute, Cary, North Carolina). The serum vitamin D levels were analyzed as both a continuous variable and a dichotomous variable (normal [≥30 ng/ml] vs deficient [<30 ng/ml]). Most analyses were done with vitamin D deficiency as a dichotomous variable, except as noted. However, repeating the analyses with continuous vitamin D levels did not change the outcomes. Descriptive statistics, univariate analysis (unpaired t tests for continuous variables, chi-square analysis for categorical variables), multivariate logistic regression analysis for odds ratios (OR) and 95% confidence intervals (CIs), survival analysis, and Cox proportional hazard modeling were performed. Interaction between vitamin D deficiency and supplementation was studied by development of an interaction variable for retesting the outcomes models, with confirmation using the Wald statistic and likelihood ratio test and proportionality testing. A 0.05 level of significance was used throughout.
Results
During the study period, a total of 24,895 samples from 14,261 unique patients were tested for vitamin D concentrations in the University of Kansas Hospital laboratory. The lowest recorded value for patients with multiple measurements was used for analysis. Database query yielded information for 11,017 matching patients. After excluding patients who were <18 years old, 10,899 patients were available for analysis.
The baseline characteristics for the patients with and without vitamin D deficiency are listed in Table 1 . The cohort’s mean age was 58.3 ± 14.9 years, 71% were women (n = 7,758), and the mean body mass index (BMI) was 29.9 ± 7.7 kg/m 2 . The mean left ventricular ejection fraction (by echocardiogram) was 57 ± 10%. The mean and median serum vitamin D value was 24.1 ± 13.6 ng/ml and 22.5 ng/ml, respectively. A total of 3,234 patients (29.7%) were in the predefined normal range for vitamin D (≥30 ng/ml) and 7,665 (70.3%) patients were deficient (<30 ng/ml).
| Variable | Vitamin D Deficient | p Value | |
|---|---|---|---|
| No (n = 3,234) | Yes (n = 7,665) | ||
| Age (years) | 60 ± 15 | 58 ± 15 | <0.0001 |
| Women | 2,503 (77%) | 5,255 (69%) | <0.0001 |
| Body mass index (kg/m 2 ) | 28 ± 7 | 31 ± 8 | <0.0001 |
| Coronary artery disease (>70% stenosis of ≥1 coronary artery) | 306 (9%) | 830 (11%) | 0.03 |
| Cardiomyopathy (clinical diagnosis of dilated, hypertrophic or restrictive myocardial disease) | 95 (3%) | 288 (4%) | 0.03 |
| Hypertension (blood pressure >140/90 mm Hg) | 938 (29%) | 2,795 (36%) | <0.0001 |
| Diabetes mellitus | 294 (9%) | 1,436 (19%) | <0.0001 |
| Valvular heart disease (stenosis or regurgitation greater than moderate or valve replacement) | 181 (6%) | 487 (6%) | NS |
| Atrial fibrillation | 201 (6%) | 398 (5%) | 0.03 |
| Ejection fraction (%) | 58 ± 10 | 57 ± 10 | NS |
| Aspirin use | 997 (31%) | 2,254 (29%) | NS |
| Angiotensin-converting enzyme inhibitor use | 725 (22%) | 2,089 (27%) | <0.0001 |
| Statin use | 1,098 (34%) | 2,611 (34%) | NS |
| Calcium (mg/dl) | 9.17 ± 0.59 | 9.31 ± 0.49 | <0.0001 |
| Creatinine (mg/dl) | 1.38 ± 6.29 | 1.58 ± 2.98 | NS |
| Total cholesterol (mg/dl) | 166 ± 43 | 171 ± 54 | 0.02 |
| High-density lipoprotein cholesterol (mg/dl) | 52 ± 18 | 47 ± 16 | <0.0001 |
| Low-density lipoprotein cholesterol (mg/dl) | 92 ± 33 | 97 ± 39 | 0.002 |
| Triglycerides (mg/dl) | 115 ± 80 | 140 ± 126 | <0.0001 |
| Vitamin D (ng/ml) | 40 ± 11 | 17 ± 7 | <0.0001 |
| Vitamin D supplement use | 689 (21%) | 2,423 (32%) | <0.0001 |
| Vitamin supplement (any) | 1,097 (34%) | 2,992 (39%) | <0.0001 |
| Death | 43 (1%) | 293 (4%) | <0.0001 |
On univariate analysis, vitamin D deficiency as a dichotomous variable was significantly associated with an increased risk of several cardiovascular disease states, including coronary artery disease (OR 1.16, 95% CI 1.012 to 1.334), diabetes (OR 2.31, 95% CI 2.018 to 2.633), cardiomyopathy (OR 1.29, 95% CI 1.019 to 1.633), and hypertension (OR 1.40, 95% CI 1.285 to 1.536) and all-cause death (OR 2.95, 95% CI 2.135 to 4.073). In contrast, vitamin D deficiency was negatively associated with the risk of atrial fibrillation (OR 0.83, 95% CI 0.693 to 0.984; Table 2 ). On stepwise multivariate logistic regression analysis, vitamin D deficiency (dichotomous variable) was a strong independent predictor of all-cause death (OR 2.64, 95% CI 1.901 to 3.662, p <0.0001; Table 3 ).
| Event | OR | 95% CI | p Value |
|---|---|---|---|
| Death | 2.95 | 2.135–4.073 | <0.001 |
| Coronary artery disease | 1.16 | 1.012–1.334 | 0.03 |
| Atrial fibrillation | 0.83 | 0.693–0.984 | 0.03 |
| Diabetes mellitus | 2.31 | 2.018–2.633 | <0.001 |
| Cardiomyopathy | 1.29 | 1.019–1.633 | 0.03 |
| Hypertension | 1.40 | 1.285–1.536 | <0.0001 |
| Predictor | OR | 95% CI | p Value |
|---|---|---|---|
| Coronary artery disease | 2.71 | 2.062–3.573 | <0.0001 |
| Vitamin D deficiency | 2.64 | 1.901–3.662 | <0.0001 |
| Diabetes mellitus | 1.45 | 1.114–1.891 | 0.006 |
| Cardiomyopathy | 3.29 | 2.359–4.596 | <0.0001 |
| Hypertension | 1.53 | 1.183–1.969 | 0.001 |
Survival was calculated using the interval between the date of sample collection and the date of death or the end of study period. The survival curve is shown in Figure 1 (p <0.0001 for homogeneity of strata). Hazard ratios using the Cox proportional hazards model were calculated, with vitamin D deficiency as a dichotomous variable showing a hazards ratio of 2.48 (p <0.0001) for reduced survival. Hazard function analysis with additional predictive variables (disease states, age, BMI, ejection fraction, gender) was then performed with overall p <0.0001; vitamin D deficiency remaining a significant independent predictor of decreased survival. Vitamin D deficiency remained the variable with the highest OR (2.29).
Vitamin D supplementation improved survival overall but to a significant degree only in deficient patients. On univariate analysis, the overall risk of all-cause death was reduced for subjects taking vitamin D supplements with an OR of 0.62 (95% CI 0.469 to 0.806, p = 0.0004). On stepwise regression analysis for death, the OR for vitamin D supplementation was 0.44 (p <0.0001; Table 4 ), suggesting an association with a significantly lower occurrence of death. In a hazard model, the hazard ratio for death was 0.40 (95% CI 0.335 to 0.576) for subjects receiving supplementation (p <0.0001). This strong relation persisted in additional models with the clinical variables added.
