ABSTRACT
Objective: The effect of vitamin D status on adult COVID-19 pneumonia induced by Delta variant remains to be further explored. Methods: A longitudinal, real-world cohort study was performed. Artificial intelligence (AI) was used to identify and measure pneumonia lesions. All cases with pneumonia were divided into the vitamin D deficiency (VDD) and control groups according to serum 25-hydroxyvitamin D concentration. Lesion dynamics were observed within six time periods after the onset of pneumonia. Results: A total of 161 cases were included, of which 101 (63%) were male and 46 (29%) presented with pneumonia. The median age and baseline 25-hydroxyvitamin D concentrations were 37 years and 21 ng/ml, respectively. Age, fibrinogen, and SARS-CoV-2 IgG titer on admission were independent predictors for the onset of pneumonia. After the onset of pneumonia, patients in the VDD group (n = 18) had higher percentage of fever (33 vs. 7.1%; p = 0.04) than those in the control group (n = 28); the interval of pneumonia resolution was longer (28 vs. 21 days; p = 0.02); lesions progressed more rapidly (p = 0.01) within 3 to 7 days and improved more slowly (p = 0.007) within more than 28 days; notably, simultaneous interleukin-6 (18.7 vs. 14.6 pg/ml; p = 0.04) levels were higher, and cycle thresholds for N gene (22.8 vs. 31.3; p = 0.04) and ORF1ab gene (20.9 vs. 28.7; p = 0.03) were lower within 3 to 7 days. Conclusion: Vitamin D status may have effects on the progression and resolution, but not the onset of Delta variant-induced pneumonia in adults. Computed tomography image diagnosis system based on AI may have promising applications in the surveillance and diagnosis of novel SARS-CoV-2 variant-induced pneumonia.