Vitamin D and risk of hypertension among Black women

Abstract

Evidence of an association between plasma 25-hydroxyvitamin D [25(OH)D] levels and risk of hypertension, predominantly from studies of White individuals, suggests an inverse relationship. Limited data are available on Black individuals, who are more likely to have vitamin D deficiency. In the Black Women's Health Study (BWHS), a prospective study of 59 000 self-identified Black women from across the US, we assessed levels of a validated predicted vitamin D score in relation to incident hypertension. We followed 42 239 participants who were free of cardiovascular disease and cancer from 1995 to 2019, during which time 19 505 incident cases of hypertension were identified. Cox proportional hazards model were used to calculate multivariable-adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) for the association of predicted vitamin D with the risk of incident hypertension. In age-adjusted analyses, there was a strong inverse dose-response association between predicted vitamin D score and hypertension risk, with an HR of .66 (95% CI: .63-.68, p trend < .0001) for the highest quartile of predicted vitamin D relative to the lowest. After control for potential confounders including body mass index, physical activity, and cigarette smoking, the HR was attenuated to .91 (95% CI: .87–.95, p trend = .002). In this prospective cohort study of Black women, predicted vitamin D score was weakly inversely associated with the incidence of hypertension. This observed association may reflect an inability to fully control for confounding factors.

INTRODUCTION

Black Americans have an earlier onset and a higher incidence of hypertension, and have the highest rate of complications and deaths related to hypertension compared to other racial/ethnic groups.^1–3 In particular, Black women in the U.S. have the highest hypertension prevalence (57%).^3,4 Evidence from in vitro studies, animal studies, and clinical trials suggests that vitamin D regulates blood pressure (BP) by inhibiting renin-angiotensin-aldosterone (RAAS) system activity, reducing vascular oxidative stress, and modulating the function of the vascular wall.^5–7 Some evidence from epidemiological studies has suggested that lower plasma 25-hydroxyvitamin D [25(OH)D] levels may be associated with a higher risk of incident hypertension and cardiovascular events.^8–15

Black Americans on average have lower circulating levels of vitamin D than Whites.^16–21 In 2011, the Institute of Medicine reported that 54% of Black Americans had vitamin D levels < 16 ng/ml (40 nmol/l), as compared to 27% of Hispanic and 10% of White Americans.²² Darker skin pigmentation among Black Americans results in lower production of vitamin D3 in the skin.^23,24 In addition, Black Americans generally consume less vitamin D from dietary sources than White individuals.²⁵ Vitamin D deficiency may be associated with hypertension risk by activating the RAAS system⁷ and increasing renin and aldosterone levels.²⁶ Aldosterone results in a greater increase in blood pressure in Black than White individuals.^26,27

Prior studies of vitamin D and hypertension have been focused on White populations and have yielded inconsistent findings.^16,28–33 Only a few studies report results from Black individuals, who have disproportionately higher burdens of vitamin D deficiency and hypertension. The VITamin D and OmegA-3 TriaL (VITAL), the first randomized intervention trial with la arge number of Black Americans (n = 5106), found that Vitamin D did not reduce risk of major CVD events in any racial subgroup, but the trial has not published results on risk of hypertension.^34,35 A cross-sectional study found an association between low vitamin D and higher prevalence of hypertension among White but not Black participants.³⁶ A small randomized trial found that vitamin D supplementation for 3 months was associated with modestly lower systolic pressure among Black participants.³⁷

In the present study, based on 24 years of data collection in the Black Women's Health Study (BWHS), we assessed the relation between a validated predicted vitamin D score and hypertension risk. We hypothesized that higher predicted vitamin D score is associated with lower risk of hypertension.

METHODS Study population

The BWHS is a prospective cohort study of 59 000 self-identified Black women enrolled from across the United States in 1995.³⁸ At baseline in 1995, participants were 21–69 years of age, with a median age of 38. Participants have provided information on biennial questionnaires regarding demographic characteristics, lifestyle and socioeconomic factors, and medical conditions. Follow-up has been successful for 85% of potential person-years through 11 completed biennial rounds of follow-up. All but 5% of BWHS participants were born in the US.

The analytic sample consisted of 42 239 women after exclusion of 9294 women with prevalent hypertension at baseline in 1995, 563 women with cardiovascular disease at baseline, 482 women with cancer at baseline, and 6395 women for whom predicted vitamin D score could not be calculated due to missing data on predictors for the vitamin D predicted score.

The Boston University Medical Campus Institutional Review Board approved the study.

Exposure

The primary exposure was predicted vitamin D score, which was developed and validated against directly measured levels of serum vitamin D in the BWHS.³⁹ In addition to maximizing sample size for statistical analysis, using a predicted vitamin D score can overcome drawbacks of prior studies that used a single measure of vitamin D. Serum vitamin D levels vary over time and are influenced by many factors, including season, age, body mass index (BMI), recent sun exposure, skin pigmentation, and dietary intake.⁴⁰ Studies that use a single vitamin D measure cannot take that variation into account. Case-control studies using serum vitamin D as an exposure could have the additional problem of reverse causation. A predicted vitamin D score in a follow-up study with repeated data collection can be derived for multiple longitudinal time points, thus considering changes in lifestyle and environmental factors that may not be captured in a single blood sample.⁴⁰

The predicted vitamin D score was developed and validated against directly measured levels of plasma 25(OH)D in 2856 BWHS participants.³⁹ This validated predicted score of 25(OH)D has been inversely associated with risk of breast cancer and colorectal cancer in studies of BWHS participants.^39,41 The score was computed for each participant at each questionnaire cycle based on their values for the predictors at that point in time. Predictors include vitamin D supplementation, multivitamin use, BMI, dietary vitamin D intake, postmenopausal hormone use, vigorous physical activity, alcohol consumption, smoking, and oral contraceptive use. The correlation coefficient between predicted and measured plasma 25(OH)D was .49 (SD = .026). Vitamin D supplementation was the strongest predictor and accounted for 10% of the total variation. For our primary analyses, we modeled the predicted vitamin D score as a cumulative average⁴¹ and categorized predicted vitamin D score into quartiles. The cutoff points for predicted vitamin D score quartiles correspond to the cutoff points for quartiles of plasma 25(OH)D measured in the BWHS blood samples (21, 31, and 40 ng/ml for quartiles 2, 3, and 4, respectively).³⁹

Outcome

Incident hypertension cases were identified through self-reports on biennial questionnaires from 1995 through 2019. We considered a participant to have hypertension if she reported use of one or more antihypertensive medications or a physician diagnosis of hypertension together with diuretic use.⁴² In a validation study of self-reported hypertension in the BWHS, 138 (99%) of 139 reports of hypertension for which we obtained medical records or physician checklists were confirmed, with all systolic pressures being 140 mmHg or higher and diastolic pressures being 90 mmHg or higher.⁴³

Covariates

Hypertension risk factors to be controlled in the analysis were selected a priori, and included age (continuous, years), BMI (continuous, kg/m2), completed education (< 12, 13–15, 16, ≥17 years), geographic region (Northeast, South, Midwest, West), neighborhood socioeconomic status (SES, quintiles),⁴⁴ physical activity level (continuous, hours of metabolic equivalent tasks per week based on both vigorous physical activity and walking),⁴⁵ cigarette smoking (never smoker, past smoker <10 pack years, current smoker <10 pack years, past smoker 10+ pack years, current smoker 10+ pack years), history of type 2 diabetes (yes, no), and alcohol consumption (continuous, drinks/week).

For stratified analyses, we considered age (<50, 50–69, ≥70 years), BMI (<25, 25–29, ≥30kg/m2), diabetes (yes, no), smoking (never, ever), education (<12, 13–15, 16, ≥17 years), neighborhood SES (quintile 1, quintile 5), and Vit D supplement use (yes, no). All covariate information was based on information updated at each questionnaire cycle.

Statistical analyses

For our main analyses, we examined the relation between cumulative average predicted vitamin D score with risk of hypertension. We fit Cox proportional hazards models, stratified by age in 1-year intervals and questionnaire cycle, with each participant contributing person-years from baseline in 1995 until diagnosis of incident hypertension, death, loss to follow up, or end of follow up in 2019, whichever occurred first. Multivariable adjusted hazard ratios (HRs) and 95% confidence intervals (CIs), adjusted for the covariates listed above, were calculated comparing each quartile with the lowest quartile of predicted vitamin D score. We used the missing data indicator method for missing data in covariates. P for trend was tested using the median of the quartiles of predicted vitamin D score. We examined potential effect measure modification by age (<50, 50–69, ≥70 years), BMI (<25, 25–29, ≥30 kg/m2), diabetes (yes, no), smoking (never, ever), education (<12, 13–15, 16, ≥17 years), and neighborhood SES (quintile 1, quintile 5). Effect modification was evaluated using the Wald test.

RESULTS

During 583 296 person-years of follow up, 19 505 incident cases of hypertension were identified. Compared with Black women in the lowest quartile of predicted vitamin D score at baseline, those in the highest quartile were younger, had a lower BMI, consumed less alcohol, were less likely to smoke, were more physically active, had higher education, and were more likely to live in a neighborhood with high SES (Table 1). Those in the highest quartile were also more likely to use vitamin D supplements.

TABLE 1 Age-standardized baseline characteristics in Black Women's health study participants by quartiles of predicted vitamin D score

	Predicted vitamin D score
	Q1	Q2	Q3	Q4
n	9774	10354	10848	11263
Age, year	38.4(9.3)	37.3(9.7)	36.7(9.8)	34.8(9.4)
Body Mass Index, kg/m2	31.4(7.2)	27.6(6.1)	25.9(5.0)	24.6(4.5)
Current alcohol use, drinks/week	2.8(6.1)	1.7(4.4)	1.2(3.4)	.8(3.0)
Smoking, pack years	4.9(9.3)	3.7(8.3)	3.1(7.6)	2.5(7.1)
Physical activity, MET hours/week	12.8(15.1)	16.4(16.4)	19.3(17.4)	24.3(17.9)
Type 2 diabetes, %	3.1	2.2	1.8	1.4
Education, %
12 or less years	23.3	16.5	13.7	10.7
13–15 years	38.8	37.7	35.2	34.7
16 years	21.3	24.7	27.4	28.5
17+ years	16.3	20.9	23.6	26.0
Neighborhood SES: Highest quintile, %	14.8	17.6	21.4	23.5
Geographic region, %
Northeast	25.3	27.9	28.0	28.7
South	31.1	30.8	30.1	31.3
Midwest	22.0	22.5	22.5	23.5
West	21.4	18.6	19.3	16.3
Current use of vitamin D supplement, %	.0	.08	.7	26.3

Values are means (± standard deviations) or percentages and are standardized to the age distribution of the study.

Abbreviations: MET, metabolic equivalent of task; Q, Quartiles; SES, Socioeconomic status.

There was an inverse dose-response relationship between cumulative average predicted vitamin D score and risk of hypertension in analyses controlled only for age (Table 2). Compared to quartile 1 (the lowest predicted vitamin D score), HRs were .82 (95% CI: .79–.85), .72 (95% CI: .70–.75), and .66 (95% CI: .63–.68) for quartiles 2, 3, and 4, respectively (p trend < .0001, Table 2). Multivariable adjustment attenuated the associations: HRs were .98 (95% CI: .95–1.02), .98 (95% CI: .93–1.02), and .91 (95% CI: .87–.95) for quartiles 2, 3, and 4, respectively (p trend = .002, Table 2). BMI appeared to be the strongest confounder for this association. In the age adjusted model (Model 1), after adjustment in addition for BMI, HRs were .96 (95% CI: .92–.99), .93 (95% CI: .89–.97), and .85 (95% CI: .82–.89) for quartiles 2, 3, and 4, respectively. In the multivariate fully adjusted model, after further adjustment for supplemental vitamin D use, results remained similar. HRs were .98 (95% CI: .95–1.02), .98 (95% CI: .93–1.02), and .93(95% CI: .88–.98) for quartiles 2, 3, and 4). When restricting to participants with no use of supplemental vitamin D, there was no association between predicted vitamin D score with incident hypertension. HRs were .99 (95% CI: .95–1.03), .99 (95% CI: .94–1.03), .95 (95% CI: .89–1.01) for quartiles 2, 3, and 4, respectively (p = .14).

TABLE 2 Age- and multivariable-adjusted hazard ratios for association of cumulative average predicted vitamin D score with risk of hypertension

Cumulative average predicted vitamin D
	Quartile 1	Quartile 2	Quartile 3	Quartile 4	p trend
Incident hypertension cases	5174	5354	4842	4135
Person years	116778	150534	160334	155650
Age-adjusted HR (95% CI)a	Ref	.82(.79, .85)	.72 (.70, .75)	.66 (.63, .68)	<.0001
Multivariable-adjusted HR (95% CI)b	Ref	.98(.95, 1.02)	.98 (.93, 1.02)	.91 (.87, .95)	.002

P for trend was tested using the median of the quartiles of predicted vitamin D score.

Abbreviations: CI, confidence interval; HR, hazard ratio.

a

Stratified by age (continuous), questionnaire cycle.

b

Further adjusted for body mass index (continuous, kg/m2), educational level (<12, 13–15, 16, ≥17 years), geographic region (Northeast, South, Midwest, West), neighborhood SES (quintiles), physical activity level (continuous, hours of metabolic equivalent tasks per week), cigarette smoking (pack years of smoking, continuous, never smoker, past smoker <10 pack years, current smoker <10 pack years, past smoker 10+ pack years, current smoker 10+ pack years), history of diabetes (yes, no), current alcohol consumption (continuous, drinks/week).

In analyses stratified by age, BMI, smoking status, geographic region, education, neighborhood SES, and diabetes status, similar patterns to the overall association were observed (Table 3). No significant interactions were identified.

TABLE 3 Associations of predicted vitamin D with risk of hypertension by subgroup

	Ncases/py	Quartile 1	Quartile 2	Quartile 3	Quartile 4	p int
Age, years
<50	11118/428425	Ref	.96 (.92, 1.01)	.93 (.88, .99)	.88 (.83, .93)	.11
50–69	7938/147689	Ref	1.01 (.95, 1.08)	1.02 (.95, 1.09)	.94 (.88, 1.00)
≥70	449/7182	Ref	1.07 (.78, 1.46)	1.11 (.82, 1.53)	1.11 (.85, 1.45)
BMI, kg/m2
<25	3299/198591	Ref	.92 (.82, 1.04)	.98 (.87, 1.10)	.94 (.84, 1.06)	.14
25–29	6637/200071	Ref	1.03 (.95, 1.12)	1.00 (.99, 1.02)	.91 (.83, .99)
≥30	9569/184633	Ref	1.02 (.97, 1.08)	1.03 (.97, 1.10)	.95 (.89, 1.01)
Never smoking	11829/398465	Ref	1.00 (.95, 1.05)	.99 (.93, 1.04)	.95 (.90, .99)	.83
Ever smoking	7676/184832	Ref	.97 (.91, 1.03)	1.01 (.94, 1.08)	.97 (.91, 1.03)
Neighborhood SES Q1	3305/86730	Ref	1.02 (.93, 1.11)	1.04 (.94, 1.16)	.93 (.84, 1.03)	.88
Neighborhood SES Q5	3499/123227	Ref	.98 (.94, 1.03)	.98 (.94, 1.03)	.95 (.91, 1.00)
Diabetes	2058/24490	Ref	.90 (.80, 1.01)	.89 (.78, 1.02)	.91 (.81, 1.03)	.85
No diabetes	17447/558806	Ref	1.00 (.96, 1.05)	1.01 (.96, 1.05)	.96 (.92, 1.00)
Geographic region
Northeast	4483/147573	Ref	1.04 (.96, 1.12)	1.01 (.92, 1.10)	.99 (.90, 1.07)	.93
South	7396/205914	Ref	.99 (.93, 1.06)	1.00 (.93, 1.07)	.96 (.90, 1.03)
Midwest	4274/122775	Ref	.94 (.87, 1.02)	.95 (.86, 1.04)	.90 (.82, .98)
West	3352/107034	Ref	.99 (.90, 1.08)	1.02 (.92, 1.13)	.95 (.86, 1.05)
Education
<12	3414/81145	Ref	1.04 (.95, 1.13)	1.02 (.92, 1.13)	1.03 (.93, 1.14)	.34
13–15	7148/210184	Ref	.96 (.90, 1.02)	.94 (.88, 1.01)	.89 (.83, .96)
16 years	4696/158883	Ref	.98 (.91, 1.07)	1.00 (.92, 1.09)	.95 (.87, 1.03)
≥17 years	4247/133084	Ref	1.02 (.93, 1.11)	1.06 (.96, 1.16)	1.01 (.92, 1.10)

Model adjusted for age (continuous), questionnaire cycle (continuous), body mass index (continuous, kg/m²), educational level (< 12, 13–15, 16, ≥ 17 years), geographic region (Northeast, South, Midwest, West), neighborhood SES (quintiles), physical activity level (continuous, hours of metabolic equivalent tasks per week), cigarette smoking (pack years of smoking, continuous, never smoker, past smoker < 10 pack years, current smoker < 10 pack years, past smoker 10+ pack years, current smoker 10+ pack years), history of diabetes (yes, no), current alcohol consumption (continuous, drinks/week).

Abbreviations: CI, confidence interval; HR, hazard ratio; p int, p for interaction; py, person years.

DISCUSSION

In this prospective cohort study of 42 239 Black participants in the BWHS that included 19 505 cases of incident hypertension, there was a very weak association of predicted vitamin D score with risk of hypertension, with a HR of .91 (95%CI: .87–.95) for the highest quartile relative to the lowest. Because control for potential confounders such as BMI markedly reduced the HRs from what was observed in age-adjusted analyses, it is possible the weak association remaining after such control is due to residual confounding from important risk factors.

Compared with Whites, Black individuals have lower serum vitamin D levels due to reasons that include darker skin pigmentation, genetics,⁴⁶ diet, lifestyle and environmental factors.⁴⁷ Current epidemiologic evidence on vitamin D and hypertension is predominantly based on studies of Whites. A meta analyses reported a RR of .88 (.81, .97) for incident hypertension per 10 ng/ml increment in baseline 25(OH)D levels.³² Comparing high versus low 25(OH)D level, the pooled relative risk was .70 (.58–.86) for risk of hypertension in another meta analyses.⁹ There is little evidence specifically for Black Americans.^16,28–33 It is unclear as to the optimal dosage for cardiovascular health benefits from vitamin D^33,35 as well as race/ethnicity-specific thresholds for vitamin D deficiency.^36,48 Evidence from randomized clinical trials failed to show a benefit of vitamin D supplementation on intermediate cardiovascular endpoints: there were no changes or only small reductions in blood pressure.^5,31,37 However, the sample sizes in these trials were small, follow up was short, and vitamin D dosages differed. The VITAL trail, the first trial with a large number of Black participants (n = 5106), showed that Vitamin D did not reduce risk of major CVD events in any racial subgroup.^34,35 Results on hypertension were not reported separately. In a randomized clinical trial of 283 Black Americans, vitamin D supplementation for 3 months resulted in modestly lower systolic pressure.³⁷ A cross-sectional study found a null association between serum vitamin D and prevalence of hypertension among Black participants.³⁶

Strengths of our study include the prospective cohort design, large sample size of Black women, large number of incident cases of hypertension, and the use of predicted vitamin D score to represent long-term vitamin D status. The total of 19 505 incident cases of hypertension provided sufficient statistical power for this investigation. The 42 239 Black women in this study were from across the U.S with a wide range of ages, geographic distribution, and demographic characteristics. The vitamin D prediction score has been previously validated and shown to be associated with breast cancer and colorectal cancer incidence in the BWHS.^39,41 Compared with traditional studies using a single plasma vitamin D level, the predicted vitamin D score, derived repeatedly for multiple time points, represented long-term exposures that a single blood sample cannot capture because of the relatively short half-life of 25(OH)D.^39,40

Our study has several limitations. First, despite our effort to control for confounding factors, our study was observational in nature and still subject to unmeasured and residual confounding. While we can distinguish relatively low and high levels using the predicted vitamin D score, the distinction between deficient (<20 ng/ml) and insufficient (<30 ng/ml) in terms of measured plasma 25(OH)D level is uncertain because of the variability in the prediction model (prediction model R-square = .25).³⁹ Biologically, both vitamin D and parathyroid hormone level (PTH) influence calcium metabolism, and activation of the RAAS, and endothelial dysfunction.⁴⁹ Vitamin D deficiency leads to elevated PTH and Black individuals have lower vitamin D and higher PTH levels compared with Whites.⁴⁹ Observational studies suggest a synergistic association among Whites between vitamin D and PTH with risk of hypertension and cardiometabolic outcomes.³⁶ In the BWHS, we did not collect information on PTH and therefore were not be able to assess the joint association of vitamin D and PTH. Renal function status can affect the relationship between vitamin D and hypertension. We did not have relevant information on renal function, such as eGFR and serum creatinine. Information on hypertension in the BWHS was self-reported and did not allow for differentiation between essential hypertension and secondary hypertension. Since we did not have blood pressure measurements, we were unable to assess a possible dose response relationship with blood pressure, and were unable to examine the robustness of our results based on different definitions of hypertension (e.g., BP≥140/≥90 mmHg, vs. BP≥130/≥80 mmHg). Due to the asymptomatic nature of hypertension, the incidence of hypertension assessed in the BWHS is likely an underestimation of the actual incidence. However, given the prospective nature of this investigation, the measurement error in hypertension is likely to be non-differential and unlikely to be correlated with the exposure (predicted vitamin D). Thus, the measurement error in hypertension may reduce statistical precision but is unlikely to introduce bias into our estimation.

In this large prospective investigation of predicted vitamin D and hypertension among Black women, predicted vitamin D score was very weakly inversely associated with incident risk of hypertension after control for confounding factors. Further studies on serum vitamin D levels are needed to assess the association of vitamin D with risk of hypertension in Black individuals. Measurements of vitamin D levels over time would be desirable, though very expensive to carry out. Some follow-up studies currently in progress have the ability to determine new cases of hypertension, and the previous collection of blood samples could be leveraged for assays of serum vitamin D levels at the time of sample selection in relation to risk of hypertension subsequently.⁵⁰ If an inverse association is confirmed going forward, then attention by physicians to consideration of “sufficient” levels of vitamin D among Black individuals will be desirable.

AUTHORS CONTRIBUTION

Study concept and design: SS, JRP, LR. Acquisition of data: LR, JRP. Analysis and interpretation of data: SS, LR, JRP. Drafting of the manuscript: SS. Critical revision of the manuscript for important intellectual content: SS, LR, JRP, YZ, KAB. Statistical analysis: SS. Administrative, technical, or material support: LR, JRP. Study supervision: LR, JRP. SS had primary responsibility for final content. All authors have read and approved the final manuscript.

ACKNOWLEDGMENTS

We thank the staff and participants in the BWHS study. S. Sheehy had full access to all of 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. This study was funded by National Institute of Health, R01CA058420, U01CA164974, and R01MD015085.

CONFLICTS OF INTEREST

The authors have no conflict of interests. The sole role of the funders was to support data collection. The funders had no role in design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.