The study population included 43.8% men, mean age was 62.1 years. Among the 9548 participants included in the study, 4186 (43.8%) had vitamin D insufficiency (25(OH)D levels of 30–<50 nmol/L) and 1438 (15.1%) had vitamin D deficiency (25(OH)D levels <30 nmol/L) (Figure 1
). Furthermore, 13 (0.1%) had too high 25(OH)D levels >200 nmol/L.
provides a description of the characteristics of the study population at baseline as well as prevalence of vitamin D insufficiency and deficiency and median 25(OH)D levels according to those characteristics. Both vitamin D insufficiency and deficiency were more frequent among females with higher age and BMI. In subjects with low physical activity and those who consumed fish less than once per week, median 25(OH)D levels were correspondingly lower. A seasonal variation with lower 25(OH)D levels in winter than in summer months was also observed. Moreover, 25(OH)D levels were particularly low among subjects with low education and current smokers.
Overall, 2363 (24.7%) study participants died during a median of 15.3 years of follow-up, of whom 815, 825 and 123 died from cardiovascular disease (CVD), cancer and respiratory disease, respectively. Figure 2
shows the Kaplan-Meier curves for deaths from respiratory disease according to vitamin D status. Mortality from respiratory diseases was consistently highest among participants with vitamin D deficiency and consistently lowest among those with sufficient vitamin D levels throughout up to 16.5 years of follow-up. The log-rank test indicated statistically significant survival differences with respect to respiratory disease mortality between the groups with vitamin D deficiency and sufficient vitamin D (p
< 0.0001), as well as for the comparison of subjects with vitamin D insufficiency and sufficient vitamin D (p
presents results on the adjusted dose–response relationship between 25(OH)D levels and respiratory disease mortality. Mortality strongly increased with decreasing 25(OH)D levels below 50 nmol/L and even more so below 30 nmol/L, i.e., in the vitamin D insufficiency and deficiency range. It is not statistically significant that 25(OH)D levels >75 nmol/L are associated with further decreasing respiratory disease mortality because the confidence interval is large and includes the null effect value of HR = 1.
Curves were assessed by using restricted cubic splines with knots at 25-hydroxyvitamin D concentrations of 30, 60, 90 and 120 nmol/L, and a 25-hydroxyvitamin D concentration of 75 nmol/L was used as the reference. The Cox proportional hazards regression model was adjusted for sex, age, season of blood draw, school education, smoking, BMI, physical activity, and fish consumption.
shows the associations of vitamin D status with all-cause, CVD, cancer and respiratory disease mortality after adjustment for multiple potential confounders. Vitamin D insufficiency and deficiency were associated with significantly increased all-cause mortality compared to sufficient vitamin D status (full model HRs (95%CI): 1.2 (1.1–1.3) and 1.7 (1.5–1.9), respectively) (Table 2
). Vitamin D deficiency was also associated with significant increases in CVD and cancer mortality by 52% and 38%, respectively (full model results). However, vitamin D insufficiency and deficiency were particularly strongly associated with respiratory disease mortality with full model HRs of 2.1 (95%CI: 1.3–3.2) and 3.0 (95%CI: 1.8–5.2), respectively. Overall, 41% (95%CI: 20–58%]) of all deaths from respiratory diseases were statistically attributable to 25(OH)D levels <50 nmol/L.
shows the results of the sex-specific analyses. For all-cause, cardiovascular disease and cancer mortality, only modest, statistically non-significant differences were seen between women and men. Although significant increases were seen for respiratory disease mortality in both women and men, they were much stronger among women, with 8.5 (95% CI 2.4–30.1) and 2.3 (95% CI 1.1–4.4)-fold increase of respiratory disease mortality in the case of vitamin D deficiency among women and men, respectively. However, the number of respiratory deaths among women was small (n
= 42 overall), especially in the reference groups of those with 25(OH)D > 50 nmol/L (n
= 3), which resulted in very wide confidence intervals of the estimated hazard ratios.
In this large population-based cohort study from Saarland, Germany, the majority of participants aged 50–75 years at baseline had vitamin D insufficiency or deficiency, and these conditions were associated with increased mortality. In particular, mortality from respiratory diseases was increased by 2.1- and 3.0-fold in subjects with vitamin D insufficiency or deficiency, respectively, compared to participants with sufficient vitamin D status. Significant associations with respiratory disease mortality were seen among both women and men, but they were particularly strong for women. Overall, 41% of deaths from respiratory diseases were statistically attributable to vitamin D insufficiency or deficiency and could possibly be avoided by overcoming these conditions, assuming causality of the association.
The assumption of causality of vitamin D3
effects on mortality obviously requires most careful discussion. Although we made the best attempts to adjust potential confounding factors, we cannot exclude the possibility of residual confounding by imperfect measurement of confounding variables, such as smoking or physical activity, or omission of unknown confounders. As addressed in detail elsewhere [6
], interpretation of the evidence is further complicated by the fact that vitamin D deficiency could be considered both a consequence of poor health as well as a risk factor for increased vulnerability to acute disease and poor outcomes of chronic diseases among people with poor health. Our findings therefore require critical discussion in the light of additional criteria and evidence, such as biological mechanisms and plausibility, and, in particular, in the light of data from RCTs providing vitamin D3
Deaths from respiratory disease are mostly deaths from lower respiratory infections [13
]. Vitamin D3
is thought to protect from occurrence and poor outcomes of respiratory infections by several mechanisms, including enhanced physical barriers (maintenance of junction integrity), cellular innate immunity, and adaptive immunity [1
Innate and adaptive immunity are being influenced by sex hormones [14
], which may explain the observed interaction of sex and 25(OH)D levels with respiratory disease mortality. According to data from the US-American National Health and Nutrition Survey, women have a higher inflammation burden than men (the age range was 40–90+ years [15
]). Especially postmenopausal women, like those included in the ESTHER study, have a high inflammatory burden because a decline in estrogen levels during menopause is associated with an increased expression of pro-inflammatory cytokines, including interleukin 6 and tumor necrosis factor (TNF) α [14
A cytokine storm as an adverse immune response to a SARS-CoV-2 infection is currently a major hypothesis for the underlying cause of a large proportion of COVID-19 deaths [18
]. Sufficient 25(OH)D levels are suggested to contribute to prevention of the cytokine storm [1
]. Vitamin D is known to interact with the angiotensin-converting enzyme 2 (ACE2), which is both used by SARS-CoV-2 as an entry receptor and is an important protein on an anti-inflammatory pathway [21
]. While SARS-CoV-2 downregulates the expression of ACE2, vitamin D upregulates it.
In a meta-analysis of individual participant data of 25 RCTs that included 11,321 participants aged 0–95 years, vitamin D3
supplementation was shown to reduce the risk of acute respiratory tract infection (OR 0.88, 95% CI 0.81–0.96) [2
]. The best effects were shown for daily or weekly vitamin D3
supplementation without additional bolus doses (OR 0.81, 95% CI 0.72 to 0.91). The protective effects were particularly strong in those with baseline 25-hydroxyvitamin D levels <25 nmol/L (OR 0.30, 95% CI 0.17–0.53). In a recent meta-analysis of RCTs on vitamin D3
supplementation for patients with chronic obstructive pulmonary disease (COPD), the risk of acute exacerbations was estimated to be reduced by 61% (95% CI 36–77%) [22
]. In remarkable consistency with our results, these meta-analyses of RCTs provide strong evidence for the preventive potential of vitamin D3
supplementation against acute respiratory infections and COPD exacerbations in particular.
It appears plausible to assume that the anti-inflammatory mechanisms of vitamin D would be relevant for SARS-CoV-2 infections in a similar manner as for other severe viral respiratory diseases, such as influenza. A first study posted on a pre-print server on 13 May 2020 suggests that the protective effects of vitamin D3
on other acute respiratory tract infections may be translated to COVID-19 infections [23
]. Vitamin D deficiency and vitamin D3
treatment data were available for 499 COVID-19 patients from Chicago for the year prior COVID-19 testing. Being likely vitamin D deficient (defined as being vitamin D deficient at last available time point without increase of vitamin D treatment) at the time of COVID-19 testing was associated with a 1.8-fold increased risk of being tested positive for COVID-19 (p
< 0.02) as compared to likely vitamin D sufficient.
It is worth noting that beneficial effects of vitamin D3 supplementation against manifestation or exacerbation of acute respiratory infection during an epidemic would be expected to go beyond individual protection of those using supplementation, as limiting such manifestation and exacerbation would also be expected to reduce the potential of spread of the disease to other persons and relieve the overload of the medical system by the epidemic.
To our knowledge, no previous vitamin D3 supplementation RCTs have addressed mortality from respiratory disease as the primary endpoint, and no meta-analysis of results for this specific endpoint have been reported, which most likely reflects the relatively small share of deaths from respiratory diseases among all deaths. In our cohort of older adults, these deaths accounted for 5.2% of all deaths. Even though this proportion is expected to be higher during the COVID-19 pandemic, the majority of deaths still occur from other diseases, and the summary effect, benefit–harm ratio and cost-effectiveness with respect to all relevant outcomes therefore deserve most careful attention for any general prevention efforts.
In that respect, vitamin D3
supplementation appears to be a particularly promising approach, especially for population groups with high prevalence of vitamin D insufficiency or deficiency, such as the elderly and those with severe comorbidities (which essentially coincide with population groups at highest risk of severe course and death from SARS-CoV-2 infection [24
]). The personal, health care and societal costs of the vitamin D3
intervention are negligibly low compared to the very high costs of currently employed “general population measures,” such as extensive testing for the infection and the lockdown of large proportions of economic and social life, including the delay or omission of much of routine medical care for other relevant diseases. In fact, some of these measures are expected to severely aggravate vitamin D insufficiency or deficiency, especially in high risk groups, such as restrictions of spending time outdoors for the total population (as practiced, for example, in France and Spain) or certain high risk groups, such as nursing home residents (as practiced in many countries, including Germany). Such restrictions dramatically reduce opportunities to maintain adequate vitamin D levels through endogenous synthesis by relevant sun exposure. Avoidance of sun exposure has been shown to be associated with increased mortality in epidemiological studies [25
supplementation has been demonstrated to be safe in numerous large-scale studies, and the risk of harm seems to be negligible compared to the risk of harmful side effects of the aforementioned and other general population measures, such as delayed diagnosis and treatment of cancer, myocardial infarction or stroke, withheld or deferred delivery of surgical or other medical services, or health risks related to unemployment and loneliness [26
]. On the contrary, one expected “side effect” would be reducing total cancer mortality by 13%, as suggested by a recent meta-analysis of RCTs [3
]. For Germany, with currently approximately 230,000 deaths from cancer per year [30
], this would translate in prevention of approximately 30,000 cancer deaths each year, suggesting substantial additional benefit besides lowering the COVID-19 burden during the COVID-19 pandemic and beyond.