Seasonal affective disorder (SAD)[1] manifests as a clinical syndrome with symptoms of depression, carbohydrate craving, hypersomnia, lethargy and changes in circadian rhythms. The symptoms recur each winter and disappear again during spring or early summer. Women are more affected than men, and the prevalence at high latitudes ranges from 1 to 10%[2]. It has been suggested that the seasonality and symptoms of SAD may be due to low levels of serum 25-hydroxyvitamin D levels (25(OH)D)[3]. In the brain 1 alpha-hydroxylase converts 25-hydroxyvitamin D to the active vitamin D (1,25(OH)D). Further vitamin D receptors (VDR) are widespread in the human brain. Thus, it is proposed that vitamin D operates as a neurosteroid and thought to be involved in the pathogenesis of depression[3, 4].

Ultraviolet B shortwave (UVB) irradiation of the skin is the most important source of vitamin D. Irradiation on a summer day between 20–30 minutes on the face and arms results in the synthesis of 100–250 μg of vitamin D, which is subsequently activated in the liver to 25-hydroxyvitamin D. Solar UVB content at high latitudes (above 37 N) is minimal between October and March, and consequently, the synthesis and stores of vitamin D during winter and early spring are low[5]. In Denmark, suboptimal 25(OH)D status is common; thus, the prevalence of vitamin D deficiency among blood donors is 18%, whereas 47% have vitamin D insufficiency[6]. Normal 25(OH)D vary between 50–160 nmol/L, whereas values between 25–50 nmol/L are defined as vitamin D insufficiency, and deficiency is defined as less than 25 nmol/L. Additionally, 25(OH)D are lower in persons with darker skin (types III–VI) as assessed by the Fitzpatrick classification[7, 8].

Vitamin D supplementation in relation to well-being during the winter season has been studied in three small studies. A positive effect on well-being and symptoms of depression was demonstrated, when high doses of vitamin D (≥100 μg D3 daily) were given for 1 to 3 months[9–11]. One large study (n = 441) demonstrated similar significant improvement in Beck Depression Inventory (BDI) scores in treatment groups receiving 70 μg and 140 μg compared to placebo during a 1 year period[12]. Other large studies have failed to demonstrate effects on symptoms of depression when low dose vitamin D (≤20 μg daily) was given[13, 14].

Indoor work has been hypothesized as a risk factor for mood difficulties among civil servants in a Danish cross-sectional study (latitude 55-57N°), a significant risk reduction (odds ratio = 0.63) being found in outdoor workers (>2 hours/day)[15]. Thus, indoor work might be a significant risk factor for SAD, since indoor work is by far the most prevalent form of work among the total work force at temperate latitudes[16].

The most efficient and well-documented therapy for SAD is light therapy (LT)[17]. However, the neurobiological impact and mechanism of LT is not fully understood. SAD was at first believed to be related to abnormal melatonin metabolism, but later findings did not support this hypothesis. Thus, LT during two weeks can reduce subjective sleepiness, but the effect is not associated with a rise in levels of melatonin in saliva or 25(OH)D[18]. However, LT can be challenging to apply, and compliance might be reduced due to lack of time and social support[19, 20]. Tablet treatment can therefore be attractive for some SAD patients. Although studies of brain serotonin function support the hypothesis of disturbed activity[20], there is still too little evidence to draw any overall conclusion regarding the use of second generation antidepressants for SAD[21]. Bases on estimates of high prevalence of vitamin D insufficiency in the general Danish population and increased mood difficulties among Danish indoor compared to outdoor working personnel: vitamin D supplementation might be a solution. Thus it could be anticipated that a majority of the participants with known symptoms of SAD would develop vitamin D insufficiency during wintertime and thereby benefit from vitamin D supplementation, which in the above-mentioned winter season studies has shown positive effect on well-being.

The purpose of this study was to examine whether vitamin D supplementation would reduce SAD symptoms among indoor workers who had experienced winter depressive symptoms in the past. The secondary objective was to examine correlations between 25(OH)D, SAD and mental well-being.

The objective of this study was to investigate whether employees with indoor work and SAD symptoms would benefit from vitamin D supplementation during the winter season, i.e. whether vitamin D would alleviate symptoms of depression during the winter and early spring. The study is qualified due to the block randomisation, which is crucial when including participants during the wintertime. Additionally, the double-blinded randomised design is the best suited to demonstrate an effect of the intervention. Finally, the trial included both psychometric assessments as well as levels of the active treatment. However the design did not favour inclusion of participants with lower 25(OH)D, thus the study did not allow an investigation of the ability of vitamin D to improve mood in those with low 25(OH)D at baseline. Thus, the overall most significant finding in this study was that vitamin D had no effect on SAD symptoms.

The findings are different from those achieved in previous small studies[9, 10] and a larger Norwegian study[12]. In the small studies, the inclusion criterion was 25(OH)D below 40 nmol/L[9, 10], and in the Norwegian study[12], mean baseline 25(OH)D was 52.5 nmol/L. This indicates that vitamin D supplementation might only be relevant for those with lower 25(OH)D. We chose a SPAQ-SAD cut-off at 8 in order to include more participants with SAD symptoms, thus we included participants with less severe symptoms than recommended by Kasper et al.[22]. The lower SPAQ-SAD cut-off might have influenced the results negatively due to a reduced effect on participants with limited symptoms according to the low sensitivity in SPAQ-SAD[28]. The magnitude of this effect is unknown since we did not have scores from the summer months.

Both weight and waist-circumference among the participants were measured since seasonal affective disorder is related to an increased appetite, which again results in an increased risk of the metabolic syndrome. We did not find any improvement in the blood pressure or a positive effect on weight. These findings are in accordance with two other trials which showed no effect of vitamin D supplementation on weight loss among obese patients[29].

There were five main limitations of the study[1]. High PTH was an exclusion criterion for participation. High PTH is correlated with low 25(OH)D, and seven participants were excluded for this reason. This limitation was defined by the legislative authorities because of the risk of placebo treatment of secondary hyperparathyroidism[2]. The drop-outs had a significantly lower 25(OH)D, which reduced the possibility to demonstrate an effect of the intervention[3]. We did not reach the calculated sample size. Thus, the results from the study might have been subject to a type II error. The study was grossly underpowered in a way that detection of clinically significantly results was unlikely. We had expected more participants to enter the study; the most prominent factor for the women was the criterion that anti-conception be practised, which many could not meet. The smaller number of participants, especially among young women, might weaken the generalizability of the study. This apart, we consider these data as being potentially relevant for other staff categories with indoor work[4]. The study was carried out using posters and direct e-mails to the staff at the two hospitals. Thus, we do not know whether the participants were representative of people with SAD symptoms[5]. Finally, we included participants according to their scores on a single question in the SPAQ-SAD rather than the full definition of SAD, but all participants except four had abnormal SIGH-SADs at baseline. Considering this limitation, we did not find a difference between the vitamin D and placebo group but a significant decline of SIGH-SAD in both groups, which could be due to placebo respond. We could not explain it by the duration of daylight since time from solstice did not have an impact on SIGH-SAD.

The participants were respondents of announcements: neither we do not know the prevalence of winter depression among health care professionals, nor we do not know whether the participants had more or fewer symptoms than those who did not respond, reducing the external validity of the study. Additionally the trial might only find relevance for those with milder symptoms and declined by those with more severe symptoms.

The decision to include participants with 25(OH)D between 10–160 nmol/L was taken not only to study the relation between low 25(OH)D status and symptoms of depression but also to test whether a decrease in 25(OH)D during the winter season could be a risk factor for depressive symptoms in those who had previously experienced these symptoms. However, we did not find the anticipated prevalent low 25(OH)D during the wintertime in this study, where those participants thereby could benefit from vitamin D supplementation, which could be explained by the high mean status at baseline.

The study had a clinical perspective focusing on participants who had mild to more severe symptoms of SAD. Unfortunately, we did not add a healthy control group, which could have had a more public health perspective.

The participants were allowed to use LT, and in the written information they were informed that LT has shown a positive effect on SAD. However, only 4 participants used LT, which might be the explanation of no confounder effect on any outcome.

The study does not exclude a positive effect of vitamin D supplementation on SAD, especially in patients with low 25(OH)D. Further investigation of SAD and vitamin D supplementation must therefore include participants with low 25(OH)D. Moreover, studies must be concluded during the early winter months in order to ensure no influence of the increased sunlight after the winter solstice.

There were no significant between-group differences in the primary (SIGH-SAD) and secondary (WH0-5) as well as the exploratory outcome measures (weight, waist circumference, blood pressure, absenteeism from work and 25(OH)D). Thus, the study failed to demonstrate an effect of vitamin D on SAD symptoms. These finding may be limited by the fact that the study was underpowered, and the inclusion did not favour participants with low 25(OH)D at baseline.