Background
Type 2 diabetes mellitus (T2DM) is a serious and growing global health problem. In 2013, there were 382 million people with diabetes; this number is estimated to grow to 592 million by 2035. The highest prevalence rates of T2DM are found in developing countries [
1].
Hypovitaminosis D is an emerging health problem that affects approximately one billion people worldwide [
2] and might be increasing in frequency [
3]. Vitamin D is an important hormone for mineral homeostasis and bone integrity but also has several pleiotropic effects outside the skeleton, including the endocrine system [
4]. Recent evidence suggests that hypovitaminosis D is common in patients with type 2 diabetes mellitus [
5].
Hypovitaminosis D may be a neglected cardiovascular risk factor in patients with type 2 diabetes mellitus [
6,
7] as vitamin D appears to influence several pathways that have been linked to coronary artery disease, such as inflammation, vascular calcification, proliferation of smooth muscle cells in vascular tissue, myocyte hypertrophy, arterial intimal thickness, renin-angiotensin system, blood pressure control and insulin resistance [
8‐
10].
To further examine this issue, we sought to determine the prevalence of hypovitaminosis D in patients with type 2 diabetes mellitus and investigate the relationship between vitamin D levels and variables indicative of glycemic control and cardiovascular risk factors.
Methods
Study design
Observational, cross-sectional study. Patients were recruited between October 2012 and November 2013.
Study site
Endocrinology outpatient clinic affiliated with the Hospital of the Federal University of Bahia, located in Salvador, Bahia—Brazil (latitude 12°58′S).
Study population
Adult patients with type 2 diabetes mellitus. We did not include patients with other types of diabetes, current or previous use of vitamin D or multivitamins, pregnant women, patients with chronic kidney disease, patients with malabsorptive intestinal diseases or status post bariatric surgery and patients taking anticonvulsants, drugs for the treatment of HIV-AIDS, steroids, rifampicin, cholestyramine or orlistat.
Clinical variables
We conducted individualized interviews to collect data regarding age, sex, self-reported race, time of diagnosis of diabetes, medications in use, and comorbidities. Height to the nearest 0.1 cm and weight to the nearest 0.1 kg were measured with participants wearing light clothing and no shoes. Participants with a body mass index (BMI) of 25.0–29.9 kg/m2 were classified as overweight, and those with BMI ≥ 30.0 kg/m2 were classified as obese.
Laboratory tests
All patients were asked to stop oral antidiabetic drugs and insulin for at least 12 h prior to examination. Patients were also asked to refrain from strenuous physical exercise and alcohol intake during 3 days and fast for at least 10 h before the serum tests. Blood samples were processed for storage within 24 h of collection. Serum 25-hydroxyvitamin D [25(OH)D] was determined using a chemiluminescence kit (DiaSorin LIAISON, MN, USA) that recognizes both vitamin D2 and D3 equally. Two levels of controls provided by the manufacturer were run in each assay. Inter- and intra-assay coefficients of variation based on repeated analysis of a pooled control were 15 and 13%, respectively. The lower limit of detection of the assay was 2.8 ng/mL. Levels of 25-hydroxyvitamin D were stratified according to the classification of the Endocrine Society 2011 [
11]: deficiency (<20 ng/mL), insufficiency (20–29 ng/mL) and sufficiency (≥30 ng/mL). Serum glycated hemoglobin (HbA1c) was determined using HPLC assay on automated analyzer. Serum calcium, phosphorus, alkaline phosphatase, parathyroid hormone, creatinine, total cholesterol, high-density lipoprotein-cholesterol (HDL-c), low-density lipoprotein-cholesterol (LDL-c), triglycerides, uric acid, ultra-sensible C-reactive protein (CRP), glucose concentrations and microalbuminuria were measured using standardized and automated assays.
Sample size
Sample size was calculated using the software OpenEpi (
http://www.openepi.com). The minimum estimated sample size to detect a prevalence of hypovitaminosis D of 50%, with a precision of 10–95% confidence limits was 97 patients.
Statistical analyses
Continuous data were presented as mean (±standard deviation) or median (inter-quartile range). Categorical variables were summarized using absolute and relative frequencies and compared using the Chi-square test. The relationship between continuous variables indicative of glycemic control and cardiovascular risk factors with 25(OH)D levels was evaluated using Pearson’s correlation analyses. Subsequently, variables that were correlated with 25(OH)D with p < 0.10 were entered into a backward multiple linear regression model (with vitamin D level as the dependent variable). A similar strategy was used to identify predictors of hypovitaminosis D (a binary event) by logistic regression. Variables that were associated with hypovitaminosis D (p < 0.10) on univariate analyses were entered into a backward, multivariate logistic regression model. A two-sided p value <0.05 in final analyses was deemed significant. All statistical analyses were performed using IBM SPSS for Windows version 20.0 (SPSS Inc., Chicago, IL).
Discussion
There were no prior studies on the prevalence of hypovitaminosis D in patients with type 2 diabetes mellitus in Brazil. Herein, we showed that hypovitaminosis D was present in 62% of patients with type 2 diabetes and it was associated with female gender, obesity and dyslipidemia. Since the current study was conducted in Salvador—Bahia, located at 12°58′S, an area that receives adequate sunlight throughout the year, the high prevalence of hypovitaminosis D among individuals with type 2 diabetes encountered herein could be considered unexpected. In fact, our findings are comparable to those found in diabetic populations of non-tropical regions [
12‐
14]. Certain characteristics of our sample may have contributed to the high prevalence of hypovitaminosis D.
First, our sample was comprised mostly of non-white participants, reflecting the demographics of Bahia, which is the Brazilian state with the highest percentage of non-white population (77.8%). There is evidence that melanin can hinder the penetration of sunlight on the skin and vitamin D conversion [
15].
Second, most participants were overweight or obese. The prevalence of obesity was high in our population, corroborating the already established link between excess weight and type 2 diabetes [
1,
16]. Several studies in diabetic patients report an association between obesity and hypovitaminosis D [
14,
17,
18]. Since vitamin D can be sequestered in body fat, obese patients tend to have lower serum levels. A potential confounder is that obesity is also linked to an unhealthier lifestyle, characterized by less physical activity, less sun exposure and, hence, lower vitamin D levels and worse clinical outcomes [
19].
Third, the majority of our participants were women. Hoteit et al. and van der Meer et al. [
20,
21] suggested that female gender is an independent predictor of vitamin D deficiency. This finding is supported by other studies involving type 2 diabetes [
14,
17,
22]. Indeed, female gender was an independent predictor of hypovitaminosis D in our logistic regression model.
Obesity varies between the sexes and between different races and is a recognized risk factor for vitamin D deficiency and type 2 diabetes [
11,
23‐
27]. In our study, the women had BMIs greater than the men. Perhaps the excess body fat in females might have contributed to the observed relationship between gender and hypovitaminosis D.
Dyslipidemia was an independent predictor of hypovitaminosis D and total cholesterol was inversely correlated with levels of 25(OH)D even after controlling for confounders. Our data are consistent with other studies in the literature that evaluated cardiovascular risk in type 2 diabetes patients from sunny regions [
17,
28]. Perhaps lipid levels might be the intervenient variable that explains the link between hypovitaminosis D and cardiovascular disease in patients with type 2 diabetes. However, the literature on this subject is not homogeneous and interventional studies have failed to demonstrate that raising the levels of vitamin D resulted in improvement in the lipid profile of type 2 diabetes patients [
29‐
32].
The association between HbA1c and 25(OH)D is controversial. We found a significant weak negative correlation between HbA1C and 25(OH)D, similar to other previously published studies [
12,
18,
23,
28,
33,
34]. However glycemic control was not associated with vitamin D when we controlled for confounding factors, as verified by Luo et al. and Al-Shoumer et al. [
35,
36].
The Institute of Medicine suggests that the prevalence of hypovitaminosis D in the population has been overestimated due to improper use of a high cutoff level of 25(OH)D. This Institute alerts that concentrations of 25(OH)D above 30 ng/mL were not consistent with increased clinical benefit and calls for a consensus to determine the appropriate levels of vitamin D to avoid under and overtreatment [
37]. Considering that our mean 25(OH)D was very close to 30 ng/mL, more studies are needed to evaluate the adequacy of the current vitamin D cutoffs in Brazilians with type 2 diabetes. In this study, a lower 25(OH)D cutoff for normality would certainly have resulted in a much lower prevalence of hypovitaminosis D.
Our work has important limitations. At our center, we use the chemiluminescence method Diasorin LIAISON to assess 25(OH)D status due to convenience, turnaround time and cost. Although the gold standard for dosing of vitamin D is the LC-MS/MS, several studies have shown that Diasorin LIAISON shows acceptable correlation with LC-MS/MS [
38,
39]. We did not formally collect data on financial income, sun exposure time in outdoor activities or other characteristics predictive of outdoor behavior. We also don’t have data about the season that the blood sample was drawn but in Salvador-Bahia, temperatures are relatively constant, with about 2500 h of sunshine a year [
40]; therefore, we believe that season did not play an important role in our findings. The observational, cross-sectional design precludes any conclusions regarding causality or even the direction of the association between vitamin D deficiency and cardiovascular risk factors. Moreover, our sample is small and of peculiar characteristics, which may limit the generalizability of our findings. The relationship between HbA1C and serum lipids with 25(OH)D levels may have been affected by the use of antidiabetic and lipid-lowering medications, as these drugs lower HbA1C and lipid levels regardless of the vitamin D status; similar studies in patients with early type 2 diabetes before initiating treatment could help clarify this relationship.
Conclusion
In conclusion, we have identified a high prevalence (62%) of hypovitaminosis D among Brazilians with type 2 diabetes and an association between hypovitaminosis D with female gender, obesity and dyslipidemia. Our data indicates that availability of adequate sunlight alone is not sufficient for the prevention of vitamin D deficiency and raises awareness of hypovitaminosis D in the diabetic population, regardless of geographic location.
Authors’ contributions
MCR researched data and wrote the manuscript. BMS, GC and PNR researched data, reviewed, edited the manuscript and contributed to discussion. PNR is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. All authors read and approved the final manuscript.