Literature search and included studies
After searching the seven databases and removing duplicates, 1318 potentially eligible articles were reviewed. After eliminating 859 articles based on the title and abstract, the remaining 459 were read in full text and 441 were excluded.4 clinical trials studies[10, 11, 13, 14] and 3 observational studies[15–17] were finally included in the systematic review. The four studies involving 2,279 individuals were selected for meta-analysis, with follow-up period ranging from 14 days to 28 days. The flow chart was shown in Fig. 1. Details on the individual included studies are listed (see Additional file 1: Table S1).
Clinical improvement of short-term, mid-term and long-term
The meta-analysis of three studies (n = 1,691)[10, 11, 13] showed that as compared with placebo, 10-day remdesivir was associated with significant increased clinical improvement on days 14 (mid-term) and 28 (long-term) with RR 1.19 (95%CI 1.09–1.30) and RR 1.09 (95%CI 1.03–1.16), while no-significant difference was found between 10-day remdesivir group and placebo on the short-term clinical improvement (day 7) by the pooled result of two studies (RR = 1.01; 95%CI 0.82–1.25). Among the 3 studies reporting clinical improvement between 10-day remdesivir treatment and placebo, no heterogeneity was detected on days 7 (I2 = 0%; p = 0.97), 14 (I2 = 0%; p = 0.68) and 28 (I2 = 0%; p = 0.94), so fixed-effect models were used to meta-analyze the data (Fig. 2).
Two studies (n = 781)[13, 14] investigated clinical improvement difference between 10-day remdesivir group and 5-day remdesivir group, the meta-analysis demonstrated that 5-day remdesivir had better short-term (day 7) clinical improvement than 10-day remdesivir group with RR 1.20 (95%CI 1.02–1.41), but the difference of long-term (day 14) clinical improvement between these two groups was not statistically significant (RR 1.08; 95%CI 0.90–1.29). Heterogeneity of these two studies was not significant on days 7 (I2 = 0%; p = 0.55), but moderate heterogeneity existed on days 14 (I2 = 69%; p = 0.07), so a random-effect model was used to meta-analyze the data (Fig. 3).
Adverse events rates
Three studies (n = 1,865) [10, 11, 13] show adverse events rates of remdesivir arm was not significantly different as compared with placebo (RR = 1.02; 95%CI 0.87–1.20). Moderate heterogeneity was detected among included studies (I2 = 69%; p = 0.04), and a random-effect model was used to analyze the data (Fig. 4).
Two studies (n = 781) [13, 14]observed the adverse events rates of 10-day remdesivir group and 5-day remdesivir group. The pooled result illustrated no significant difference existed between these two treatment groups (RR = 0.92;95%CI 0.83–1.02). Heterogeneity was not significant between studies (I2 = 0%; p = 0.41), and a fixed-effect model was used for the data (Fig. 5).
Serious adverse events rates
As shown in Fig. 6, the meta-analysis result of three studies (n = 1,865) [10, 11, 13]shown that the serious adverse events rate of remdesivir treatment was obviously lower than that of placebo, with significant difference (RR = 0.75;95%CI 0.63–0.89). No heterogeneity was detected among three studies (I2 = 0%; p = 0.56), and a fixed-effect model was used to meta-analyze the data. As compared with 10-day remdesivir, 5-day remdesivir treatment reflected a lower serious adverse events rate (RR = 0.65;95%CI 0.47–0.88) by the pooled result of two studies (n = 781) [13, 14]. There was no heterogeneity between two studies (I2 = 0%; p = 0.40), and a fixed-effect model was used in this analysis (Fig. 7).
Grade 3 or 4 adverse events rates
The pooled result from three studies (n = 1,865) [10, 11, 13] that reported the grade 3 or 4 adverse events rates of remdesivir and placebo indicated that COVID-19 patients who treated by remdesivir suffered significantly lower risk of grade 3 or 4 adverse events than patients of placebo group (RR = 0.89;95%CI 0.80–0.99). No heterogeneity was detected (I2 = 0%; p = 0.55), so a fixed-effect model was used in this analysis (Fig. 8).
By the pooled result of two studies (n = 781) [13, 14] with RR 0.74 (0.58–0.95), the difference of grade 3 or 4 adverse events rates between 10-day remdesivir and 5-day remdesivir was statistically significant. No heterogeneity was detected(I2 = 0%;p = 0.60), and a fixed-effect model was used (Fig. 9).
Mortality
As illustrated in Fig. 10, the mortality of remdesivir and placebo was similar with RR 0.78 (0.59–1.03), no heterogeneity was detected among three studies (n = 1,882) [10, 11, 13]that investigated the mortality of remdesivir and placebo (I2 = 0%; p = 0.59). The mortality difference between 10-day remdesivir and 5-day remdesivir was also not statistically significant (RR = 0.74;95%CI 0.41–1.33), in the view of no heterogeneity between included two studies (n = 781) [13, 14], we used a fixed-effect model to analyze the data (Fig. 11).