Vitamin D: determinants of status, indications for testing and knowledge in a convenience sample of Irish adults

Abstract Vitamin D deficiency is common in Irish adults, though there is limited research on its determinants, knowledge of vitamin D or indications for testing. We aimed to explore the determinants of vitamin D status in adults and examine knowledge and reasons for testing. The study population comprised adults who had serum 25-hydroxyvitamin D tested by general practitioners request at a Dublin Hospital in 2020. Questionnaires detailing dietary intake, sun exposure, ethnicity, biophysical factors and vitamin D knowledge were sent to a sample stratified by age, sex and vitamin D status. In total, there were 383 participants, mean age 56·0 (sd 16·6) years. Wintertime deficiency disproportionally affected non-white v. white (60 % v. 24 %, P < 0·001). The greatest predictors of deficiency were low vitamin D intake (< 10 μg/d) (P < 0·001) and non-white ethnicity (P = 0·006), followed by sun avoidance (P = 0·022). It was also more prevalent in those with lower body exposure when outdoors. The majority (86 %) identified vitamin D as important for bone health. However, 40 % were tested for non-clinical indications and half were not aware of the recommended daily allowance (RDA). Low vitamin D intake was the most important determinant of deficiency, but ethnicity and sun exposure habits were also significant predictors. The majority had no clear indication for testing and were not aware of the RDA. Public health policies to improve knowledge and vitamin D intake, especially for those of non-white ethnicity and with reduced sun exposure, should be considered.

Vitamin D is primarily derived (80-90 %) from the action of UVB sunlight on dehydrocholesterol in the skin and apart from supplement use only a small proportion is obtained from dietary sources (1) . However, cutaneous synthesis is negligible between October and March in Ireland which results in a dependency on dietary vitamin D in winter months (2) . Apart from season, UVB exposure also depends on latitude, cloud cover, air pollution, sunscreen use and clothing while biophysical factors such as skin type and ageing can affect cutaneous synthesis (3,4) . However, the Irish diet is characteristically low in sources of vitamin D including cod liver oil and oily fish, with 87 % of men and 77 % of women not meeting the recommended intake (10 μg) (5,6) . Furthermore, only 10-17 % of Irish adults consume a supplement, yet this is the most consistent way of achieving adequate intake (5,7) .
To date, most research on the determinants of vitamin D status in the Irish population has focused on older adults (7)(8)(9)(10) . Overall, studies point to supplement use as the most important determinant (7,9,10) . Several have identified a characteristic seasonal variation and found positive associations with proxy measures of sun exposure (sun enjoyment, sun holiday travel, geographical UVB irradiation and sunshine hours) (4,10,11) . However, they did not specifically assess body skin exposure. Lower physical activity and frailty which may be indirectly linked to sun exposure have also been associated with lower vitamin D status (9,10) . Only four studies have investigated vitamin D status in a non-European ethnic demographic, finding lower 25-hydroxyvitamin D (25(OH)D) and a high prevalence of deficiency (< 30 nmol/l) between 57 and 88 % (12)(13)(14)(15) . Few have examined the association with dietary or specific food intakes, though fortified milk, fish and egg consumption were found to be positive determinants in older adults and adolescents (4,10,16) . Biophysical factors such as increased BMI and female sex were also associated with lower vitamin D status in children and older Irish adults (7,9,(16)(17)(18) while smoking, living alone and lower socio-economic status have been found to be negative predictors (9,10,19) .
Despite an increase in referrals for vitamin D testing in Ireland and evidence of up to a third being done inappropriately (too early, too frequently or in replete individuals), no studies have explored the indications for these 25(OH)D assessments (20)(21)(22) . Furthermore, just one study investigated knowledge regarding vitamin D, but only in pregnant women where 70 % had little awareness of dietary sources (12) . Given the lack of studies, we aimed to explore in detail the biophysical, lifestyle and dietary determinants of vitamin D status in a diverse population of adults. Furthermore, we aimed to explore for the first time in Ireland indications for vitamin D testing, as well as adult knowledge of vitamin D.

Data collection
Data were collected at St James's Hospital (SJH), Dublin, Republic of Ireland (53°Northern latitude) which receives referrals primarily from Dublin city and surrounds. A search was completed for vitamin D requests from primary care general practitioners in 2020 using the laboratory information system (iSOFT Telepath®) at SJH Biochemistry Department. A convenience sample was identified using the exclusion criteria: age < 18 years, incomplete or missing demographic data, non-community address (e.g. hospital, nursing home, convent) or location outside the Republic of Ireland.

Participant screening and selection
Of the 13 669 results collected, 1639 were excluded due to incomplete data (n 423), age < 18 (n 262), non-community address (n 239) and repeat samples (n 715) (Fig. 1). This left a sample size of 12 030 from which we randomly selected 1260 adults initially stratified by season (winter: n 630, summer: n 630). Within each season, we further stratified by age (above 50 years, n 315, below 50 years, n 315) and then by vitamin D status (< 30 nmol/l, n 100; 30-49 nmol/l, n 100; 50-124 nmol/l, n 100 nmol/l; > 125 nmol/l, n 15). In this way, we were left with a sample of 1260 adults with an equal distribution of age, vitamin D status and season of testing to which questionnaires were sent. Participants were contacted via postal address with an information sheet, consent form and questionnaire which could be completed online (via link to survey monkey) or sent back manually in hard copy form in a pre-stamped, self-addressed envelope.

Questionnaire
The questionnaire we designed included thirty-three questions detailing medical information (indications for testing and pre-existing conditions that might affect vitamin D), biophysical (ethnicity, BMI, body size (23) , socio-economic status (education level; third level or below), vitamin D intake (supplement and dietary intake) as well as dietary Ca intake using a FFQ, lifestyle (smoking, alcohol intake) and sun exposure (time spent in peak sunshine, sunscreen use, body exposure, sun-seeking habits). Information also included data on participants vitamin D knowledge (familiarity, awareness of health benefits and recommended daily allowance (RDA)). Questionnaires were sent to individuals between March and June 2022 and answers to our survey related to the period in which they had serum 25(OH) D tested. Reasons for vitamin D testing were queried, with routine health checks, patient requests and fatigue considered inappropriate.
Ethnicity was dichotomised into white and non-white (Black, Asian-Chinese, Asian-other and mixed). BMI category was determined based on self-identification using a 10 image scale of body sizes representing underweight, normal weight, overweight and obese as validated by Harris et al. (23) . We asked if participants had any of the following conditions that could affect vitamin D status (gastrointestinal conditions, e.g., Crohn's, coeliac disease, bowel/stomach surgery, inflammatory bowel disease), cystic fibrosis, liver/renal conditions, pancreatic disease and eating disorders.
Sun-seeking was categorised as no (avoid the sun) or yes (spend some time/seek the sun). Time spent outdoors was calculated based on the daily period spent outside between the hours of 13.00 and 17.00 during March to September. Body exposure was categorised as high, if more than face and hands exposed on a sunny day and otherwise as being low. Vitamin D familiarity was categorised as yes (extremely, very familiar) v. no (somewhat, not so, not at all familiar).

Vitamin D/Ca intake
Dietary vitamin D (μg) from food sources (unfortified and fortified) and Ca intake (mg) was calculated using a FFQ adapted from The Irish LongituDinal Study on Ageing questionnaire (9) . For each food consumed, an average vitamin D/Ca content per portion was estimated using food manufacturer's information and Nutritics software version 5.78 (online Supplementary  Table 1). Where an approximate size of a food portion was not specified in the FFQ, an average portion size was assumed (e.g. yogurt = 125 g pot). In order to estimate daily dietary Ca and vitamin D intake, we initially calculated total weekly intake as follows: once per week (1 × unit food), 2-4 times per week (3 × unit food), 5-6 times per week (5·5 × unit food), once per day (7 × unit food), 2-3 times per day (2·5 × unit × 7) and 4-5 times per day (4·5 × unit × 7). The weekly total was then divided by 7 to give the total daily intake for vitamin D and Ca. We also dichotomised daily vitamin D intake from unfortified or fortified sources. The daily vitamin D intake from supplements (cod-liver oil, vitamin D only supplement, multivitamin containing vitamin D) was also calculated. Total daily vitamin D intake was then estimated by combining supplemental and dietary intake, and those who met the RDA were identified (10 μg/d as per advised by the Food Safety Authority of Ireland (FSAI) at the time of vitamin D sampling). We also identified those who exceeded the tolerable upper intake level for

Statistics
Data were checked for normality by the Kolmogorov-Smirnov test. Geometric mean with standard deviation was reported in tables. Median and interquartile range were used to report dietary intakes. Categorical variables were tested using Chi-squared, with independent sample t-tests, Mann-Whitney and Kruskal-Wallis test for continuous variables. Independent factors associated with vitamin D deficiency (< 30 nmol/l) were explored in binary logistic regression models using the following variables and reference categories: age (≥ 50 years), sex (male), BMI category (normal weight), season of sampling (summer), ethnicity (white), smoking (non-smoker), alcohol (alcohol consumer), sun habits (sun seeker), education (third level) and adherence to vitamin D RDA (yes/no). Body exposure and hours spent in peak sunshine were co-correlated with sun-seeking behaviour and were therefore not included in the model. Statistical analysis was carried out using SPSS (Version 26, IBM Corp). Statistical significance was accepted when P < 0·05.

Demographics
Questionnaires were completed by 383 (32 %) of the contacted participants. In fifty-seven cases (4·5 %), they were not received by the participant due to a change of address, and in fifteen (1·1 %), they were not completed due to death or illness ( Fig. 1). Characteristics of the sample are shown in Table 1.
The average age was 56·0 (SD 16·6) years, 60 % were female and 90 % were of white ethnicity. Two-thirds (67 %) had third level education and one-fifth (21 %) identified as having a condition that could predispose to lower vitamin D. The majority of the population were overweight or obese (58 %), with 36 % normal and 7 % underweight. Most participants were sun seekers (74 %), had a high UV body exposure (81 %) and used sunscreen (71 %) and were not familiar with vitamin D (70 %). About 50 % (192/383) were taking a vitamin D supplement with precise data on vitamin D content and dosing frequency available in 79 % (151/191). For this reason, the sample size (n 338) on which there was estimation of total vitamin D intake and analysis of RDA was smaller (Table 3). There was a near equal split between  (23) on body size categorised as underweight, normal weight, overweight and obese. Sun-seeking was categorised as no (avoid the sun) or yes (spend some time/seek the sun). Body exposure was categorised as low if only face and hands or higher if additional body parts exposed on a sunny day. Time spent outdoors calculated based on the daily period spent outside between the hours of 13.00 and 17.00 during March to September. Vitamin D familiarity was defined as yes (extremely, very) v. no (somewhat, not so, not at all familiar). *P < 0·05.  Tables 2 and 3.

Dietary intakes
The overall contribution of diet to vitamin D intake was low with half of all participants consuming less than 4·5 μg (180 μg) per day. There was a trend for better vitamin D status with higher levels of vitamin D intake from either unfortified or fortified sources (Table 4). However, total dietary vitamin D intake (combining unfortified and fortified foods) was significantly lower in those who were deficient v. sufficient (4·0 v. 5·2 μg/d, P = 0·044). We also identified that those who were over 50 had higher dietary intakes (median 5·4 v. 3·7 μg/d, P < 0·001) and were more likely to consume oily fish on a weekly basis (60 % v. 30 %, P < 0·001). However, there was no difference in dietary intake by sex. We also found that the median dietary Ca intake was 658 mg/d and was significantly different by vitamin D status (P = 0·004): lowest in those with deficiency (527 mg/d) and highest with sufficiency (768 mg/d).

Vitamin D excess
Serum 25(OH)D levels ≥ 125 nmol/l were identified in nineteen respondents and were more likely in those aged <50 years (P = 0·020) and in supplement users (P = 0·001). The median total vitamin D intake in those with a level ≥ 125 nmol/l was 27·5 μg (1100 μg)/d) with the highest intake of 145 μg/d (5800 μg) in a patient with a serum concentration of 131 nmol/l. Overall, 1·5 % (n 5) had an intake above the tolerable upper intake level of 100 μg (4000 μg) per day (2) and the highest 25(OH)HD level identified in this group was 193 nmol/l.

Vitamin D knowledge and testing indications
The primary reason, in more than a third (34 %) of patients for testing, was for a routine health check. Appropriate reasons for testing included unexplained aches and pains (21 %), brittle bones (10 %) and limited sun exposure (9 %), though 19 % reported 'other' which included requests due to patient request (n 13), fatigue (n 7) and immunity/COVID (n 6) (online Supplementary Fig. 2). There was a lack of awareness of current vitamin D guidelines, with nearly half (46 %) not knowing, onethird (32 %) believing the RDA was more than 20 μg (1000 μg)/d and just 12 % correctly identifying 10-15 μg (400-600 μg)/d (online Supplementary Fig. 3). The vast majority (86 %) of respondents cited vitamin D as being important for bone health with 66 % citing immunity/COVID, 47 % heart health and 40 % mental health (online Supplementary Table 4). There was no difference in vitamin D status in those who were familiar v. not familiar with vitamin D (61·9 v. 55·5 nmol/l, P = 0·097). Vitamin D familiarity was predicted by education in binary logistic regression, with no effect found for age, sex, season or ethnicity. A total of 40 % (n 152) of referrals were inappropriate, including for routine health checks (n 132), patient request (n 13) and fatigue (n 7).

Discussion
This is the first study to investigate in detail the determinants of vitamin D status in Irish adults and to explore indications for testing as well as knowledge of vitamin D's role in health and its RDA. The strongest predicators for deficiency were low vitamin D intake (< 10 μg/d) and non-white ethnicity, and it was also twice as likely in sun avoiders. The contribution of dietary sources to overall intake was small, but it was still positively associated with better vitamin D status. However, the vast majority who met the RDA were taking supplements. More than a third had vitamin D testing for inappropriate reasons and less than 12 % could correctly identify the recommended dietary intake.

Vitamin D intake
The overall contribution of diet to vitamin D intake was low with half of all participants consuming less than 4·5 μg (180 μg) per day. The median intake due to supplements was 10·0 μg (400 μg) per day, and those taking supplements were about three *Supplement intake dose (total cod-liver oil, vitamin D and multivitamin containing vitamin D) available for n 151. P-value determined by Kruskal-Wallis test, significant at P < 0·05. Values reported as median intake (interquartile range) in micrograms for vitamin D and milligrams for Ca.
times less likely to be deficient in summer. The mean difference in serum 25(OH)D in users v. non-users of supplements was 21·7 nmol/l, which is similar to that found previously in older Irish adults and pregnant women (10,14,27) . Older adults had both higher dietary and total vitamin D intakes. These findings are in keeping with other dietary surveys in Ireland that found intakes between 3·0 and 6·9 μg/d, though being lower in younger (18-35 years) v. older adults (> 65 years) (6,28) . We found similar rates of supplement use by age in this study which contrasts to findings elsewhere (2,5) . However, oily fish consumption was more frequent in those > 50 in our survey which may partly explain their higher intake. Nearly half (43 %) of adults did not meet the RDA for vitamin D while in those taking supplements this was lower at 19 %. However, some supplements, especially those over the counter, contain relatively small amounts of vitamin D and/or Ca. Importantly, those achieving the vitamin D RDA were 72 % less likely to be deficient though this still occurred in 12 % of our population. Previous meta-analysis studies estimated that 12-13 μg/d per day is required for the general population living ≥40°N to maintain wintertime vitamin D status ≥ 30 nmol/l (29,30) . However, previous dietary surveys in Ireland have found that just 10 % of adults meet the 10 μg/d level, indicating that fortification may be required to achieve adequate vitamin D intakes in the population (5,29) . In addition, 10 % of our survey participants were of non-white ethnicity, for whom studies suggest higher vitamin D intakes to optimise status (31) . Furthermore, the RDA (10 μg/d) on which we based our analysis was the recommendation at the time participants had their serum 25(OH)D tested. However, the FSAI more recently advised on a higher daily intake (15 μg/d) for older adults (aged >65) which constitute 32 % of our sample (32) . We found that 1·5 % of participants exceeded the tolerable upper intake level of 100 μg (4000 μg) per day, but the highest 25(OH)D level identified was below that which predisposes to acute vitamin D toxicity.

Ethnicity
Non-white ethnicity was associated with a very high prevalence of winter deficiency of 60 % v. only 24 % in white participants. Furthermore, 80 % of non-white ethnicity had levels <50 nmol/l in wintertime. The proportion in our survey who were non-white is also similar to that found in a recent census of the Dublin urban area (33) . There is very limited research on vitamin D status in ethnic populations in Ireland with only four studies published (12)(13)(14)(15) . In South-East Asian adults (n 186) living in Dublin, 67 % had 25(OH)D < 30 nmol/l (13) . A high prevalence of deficiency (<30 nmol/l) was also identified in eighty-one pregnant women of Middle Eastern and African (88 %), Sub-Saharan (68 %) and Asian origin (59 %) v. Thirty-one indigenous Irish (36 %) living in Ireland (12) . A larger study of pregnant woman in Ireland found that those of non-white ethnicity had a mean 25(OH)D that was 19·3 nmol/l lower (14) . African ethnicity was also a significant determinant of vitamin D status in a small sample (n 7) of Irish children (15) . We found no difference in vitamin D intake, supplement use, education or body exposure between white and non-white participants suggesting that ethnic difference in skin pigmentation is having a dramatic effect on vitamin D status. However, we did not look at sun holiday travel which could explain some of the variation and has been associated with better vitamin D status in older Irish adults (10,34) . Similar to our study, non-white ethnicity has been found to predict lower rates of deficiency in England (34) and better vitamin D status in European populations at a similar latitude (35) .
In Ireland, overall, about 5 % of the population are non-white and this demographic has increased in recent years (36) . Routine vitamin D supplementation for this section of the population is advisable as it has been found to be more effective than sunlight exposure for treating deficiency (37) and is currently recommended by the European Calcified Tissue Society (38) . Importantly, the vitamin D requirements for non-whites have been estimated to be much higher than the standard RDA advised in Ireland and by most international agencies. For example, maintaining a winter serum 25(OH)D ≥ 30 nmol/l in 97·5 % of individuals who are of South Asian and Black ethnicity would require an estimated respective daily vitamin D intake of 27·3 μg (1092 μg) and 33·2 μg (1328 μg) (31) . Public health information promoting dietary and supplement advice targeting this ethnic population in Ireland may be needed to address this deficiency.

Sun exposure
We found those who avoided sun exposure were up to twice as likely to be deficient while conversely greater body exposure when outside was associated with higher 25(OH)D concentrations and less deficiency in summer. This is in keeping with other Irish research which found that sun enjoyment was predictive of vitamin D status in older adults (4,10) and in patients with lupus (39) . Sun-seeking behaviours have also been identified as influencing vitamin D status in Irish and European women and children (40,41) . Our study indicates that summertime deficiency was halved in those with high v. low body exposure. Body exposure (days with sun exposed upper body) has been positively correlated with 25(OH)D at a similar latitude (42) . While there are concerns about skin cancer risk, moderate sun exposure has been shown to make up for deficiency in those who consume relatively low vitamin D (43) . Furthermore, for white-skinned people in the UK and similar latitudes, spending 9 min outdoors at lunchtime from March to September was estimated to be sufficient to maintain 25 nmol/l throughout winter (44) . Consistent with this, we found no difference in vitamin D status in those who spent more than 30 min in peak sunshine in the same period. We also identified that sunscreen users had better vitamin D status which can be considered a proxy for sun exposure with similar findings also reported in the Irish population (15,45,46) and at similar European latitudes (47,48) . While our study only explored vitamin D status in Dublin, other Irish studies have detected variations in deficiency by geographical location (7) that could be explained by differences in UVB availability due to latitude (4) .

Vitamin D knowledge and indications for testing
Despite a surge for vitamin D testing and increasing costs, there remains little evidence on the indications for assessing 25(OH)D status. In a recent Irish study, a high proportion (a third) of vitamin D retests were found to be inappropriate, resulting in Vitamin D: determinants of status in Ireland considerable unnecessary expenditure; however, no information was available on testing indications (21) . In this study, routine health checks accounted for a third of the reasons for testing, though this is not recommended and is considered inappropriate (49) . Additionally, 19 % reported other reasons including fatigue which are also not recognised as a valid clinical indication. Our results are similar to the UK and the Netherlands where 70-77 % of testing was considered inappropriate (50,51) . Patient reassurance has also been found to be a key driver of testing by general practitioners which is consistent with our finding that 'patient requests' were the most frequently declared other reason for testing (51) .
We found that half (46 %) had no knowledge of any RDA recommendations, though a third (32 %) felt it was higher (≥ 20 μg/d) and 4 % lower (≤ 5 μg/d). Better vitamin D knowledge has been associated with increased likelihood of taking supplements (52) , though supplement use has been found to be relatively low (10-17 %) in Ireland (5,7) suggesting a low level of concern for deficiency. However, during the COVID pandemic there is some evidence to suggest increased supplement use in Irish adults and possible improvement in vitamin D status (53) . Indeed, a publicised report by Irish researchers in April 2020 recommended a higher daily vitamin D intake of 20-25 μg (800-1000 μg) during COVID for adults aged > 70 (54) so some knowledge of higher RDA's than advised by the FSAI might be expected. The majority (86 %) of respondents cited vitamin D as being important for bone health, similar to other studies (52,55) . Perhaps surprisingly, the second most common health association (66 %) was for immunity/COVID. This likely reflects media coverage during the pandemic of research on vitamin D's possible beneficial effects on COVID infection (56) . Indeed, trend analysis indicates there was a peak in Google searches for vitamin D coinciding with the first COVID wave in Ireland (March 2020) and during a subsequent wave (January 2021) (57) . The only other research was based on a small sample (n 112) of pregnant women attending a maternity hospital and found that 71 % had insufficient knowledge, with just 10 % recognising supplements as a source (12) . While there was good awareness of the benefits for bone and immune health, there is poor knowledge of the vitamin D RDA and little understanding of the indications for testing. This suggests that better awareness may help to improve vitamin D intake and status.

Strengths and limitations
This is the first study of its kind to explore multiple determinants of serum vitamin D in Irish adults including dietary intake, ethnicity and measures of sun exposure. It also adds to the limited research on adult knowledge and perceptions of vitamin D in Ireland and is the first to investigate indications for testing. However, as the study participants were selected from a sample of patients who had their vitamin D tested by their general practitioner, it may not be representative of the wider population. In particular, there may have been information bias as participants may have been aware of their vitamin D test results. Additionally, there may be also exclusion bias given that a significant proportion of adults did not return our questionnaire, though our response rate is in keeping with other studies using a similar methodology (58) . Finally, there may be recall bias as regards the recollection of food and supplement intakes when completing the FFQ.

Conclusion
We found, in a convenience sample of Irish adults, the biggest predictors of deficiency were low vitamin D intake (< 10 μg/d) (P < 0·001) and non-white ethnicity (P = 0·006), while it was twice as likely in those who were sun avoiders (P = 0·022). In particular, deficiency in winter was twice as likely in those who of non-white ethnicity and was also more prevalent in those with lower body exposure when outside. Dietary sources of intake were small but still associated with better vitamin D status. However, the vast majority (81 %) who met the RDA were taking supplements. More than a third of vitamin D testing was for non-clinical indications, and the majority were not aware of the current RDA. Public health policy should be considered to improve vitamin D intake, especially in those of non-white ethnicity and with reduced sun exposure.