25(OH)D levels during the COVID-19 pandemic: impact of lockdown and ultraviolet radiation

Background

Vitamin D₃ (cholecalciferol), the natural form of vitamin D, is produced in the skin from 7-dehydrocholesterol. After solar irradiation (between 290 and 315 nm solar ultraviolet B [UVB] radiation), 7-dehydrocholesterol produces previtamin D₃, which undergoes a temperature-sensitive rearrangement of three double bonds to form vitamin D₃. In the liver, vitamin D₃ is hydroxylated at carbon 25 to produce 25(OH)D₃ (25-hydroxyvitamin D₃), vitamin D main circulating form.

In the proximal renal tubule, 25(OH)D₃ is hydroxylated at position 1 of the carbon ring, which results in the formation of 1,25(OH)D₃, the functional and hormonally active form of vitamin D, responsible for most biological actions of vitamin D;¹ 25(OH)D is considered the best indicator of vitamin D status,² although definitive reference ranges are still under debate. The US Institute of Medicine has set the sufficiency threshold at 20 ng/dL, and the Endocrine Society at 30 ng/dL.^3,4

Although vitamin D can be obtained from diet and supplements, its synthesis in the skin is the most important source and depends on ultraviolet (UV) radiation intensity, which in turn is contingent on season and latitude.^5-7 Biological weighting of solar UV irradiance spectra with vitamin D synthesis action spectrum has been used to predict the ability to synthesize vitamin D with variations by latitude, season, and time of day. It has also been used to assess the effect of clouds on vitamin D production. For this, a cloud modification factor is applied to the UV radiation results to calculate the action spectrum for vitamin D biosynthesis.^8,9 Finally, ambient temperature would also be a contributory factor in the maintenance of adequate levels of vitamin D.¹⁰

In 2019, a new coronavirus, designated as SARS-CoV-2, emerged in the city of Wuhan, China, causing an unusual outbreak of viral pneumonia. Being highly transmissible, this new coronavirus disease, also known as coronavirus disease 2019 (COVID-19), has rapidly spread throughout the world.^11,12 Since then, the pandemic has had a significant impact on numerous health care systems from all over the world, which has resulted in significant morbidity and mortality, and has turned into the most important global health crisis since the 1918 influenza pandemic. Home confinement was introduced in different parts of the world to curb the spread of COVID-19. However, this strategy, by reducing exposure to UV radiation, could have negatively influenced vitamin D levels, as demonstrated in some populations.¹³ In other investigations, said association was not identified.¹⁴ No analyses have been carried out in Argentina on the subject. On the other hand, low vitamin D levels have been reported to be associated with a higher susceptibility to COVID-19.^15-19

Material and methods

Retrospective, analytical study of 15,908 patients older than 18 years, in whom the level of 25(OH)D was determined between January 1, 2019 and December 31, 2021 in a centralized laboratory of the city of Rosario, Argentina.

Total 25(OH)D level (vitamins D₂ and D₃ in ng/mL) was determined by electrochemiluminescence (Centauro^® Siemens) in a centralized laboratory. 25(OH)D levels were analyzed as a continuous variable, and as categorical variable, patients were classified as with optimal 25(OH)D levels (> 30 ng/mL), with insufficiency (20-30 ng/mL), with deficiency (10-20 ng/mL) or with severe deficiency (10 ng/mL).⁴

Six periods were established and analyzed:

To assess the effect of confinement, the post-MPSI period and 25(OH)D level were analyzed week by week. The same was done for 2019.

Gender, age and measurement date were recorded. People younger than 18 years and quantifications carried out during periods not defined in the study were excluded.

Time series of UV radiation dose data were obtained from the Tropospheric Emission Monitoring Internet Service (TEMIS), formerly part of the European Space Agency Data User Program. These data are available at http://www.temis.nl/uvradiation/ and have global geographic coverage.²⁰ Information related to UV radiation can be obtained from TEMIS using the estimate parameterized by Allaart et al.,²¹ which is based on the use of total ozone column and the zenith angle for solar noon. This value is then weighted by the specific action (or response) spectrum to be analyzed. The global distribution of assimilated ozone used by TEMIS was first based on the SCIAMACHY instrument on board of ENVISAT satellite, and then GOME-2²² UVDVC data were used for our analysis. To obtain this value, the algorithm makes use of global ozone assimilation fields, surface albedo, aerosol estimation and considers Sun-Earth distance variation, as well as the coordinates of the site in question. The UVDVC data used in this work were obtained for the coordinates corresponding to the city of Rosario (latitude, −32.9587°; longitude, −60.6930°; height above sea level, 25 m). In addition, total average temperature and average daytime temperature, measured by the National Meteorological Service, were obtained for said locality.

Statistical analysis was carried out with the R software, version 4.1.0. Parametric tests were performed (Student’s t-test for two-group comparison, analysis of variance with one classification criterion [ANOVA], Bonferroni’s post-test for independent data for comparison of more than two groups) and the results were expressed as the mean ± SD. To analyze the effect of a continuous variable that could influence the response-dependent variable, an analysis of covariance was performed. For the correlation analysis, Pearson’s or Spearman’s linear correlation coefficient was calculated, depending on the normality of variables. Categorical variables were expressed as the number (percentage) and the contingency tables were analyzed using the chi-square test or Fisher’s test, according to the expected frequencies. A p-value < 0.05 was considered to be statistically significant.

The study was conducted in accordance with the guiding principles of the Declaration of Helsinki and was approved by the Ethics Committee of the institution where it was carried out. Owing the retrospective nature of the research, exclusive use of secondary data deriving from routine care, and the absence of risk for participants, the consent requirement was waived. In addition, information on patient identity was eliminated from final database.

Results

25(OH)D levels by cohort

Of the total number of patients in whom 25(OH)D was determined (n = 15,908), 7,440 were analyzed during the different study periods. Figure 1 shows the flowchart for inclusion of subjects in the study. Median age was 54 years (range: 40-65 years), and was higher in the 2019 cohort (59 years [45-69]; 2020, 54 years [41-66]; 2021, 50 years [38-63], p < 0.001). Of the total number of patients, 78.90% were females. A higher frequency of men was found in the 2021 cohort (2019, 14.3%; 2020, 18.4%; 2021, 26.8%; p < 0.001).

Figure 1. Algorithm for incorporating patients into the study and the cohorts that were formed. MPSI: mandatory preventive social isolation.

 

In the global analysis, 25(OH)D levels were 32.69 ± 12.96 ng/dL, and were lower during 2020 with regard to 2019 and 2021, even after an age-adjusted analysis of covariance was performed.

The frequency of patients with 25(OH)D deficiency was higher in 2020, with the proportion of patients with optimal values being lower (Table 1). A positive correlation was found between age and 25(OH)D levels (rho = 0.037, p < 0.0001): these were lower in subjects younger than 40 years in comparison with individuals aged 40 to 65 years and those older than 65 years (< 40 years, 31.98 ± 12.82 ng/dL; from 40 to 65 years, 33.73 ± 13.07 ng/dL; older than 65 years, 33.73 ± 13.77 ng/dL, p < 0.001).

Table 1. 25(OH) D status by analyzed year

25(OH)D	2019 (n = 1,696)		2020 (n = 2,538)		2021 (n = 3,206)
	n	%	n	%	n	%
Severe deficiency	35	2.1	56	2.2	49	1.5
Deficiency	253	14.9	456	18	427	13.3
Insufficiency	486	28.7	823	32.4	948	29.6
Sufficiency	922	54.4	1203	47.4	1782	55.6

Chi-square, p < 0.0001. Post-Bonferroni’s test: p = 0.0004 when comparing 2020 vs. 2019; p < 0.0001 when comparing 2020 vs. 2021.

 

25(OH)D levels varied according to the different periods of the pandemic and were lower in the first wave. Given that age varied according to the analyzed cohort, an analysis of covariance was carried out, in which the differences were maintained (Table 2). The frequency of deficiency and severe deficiency was higher in the first wave (Table 3).

Table 2. 25(OH)D levels (ng/dL) during COVID-19 pandemic phases

MPSI (n = 299)	First wave (n = 1,331)	Second wave (n = 3,206)	p-value
			Unadjusted	Age-adjusted
36.57 ± 13.27	29.38 ± 12.01	33.54 ± 13.05	< 0.001*	< 0.001†

* ANOVA (post-Bonferroni’s test). p < 0.001 when comparing MPSI period vs. first wave; p = 0.0002 when comparing MPSI period vs. second wave; p < 0.0001 when comparing first wave vs. second wave.

† Analysis of covariance.

 

Table 3. 25(OH)D status during the COVID-19 pandemic phases

25(OH)D	MPSI (n = 299)		First wave (n = 1,331)		Second wave (n = 3,206)
	n	%	n	%	n	%
Severe deficiency	3	1	36	2.7	49	1.3
Deficiency	31	10.4	291	21.9	427	13.3
Insufficiency	74	24.7	459	34.5	948	29.6
Sufficiency	191	63.9	545	40.9	1,782	55.6

Chi-square, p < 0.0001. Post-Bonferroni’s test: p = 0.0004 when comparing MPSI period vs. first wave; p < 0.0001 when comparing first wave vs. second wave; p = 0.15 when comparing MPSI period vs. second wave.

 

Confinement effect

To verify if the containment strategy implemented by the Argentinian government between March 20 and June 4, 2020 had any effect on 25(OH)D levels, we first compared the years 2019 and 2020 in terms of UVDVC in the Rosario area between January 1 and August 31. We also analyzed daily temperature and daytime temperature. Figure 2A shows that the UV radiation suitable for vitamin D biosynthesis that arrived in the area within the first eight months of 2020 did not significantly differ from that of 2019 (5.41 ± 4.46 kJ/m² in 2019 vs. 5.60 ± 4.49 kJ/m² in 2020, p = 0.60). The same occurred with total temperature (17.08 ± 5.98 ^oC in 2019 vs. 17.29 ± 6.53 ^oC in 2020, p = 0.71), as shown in figure 2B, and daytime temperature (19.21 ± 5.78 ^oC in 2019 vs. 19.92 ± 6.28 ^oC in 2020, p = 0.71). UVDVC was significantly lower in the post-MPSI period in comparison with the MPSI period (31.91 ± 4.69 vs. 35.19 ± 8.29, p = 0.03). There was no difference in UVDVC between post-MPSI period and the same period in 2019 (31.91 ± 4.69 kJ/m² vs. 32.90 ± 5.11 kJ/m², p = ns).

Figure 2. A: UVDVC (cloud-modified UV radiation recommended dose for vitamin D synthesis) weekly variability between January 1 and August 31, 2019 (black line) and 2020 (magenta line). B: mean daily temperature weekly variability; the blue area represents the mandatory preventive social isolation (MPSI) period, from March 20 to June 6, 2020.

 

25(OH)D levels were higher during the MPSI period in comparison with the same period in 2019 (36.58 ± 13.28 vs. 33.98 ± 13.52, p = 0.018). No differences were identified in 25(OH)D levels between the post-MPSI period and the same period in 2019 (32.57 ± 12.85 vs. 33.8 ± 12.97, p = ns). Lower levels of 25(OH)D were found in the post-MPSI period in comparison with the MPSI period (p < 0.00001), but not between the same periods of 2019. The frequency of hypovitaminosis D was also higher in the post-MPSI period in comparison with the MPSI period (48.6% vs. 36.1%, p = 0.01), with no differences being found in the comparison between similar periods of 2019. Figure 3 shows the comparison of weekly-averaged 25(OH)D levels between March 20 and August 31, 2019 and 2020. Figure 3 also shows that, during the period between the second half of July and the first half of August 2020, 25(OH)D levels were lower with regard to the MPSI period. That time corresponds to weeks 6 to 9 after MPSI conclusion.

Figure 3. 25(OH)D weekly levels variability from March 20 to August 31. Comparison between 2019 and 2020. The blue area represents the mandatory preventive social isolation (MPSI) period, from March 20 to June 6, 2020.

 

A positive correlation was found between 25(OH)D levels and UVDVC in 2019 (r = 0.116, p < 0.0001). This correlation was not demonstrated in 2020 (r = −00218, p = 0.433).

Discussion

Our study had the purpose to evaluate the possible association between MPSI and 25(OH)D plasma levels in an adult population from the city of Rosario and relate them to UVDVC. We demonstrated that mean 25(OH)D levels were lower during 2020, mainly during the first wave of COVID-19, which occurred between September and November. In addition, we found a decrease in 25(OH)D levels in the post-MPSI period in comparison with the MPSI period. This finding is not reproduced during the same periods of 2019, which suggests the hypothesis that 25(OH)D lower levels were due to lower biosynthesis associated with lower sun exposure induced by the strict confinement during MPSI, since both UVDVC and ambient temperature did not vary. These results are similar to those reported by other authors who analyzed 25(OH)D levels in children,^13,23 but differ from those reported by Ferrari et al.¹⁴ in a cohort of adult patients.

The MPSI was established with the purpose to flatten the curve of new infections, increase public awareness on standards of care, and prepare health services with staff training and expansion of technical and technological capacity.²⁴ However, its implementation also reduced vitamin D levels, a situation that could lead to a higher risk of respiratory infections, including COVID-19.²⁵ In this sense, we demonstrated lower levels of vitamin D during the first wave of COVID-19 in comparison with the MPSI period and the second wave, together with a higher prevalence of 25(OH)D deficiency and insufficiency. This situation was also demonstrated by Ferrari et al.¹⁴

As for the association of SARS-CoV-2 infection and COVID-19 mortality rates with exposure to sunlight, there is a factor involved in vitamin D endogenous biosynthesis that has almost been neglected. The few published studies indicate the existence of an association between the disease outbreak and latitude, although they are limited to the first months of the pandemic.^15,26 In our analysis, we demonstrated a lower UVDVC in the post-MPSI period in comparison with the post-MPSI period during 2020. In this phase, the first wave of COVID-19 did develop, with a significant increase in the number of cases and mortality rates. Since UV radiation participates in vitamin D biosynthesis, the increase in the number of COVID-19 patients with a higher prevalence of hypovitaminosis D could be justified.

A more updated analysis has demonstrated a possible correlation between COVID-19 cases per million population and 25(OH)D levels, but not with COVID-19-related mortality rates.²⁷ Similarly, Moozhipurath et al. identified a negative association between UV irradiance intensity and COVID-19 deaths.²⁸ However, we cannot establish a causal relationship because we did not analyze 25(OH)D or UVDVC levels based on the presence of COVID-19. In this sense, in a cohort of adult patients, no differences were observed in vitamin D levels between individuals with a positive or negative result for SARS-CoV-2 infection.¹⁴ In contrast, in a study carried out in pediatric patients with COVID-19, lower levels of vitamin D were found in comparison with the control group.²⁹

Finally, the possibility for vitamin D to be only a marker of severity and that there is no causality cannot be disdained. In this regard, a large Mendelian randomization study of vitamin D evaluated the evidence of causality between vitamin D levels and susceptibility to SARS-CoV-2 infection and its severity; after all possible confounding factors were excluded, no association was found.³⁰

A seasonal variability of serum vitamin D levels is commonly observed in many countries of the world;⁷ it is mainly related to the different characteristics of sun exposure during the year and, more specifically, to UV radiation intensity.^5-7 However, cutaneous synthesis of vitamin D can be affected by other factors, such as climate and culture.^31,32 We did not observe vitamin D characteristic seasonal variability in the period prior to the onset of the pandemic, but we did observe lower levels of 25(OH)D and a higher prevalence of hypovitaminosis D in the post-MPSI period. Given that the effect of UV radiation on 25(OH)D levels is usually detectable after two months,³³ any effect caused by the confinement related to COVID-19 should have been observed approximately between May and September 2020. The lowest levels of 25(OH)D were precisely recorded during the month of July and the first week of August 2020, corresponding to weeks 6 to 9 after MPSI conclusion.

These findings would not be explained by UVDVC profiles, which overlapped during the analyzed periods (Fig. 2A). Neither are they explained by ambient temperature. Furthermore, the lack of a correlation between the levels of UVDVC and 25(OH)D during 2020 shows that although the UVDVC dose did not vary between 2019 and 2020, the confinement might have entailed lower exposure to sunlight and a decrease in vitamin D cutaneous synthesis.

Finally, we found a positive correlation between 25(OH)D levels and age, along with lower 25(OH)D levels in people younger than 40 years. Although it is a mere speculation, it is possible to hypothesize that during the MPSI period, and in order to limit the negative effects of confinement, systematic vitamin D supplementation in older people was suggested, as it has been demonstrated in previous studies.³⁴ However, it cannot be ignored that the correlation documented in our analysis was too weak (rho = 0.037) for drawing categorical conclusions in this regard.

Our study has limitations that might have resulted in biased estimates. First of all, average vitamin D intake was not considered due to a lack of information on supplementation with said vitamin or patients’ dietary habits. In addition, it was impossible to determine the exact duration and the area of skin exposed to sunlight of participants. However, the number of patients analyzed in our cohort is a representative and reliable sample of the general population, and the presented results are a starting point that allows to assess more specific aspects of the impact of the pandemic on vitamin D levels and the relationship between vitamin D and SARS-CoV-2 infection evolution.

Conclusions

The strict confinement implemented during the first stage of the COVID-19 pandemic led, regardless of UVDVC and ambient temperature, to a higher frequency of hypovitaminosis D during subsequent periods.