Rotating shift workers with vitamin D deficiency have a higher risk of obstructive sleep apnea

The study aimed to evaluate the association between obstructive sleep apnea (OSA) and vitamin D deficiency (VDD) in shift workers. This cross-sectional study included male rotating shift workers in an iron ore extraction company. Participants were classified as VDD when 25(OH)D < 20 ng/mL for a healthy population and 25(OH)D < 30 ng/mL for groups at risk for VDD. Risk of developing OSA was classified by Berlin questionnaire (BQ) and NoSAS score. Data were compared using chi-square analysis with Cramer’s V as effect size, and Bonferroni correction. Multivariate logistic regression analysis was performed to investigate whether or not VDD was associated with OSA risk assessment. Among 1423 male workers, mostly younger, aged 30 to 39 years (53%), worked shifts for more than 5 years (76%). The prevalence of high risk of OSA by BQ was 16%, and 33% by NoSAS score. Additionally, 29% had VDD. In multivariate analysis, controlled for confounding factors, workers with VDD had a 52% increased chance of OSA by BQ (OR 1.52; CI95% 1.06–2.18) and a 64% increased chance of OSA by NoSAS score (OR 1.64; CI95% 1.09–2.48). After subgroup analyses, similar results were not observed in workers aged 20–29 and 30–39 years. Rotating shift workers with vitamin D deficiency are more likely to have obstructive sleep apnea, assessed by the Berlin questionnaire and NoSAS score.


Introduction
Obstructive sleep apnea (OSA) is a common sleep problem in adults of all age groups [1]. It is related to a variety of health-related consequences, including increased risk of cognitive impairment, depression, metabolic disorders, hypertension, cardiovascular disease, atrial fibrillation, and mortality [2,3]. Population studies show that the prevalence of OSA in the general population reaches more than 30% of the population and almost 50% in men [1,4]. The estimated global prevalence of obstructive sleep apnea is that about 900 million adults have some degree of OSA [5].
As a result of airway collapse, hypoxia, and sleep fragmentation, individuals with OSA have increased activation of the sympathetic nervous system during sleep, which leads to daytime sleepiness, increased heart rate, arrhythmias, awakenings, and other clinical manifestations [6].

3
The prevalence of OSA can be as high as 90% in some groups [7], can vary according to gender, age, and associated comorbidities [8]. In addition, previous studies have hypothesized that OSA might have a higher incidence and severity in winter [8]. From this, it was considered that there might be some relationship between OSA and vitamin D deficiency (VDD), which is more prevalent in periods of low sunlight [9]. This hypothesis has been supported by recent studies showing that decreased vitamin D levels and sleep disturbances are associated in different populations [10][11][12][13]. The possible mediating or causal effects of vitamin D deficiency in sleep apnea are related to the accumulation of adipose tissue and muscle decline caused by VDD, increasing the chances of airway obstruction during sleep [13,14]. And it may also be related to higher pro-inflammatory states and is commonly associated with OSA [15]. And finally, studies show that VDD can increase circulating prostaglandin D2, increasing sleep complaints and problems [16,17]. All of these elements are reported to be involved in the pathophysiology of OSA.
These hypotheses were corroborated by other studies in the literature that demonstrate decreasing in vitamin D with increasing severity in OSA [18][19][20][21][22]. Furthermore, Liguori et al. (2015) demonstrated that the commonly used therapy for OSA, continuous positive airway pressure (CPAP) therapy, can increase vitamin D levels in patients with OSA [14].
Epidemiological studies have shown that certain groups of individuals or populations are at the same time more susceptible to both sleep disorders and VDD, for example, individuals with obesity [23], afro-descendants [24], and also certain occupations and work shifts [25][26][27]. The alternating shift workers are a group at risk for both situations. Vitamin D deficiency is identified in up to 80% of alternating shift workers [25], and alterations in sleep architecture are common in these workers since this work schedule leads to changes in circadian rhythm, daily routine, and hormonal changes [28].
However, to our knowledge, there is no study evaluating how vitamin D deficiency is associated with sleep apnea in shift workers. Therefore, this study aimed to verify the association of vitamin D deficiency with OSA evaluated by the Berlin questionnaire and NoSAS score in alternating shift workers.

Design and participants
The current study was conducted with a population of shift workers from an iron ore extraction company in Brazil, at Minas Gerais and Pará. Three cross-sectional studies were conducted, and the population of shift workers with the position of operators was invited to participate: (a) The first was carried out in 2012, with 337 shifts workers from four mines in the Iron Quadrangle region; (b) the second study was carried out in 2015 with 192 shifts workers from another mine in the Iron Quadrangle region; (c) the third study was carried out in 2018 with 932 shifts workers in the southern region of Pará [29]. Therefore, in total, 1461 shift workers were evaluated, Of these only 38 were women (3%). For this study, we considered only men (n = 1423). The Strengthening Reporting of Observational Studies in Epidemiology (STROBE) guidelines were used in the design and construction of the study.
The workers had two types of alternating shifts, as follows: (a) In Minas Gerais, the workers had rotating shifts of four days worked and one day off, in this way they worked for six hours, followed by 12 h rest. After finishing the weekly four-shift cycle, they received one day off. (b) In Pará, the workers had a rotating shift of 5 days worked and 2 days off, so they worked for eight hours followed by 24 h rest. After finishing the five-shift weekly cycle, they received 2 days off.

Data collect
In all studies, data collection was performed face-to-face by teams trained to administer the questionnaires, measure anthropometric data, and collect biological samples.
Sociodemographic variables (sex, age, skin color, education, and shift schedule) were assessed. Age was categorized into age groups (20-29; 30-39; 40-49; 50-59, and 60 years or older); skin color was self-reported and grouped as non-white (black, brown, yellow, or indigenous) and white; education was categorized according to the degree of schooling (1st-degree completed, 2nd-degree completed, technical, graduate, and postgraduate); and shift work time was grouped into ranges (< 5 years and ≥ 5 years).
The clinical evaluation was carried out employing a questionnaire about pre-existing diseases, use of medication, smoking, alcohol consumption, physical activity; and measuring blood pressure. The Alcohol Use Disorders Identification Test (AUDIT) was used to evaluate alcohol use dependence [30]. Individuals who reported that they had never smoked, or quit smoking more than 6 months ago, were classified as nonsmokers, and individuals who reported that they currently smoke or quit smoking less than six months ago were classified as smokers. Furthermore, the Fagerstrom Test for Nicotine Dependence (FTND) was used to evaluate nicotine dependence; 6 questions are used to produce a score from 0 to 10 points [31]. For analysis purposes, participants with no risk and low risk were grouped into the same category (AUDIT 0-7 points; FTND: 0-5 points). The International Physical Activity Questionnaire (IPAQ) was used to assess the level of physical activity of the study individuals. The workers were classified as low physical activity < 600 measure total energy (MET)-min/week [32].
The pre-existing diseases evaluated were: cardiovascular diseases, respiratory diseases, and chronic kidney disease. To evaluate hypertension, the workers' blood pressure was measured with a semi-automatic digital device in triplicate. The measurement protocol followed the recommendations of the Brazilian Society of Cardiology, and workers who had systolic blood pressure (SBP) ≥ 140 mmHg or diastolic blood pressure (DBP) ≥ 90 mmHg were classified as having hypertension [33].

Biochemical data
The evaluation of the biochemical profile was performed by analysis of the lipid profile, vitamin D, and glycemic. In the first two studies (2012 and 2015), blood samples were collected after a 10-h fast and in 2018 it was collected without a previous fast.
The lipid profile of the workers was determined by the enzymatic colorimetric method. Low-density lipoprotein cholesterol (LDL) was calculated using the Friedewald Eq. (1972), by the formula: total cholesterol-([HDL + (triglycerides/5)]. The total cholesterol (TC) was classified as normal < 190 mg/ dL, HDL > 40 mg/dL, LDL < 130 mg/dL, and triglycerides (TG) < 150 mg/dL with fasting and < 175 mg/dL without fasting [34]. Dyslipidemia is classified when at least one of the parameters was altered or uses lipid-lowering drugs. Glycemic index was evaluated by fasting plasma glucose (FPG) in Minas Gerais, and glycosylated hemoglobin (HbA1c) in Pará. Workers were classified as hyperglycemic with FPG values ≥ 100 mg/dL or HbA1c ≥ 5.7% [35]. Vitamin D was determined by the chemiluminescence method and classified as deficiency 25(OH)D < 20 ng/mL to a healthy population and 25(OH)D <≤ 30 ng/mL for groups at risk for VDD (body mass index ≥ 30 kg/m 2 , age ≥ 60 years, and presence of chronic kidney diseases) [36].

Anthropometric data
To evaluate body adiposity, we used a general adiposity indicator, the body mass index (BMI), and two indices of peripheral adiposity, waist circumference (WC), and neck circumference (NC). Height was measured using a stadiometer with a scale in centimeters and accuracy of 1 cm. Weight was measured on a portable body composition monitor. BMI was calculated and classified according to the World Health Organization (WHO) as eutrophic (BMI 18.5-24.9 kg/ m 2 ), overweight (BMI 25.0-29.9 kg/m 2 ), and obesity (BMI ≥ 30.0 kg/m 2 ) [37]. WC was measured, in triplicate, with a simple and inelastic measuring tape at the midpoint between the iliac crest and the last costal arch, and classified as central obesity values ≥ 90 cm [38]. NC was measured at the level of the cartilage, just above the laryngeal prominence, and classified as increased values ≥ 40 cm [39].

Berlin questionnaire
The Berlin questionnaire adapted and validated for Brazil [40] was applied to all workers. The BQ has three categories: witnessed snoring and apnea, excessive daytime sleepiness, and OSA-related comorbidities. The cut-off point for highrisk classification for OSA was two positive [40].

Statistical analysis
The Stata program (version 15.0) was used to assist the statistical analyses, considering an alpha of 5%. Pearson's chisquare with Bonferroni correction was performed to verify the difference between the groups. Cramer's V was used for Chi-squared analyses as an effect size, with the thresholds 0.10 (small), 0.30 (medium), and 0.50 (large).
To investigate whether VDD was associated with risk for OSA, we performed multivariate logistic regression from a hierarchical model of determination. This type of analysis considers the association of each variable with the outcome, controlling for possible confounder effects between proximal and distal variables. We developed our conceptual model to integrate sociodemographic, clinical, and anthropometric factors to explain their relations of OSA with VDD. Adjusted model 1 included seasonality, age, years of shift work, scholarly, skin color, and geographic location; model 2 adjusted by model 1 in addition to hyperglycemia, dyslipidemia, and hypertension; and model 3 was adjusted by the variables from model 2 in addition to adiposity variables, body mass index, waist circumference, and neck circumference. Collinearity among the covariates was evaluated. Hosmer-Lemeshow test and Akaike information criterion (AIC) were used to assess the goodness-of-fit of the models.
Using the program G*Power (version 3.1.9.2), we evaluated the a posteriori sampling power. We considered an alpha level of 0.05 (using a two-tailed test) and estimated a sample power of 0.98.

Results
The study included 1423 male rotating shift workers from two different mining regions in Brazil, and 29% had VDD, with a median 25(OH)D of 26.3 ng/mL, a minimum of 5.0 ng/mL, and a maximum of 65.7 ng/mL. Most workers were younger, aged 30 to 39 years (53%), with a minimum of 20 and maximum of 65 years, 73% self-reported as non-white, had up to 2nd-degree complete (72%), and worked shifts for more than 5 years (76%). Regarding anthropometric, clinical, and behavioral variables, most workers had high WC (63%), 39% had high NC, 78% had dyslipidemia, almost half had a low level of physical activity (49%), and 2% had medium and high-risk nicotine consumption by Fagerstrom test, and 10% had medium and high-risk alcohol consumption by AUDIT ( Table 1).
The elevated risk for OSA among alternating shift workers was 16 to 33%, considering the BQ and NoSAS score, respectively. Besides, workers with VDD had a higher prevalence of OSA (p < 0.001) ( Table 1).
Workers with VDD had a higher prevalence of BQ categories 1 (snoring) and 3 (comorbidities) when compared to workers without VDD (p < 0.001). Category 2 (excessive daytime sleepiness) had the lowest percentage of positive scores (5%). For the NoSAS score, workers with VDD had high positive criteria for overweight, obesity, high NC, and the presence of snoring (p < 0.001) ( Table 2).
When we evaluated the association of VDD with OSA in multivariate analysis, controlled for all confounding factors (model 3), workers with VDD were 52% more likely to be at high risk for OSA, measured by BQ (OR 1.52; CI95% 1.06-2.18) and 64% increased chance of high risk for OSA from NoSAS score (OR 1.64; CI95% 1.09-2.48) ( Table 3).
We also evaluated the association of VDD with risk for OSA stratified by age groups. In workers 40 and older, those with VDD had a 69% and 109% increased chance of high risk for OSA, on BQ and NoSAS, respectively (Table 4).

Discussion
This is the first study of male rotating shift workers in an iron ore mining company to examine the association of VDD and OSA risk. Our findings showed that shift workers who were vitamin D deficient were at higher risk of obstructive sleep apnea, as evaluated by the Berlin questionnaire and NoSAS score.
We found that 29% of the workers evaluated had VDD, and was more prevalent in workers older than 40 years, with longer working time in rotating shifts, high WC and NC, with chronic diseases, and in workers at high risk for OSA. The shift workers evaluated had vitamin D deficiency similar to that found in the general population of Brazil, which was 33% [41]. However, below the prevalence found in a systematic review study by Sowah et al. (2017), in which up to 80% of rotating shift workers have VDD [25]. Furthermore, Coppeta et al. (2018) in their systematic review found that of the occupations evaluated, shift workers and indoor workers are at high risk of having vitamin D deficiency [27]. Among the hypotheses of this risk factor, studies show that internal and alternating shift workers may be less exposed to sunlight due to occupational characteristics, leading to lower vitamin D synthesis in the epidermis [25]. The lower prevalence of VDD in our study may be partially explained due to the workday, where the work schedule provides periods of day shift, or on their days off from work, in which the individual can perform outdoor activities, with greater exposure to the sun.
We also found that 16% and 33% of workers were classified as high risk for OSA by BQ and NoSAS, respectively. And it was almost twice as high in workers with VDD, with 30% and 61% at high risk for OSA by BQ and NoSAS. The gold standard for diagnosing OSA, polysomnography, is not accessible to the entire population of shift workers. Therefore, it is estimated that about 80% of individuals are not correctly diagnosed with OSA [42]. Thus, alternative, more accessible, and simplified instruments, have been used in the screening of OSA [43,44] and perhaps be of interest in rotating shift workers and other individuals at high risk for OSA [45].
Recent studies have shown that 1 to 4 people in the general population are at high risk for OSA, depending on gender, age, and diagnostic method used [7,39]. Two large population-based cohorts, the HypnoLaus and EPISONO, entered that the high risk of OSA measured by the BQ ranged from 25 to 30%, and when assessed by the NoSAS score ranged from 35 and 43% [39]. In both studies, the prevalence measured by the BQ was at least two times higher than that found in our study. This difference may be related to the age of the participants, in which 43% of the subjects assessed were 60 years or older, an age group different from The data were compared using the chi-square analyses with Bonferroni correction. Cramer's V was used as an effect size 1 Hyperglycemia, FPG ≥ 100 mg/dL or HbA1c ≥ 5.7%; 2 dyslipidemia is classified when at least one of the parameters (TC, LDL, HDL, and TG) was altered or uses lipid-lowering drugs; 3 hypertension, SBP > 140 mmHg or DBP > 90 mmHg; 4 low physical activity (< 600 measure energy total-min/week); 5 Fagerstrom test, medium-and high-risk nicotine consumption; 6 AUDIT medium-and high-risk alcohol consumption tons of mining material), and may have underreported this information due to their occupational role [45]. OSA is a multifactorial chronic disease, with risk factors well described in the literature, such as gender, age, obesity, alcohol, and tobacco consumption [7]. However, some recent studies have shown that vitamin D deficiency may be an important risk factor [21]. In our study, we found that, after adjusting for confounding variables, workers with vitamin D deficiency were 52 to 85% more likely to have a high risk for OSA by QB and NoSAS Table 2 Berlin Questionnaire values (workers positive in each category) and NoSAS score criterion (workers positive in each criterion) of rotating shift workers according to vitamin D BQ, Berlin questionnaire; NC, neck circumference; OR, odds ratio; CI, confidence interval; p, p-value of Pearson's chi-square test; V, value of Cramer's V test The data were compared using the chi-square analyses with Bonferroni correction. Cramer's V was used as an effect size. p-values in bold are the significant associations according to the chi-square test   score, respectively. Similar results were found by other studies, in which in patients with OSA, the higher the apnea-hypopnea index (AHI), the lower were the 25(OH) D levels [21]. However, caution is needed in interpreting the data, as there is an inverse association between vitamin D metabolites and excess body fat [9]. Goswami et al. (2016) discuss that the association between serum 25(OH)D (vitamin D) levels < 20 ng/mL and OSA may be explained by some confounding factors, mainly BMI and neck circumference [47]. In this regard, it is important to mention that workers with vitamin D deficiency were more likely to have witnessed snoring and apnea (BQ category 1). This finding is corroborated by other studies with objective measures of OSA, such as polysomnography [48]. Furthermore, when assessing NoSAS scores, we observed that workers with VDD were 59% more likely to self-report the presence of snoring during sleep, the main clinical predictor signal for OSA [43]. In addition to obesity, another extremely relevant risk factor for OSA occurrence is aging. Piovezan et al. (2017) verifying the association of vitamin D levels and apnea, found that individuals with 25(OH)D levels < 30 ng/mL had a higher percentage of OSA than individuals with 25(OH)D ≥ 30 ng/mL, only in participants over 50 years old [48]. This study demonstrated the association between vitamin D and OSA starting earlier (age ≥ 40 years) in shift workers. Also, in concordance with previous studies, the greater the chances of OSA with increasing age. Shift workers are at a higher risk of developing OSA because of the work schedule, which appears to contribute to the increased risk in younger adults. Vitamin D deficiency can lead to airway muscle hypertrophy and myopathy, consequently increasing the risk of developing obstructive sleep apnea [13]. In addition to OSA, aging also affects endogenous vitamin D synthesis. There is a decrease in concentrations of the precursor (7-dehydrocholesterol) of vitamin D in the skin with aging, decreasing its production from the sun's rays [49].
Therefore, after adjusting for confounding variables, we concluded that workers with vitamin D deficiency are more likely to be at high risk for OSA, as measured by QB or NoSAS. We agree that the gold standard method for diagnosing OSA is polysomnography; however, it is an expensive test that requires qualified professionals and structure to be conducted. Thus, subjective analysis, through questionnaires or indicators, can provide very useful data for a better understanding of the presence of sleep apnea risk and its determinants.
The main limitations of this study are the variables obtained by self-report, which can lead to underestimation of risk behaviors or overestimation of protective behaviors. And also the lack of a group of regular daily workers to compare the results found in alternating shift workers. Additionally, the BQ use in shift workers may be a limitation of our study, due to the possibility that these workers could have fatigue and sleepiness due to the underlying shift work, and confound the findings. However, category 2 of BQ (which measures daytime sleepiness and fatigue) had the lowest percentage of positive scores (5%). To adjust for this, we also assessed the risk for OSA by the NoSAS score, which has no biases related to tiredness and sleepiness and has good predictive values for OSA [45,[50][51][52]. And finally, we included potential confounding variables to reduce measurement errors.
We consider as strengths of this study, the large sample of shift workers, as this is a difficult population to evaluate and is poorly studied. Furthermore, very little is explored in the literature about OSA in rotating shift workers, and also its relationship with deficient vitamin D levels. This is a population at risk for several comorbidities [29] and we encourage that similar studies be conducted with this population. Multivariate logistic regression analysis to estimate the odds ratio of OSA risk in workers with vitamin D deficiency. Hosmer-Lemeshow test and Akaike information criterion (AIC) were used to assess the goodness-of-fit of the models. p-values in bold are the significant associations according to the logistic regression Analysis adjusted for seasonality, skin color, age, years of shift work, geographic location, hyperglycemia, dyslipidemia, hypertension, body mass index, waist circumference, and neck circumference OSA risk assessment n (%)

Conclusion
Rotating shift workers have a high risk for obstructive sleep apnea, and workers with vitamin D deficiency have higher chances of apnea, even after adjustment for confounding factors. However, only workers over 40 years with vitamin D deficiency were at increased risk for OSA. The impact of these comorbidities reinforces the need for prevention, early identification, and intervention on the potential health harms of these workers. More research should be conducted using direct methods, such as polysomnography examination to assess sleep in populations prone to OSA, as a way to promote healthy habits and prevent accidents.