Fig 1.
The cosinor 25(OH)D model for all months.
The predicted 25(OH)D mean value of each month through the annum can be extracted from the graph. This figure demonstrates the predicted peak of 25(OH)D concentration in June and the expected trough in December. The predicted 25(OH)D mean value of each month can be utilized to predict the 25(OH)D for the hospitalization month of any given patient (Eq 3). M = mesor, A = amplitude, P = Phase.
Table 1.
Characteristics of hospitalized COVID-19 patients stratified by pre-infection 25(OH)D level measured 14 to 730 days before positive COVID-19 test.
Table 2.
Characteristics of hospitalized COVID-19 patients and their pre-infection measurement of 25(OH)D levels, stratified by COVID-19 disease severity category.
Table 3.
25(OH)D level stratified by COVID-19 disease severity category before and after cosinor correction models for the annual mean as well as each patient’s date of hospitalization.
Table 4.
Multivariable logistic regression analysis of possible predictors of severe or critical COVID-19 disease among hospitalized patients with a pre-infection measurement of 25(OH)D level.
The multivariable analyses before and after cosinor correction models for the annual mean as well as each patient’s date of hospitalization are shown. This is the final step in the logistic regression (backward elimination).
Fig 2.
Box-and-whisker plots of the most recent pre-infection serum 25(OH)D levels before hospitalization were collected as a baseline (N = 253).
The mean vitamin level was compared between the four categories of COVID-19 disease severity as determined by the WHO definition (WHO/2019-nCoV/clinical/2020.5). A Kruskal-Wallis test for multiple-category comparison shows a significant difference between groups p < 0.001. A Mann-Whitney test compared vitamin D mean rank of two neighboring categories sequentially; mild compared with moderate (mean difference, 12.96 ng/mL; [Rank difference 26.65] p < 0.001); moderate compared with severe (mean difference, 10.72 ng/mL [Rank difference 23.08]; p < 0.002); Severe compared with Critical (mean difference, 3.96 ng/mL [Rank difference 6.33]; p = 0.40). The boxes present the range of vitamin D values within the interquartile range (50% of the cases). The whiskers outside the box mark the most upper and lower values within 1.5 times the interquartile range. Outliers’ values in each group are represented with empty circles.
Fig 3.
Correlation between pre-infection vitamin D deficiency status and COVID-19 disease severity stratified by three different groups of age (<50, 50–64, ≥65).
The severity of illness (critical, severe, moderate, and mild, as determined by the WHO definition of COVID-19 disease severity (WHO/2019-nCoV/clinical/2020.5)). The regression fit for each age group is shown. Values that are more than two SD away from the regression fit to the rest of the data were omitted from the analysis (applicable for two values appear on <50 yr graph that omitted from regression fit calculation).