Association of lockdowns with the protective role of ultraviolet-B (UVB) radiation in reducing COVID-19 deaths

Nations are imposing unprecedented measures at a large scale to contain the spread of the COVID-19 pandemic. While recent studies show that non-pharmaceutical intervention measures such as lockdowns may have mitigated the spread of COVID-19, those measures also lead to substantial economic and social costs, and might limit exposure to ultraviolet-B radiation (UVB). Emerging observational evidence indicates the protective role of UVB and vitamin D in reducing the severity and mortality of COVID-19 deaths. This observational study empirically outlines the protective roles of lockdown and UVB exposure as measured by the ultraviolet index (UVI). Specifically, we examine whether the severity of lockdown is associated with a reduction in the protective role of UVB exposure. We use a log-linear fixed-effects model on a panel dataset of secondary data of 155 countries from 22 January 2020 until 7 October 2020 (n = 29,327). We use the cumulative number of COVID-19 deaths as the dependent variable and isolate the mitigating influence of lockdown severity on the association between UVI and growth rates of COVID-19 deaths from time-constant country-specific and time-varying country-specific potentially confounding factors. After controlling for time-constant and time-varying factors, we find that a unit increase in UVI and lockdown severity are independently associated with − 0.85 percentage points (p.p) and − 4.7 p.p decline in COVID-19 deaths growth rate, indicating their respective protective roles. The change of UVI over time is typically large (e.g., on average, UVI in New York City increases up to 6 units between January until June), indicating that the protective role of UVI might be substantial. However, the widely utilized and least severe lockdown (governmental recommendation to not leave the house) is associated with the mitigation of the protective role of UVI by 81% (0.76 p.p), which indicates a downside risk associated with its widespread use. We find that lockdown severity and UVI are independently associated with a slowdown in the daily growth rates of cumulative COVID-19 deaths. However, we find evidence that an increase in lockdown severity is associated with significant mitigation in the protective role of UVI in reducing COVID-19 deaths. Our results suggest that lockdowns in conjunction with adequate exposure to UVB radiation might have even reduced the number of COVID-19 deaths more strongly than lockdowns alone. For example, we estimate that there would be 11% fewer deaths on average with sufficient UVB exposure during the period people were recommended not to leave their house. Therefore, our study outlines the importance of considering UVB exposure, especially while implementing lockdowns, and could inspire further clinical studies that may support policy decision-making in countries imposing such measures.

Scientific Reports | (2021) 11:22851 | https://doi.org/10.1038/s41598-021-01908-w www.nature.com/scientificreports/ of lockdown is the likelihood of limited exposure to UVB. However, to the best of our knowledge, no empirical study has explored the association between the severity of lockdown, the subsequent reduction in UVB exposure, and the number of deaths attributed to COVID-19 (COVID-19 deaths).
In this observational study of secondary data, we empirically outline the independent protective roles of lockdown and UVB as measured by ultraviolet index (UVI) and subsequently examine whether the severity of lockdown is associated with a reduction in the protective role of UVB. After controlling for time-constant and time-varying factors, we find that a unit increase in UVI and lockdown severity are independently associated with 0.85 percentage points (p.p) and 4.7 p.p decline in the growth rate of COVID-19 deaths. The change of UVI over time is typically large (e.g., Fig. S1 in Supplementary Appendix on an average UVI in New York City increases up to 6 units between January until June), indicating that the protective role of UVI might be substantial. These declines indicate the protective roles of UVI and lockdowns. Surprisingly, the widely utilized lockdown with the least severity (e.g., recommendation to not leave the house) is already associated with almost complete mitigation of the protective role of UVI by 0.76 p.p, which represents a decline of 81%.

Association of lockdown severity and UVB radiation with cumulative COVID-19 deaths
In general, non-pharmaceutical interventions such as the lockdowns aim to reduce the likelihood of transmission of the virus by limiting the movement of people, reducing the contact among individuals via restricting economic activities such as closing restaurants 1 . Earlier studies indicate that such large-scale non-pharmaceutical intervention measures might have slowed down COVID-19 spread, indicating their protective role, thereby providing public health benefits 1,3,9 . However, such policies' unintended health consequences (e.g., reduced exposure to UVB radiation) are largely unknown.
Prior studies indicate that UVB radiation plays a protective role in human health [10][11][12][13][14] . Humans get vitamin D either via diet (natural food, fortified food, or supplements) or skin synthesis by UVB radiation exposure 15 . Likelihood of UVB exposure and subsequent vitamin D synthesis undergo substantial variation according to several time-varying and time-constant factors such as latitude 15 , seasons 15 , time of the day 15 , lifestyle 16,17 , mobility 18 , age 15 , skin pigmentation 15 , and obesity 19 . Prior studies associate vitamin D deficiency with the likelihood of weakened immune response [20][21][22] , infectious respiratory diseases 15,23,24 , and the severity and mortality 25 .
Early evidence indicates that lockdown severity 2 and weather factors such as temperature and humidity may reduce the likelihood of transmission of the SARS-CoV-2 virus (Severe Acute Respiratory Syndrome Coronavirus 2), which causes COVID-19 26 . Even though UV radiation may help in reducing the likelihood of transmission by inactivating viruses in fomite transmission 27 , emerging epidemiological evidence related to COVID-19 suggests a protective role of UVB and the plausible role of vitamin D in improving immunity and decreasing the likelihood of COVID-19 severity and mortality [4][5][6][7][8]28 .
Specifically, emerging COVID-19 studies provide evidence of the protective role of UVB as well as vitamin D 8,[29][30][31] . Studies indicate that 1,25-dihydroxy vitamin D [1,25 (OH) 2 D], one of vitamin D's active forms, modulates innate and adaptive immune systems 4,32 as well as renin-angiotensin system (RAS) [32][33][34] . Further, studies note that it may reduce the risk of cytokine storm 4,32 by modulating the inflammatory response. Furthermore, studies indicate that it plays a role in stimulating antimicrobial peptides with antiviral properties such as defensins and human cathelicidin 4,32,35,36 . Emerging COVID-19 studies also provide evidence of the role of vitamin D deficiency in the incidence 29,30,37,38 , severity 37,39 , and mortality 31,40-42 associated with COVID-19 43 . A quasi-experimental study notes that vitamin D supplementation is associated with a higher chance for survival in elderly hospitalized patients 44 . Further, a recent randomized placebo-controlled study provides early evidence that vitamin D supplementation may help in SARS-CoV-2 (Severe Acute Respiratory Syndrome Coronavirus 2) viral clearance 45 . Furthermore, a pilot randomized clinical study by Castillo et al. 41 notes that Calcifediol, a metabolite of vitamin D, seems to mitigate the severity of COVID-19.
We explain the plausible association between the severity of lockdown and the subsequent reduction in the likelihood of UVB exposure with the COVID-19 deaths in Fig. 1. Figure 1 also summarizes various timevarying (e.g., weather, season, etc.) and time-constant factors (age, skin pigmentation, latitude, etc.) affecting the likelihood of UVB exposure and subsequent vitamin D synthesis. In light of the evidence from the emerging studies regarding the role of vitamin D 4-8 , we anticipate that an increased UVB radiation might be associated with a reduction in the number of COVID-19 deaths as it might affect the likelihood of vitamin D deficiency 8 and the transmission of COVID-19. Although lockdowns might help reduce the likelihood of transmission of SARS-CoV-2 virus 2 , they potentially also reduce the likelihood of UVB radiation exposure, increasing the likelihood of vitamin D deficiency, as indicated in Fig. 1 8 . Therefore, we anticipate that the severity of lockdown and UVB radiation are independently associated with a reduction in the number of COVID-19 deaths. However, an increased lockdown severity might also be associated with a reduction in the protective role of UVI in mitigating COVID-19 deaths.
We extend the theoretical and methodological framework of Moozhipurath et al. 8 by exploring the protective role of lockdown in the context of UVB exposure. Although Moozhipurath et al. 8 study the protective role of UVB radiation, it is not yet clear from their results whether the severity of lockdown is associated with a reduction in the protective role of UVB due to reduced likelihood of exposure. We also empirically explore in detail the different types of lockdown severities and whether and which lockdown severity mitigates the protective role of UVI. Thus, the results of this study aim to support the policy decision-making related to COVID-19 in countries that are currently implementing lockdowns or are considering them.
Furthermore, Moozhipurath et al. 8 investigated the protective role of UVB radiation when the Northern Hemisphere transitioned to spring and Southern Hemisphere countries transitioned to their autumn season. During their study period until 8 May 2020, the COVID-19 outbreak was focused mainly on European countries 8 .

Data and methods
Description of data. To empirically estimate the independent protective roles of lockdown, UVB, and the mitigating influence of lockdown severity on the protective role of UVB in reducing COVID-19 deaths, we use the dataset based on secondary data 46 as outlined in Table 1    The country-level dataset consists of daily observations of cumulative COVID-19 infections and deaths, ultraviolet index (UVI-closely associated with UVB spectrum of solar radiation), and a set of weather factors that we use as control variables-precipitation index, cloud index, ozone level, visibility level, humidity level, as well as minimum and maximum temperature on a daily basis. We merged this dataset with another dataset containing the severity of lockdown enforced by various governments. We describe each dataset in Table 1. We use the interaction of the lockdown severity with UVI to examine whether higher lockdown severity is associated with a reduction in the protective role of UVI in mitigating COVID-19 deaths. Table S3 in the Supplementary Appendix shows the number of observations of countries used in the analysis, whereas Table S4 shows the countries' latitude and longitude information used in the analysis.
In Table 2, we present the descriptive statistics of the dataset. On 7 October 2020, the average growth rate of COVID-19 deaths per country was 0.76%, and there were on average 6789 cumulative COVID-19 deaths across these 155 countries. The daily average COVID-19 deaths growth rate per country was 3.4% across the duration of the study. The corresponding average for UVI was 7.44, indicating a moderate to high risk of harm from sun exposure. We use cumulative COVID-19 deaths as the primary dependent variable to test our hypothesis that lockdowns mitigate the protective role of UVB radiation. Based on the severity of governmental advice and restrictions, there are four different severity levels that could decrease skin exposure to UVB radiation ranging from "no measures" to "requiring the population not to leave the house with minimal exceptions". On 7 October 2020, 54 countries had not implemented a lockdown, whereas 56 countries recommended their people not to leave the house. 43 countries required people to not leave the house except for daily exercises, grocery shopping, and essential trips, whereas 2 countries implemented a severe lockdown and required their people not to leave the house with minimal exceptions.

Summary of method.
We apply a log-linear fixed-effects model to estimate the mitigating influence of lockdown severity on the association between UVI and growth rates of COVID-19 deaths 8 . We describe and discuss the log-linear fixed-effect model and the structural equation, which we estimate, in the Supplementary Appendix Section 1, building upon and extending Moozhipurath et al. 8 as well as Hsiang et al. 1 . We chose this type of model since (i) a log-linear model considers percentage rather than absolute changes, the changes in COVID-19 deaths are more comparable over time across countries 8 51 . Such models are also used in the emerging literature 50 to estimate climate change's impact on mortality 52 and the migration pattern of humans 53 .
To assess the respective protective roles of lockdown and UVI in mitigating the growth rates of COVID-19 deaths and subsequently determine whether and which lockdown severity mitigates the protective role of UVI, we estimate three versions of the log-linear fixed-effects model. Model 1 outlines whether a unit increase in the lockdown severity mitigates the association between UVI and the growth rates of COVID-19. Model 2 and model 3 outline whether a more severe lockdown measure (e.g., LD severity 2 or 3 and LD severity 3, respectively) mitigates this association more strongly than a less severe lockdown (LD severity 1 and LD severity 1 or 2, respectively).

Results
We present our results in Table 3. After controlling for all time-constant and various time-varying factors 8 , we find that unit increases in UVI and lockdown severity are independently associated with 0.85 percentage points (p.p) and 4.7 p.p decline in COVID-19 deaths growth rate, indicating their respective protective roles. However, we find a significant mitigating influence of lockdown severity on the protective role of UVI in reducing the growth rates of COVID-19 deaths. A unit increase of the lockdown severity weakens the association of UVI in reducing the growth rates of COVID-19 deaths by − 44% (= 0.0037/− 0.0085). This decrease represents the average mitigation of a unit increase of the lockdown severity from 0 to 1, 1 to 2, and 2 to 3.
Surprisingly, Model 2 and Model 3 outline that the mitigation effect is mostly associated with lockdown severity of level 1 rather than level 2 or level 3 (stricter lockdowns) as the interaction of lockdown severity of level 2 or 3 with UVI is insignificant. Besides, the lockdown severity of level 1 mitigates the association of UVI and growth rates of COVID-19 deaths by − 81% (0.0076/− 0.0094) and, thus, almost completely mitigates the association. Finally, all models outline the significant negative association of lockdown severity and UVI with growth rates of COVID-19 deaths, indicating their protective roles.
To assess the robustness of our results of the primary model-Model 1-we isolate the mitigating influence of lockdown severity on the association of UVI and growth rates of COVID-19 deaths from time trends in flexible ways. Models 4-9 in Tables S1 and S2 in the Supplementary Appendix isolate our findings from linear, square and exponential time trends, which may be similar across countries or even country-specific. Overall, we find consistent results across different model specifications.
Next, we compare two scenarios to illustrate the mitigated protective role of a unit increase of UVI on the cumulative COVID-19 deaths. In scenario 1, UVI's protective role is not mitigated by lockdown, whereas in Table 3. Results of log-linear fixed-effects model. LD Lockdown severity. + : p < 0.10; *: p < 0.05; **: p < 0.01; ***: p < 0.001. F-statistic for long-run coefficients in parentheses. www.nature.com/scientificreports/ scenario 2, UVI's protective role is fully mitigated. To relate the mitigated protective role with COVID-19 deaths, we take the average number of COVID-19 deaths at the end of the observational period, i.e., 6,789, as cumulative COVID-19 deaths at day 0 shown in Fig. 2. In the full mitigation scenario (Scenario 2), we use the growth rate of COVID-19 deaths in our sample, i.e., 3.4%. Scenario 1, where UVI's role is not mitigated, uses an average growth rate of 2.55% (i.e., 3.4%-0.85 p.p). Figure 2 outlines that this mitigating influence of lockdown on the protective role of UVI translates into 1,183 or 11% fewer COVID-19 deaths after 14 days.

Discussion
Our empirical results indicate that large-scale lockdowns are associated with a substantial slowdown in the daily growth rates of COVID-19 deaths consistent with prior studies 1 . However, such measures also significantly reduce the protective role of UVB in COVID-19 deaths. Specifically, we find that unit increases in UVI and lockdown severity are independently associated with a decline in the growth rate of COVID-19 deaths, indicating their respective protective roles. However, the lowest lockdown severity (recommendation not to leave the house) is already associated with almost complete mitigation of the protective role of UVI in reducing the growth rate of COVID-19 deaths via a reduction of 0.76 percentage points or -81% (p < 0.001). Our results are consistent across different model specifications.
Our results suggest that lockdowns in conjunction with adequate exposure to UVB radiation might have provided even more substantial health benefits than lockdowns alone. For example, we estimate that there would be 11% fewer deaths on average with more UVB exposure while people were recommended not to leave their houses.
Our contributions are three-fold. First, to the best of our knowledge, this study is one of the first ones that outlines the association between the severity of lockdown, the subsequent reduction in UVB exposure, and COVID-19 deaths. Second, our study outlines the need for further large-scale clinical studies exploring the role of vitamin D in mitigating the pandemic. Third, even though emerging studies suggest the need for continued large-scale interventions 1,2 , in addition to substantial economic and social costs, the findings of our study indicate that an unintended consequence is the limited UVB exposure, which plausibly increases the risk of COVID-19 deaths. The results of our study can therefore inspire observational or experimental clinical studies that can further support COVID-19 related policy decision-making in countries that are currently implementing lockdowns or are considering them in the future to slow down COVID-19 growth. Specifically, such clinical studies may investigate if sensible sunlight exposure in conjunction with lockdown or with proper social distancing can mitigate COVID-19 deaths. Sensible UVB exposure is possible during lockdown by spending time outside in a garden, on balconies, or by exposing to sunlight through open windows. Further, the results of large scale clinical studies could help guide nations to create awareness regarding the importance of sensible sunlight exposure and also to assist vulnerable populations at a higher risk of vitamin D deficiency-e.g., darker-skinned people living in high latitudes, people with limited mobility or indoor lifestyle (nursing home residents), and vegetarians 8 .

Limitations
We follow a macro-level statistical backward-looking approach that captures real-life behavior without making any specific assumptions regarding epidemiological parameters 1 . Although this macro-level approach is a crucial strength of the study, the results cannot be interpreted as health guidance, which often comes from clinical studies 1 . Therefore, further clinical studies are needed to establish a causal relationship between UVB-induced vitamin D and COVID-19 deaths.  We use a fixed-effects model that isolates the effect of relevant weather parameters from country-specific time-constant factors 8 . Such country-specific time-constant factors consist of various economic, social, and health factors that are likely to remain relatively stable over the period of our study 8 . The time constant factors include the location (e.g., latitude and longitude), demographics, age composition, gender, genetics, and culture at a country level 8 . More importantly, such time-constant factors include factors that are closely associated with the severity of COVID-19, such as age, gender, mobility, lifestyle of the population, the prevalence of co-morbidities (e.g., obesity, hypertension, etc.), and skin pigmentation 8 . Fixed-effects also may capture factors associated with regular habits such as regular dietary patterns, the proportion of vegetarians in the population, regular dietary supplement consumption, and food fortification that may affect COVID-19 severity 8 .
Our methodology also flexibly controls for various time-varying factors 8 . First, our methods control for relevant confounding time-varying weather factors, including air pollution. Second, we control for various remaining time-varying factors by incorporating linear, quadratic, and exponential time-trends at a country level 8 in the robustness checks of the Supplementary Appendix. These flexible time-trends control for time-varying factors such as pressure on the health care system and exponential-shaped or s-shaped trends associated with the COVID-19 growth rate.
Although our methodology controls for all time-constant and various time-varying confounding factors, we acknowledge that our method has limitations. First, our method may be limited in capturing some of the timevarying confounding factors that may confound the results. For instance, our methodology may not capture behavioral changes of the people that are likely to be associated with seasonal changes (UVB variation), vitamin D levels, and COVID-19 deaths. For example, such time-varying behaviors include seasonal travel patterns, seasonal nutritional supplement intake, and seasonal dietary habits 8 . Second, although we use governmental measures, we do not have granular data on whether people comply with these measures. Therefore, we have limited information about whether people followed governmental instructions and stayed indoors during the lockdown. Third, we do not have data on the vitamin D level at a population level across these countries corresponding to the UVB radiation that prevents us from directly analyzing the association between COVID-19 deaths and vitamin D levels. Finally, our study is an ecological study based on country-level data and therefore has inherent limitations that are commonly associated with such ecological studies.
Even though we anticipate that reduced likelihood of skin synthesis due to lockdown plausibly explain these associations, we may not be able to rule out the possibility of other UVB-induced mediators-such as nitric oxide 8,10,54 . While we acknowledge there may be confounding factors posing challenges to our analysis, we used statistical methods to account for all time-constant and various time-varying factors as much as possible. We also acknowledge that UVB exposure may not substantially increase the vitamin D synthesis of specific categories of people, such as those wearing cultural protective clothing due to lower exposure to sunlight 15 and those who are elderly due to less efficient skin synthesis 55 . Although we do not model these factors explicitly, our fixed-effects model accounts for most of these time-constant confounding factors.

Data availability
The data used in the study are from publicly available sources. Data regarding COVID-19 are obtained on 9 October 2020 from COVID-19 Data Repository by the Center for Systems Science and Engineering (CSSE) at Johns Hopkins University and can be accessed at https:// github. com/ CSSEG ISand Data/ COVID-19. Data regarding weather is obtained from Dark Sky on the 9 October 2020 and can be accessed at https:// darks ky. net/. Data regarding lockdown severity is obtained from https:// www. bsg. ox. ac. uk/ resea rch/ resea rch-proje cts/ coron avirusgover nment-respo nse-track er. We will make specific dataset used in this study available for any future research. Interested researchers can contact one of the authors via email to get access to the data.