Vitamin D deficiency is common in older adults, with recent studies describing deficiency in 18–25 % of adult postmenopausal women(Reference Holick, Siris, Binkley, Beard, Khan, Katzer, Poetruschke, Chen and dePapp1, Reference Lips, Hosking, Lippuner, Norquist, Wehren, Maalouf, Ragi-Els and Chandler2). Vitamin D deficiency results from several factors including inadequate sun exposure, reduced cutaneous vitamin D synthesis, poor nutrition, and certain medications and co-morbid diseases such as anticonvulsants and coeliac sprue(Reference Vieth3). The increasing measurement of serum 25-hydroxyvitamin D (25(OH)D) in older people arises from an increasing awareness of the prevalence of hypovitaminosis D and the role of vitamin D in both the prevention and management of osteoporosis(Reference Bischoff-Ferrari, Willett, Wong, Giovannucci, Dietrich and Dawson-Hughes4). However, the prevalence of hypovitaminosis D and its impact on health is less certain in adults under 50 years of age.
Healthy young adults may develop vitamin D deficiency for several reasons. First, the vitamin D intake of young adults is often below the recommended intake of 200 IU/d(Reference Gartner and Greer5, Reference Misra, Tsai, Anderson, Hubbard, Gallagher, Soyka, Miller, Herzog and Klibanski6). Second, young adults today spend less time outside than young adults one decade ago(Reference Hartman7–Reference Schofield, Freeman, Dixon, Borland and Hill9). Third, the increasing use of sunscreen to reduce skin damage or cancer may decrease or eliminate cutaneous vitamin D synthesis(Reference Cokkinides, Weinstock, Glanz, Albano, Ward and Thun8, Reference Matsuoka, Ide, Wortsman, MacLaughlin and Holick10, Reference Peacey, Steptoe, Sanderman and Wardle11). Finally, many young adults drink carbonated beverages in place of milk, thereby decreasing the intake of both Ca and vitamin D and potentially increasing the risk of fracture(Reference Tucker, Morita, Qiao, Hannan, Cupples and Kiel12).
Despite mounting evidence that young adults are at risk for vitamin D deficiency, no specific recommendations exist regarding evaluation of their vitamin D status. Such lack of guidelines may result from limited information on either the impact of vitamin D on the development of peak bone mass(Reference Valimaki, Alfthan, Lehmuskallio, Loyttyniemi, Sahi, Stenman, Suominen and Valimaki13–Reference Bischoff-Ferrari, Dietrich, Orav and Dawson-Hughes15) or the long-term safety of increasing serum 25(OH)D levels in young adults. Although epidemiological data suggest that improved vitamin D status may decrease the risk of certain cancers and autoimmune diseases(Reference Barger-Lux and Heaney16–Reference Merlino, Curtis, Mikuls, Cerhan, Criswell and Saag18), a true cause–effect relationship has not been established. Measurement of serum 25(OH)D is costly, with charges ranging from $US 45 to $US 100. A questionnaire to identify persons at high or low risk of vitamin D deficiency would be clinically useful, particularly as there is no consensus regarding the indications for measurement of 25(OH)D in young adults.
Other groups have used questionnaires to detect hypovitaminosis D. One group queried subjects on the use of multivitamins, milk and other foods containing vitamin D(Reference Tangpricha, Pearce, Chen and Holick19) and revealed a positive correlation between serum 25(OH)D levels and multivitamin intake. However, subjects did not record sun exposure. A study in Icelandic women demonstrated an association between higher serum 25(OH)D levels and sun-seeking and dietary habits, but the questionnaire itself was not published(Reference Sigurdsson, Franzson, Porgeirsdottir and Steingrimsdottir20). Utilizing questions to assess diet and sun exposure, a third group reported associations between serum 25(OH)D levels and season of measurement, BMI, age, time spent indoors, living in three southern states, vitamin D intake and creatinine. However, the study was limited to elderly subjects(Reference Jacques, Felson, Tucker, Mahnken, Wilson, Rosenberg and Rush21).
We hypothesized that a simple questionnaire could identify young adults with a high and low likelihood of vitamin D deficiency. We designed a series of questions to assess use of vitamin D-containing supplements, milk and sun exposure in order to test this hypothesis.
Materials and methods
Between January and May 2004 we recruited 184 men and women between the ages of 18 and 40 years into the current study. The purposes of the study were twofold. First, we wished to estimate the prevalence of vitamin D deficiency in young adults. Second, we queried whether a questionnaire could identify subjects at high or low risk of vitamin D deficiency. We excluded individuals with pre-existing conditions affecting vitamin D and/or Ca metabolism including liver or kidney disease, eating disorders, skin diseases and use of oral corticosteroids, anticonvulsants, insulin or bisphosphonates. We paid volunteers $US 20 for the single study visit and notified them of vitamin D test results by mail.
The Human Subjects Committee of the University of Wisconsin approved the study protocol. Participants received verbal and written descriptions of the study, signed the consent form and retained a copy for their records. We recorded the age, gender and self-reported race of each person at the study visit. Each subject completed a questionnaire designed to quantify intake of vitamin D through diet and sun exposure and to record the presence of conditions or medications known to affect vitamin D stores (Table 1).
Table 1 The Vitamin D Questionnaire
To measure 25(OH)D, we collected blood from non-fasting participants(Reference Juttmann, Visser, Buurman, de Kam and Birkenhager22) and transported samples, without exposure to light, to a central laboratory at the University of Wisconsin. Samples were stored at −70°C until analysis. Subsequently, we measured serum 25(OH)D using a Liaison chemiluminescence assay (DiaSorin Inc., Stillwater, MN, USA). The chemiluminescence assay is an accurate, rapid and precise method for vitamin D measurement, correlating well with traditional radioimmunoassay but overestimating levels by 3·9 ng/ml compared with HPLC(Reference Lensmeyer, Wiebe, Binkley, Drezner, Singh and Darcy23–Reference Ersfeld, Rao, Body, Sackrison, Miller, Parikh, Eskridge, Polinske, Olson and MacFarlane25). In a study of 329 clinical samples, the intra- and inter-assay CV for this assay were 8–13 % and 8–15 %, respectively(Reference Ersfeld, Rao, Body, Sackrison, Miller, Parikh, Eskridge, Polinske, Olson and MacFarlane25).
The precise cut-off points used to define vitamin D adequacy, insufficiency and deficiency vary, depending on the assay utilized and the investigator. However, many experts use a 25(OH)D level <30 ng/ml (75 nmol/l) to define vitamin D insufficiency and a level <16 ng/ml (40 nmol/l) to define deficiency(Reference Vieth3). For the purposes of the present study, we used serum 25(OH)D < 16 ng/ml to categorize individuals as vitamin D-deficient.
Statistical analysis
We summarized data as the mean and standard deviation for continuous variables and as frequencies and percentages for categorical variables. We compared vitamin D-sufficient and -deficient subjects using the Wilcoxon rank-sum test for continuous variables and the χ 2 or Fisher’s exact test for categorical variables. We used univariate and multivariate logistic regression models to evaluate the effects of questionnaire responses on odds ratios for vitamin D deficiency. We controlled for age and gender in multivariate analysis. This is because age and gender are significantly associated with the likelihood of vitamin D deficiency and both likely affect nutritional and sun-seeking habits. We assessed the sensitivity and specificity of combinations of questions in their ability to identify subjects with vitamin D deficiency. We completed analyses using SAS version 9·1 (SAS Institute, Cary, NC, USA) and R version 2·4·0 (The R Project for Statistical Computing, http://www.r-project.org) statistical software packages.
Results
We recruited 184 subjects for the study. Two-thirds of subjects (n 124) participated during winter and 33 % (n 60) during the spring. Subjects’ mean (sd) age was 24 (4) years while the median age and range were 22·4 years and 18 to 40 years, respectively. Over half of the subjects (53 %, n 98) were female and 90 % (n 165) were Caucasian (Table 2). Mean (sd) serum 25(OH)D levels were 25 (11) ng/ml (range 4 to 52 ng/ml). Nearly three in four subjects (71 %, n 130) had serum 25(OH)D level <30 ng/ml and one in four (26 %, n 48) subjects were vitamin D-deficient.
Table 2 Participants’ demographics and answers to the vitamin D questionnaire: young adults (n 184) living in Madison, Wisconsin, January–May 2004
25(OH)D, 25-hydroxyvitamin D.
Data are shown as mean values with their standard deviation (indicated by *) or as numbers and percentage.
Table 2 summarizes the entire group’s answers to the questionnaire. Participants reported drinking an average of 1·9 (1·5) servings of milk daily. Nearly half of the subjects (46 %, n 84) ingested a daily multivitamin but only 6 % (n 11) took an additional vitamin D supplement and 3 % (n 5) reported daily use of cod-liver oil. Subjects’ mean vitamin D intake through milk was 188 (148) IU and through supplements was 77 (238) IU daily. Eighty-five per cent of responders (n 157) reported sun tanning, 88 % (n 161) reported sunscreen use, 29 % (n 53) reported over 30 min of sun exposure daily and 35 % (n 64) reported tanning booth use in the past year.
Several characteristics identified subjects more likely to have vitamin D deficiency (Table 3). Individuals with vitamin D deficiency reported lower milk intake (1·2 v. 2·1 servings/d, P < 0·001). Subjects with deficiency were less likely to report a suntan (71 % v. 90 %, P < 0·001) or use a tanning booth (4 % v. 46 %, P < 0·001). Individuals with vitamin D deficiency were slightly older (mean age 26·5 years v. 22·9 years, P < 0·001) and more often male (65 % v. 40 %, P = 0·005). Finally, non-Caucasian individuals were more likely to be vitamin D-deficient than Caucasians (deficiency in 67 % of non-Caucasians v. 23 % of Caucasians, P < 0·001).
Table 3 Characteristics of subjects with and without vitamin D deficiency: young adults (n 184) living in Madison, Wisconsin, January–May 2004
25(OH)D, 25-hydroxyvitamin D.
Data are shown as mean values with their standard deviation (indicated by*) or as numbers and percentage.
We performed univariate and multivariate logistic regression analyses to determine the odds ratio for vitamin D deficiency based on individual items in the questionnaire (Table 4). In univariate results, older age and male gender conferred a greater odds ratio for vitamin D deficiency. We controlled for age and gender in multivariate analyses for two reasons. First, age and gender may themselves be associated with differing nutritional and sun-seeking habits. Second, other studies disagree on whether older age and male gender are risk factors for vitamin D deficiency(Reference Jacques, Felson, Tucker, Mahnken, Wilson, Rosenberg and Rush21, Reference Looker, Dawson-Hughes, Calvo, Gunter and Sahyoun26, Reference Zadshir, Tareen, Pan, Norris and Martins27). Results from adjusted and unadjusted analyses were similar. In multivariate analyses, adults under the median age of 22·4 years (OR = 0·25, 95 % CI 0·12, 0·53, P < 0·001), those reporting a suntan (OR = 0·24, 95 % CI 0·09, 0·63, P = 0·004), tanning booth use (OR = 0·09, 95 % CI 0·02, 0·43, P = 0·002) and daily ingestion of two or more servings of milk (OR = 0·21, 95 % CI 0·09, 0·48, P < 0·001) were less likely to be deficient. In contrast, individuals more likely to be deficient were non-Caucasians (OR = 5·50, 95 % CI 1·35, 22·41, P = 0·02) and men (OR = 3·44, 95 % CI 1·60, 7·37, P = 0·002). We performed additional univariate and multivariate analyses without non-Caucasian subjects (n 19); these analyses showed virtually identical findings (Table 5).
Table 4 Univariate and multivariate odds ratios for vitamin D deficiency: young adults (n 184) living in Madison, Wisconsin, January–May 2004
*Multivariate analyses are adjusted for age and gender.
Table 5 Univariate and multivariate odds ratios for vitamin D deficiency in Caucasian subjects: young adults (n 165) living in Madison, Wisconsin, January–May 2004
*Multivariate analyses are adjusted for age and gender.
We evaluated the sensitivity and specificity of the questionnaire as a screening test for vitamin D deficiency (Table 6). Three self-reported habits (suntan, tanning booth use and drinking two or more servings of milk daily) discriminated between young adults with and without vitamin D deficiency. Using a threshold of two out of three negative responses for these habits, we obtained a sensitivity of 79 % and a specificity of 78 % for identifying persons with vitamin D deficiency. We obtained a similar sensitivity (76 %) and specificity (79 %) when excluding non-Caucasian subjects (Table 7).
Table 6 Sensitivity and specificity of three questionsFootnote * for vitamin D deficiency: young adults (n 184) living in Madison, Wisconsin, January–May 2004
* Questions: (i) Have you received a suntan in the past 12 months? Yes or No; (ii) Have you used a tanning booth in the past year? Yes or No; (iii) How many servings of milk do you get daily? Two or more servings = Yes.
Table 7 Sensitivity and specificity of three questionsFootnote * for vitamin D deficiency in Caucasian subjects: young adults (n 165) living in Madison, Wisconsin, January–May 2004
* Questions: (i) Have you received a suntan in the past 12 months? Yes or No; (ii) Have you used a tanning booth in the past year? Yes or No; (iii) How many servings of milk do you get daily? Two or more servings = Yes.
Discussion
We hypothesized that, like other age groups, young adults would commonly have vitamin D deficiency. In this study of 184 healthy young subjects, 71 % had serum 25(OH)D level <30 ng/ml and 26 % were clearly vitamin D-deficient, with 25(OH)D level below 16 ng/ml. People with vitamin D deficiency may develop osteomalacia, a disease characterized by unmineralized osteoid leading to bone pain and skeletal fragility. Higher vitamin D levels are associated with increased bone mass(Reference Valimaki, Alfthan, Lehmuskallio, Loyttyniemi, Sahi, Stenman, Suominen and Valimaki13–Reference Bischoff-Ferrari, Dietrich, Orav and Dawson-Hughes15, Reference Dawson-Hughes, Dallal, Krall, Harris, Sokoll and Falconer28). Indeed, studies suggest that preventing vitamin D deficiency may optimize Ca homeostasis and facilitate peak bone mass in young adults(Reference Valimaki, Alfthan, Lehmuskallio, Loyttyniemi, Sahi, Stenman, Suominen and Valimaki13, Reference Lehtonen-Veromaa, Mottonen, Nuotio, Irjala, Leino and Viikari14). The high prevalence of deficiency in young adults highlights the need for further research to identify the precise vitamin D level needed to optimize musculoskeletal health. Such knowledge will facilitate patient education and public policy, with the goal of achieving vitamin D adequacy.
We hypothesized that a questionnaire could identify young people at high and low risk of vitamin D deficiency. A single question did not reliably distinguish between these groups. However, those subjects who received a suntan, used a tanning booth or drank at least two servings of milk daily were significantly less likely to be deficient than subjects not reporting these habits. In combination, these three items were useful in differentiating between those with and without vitamin D deficiency. For subjects responding in the negative to any two of these three questions, we obtained a sensitivity of 79 % and specificity of 78 % for predicting vitamin D deficiency. Although the questionnaire needs further revision to improve its performance, it appears that three questions may help clinicians decide whether to pursue laboratory testing for vitamin D deficiency.
It is not surprising that sun exposure and milk ingestion may protect against vitamin D deficiency. Vitamin D fortification of milk is required in the USA, based on research carried out decades ago at the University of Wisconsin. An 8 ounce glass of milk contains ~100 IU of vitamin D. Likewise, cutaneous sun exposure increases 25(OH)D levels, unless sunscreen with a sun protection factor >15 is used(Reference Webb and Holick29). Additionally, many tanning beds emit UV-B light, which increases vitamin D synthesis(Reference Sigurdsson, Franzson, Porgeirsdottir and Steingrimsdottir20, Reference Devgun, Johnson and Paterson30, Reference Tangpricha, Turner, Spina, Decastro, Chen and Holick31). Although sun-induced summer increments in serum 25(OH)D gradually decline over the winter, women with low vitamin D intake but high summer sun exposure may maintain higher serum 25(OH)D levels in the winter as well(Reference Salamone, Dallal, Zantos, Makrauer and Dawson-Hughes32). While the explanation for this observation is unknown, summer sun exposure was associated in one study with improved vitamin D status year-round(Reference Salamone, Dallal, Zantos, Makrauer and Dawson-Hughes32).
Few studies have used questionnaires to predict low serum 25(OH)D levels. In a study by Tangpricha et al., serum 25(OH)D levels were higher in subjects taking multivitamins, but not higher in milk drinkers(Reference Tangpricha, Pearce, Chen and Holick19). The authors analysed these associations between habits and serum 25(OH)D levels(Reference Tangpricha, Pearce, Chen and Holick19), rather than stratifying subjects as sufficient or deficient as we did. Our vitamin D-deficient subjects were less likely to take multivitamins compared with sufficient subjects (35 % v. 50 %, P = 0·08) but further analyses showed no effect of multivitamin use on the odds of vitamin D deficiency. Milk consumption in Tangpricha’s study was lower (mean 1·6 (sd 1) servings/d)(Reference Tangpricha, Pearce, Chen and Holick19) than that reported by our subjects, which might explain why there was no difference in vitamin D levels between their subjects who drank and did not drink milk. In a questionnaire-based Icelandic study, older women consuming fish oil or multivitamins had higher serum 25(OH)D levels (P < 0·01) than younger women who did not report these habits(Reference Sigurdsson, Franzson, Porgeirsdottir and Steingrimsdottir20). Women whose used tanning beds (P = 0·06) or travelled to warmer climates (P < 0·01) also had higher 25(OH)D levels(Reference Sigurdsson, Franzson, Porgeirsdottir and Steingrimsdottir20). A third study demonstrated associations between serum 25(OH)D and season, BMI, age, time spent indoors, living in southern states, vitamin D intake and creatinine(Reference Jacques, Felson, Tucker, Mahnken, Wilson, Rosenberg and Rush21). Taken as a whole, many of these findings are very similar to ours and suggest that questions about sun exposure and supplemental and dietary vitamin D intake may be universally useful to identify individuals at risk for vitamin D deficiency.
Based on larger cross-sectional studies(Reference Looker, Dawson-Hughes, Calvo, Gunter and Sahyoun26), we do not believe that questions about age will prove useful to exclude or suspect vitamin D deficiency. Indeed, ours is not the first study to report vitamin D inadequacy in young adults. In Iceland, younger women had lower vitamin D levels than older women(Reference Sigurdsson, Franzson, Porgeirsdottir and Steingrimsdottir20). Nearly half of adolescent girls in Maine had hypovitaminosis D at least once during a three-year observation(Reference Sullivan, Rosen, Halteman, Chen and Holick33). Two-thirds of internal medicine residents had 25(OH)D levels below 20 ng/ml during spring months in Oregon(Reference Haney, Stadler and Bliziotes34). Together these studies indicate that hypovitaminosis D is common in young adults.
One unexpected finding in the present study was that men were more likely than women to be vitamin D-deficient. In large epidemiology studies, men typically have higher serum 25(OH)D levels than women(Reference Jacques, Felson, Tucker, Mahnken, Wilson, Rosenberg and Rush21, Reference Looker, Dawson-Hughes, Calvo, Gunter and Sahyoun26, Reference Zadshir, Tareen, Pan, Norris and Martins27). However, other studies have reported no gender difference in 25(OH)D levels(Reference Jacques, Felson, Tucker, Mahnken, Wilson, Rosenberg and Rush21, Reference Rucker, Allan, Fick and Hanley35). Based on our and other studies, we do not believe gender is a useful means of identifying persons at higher risk of vitamin D deficiency.
Strengths of our study include testing of individuals of both genders from ages 18 to 40 years and uniform measurement of serum 25(OH)D by a single assay. Study weaknesses also exist. The first is the recruitment of relatively few, predominantly Caucasian, study subjects. Aside from milk, we did not query intake of other foods that might contain vitamin D; however, very few other foods contain meaningful doses of vitamin D(Reference Vieth3, Reference Utiger36). Additionally, the chemiluminescent assay used for the study may slightly overestimate serum 25(OH)D compared with the gold standard HPLC assay. We did not record time of day in the sun, although it is known that both season and time of day influence cutaneous vitamin D synthesis(Reference Webb, Kline and Holick37). We measured 25(OH)D levels in late winter and early spring, a time of low sun exposure in Wisconsin(Reference Webb, Kline and Holick37). Thus, 25(OH)D levels were measured at a nadir in our subjects, increasing the likelihood of vitamin D deficiency.
Additional research is needed to refine the current questionnaire and provide cost-effective algorithms to identify individuals who benefit from serum 25(OH)D measurement. Ideally, a larger study performed in one season would query subjects about sun exposure, milk ingestion and intake of foods (fatty fish, liver, eggs) and supplements containing vitamin D. The study should also assess BMI and smoking, given the higher risk of deficiency reported in obese patients and smokers(Reference Jacques, Felson, Tucker, Mahnken, Wilson, Rosenberg and Rush21, Reference Bell, Epstein, Greene, Shary, Oexmann and Shaw38, Reference Isaia, Giorgino, Rini, Bevilacqua, Maugeri and Adami39). Additionally, symptoms or signs of vitamin D deficiency, such as proximal muscle weakness or tibial tenderness, might allow better identification of subjects at high risk of deficiency(Reference Van Veldhuizen, Taylor, Williamson and Drees40). Ideally, a questionnaire modified from the one herein would provide a ‘score’ with higher sensitivity and specificity for vitamin D deficiency. Such a tool would prove useful in clinical practice.
Peak bone mass occurs around 30 years of age(Reference Recker, Davies, Hinders, Heaney, Stegman and Kimmel41, Reference Lin, Lyle, Weaver, McCabe, McCabe, Johnston and Teegarden42). Models indicate that interventions to increase peak bone mass are more effective at preventing osteoporosis than interventions later in life(Reference Hernandez, Beaupre and Carter43). Early research suggests that improved vitamin D status promotes peak bone mass(Reference Valimaki, Alfthan, Lehmuskallio, Loyttyniemi, Sahi, Stenman, Suominen and Valimaki13, Reference Lehtonen-Veromaa, Mottonen, Nuotio, Irjala, Leino and Viikari14). If researchers confirm the importance of vitamin D on peak bone mass, young adults would benefit from global vitamin D fortification of food and beverages.
Acknowledgements
The authors thank the people who participated in this study. We are grateful to Dr Neil Binkley for laboratory assistance with vitamin D assays and Dr Gunnar Thomason for translation of an Icelandic journal article. We thank Andrea N. Jones for editorial assistance. K.E.H. received salary support from NIH grant 1K23 AR050995 during the conduct of this study. All authors were involved in the collection of data and its interpretation, and each author assisted meaningfully in the composition of the manuscript. The individuals cited in the acknowledgement have explicitly agreed to this designation. The authors accept the conditions for submission as explained in the instructions provided. As authors, we have no financial relationship with the sponsor of the study. Sponsors had no role in the design or conduct of the study, nor analysis of the resulting data. If the editors wish to do so, they may review the de-identified study database.
