Eligible studies and study characteristics
Of the 6055 records, 39 studies10-38 involving a total of 7119 patients were included in the final meta-analysis (Figure 1). Table 1 showed the characteristics of the included studies. These studies, with a size from 33 to 774 and a median age from 42 to 64, comprised 1 RCT, 11 case-control studies, and 17 retrospective cohort studies. One of the studies came from Brazil, one from Spain, and the rest from China. Most studies used a low dose of glucocorticoids, i.e.1-2 mg/kg/d (an equivalent of methylprednisolone) and only one study29 reported pulse use of glucocorticoids, i.e. 250-500 mg/d (an equivalent of methylprednisolone). Methylprednisolone was the most common type, followed by dexamethasone, prednisone and prednisolone, and finally hydrocortisone. The median days for glucocorticoids treatment from illness onset ranged from 1 to 13 days and the median duration of treatment from 3 to 10.8 days. The studies reported different time frames of viral clearance delay, between 5- and 45-day, and the longest reported follow-up was 50 days.
Supplementary Tables 1, 2, and 3 showed the risk bias of the included studies. Seven studies were considered as low risk, 19 as some concerns, and 3 as high risk. The average score of total risk bias for case-control studies was 6.1 and the average score for retrospective cohort studies was 6.7. The only RCT was assessed as the trial with the risk bias of some concerns, due to its deviations from intended interventions.
Risk of viral clearance delay
A total of nine studies reported HR for risk of viral clearance delay in COVID-19 patients who received glucocorticoids treatment, of which the longest follow-up was 50 days. The overall unadjusted HR (1.58, 95% CI 1.39 to 1.80, I2=13%, PI 1.32 to 1.90) (Figure 2B) and adjusted HR (1.71, 95% CI 1.51 to 1.94, I2=22%, PI 1.45 to 2.01) (Figure 2A) revealed an association between glucocorticoids treatment and increased risk of viral clearance delay in COVID-19 patients. The pooled MD of days for SARS-CoV-2 RNA shedding from illness onset (2.13, 95% CI 0.83 to 3.42, I2=73%, PI -2.66 to 6.92) (Figure 2C) and overall unadjusted RR (1.29, 95% CI 1.14 to 1.47, I2=58%, PI 0.86 to 1.95) (Figure 3C) also confirmed the delayed viral clearance in glucocorticoids treatment patients, compared to patients revived non-glucocorticoids treatment. A few studies (four studies)24,31,34,36 reported the ORs for risk of viral clearance delay, however, the time frames of viral clearance delay among these studies were substantially different (Li&Cao et al24, 11-day; Xu&Chen et al34, 15-day; Qi&Yang et al31, 17-day; Yan&Liu et al36, 23-day) (Table 1). Pooled unadjusted OR (2.08, 95% CI 0.35 to 12.41, I2=84%) (Figure 3B) and adjusted OR (1.82, 95% CI 0.70 to 4.76, I2=57%, PI 0.04 to 81.84) (Figure 3A) of these studies showed no association between glucocorticoids treatment and risk of viral clearance delay.
Influence analyses identified four studies15,23,26,32, with an excessive influence on the overall results, of which two studies with extreme sample size are the excluded studies of predesign in the sensitivity analyses (Supplementary Figure 8). All the sensitivity analyses based on adjusted HRs showed a similar result with that from the main analysis (Supplementary Figure 1-7). Funnel plot analysis showed no asymmetry on the HRs, RRs, and MDs (Supplementary Figure 9-12), and the Egger test detected no significant small-study effects. Due to a very limited number of studies reporting ORs, we failed to draw funnel plots and correspondingly failed to conduct Egger tests on these studies.
Subgroup analysis
Subgroup analysis revealed that risk of viral clearance delay was significantly higher in glucocorticoids-treated COVID-19 patients of being mild or moderate (adjusted HR 1.94, 95% CI 1.39 to 2.70, I2=52%; MD 2.59, 95% CI 1.21 to 3.97, I2=24%), but not patients of being severe or critical (adjusted HR 1.85, 95% CI 1.05 to 3.26; MD 0.22, 95% CI -1.85 to 2.29, I2=56%) (Figure 4). Only one study19 compared the risk of viral clearance delay between COVID-19 patients who received a low dose (40 mg/day, an equivalent of methylprednisolone) and those who received a high dose (80 mg/day), and another study29 reported the effects of the pulse (250-500 mg/day) use of glucocorticoids. Their results indicated that a high dose of glucocorticoids increased the risk of viral clearance delay (adjusted HR 1.49, 95% CI 1.03 to 2.15; unadjusted RR 1.47, 95% CI 1.12 to 1.94), but neither low dose (adjusted HR 1.39, 95% CI 0.93 to 2.08; unadjusted RR 1.33, 95% CI 1.00 to 1.77) nor pulse use (unadjusted RR 1.85, 95% CI 0.66 to 5.19) (Figure 5).