Risk Factors Associated with Vitamin D Status among Older Puerto Rican Adults

ABSTRACT

Background

Vitamin D deficiency has been associated with health problems globally, but there is limited information on vitamin D status and associated risk factors among adults in underserved populations.

Objective

This study aimed to identify risk factors for vitamin D deficiency/insufficiency among Puerto Rican adults from the Boston Puerto Rican Health Study (BPRHS).

Methods

A total of 822 adults (45–75 y, at baseline) were included in these analyses. Deficiency was defined as serum 25-hydroxyvitamin D [25(OH)D] <30 and insufficiency as 30 to <50 nmol/L. Dietary intake was assessed with a validated FFQ. Associations between risk factors, including dietary vitamin D, supplement use, ancestry, skin pigmentation, months in the past year spent in a southern climate, and serum 25(OH)D were assessed with multivariable general linear models.

Results

Approximately 13% of participants were deficient in 25(OH)D and another 43% insufficient. Skin pigment was associated with 25(OH)D using 3 measures, greater African ancestry (β ± SE) (–7.74 ± 2.91, P = 0.01); interviewer assessed dark or medium, compared with white, skin tone, (–5.09 ± 2.19, P = 0.02 and –5.89 ± 1.58, P < 0.001, respectively); and melanin index of the upper inner right arm, assessed using a spectrophotometer (–2.04 ± 0.84, P = 0.02). After adjusting for ancestry, factors associated with lower serum 25(OH)D included smoking (–4.49 ± 1.58, P = 0.01); BMI (–0.21 ± 0.10, P = 0.04); and spring compared with autumn blood draw (–4.66 ± 1.68, P = 0.004). Factors associated with higher serum 25(OH)D included female sex compared with male (4.03 ± 1.58, P = 0.01); dietary vitamin D intake μg/d (0.71 ± 0.25, P < 0.004); vitamin D supplement use (4.50 ± 1.87, P = 0.02); income to poverty ratio (0.01 ± 0.01, P = 0.06), and months in a southern climate during the past year (0.96 ± 0.56, P = 0.09).

Conclusions

Vitamin D deficiency/insufficiency was prevalent in this Puerto Rican population living in the northeastern USA. Several factors were associated with this, which may assist in identifying those at risk. Interventions are needed to improve serum 25(OH)D concentration, particularly among those with limited exposure to sunlight.

Introduction

Vitamin D deficiency is a significant public health problem, as this nutrient is essential for controlled cell growth, reduced inflammation, and optimal immune and neuromuscular function (1–3). More than 1 billion individuals worldwide are estimated to be deficient {25-hydroxyvitamin D [25(OH)D]: <30 nmol/L} or insufficient [25(OH)D: 30–50 nmol/L] in vitamin D (4). Vitamin D plays an essential role in regulating many physiologic pathways, including control of the renin angiotensin system (5), calcium absorption (6), insulin synthesis and glucose metabolism (7), antioxidant defense, and DNA repair mechanisms (8). Vitamin D deficiency has primarily been characterized by impaired bone metabolism, whereas insufficient vitamin D status has been associated with other conditions (9). Vitamin D sufficiency may be important for reducing the risk of developing chronic conditions, such as diabetes, hypertension, cancer, cardiovascular disease, and osteoporosis (5–8, 10).

Vitamin D is found in few food sources (11, 12) and historically, a significant amount of vitamin D in the body was produced from skin exposure to sunlight (13–15). However, in today's society with air conditioning, use of sunscreen, and limited outdoor exposure, an increasing proportion of the population is experiencing vitamin D deficiency (16). In addition, older adults are at increased risk of vitamin D deficiency/insufficiency (17, 18) due to the reduced capacity to synthesize the vitamin D precursor in skin (19). Previously identified determinants of vitamin D status include age, sex, body weight, vitamin D supplement use, multivitamin use, skin tone, skin pigmentation, latitude of residence, plasma creatinine, use of sunscreen, and skin-covering clothing (17, 18, 20–25). The majority of studies that have examined risk factors for vitamin D status have been conducted in primarily non-Hispanic white populations (18, 26, 27), with few in ethnic and minority populations (28, 29).

Hispanics have been shown to have 2.3 times higher prevalence of vitamin D insufficiency, compared with non-Hispanic whites (18). Those aged 40 y and older also have higher prevalence of insufficiency than younger groups (18). Puerto Ricans are the second largest Hispanic origin group in the USA, and have been shown to experience disparities in health outcomes such as diabetes, depression, and heart disease compared with non-Hispanic whites (30–32). These may be due, in part, to vitamin D insufficiency. To our knowledge, no studies have examined risk factors associated with vitamin D status among older Puerto Rican adults, although this group is likely at increased risk of vitamin D insufficiency. This study aimed to identify major risk factors for vitamin D deficiency/insufficiency [serum 25(OH)D <50 nmol/L] among Puerto Rican adults from the Boston Puerto Rican Health Study (BPRHS). Understanding risk factors for vitamin D inadequacy in this population may be helpful in identifying those at risk and for developing strategies for prevention including culturally relevant dietary interventions and programs.

Methods

Study design and population

This cross-sectional study included 822 (and subgroup n = 618) Puerto Rican adults, aged 47–77 y, who participated in the Boston Puerto Rican Osteoporosis Study (BPROS), an ancillary study to the BPRHS ∼2 y after the baseline visit. Participants for the parent study were recruited in Massachusetts, via door to door approaches and through community events, as described elsewhere (33). A total of 1504 participants completed baseline interviews and 1267 completed 2-y follow-up interviews, with a mean of 2.2 y ± 1.1 between visits. Upon completion of the 2-y follow-up interview, BPRHS participants were invited to the BPROS and 973 agreed to participate. This included an interview, a bone scan using DXA, and a blood draw at the Jean Mayer USDA Human Nutrition Research Center on Aging at Tufts University. Blood samples were obtained from 958 participants and 27 had missing serum for the analysis of 25(OH)D. Of these, 38 had implausible total energy intakes (<600 or >4800 kcal/d). These cut-off points were selected based on the Willett et al. recommendations of <500 or >3500 kcal/d (34) and adjusted upward with consideration of the much higher obesity in this population, 3 had missing skin tone assessment, and 161 had incomplete information on risk factors excluding melanin index. Means were used for missing covariate data; therefore, 93 (of 161 with incomplete information) were retrieved and 822 were included in the final analysis (models 1 and 2). A subset of 618 had complete data on skin pigmentation and were included in additional analysis (model 3) (Figure 1). Those included, compared with not included, in analysis had slightly higher plasma albumin (43.1 ± 3.33 g/L compared with 42.0 ± 3.68 g/L), and those excluded, compared with not excluded, in model 3 had higher acculturation (27.1 ± 24.9% compared with 21.8 ± 21.1%) (Supplemental Tables 1 and 2, respectively). Other variables did not differ by inclusion. Participants provided informed consent. This study was approved by the Institutional Review Boards at Tufts University, Northeastern University, and the University of Massachusetts Lowell.

FIGURE 1.

Flow chart of Boston Puerto Rican Health study participants. 25(OH)D, 25-hydroxyvitamin D.

Assessment of vitamin D status

Blood samples were collected by a certified bilingual phlebotomist during the BPROS study visit after an overnight fast. Upon centrifugation (3421 × g at 4°C for 15 min), aliquots were frozen at −80°C for analysis. Serum 25(OH)D concentration was measured by extraction and using a 125I radioimmunoassay Packard COBRA II Gamma Counter (DiaSorin Inc.). The inter- and intra-assay coefficients of variation were 9.4% and 10.8%, respectively. Vitamin D deficiency was defined as serum 25(OH)D <30 nmol/L and insufficiency as serum 25(OH)D of 30 to <50 nmol/L (3).

Additional biochemical measures

Plasma creatinine was estimated using a modified Jaffe colorimetric, kinetic reaction on the Olympus AU400e with Olympus Creatinine Reagents (OSR6178) (Olympus America Inc.) (35, 36). Inter- and intra-assay coefficients of variations were 4.0% and 2.0%, respectively. Plasma albumin was measured on the Olympus AU400e using Olympus Albumin Reagents (OSR6102) (Olympus America Inc.) with a dye binding, endpoint reaction (37). Inter- and intra-assay coefficients of variations were 2.4% and 1.6%, respectively.

Sociodemographic characteristics and migration history

Age (y) and sex were assessed by questionnaire at the baseline interview. The Bi-dimensional Acculturation Scale, which records preferences in language use (Spanish or English) during 7 common daily activities, was assessed at the 2-y BPRHS follow-up (summary score from 0 to 100%) (38). The income to poverty ratio was calculated at the 2-y visit using total household income divided by poverty threshold for the specific year and household size, following United States Department of Health and Human Services guidelines (39). Sociodemographic factors were considered as potential risk factors for vitamin D status, based on findings from previous studies (17, 18, 40, 41). Additional risk factors included season of blood draw and number of months in the past year spent in Puerto Rico or other southern climate areas.

Dietary intake and supplement use

Dietary intake was assessed with an FFQ at the 2-y follow-up. The FFQ used in the BPRHS was adapted from the Block/National Cancer Institute FFQ and validated for use in this population (42–45). Food lists and portion sizes were expanded using data from the Hispanic Health and Nutrition Examination Survey (HHANES) dietary recalls for Puerto Rican adults. Total daily intakes of nutrients were calculated using the University of Minnesota Nutrient Database for Research (NDS-R, 2010) (46). Dietary vitamin D intake was adjusted for total energy intake using the residual method, which helps to correct for the structure of the FFQ, and to partially standardize for body size and physical activity, as described by Willett et al. (47). Use of dietary supplements was ascertained through the FFQ. Those reporting use of vitamin D supplements, multivitamins, or calcium supplements were categorized as users (Y/N), for each.

Lifestyle behaviors

Data on the use of sunscreen was assessed by questionnaire at the BPROS, whereas smoking status (noncurrent and current), alcohol intake (none, moderate, and heavy consumption), and physical activity (continuous score) were assessed at the 2-y interview. Physical activity was measured using a modified version of the Harvard Alumni survey Paffenbarger questionnaire (48, 49). A physical activity score was estimated for participants based on the sum of hours spent sleeping or in sedentary, light, moderate, or heavy activities over a 24-h period, multiplied by weighting factors that correlate with oxygen consumption rates for corresponding activities.

Anthropometric measures

Standing height and weight were measured in duplicate and an average of the 2 measures was used (50, 51). Weight was estimated with a clinical scale (Toledo Weight Plate, Model 15S) and height was measured using a SECA 214 portable stadiometer. BMI was calculated as weight (kg) divided by height (m) squared.

Ancestry, skin tone, and skin pigmentation

A genome-wide association study of BPRHS was conducted using Affymetrix's Axiom Genome-Wide LAT Array, designed specifically for Hispanic populations and contains probe sets to genotype 817,810 single nucleotide polymorphisms (SNPs). To estimate ancestries, 50,704 SNPs were further selected based on the following standards: call rate >97%, minor allele frequency ≥5%, pair-wise linkage disequilibrium R square ≤0.1, P value of Hardy–Weinberg Equilibrium (HWE) ≥10⁻⁶. The program “ADMIXTURE” was used for ancestry estimate (52, 53).

The skin tone of participants was assessed at the 2-y interview based on the interviewers’ perception. Individuals were categorized as having dark, medium, light, or white skin tone.

Skin pigmentation was measured at 3 body sites using a CM-700D Spectrophotometer (Konica Minolta) at a later exam (wave 3) (54). The frequencies from low skin reflectance (darker pigmentation) were used to determine the total melanin concentration of the skin. The melanin index of the upper inner right arm was assessed as the mean of 3 measurements and was considered the base measurement of skin pigmentation (55).

Statistical analysis

SAS statistical software (SAS version 9.4; SAS Institute Inc.) was used to perform all analyses. Normality of distributions of all variables was examined; t tests and chi-square tests were conducted to compare differences between means and frequencies, respectively. The Spearman correlation was used to evaluate the relation between interviewer reported skin tone (dark, medium, light, and white), ancestry (African), and melanin index (upper inner right arm).

Means were used to impute missing data for BMI, physical activity, income to poverty ratio, months lived in a southern climate, and creatinine and albumin concentrations, to minimize exclusion from analyses based on missing data (56). To examine potential risk factors associated with serum 25(OH)D status, the following variables were included in multivariable linear regression models: age, sex, BMI, physical activity, energy-adjusted dietary vitamin D, vitamin D supplement use, multivitamin use, calcium supplement use, plasma creatinine, plasma albumin, smoking status, season of blood draw, number of months lived in a southern climate, sunscreen use (asked as “when you go out, do you normally use sunscreen”), income to poverty ratio, acculturation, and alcohol consumption. Model 1 included skin tone, model 2 included ancestry, and model 3 included melanin index of the upper inner right arm, in addition to the above-mentioned potential factors. In sensitivity analyses, we repeated the models with vitamin D intake not adjusting for energy intake.

Results

Vitamin D deficiency/insufficiency was prevalent; 15.9% of men and 11.2% of women had 25(OH)D <30 nmol/L; 44.7% of men and 41.5% of women were in the insufficient category (30 to <50 nmol/L) (Figure 2). Dietary vitamin D, adjusted for total energy, varied significantly between classification levels and was higher with increasing serum 25(OH)D status (Figure 3). European ancestral markers were predominant in both men and women (56 ± 16% and 57 ± 15%) compared with African (28 ± 16% and 28 ± 15%) and native American (16 ± 7% and 15 ± 6%) (57). Individuals with vitamin D deficiency/insufficiency, compared with adequate status, tended to be younger (mean ± SD) (57.8 y ± 7.3 compared with 59.4 y ± 7.34 y, P = 0.002), to have higher BMI (32.7 ± 6.56 compared with 31.6 ± 6.44, P = 0.02), and to consume less vitamin D from dietary sources (4.66 ± 2.59 μg/d compared with 5.26 ± 2.40 μg/d, P < 0.001) (Table 1). Those with vitamin D deficiency/insufficiency were also more likely to have their blood drawn during the spring season (62.4% compared with 37.6%, P = 0.05), to have a dark or medium skin tone (59.6% or 60.5%, respectively), to be current smokers (62.9% compared with 37.1%, P = 0.02), and were less likely to be taking vitamin D (57.9% compared with 39.7%, P < 0.001) and calcium (58.7% compared with 42.9%, P < 0.001) from supplements. Dietary supplements were the greatest contributor (34.1%) to total vitamin D intake, followed by consumption of milk (28.4%), fish (15.3%), cold cereal (4.4%), eggs (4.2%), and beef (2.3%) (Table 2).

FIGURE 2.

Percentage distribution of Puerto Rican adults (596 women and 226 men aged 47–77 y), by sex and serum 25(OH)D status. Deficiency was defined as serum 25(OH)D <30 nmol/L, insufficiency as 30 to <50 nmol/L, and sufficiency as ≥50 nmol/L (3). 25(OH)D, 25-hydroxyvitamin D.

FIGURE 3.

Box plot of adjusted dietary vitamin D (n = 822 participants included in the final analysis), by level of serum 25(OH)D status. Deficiency was defined as serum 25(OH)D <30 nmol/L, insufficiency as 30 to <50 nmol/L, and sufficiency as ≥50 nmol/L (3). Mean energy-adjusted vitamin D intake was highest among individuals with sufficient (n, mean ± SD) (371, 4.93 ± 2.85 μg/d), followed by insufficient (348, 4.65 ± 3.13 μg/d) and deficient (103, 4.12 ± 2.83 μg/d) (P < 0.05) serum 25(OH)D status. 1-factor ANOVA was used to test differences between group means. 25(OH)D, 25-hydroxyvitamin D.

TABLE 1.

Participant characteristics in the Boston Puerto Rican Health Study, by vitamin D sufficiency status

	Serum 25(OH)D
	Vitamin D deficiency/insufficiency	Vitamin D sufficiency
Characteristics	<50 nmol/L (n = 451)	≥50 nmol/L (n = 371)	P 1
Age, y	57.8 ± 7.302	59.4 ± 7.34	0.002
Sex, %
Male	60.6	39.4
Female	52.7	47.3	0.04
BMI, kg/m2	32.7 ± 6.56	31.6 ± 6.44	0.02
Physical activity3	31.9 ± 4.80	31.7 ± 4.44	0.64
Alcohol use, %
None	53.3	46.8
Moderate	55.7	44.3
Heavy	73	27	0.06
Smoking status (current), %
No	52.8	47.2
Yes	62.9	37.1	0.02
Vitamin D intake,4 μg/d	4.66 ± 2.59	5.26 ± 2.40	<0.001
Supplement use, %
Vitamin D
No	57.9	42.1
Yes	39.7	60.3	<0.001
Calcium
No	58.7	41.4
Yes	42.9	57.1	<0.001
Multivitamin
No	56.8	43.3
Yes	49	51	0.06
Months in PRSC5	0.32 ± 0.95	0.45 ± 1.21	0.07
Acculturation6	23.2 ± 22.3	23.1 ± 22.0	0.95
Income to poverty ratio7	113 ± 97.2	125 ± 116	0.09
Skin tone
Dark	59.6	40.4
Medium	60.5	39.5
Light	50.2	49.8
White	49.8	50.2	0.03
Ancestry, %
African	27.0 ± 22.9	24.4 ± 21.7	0.10
Native-American	13.0 ± 12.4	13.4 ± 11.7	0.61
European	60.1 ± 21.8	62.2 ± 20.3	0.15
Melanin index of the upper inner right arm8	1.50 ± 0.87	1.42 ± 0.74	0.26
Season of blood draw, %
Winter	54.6	45.4
Spring	62.4	37.6
Summer	51.2	48.9
Autumn	50.8	49.2	0.05
Sunscreen use, %
No	55.8	44.2
Yes	52.3	47.7	0.37
Plasma creatinine, mg/dL	0.83 ± 0.25	0.86 ± 0.29	0.15
Plasma albumin, g/L	43.1 ± 3.22	43.1 ± 3.45	0.99

¹ P values for difference in population means (t test) or between categories (chi-square test).

²Values are mean ± SDs unless otherwise indicated.

³Physical activity score was estimated based on sum of hours spent sleeping or in sedentary, light, moderate, or heavy activities in a 24-h period, multiplied by weighing factors associated with oxygen consumption rates for corresponding activities (range: 24.5–60.5).

⁴Dietary vitamin D was adjusted for total energy using the residual method.

⁵Months lived in Puerto Rico or another southern climate in the past 1 y.

⁶Acculturation, 100% fully accultured speak fluent English for all interactions, whereas 0% speak only Spanish (range: 0–95.8).

⁷Income to poverty ratio was calculated using total household income divided by poverty threshold for the specific year and household size (39) (range: 1.77–1520).

⁸Melanin index of the upper inner right arm was estimated from the mean of 3 measurements of melanin concentration at the upper inner right arm (<50 nmol/L, n = 332 and ≥50 nmol/L, n = 371; range: 0.58–10.9).

PRSC, Puerto Rico or other southern climate; 25(OH)D, 25-hydroxyvitamin D.

TABLE 2.

Top 10 contributors to total vitamin D intake in the Boston Puerto Rican Health Study¹

Food groups	Mean vitamin D intake (μg/d)	Contribution to total vitamin D (%)
Vitamin supplements2	2.4	34.1
Milk	2.0	28.4
Fish	1.1	15.3
Cereal (cold, ready-to-eat)3	0.3	4.4
Eggs	0.3	4.2
Beef	0.2	2.3
Chicken/turkey	0.1	2.0
Yogurt	0.1	2.0
Processed meat, sausage, franks	0.1	1.8
Pork	0.1	1.3

¹Ranking using 64 food groups and supplements; above 10 items explain 95.8% of intake.

²Multivitamins and single supplements.

³Includes fortified cereals.

As expected, positive correlations were observed between African ancestry and melanin index of the upper inner right arm (n = 618) (0.57, P < 0.001), interviewer reported skin tone and African ancestry (0.49, P < 0.001), and skin tone and melanin index of the upper inner right arm (0.50, P < 0.001). Therefore, each was included in separate models.

In fully adjusted models (Table 3), potential factors associated with serum 25(OH)D status were examined with interviewer reported skin tone (model 1), African ancestry (model 2), and skin melanin index (model 3). Results for these 3 measures were each significant, ranging from –2.04 ± 0.84 nmol/L for melanin index (basic skin tone) to –5.09 ± 2.19 (P = 0.02) and –5.89 ± 1.58 (P < 0.001) for dark and medium interviewer rated skin tone, compared with white, respectively, to –7.74 ± 2.91 (P = 0.01) for African compared with European ancestry.

TABLE 3.

Factors associated with serum 25(OH)D (nmol/L) in older Puerto Rican adults living in Massachusetts

	Model
	1 (n = 822)		2 (n = 822)		3 (n = 618)
	β ± SE	P	β ± SE	P	β ± SE	P
Age	0.12 ± 0.09	0.17	0.11 ± 0.09	0.21	0.13 ± 0.10	0.20
Female (vs. male)	3.48 ± 1.56	0.03	4.03 ± 1.58	0.01	4.68 ± 1.78	0.01
BMI	–0.24 ± 0.10	0.02	–0.21 ± 0.10	0.04	–0.21 ± 0.12	0.08
Physical activity1	–0.15 ± 0.14	0.27	–0.10 ± 0.14	0.45	–0.16 ± 0.16	0.31
Dietary vitamin D,2 μg/d	0.68 ± 0.25	0.01	0.71 ± 0.25	0.004	1.05 ± 0.29	<0.001
Vitamin D supplement use (vs. no)	4.41 ± 1.86	0.02	4.50 ± 1.87	0.02	3.13 ± 2.15	0.15
Multivitamin use (vs. no)	1.94 ± 1.45	0.18	2.30 ± 1.46	0.12	2.88 ± 1.64	0.08
Calcium supplement use (vs. no)	2.89 ± 1.71	0.09	2.67 ± 1.73	0.12	1.68 ± 1.97	0.39
Plasma creatinine, mg/dL	5.37 ± 2.43	0.03	5.49 ± 2.45	0.03	4.86 ± 2.71	0.07
Plasma albumin, g/L	–0.14 ± 0.19	0.46	–0.15 ± 0.19	0.42	–0.12 ± 0.24	0.63
Smoking (vs. no)	–4.34 ± 1.58	0.01	–4.49 ± 1.58	0.01	–3.60 ± 1.84	0.05
Acculturation3	0.02 ± 0.03	0.60	0.02 ± 0.03	0.64	0.02 ± 0.04	0.65
Sunscreen	–0.78 ± 1.39	0.58	–0.33 ± 1.41	0.82	–0.81 ± 1.60	0.61
Season (vs. Autumn)
Winter	–3.81 ± 1.91	0.05	–3.44 ± 1.92	0.07	–3.42 ± 2.17	0.12
Spring	–5.07 ± 1.67	0.003	–4.66 ± 1.68	0.01	–5.60 ± 1.94	0.004
Summer	–0.47 ± 1.64	0.78	–0.13 ± 1.65	0.94	–0.59 ± 1.92	0.76
Months in PRSC4	0.96 ± 0.56	0.09	1.07 ± 0.57	0.06	1.83 ± 0.66	0.01
Income to poverty ratio5	0.01 ± 0.01	0.06	0.01 ± 0.01	0.06	0.02 ± 0.01	0.02
Alcohol (vs. none)
Moderate	–0.98 ± 1.38	0.48	–1.36 ± 1.38	0.33	–1.01 ± 1.58	0.52
Heavy	–3.84 ± 3.01	0.20	–4.39 ± 3.02	0.15	–6.47 ± 3.57	0.07
Skin tone (vs. white)
Dark	–5.09 ± 2.19	0.02	—	—	—	—
Medium	–5.89 ± 1.58	<0.001	—	—	—	—
Light	–0.12 ± 1.71	0.94	—	—	—	—
Ancestry (vs. European)
Native American	—	—	–4.37 ± 5.43	0.42	—	—
African	—	—	–7.74 ± 2.91	0.01	—	—
Melanin index of the upper inner right arm6	—	—	—	—	–2.04 ± 0.84	0.02

¹Physical activity score was estimated based on sum of hours spent sleeping or in sedentary, light, moderate, or heavy activities in a 24-h period, multiplied by weighing factors associated with oxygen consumption rates for corresponding activities (range: 24.5–60.5).

²Adjusted for energy using the residual method; all models repeated with vitamin D intake not adjusted for energy intake remained significant (P < 0.05).

³Acculturation, 100% fully accultured speak fluent English for all interactions, whereas 0% speak only Spanish (range: 0–95.8).

⁴Months lived in Puerto Rico or other southern climate in the past year.

⁵Income to poverty ratio was calculated using total household income divided by poverty threshold for the specific year and household size (39) (range: 1.77–1520).

⁶Melanin index of the upper inner right arm was estimated from the mean of 3 measurements of melanin concentration at the upper inner right arm (range: 0.58–10.9); the model repeated with this measure at the forehead remained significant (P < 0.05), but at the hand, significance was lost (P = 0.22).

PRSC, Puerto Rico or other southern climate.

For the most part, other factors associated with serum 25(OH)D were consistent across models. In all models higher serum 25(OH)D concentration (nmol/L) was significantly associated with female sex (β ± SE) (ranging from 3.48 ± 1.56, P = 0.03 to 4.68 ± 1.78, P = 0.01), energy-adjusted vitamin D intake (μg/d) (0.68 ± 0.25, P = 0.01 to 1.05 ± 0.29, P < 0.001), smoking (–3.60 ± 1.84, P = 0.05 to –4.49 ± 1.58, P = 0.01), and measurement in the spring compared with autumn (–4.66 ± 1.68, P = 0.01 to –5.60 ± 1.94, P = 0.004). In addition, BMI was significant in models 1 and 2, and approached significance in model 3 (from –0.21 ± 0.12, P = 0.08 to –0.24 ± 0.10, P = 0.02). Intake of vitamin D supplements was significant in models 1 and 2 (4.41 ± 1.86 and 4.50 ± 1.87, each P = 0.02), whereas multivitamin use approached significance in model 3 (2.88 ± 1.64, P = 0.08). Exposure to a southern climate during the past year (months) was significantly associated with 25(OH)D in model 3 and approached significance in models 1 and 2 (ranging from 0.96 ± 0.56, P = 0.09 to 1.83 ± 0.66, P = 0.01). Similarly, the income/poverty ratio was significantly associated with 25(OH)D in model 3 and approached significance in models 1 and 2 (ranging from 0.01 ± 0.01, P = 0.06 to 0.2 ± 0.01, P = 0.02). In sensitivity analyses, replacing energy-adjusted vitamin D dietary intake with unadjusted intake, the results remained significant (P < 0.05).

In these fully adjusted models, there were no significant associations between serum 25(OH)D and physical activity, alcohol use, calcium supplement use, acculturation, or self-reported sunscreen use.

Discussion

Approximately 61% of the Puerto Rican men and 53% of women in this Boston area, Massachusetts, study were identified as having deficient/insufficient vitamin D status, with 15.9% of men and 11.2% of women deficient, based on 25(OH)D <30 nmol/L. This prevalence is based on current recommendations by the Institute of Medicine (IOM), recognizing that some others recommend higher concentrations for health. For example, the Endocrine Society recommends that the cut-off for deficiency be set at 50 nmol/L and insufficiency at 50–75 nmol/L (58). Use of these higher cut-off points would increase the level of deficiency to >2/3 the sample, with very few sufficient. Three measures of skin tone: ancestral markers, interviewer reported skin tone, and measured melanin index were each associated with 25(OH)D concentration, showing the heightened risk of dark skin in this northern latitude, as observed in prior studies (20, 59). Puerto Ricans are racially admixed and African ancestry is associated with lower 25(OH)D. Exposure to sunlight is a primary predictor of vitamin D status, and African ancestry is associated with more melanin, which absorbs UV and reduces synthesis of vitamin D (24, 60). One study showed a significant positive association between the percentage of European ancestry and 25(OH)D concentration among African-American women who were not taking supplemental vitamin D (61). The correlation between African ancestry and 25(OH)D extends beyond diet and skin pigmentation differences. Future studies should investigate additional genetic polymorphisms and variation in diet, which may affect absorption or conversion of vitamin D to its active state. For example, considerable ethnic variation has been observed in the vitamin D receptor gene (62).

Similar to other findings, (17, 18, 21, 22, 25) 25(OH)D was lowest in the spring season, after a winter with limited sun exposure. Evidence shows that photoinduced skin pigmentation causes stimulation of melanocytes in the skin by vitamin D3 (63), which may inhibit 25(OH)D serum concentration. A previous study found low vitamin D status among Hispanic adults in southern latitudes, despite high chronic sun exposure in those areas (64). In contrast, our findings show that time spent in a southern latitude increased 25(OH)D concentration in this population. The precursor of vitamin D is lower in the skin of older adults (19), thus low vitamin D status may be expected in this older population, given the latitude and sun exposure intensity in the northern USA.

Major dietary sources of vitamin D were similar to those noted in other populations (65, 66), including milk, fish, cold cereals, eggs, and beef. The mean dietary intakes of vitamin D from food sources (4.66 ± 2.59 μg/d and 5.26 ± 2.40 μg/d, for vitamin D insufficient and sufficient individuals, respectively) are less than one third of national dietary recommendations (RDA for adults aged 19–70 and >70 y: 15 μg/d and 20 μg/d, respectively) (3). The mean vitamin D intake of those with sufficient serum 25(OH)D status in this population was lower than that reported for US non-Hispanic white adults (6.75 ± 6.42 μg/d) with adequate vitamin D status of 25(OH)D (26).

After adjusting for skin tone and respective variables, dietary intake of vitamin D, supplemental vitamin D, BMI, smoking, season of blood draw, income to poverty ratio, and exposure to southern climates (17, 20, 23, 24) were each associated with serum 25(OH)D status. As shown in previous studies, smoking was associated with lower 25(OH)D (18, 21, 22). It is possible that, because smoking introduces oxidative damage that interferes with normal cell function, it may impair absorption and synthesis of vitamin D (67). Our findings are consistent with studies in other populations but highlight the importance of ancestry, skin pigmentation, and migration history for 25(OH)D status in this population. Diet, sex, smoking, season, latitude, and time spent outdoors have been reported to explain <40% of the variation in 25(OH)D (68–74), suggesting that more investigation is needed to fully understand the factors associated with vitamin D status.

This study has several strengths, which include the use of multiple measures of skin tone, including objective measures, with consistent results. It is also one of the few studies to comprehensively examine vitamin D status in the Puerto Rican population, a group affected by serious health disparities which may be influenced by vitamin D. It is limited by its cross-sectional design and, therefore, cannot determine causality. Additional limitations include exclusion of >10% of the participants in the BPRHS 2-y follow up study due primarily to missing data. However, those included did not differ from those not included, with the exception of plasma albumin concentration.

In conclusion, more than half of Puerto Rican adults in this Boston area study were serum 25(OH)D deficient or insufficient. It is important to identify risk factors associated with serum 25(OH)D status, as these will help determine interventions to prevent or treat vitamin D deficiency/insufficiency, which is associated with other chronic health conditions. Individuals with a higher proportion of African ancestry and with darker skin tone are at higher risk of vitamin D insufficiency. Dietary recommendations tailored for this population are needed to encourage increased intake of foods containing vitamin D such as salmon, milk, and eggs. In addition to improving dietary vitamin D intake, supplement use may be an important strategy for increasing circulating vitamin D concentration, especially among individuals with poor diet and less exposure to sunlight.

Notes

The Boston Puerto Rican Health Study (BPRHS) and Osteoporosis Study (BPROS) were supported by the NIH (P01 AG023394, P50 HL105185, R01 AG055948, R01 AG027087, and R01 AR072741). SEN is supported by K01 AR067894.

Author disclosures: The authors report no conflicts of interest.

Supplemental Tables 1 and 2 are available from the “Supplementary data” link in the online posting of the article and from the same link in the online table of contents at https://academic.oup.com/jn/.

OJA and C-QL are joint first authors of this manuscript.

Abbreviations used: BPRHS, Boston Puerto Rican Health Study; BPROS, Boston Puerto Rican Osteoporosis Study; SNP, single nucleotide polymorphism; 25(OH)D, 25-hydroxyvitamin D.