Supplementary materials.docx: Efficacy of chloroquine and hydroxychloroquine for the treatment of hospitalized COVID-19 patients: A meta-analysis

Table S1 PRISMA 2020 checklist [1]Table S2 MEDLINE search strategyTable S3 EMBASE search strategyTable S4 PubMed search strategyTable S5 CNKI search strategyTable S6 Wanfang search strategyTable S7 Wanfang Med Online search strategyTable S8 SinoMed search strategyTable S9 CQVIP search strategyTable S10 Characteristics of included studies and patientsTable S11 Risk of bias, randomized controlled trials (RoB2)Table S12 Risk of bias, observational studies (ROBINS-I)Figure S1 Forest plot for subgroup analysis by study design, time to negative conversion of SARS-CoV-2 tests
There is no significant difference between the pooled mean differences from randomized studies versus non-randomized studies (P=0.64).Figure S2 Forest plot for subgroup analysis by imputed versus non-imputed studies, time to negative conversion of SARS-CoV-2 tests
There is no significant difference between the pooled mean differences of imputed versus non-imputed studies (P=0.05).Figure S3 Forest plot for subgroup analysis by regimen, incidence of negative test conversion at day 7
There is no significant difference between different regimen subgroups (P=0.18).Figure S4 Forest plot for subgroup analysis by regimen, incidence of negative test conversion at day 14
There is no significant difference between different regimen subgroups (P=0.31).Figure S5 Forest plot for subgroup analysis by study design, incidence of negative test conversion at day 7
There is no significant difference between the pooled odds ratios from randomized studies versus non-randomized studies (P=0.08).Figure S6 Forest plot for subgroup analysis by study design, incidence of negative test conversion at day 14
There is no significant difference between the pooled odds ratios from randomized studies versus non-randomized studies (P=0.79).Figure S7 Forest plot for subgroup analysis by study design, length of stay
There is no significant difference between the pooled mean differences from randomized studies versus non-randomized studies (P=0.98).Figure S8 Forest plot for subgroup analysis by risk of bias ratings, length of stay
There is no significant difference between the pooled mean differences from studies with low/moderate risk of bias versus studies with high risk of bias (P=0.28).Figure S9 Forest plot for subgroup analysis by imputed versus non-imputed studies, length of stay
There is no significant difference between the pooled mean differences of imputed versus non-imputed studies (P=0.17).Figure S10 Forest plot for subgroup analysis by study design, mortality
There is no significant difference between the pooled odds ratios from randomized studies versus non-randomized studies (P=0.16).Figure S11 Forest plot for subgroup analysis by risk of bias ratings, mortality
There is no significant difference between the pooled odds ratios from studies with low/moderate risk of bias versus studies with high risk of bias (P=0.61).Figure S12 Forest plot for subgroup analysis by study design, time to fever resolution
There is no significant difference between the pooled mean differences from randomized studies versus non-randomized studies (P=0.93).Figure S13 Forest plot for subgroup analysis by imputed versus non-imputed studies, time to fever resolution
There is no significant difference between the pooled mean differences of imputed versus non-imputed studies (P=0.38).Figure S14 Forest plot for subgroup analysis by study design, incidence of mechanical ventilation
There is no significant difference between the pooled odds ratios from randomized studies versus non-randomized studies (P=0.39).Figure S15 Forest plot for subgroup analysis by risk of bias ratings, incidence of mechanical ventilation
There is no significant difference between the pooled odds ratios from studies with low/moderate risk of bias versus studies with high risk of bias (P=0.77).Figure S16 Forest plot for subgroup analysis by study design, incidence of adverse events
There is no significant difference between the pooled odds ratios from randomized studies versus non-randomized studies (P=0.88).Figure S17 Forest plot for subgroup analysis by risk of bias ratings, incidence of adverse events
There is no significant difference between the pooled odds ratios from studies with low/moderate risk of bias versus studies with high risk of bias (P=0.74).Figure S18 Forest plot for subgroup analysis by study design, incidence of QT prolongations
There is no significant difference between the pooled odds ratios from randomized studies versus non-randomized studies (P=0.89).Figure S19 Forest plot for subgroup analysis by risk of bias ratings, incidence of QT prolongations
There is no significant difference between the pooled odds ratios from studies with low/moderate risk of bias versus studies with high risk of bias (P=0.79).Figure S20 Bubble plot for meta-regression of cumulative chloroquine base dose, incidence of adverse events
There is significant correlation between the cumulative chloroquine base dose and the treatment effect (P=0.04).Table S13 P-values for meta-regression analyses
Figure S21 Funnel plot, incidence of mechanical ventilation
There is evidence of small study effects based on visual inspection and Egger’s regression test (PEgger=0.04).Figure S22 Forest plot for trim-and-fill analysis, incidence of mechanical ventilationFigure S23 Funnel plot, time to negative conversion of SARS-CoV-2 tests
There is no evidence of small study effects based on visual inspection.Figure S24 Funnel plot, incidence of negative test conversion at day 7
There is no evidence of small study effects based on visual inspection.Figure S25 Funnel plot, incidence of negative test conversion at day 14
There is no evidence of small study effects based on visual inspection.Figure S26 Funnel plot, length of stay
There is no evidence of small study effects based on visual inspection and Egger’s regression test (PEgger=0.26).Figure S27 Funnel plot, mortality
There is no evidence of small study effects based on visual inspection and Egger’s regression test (PEgger=0.83).Figure S28 Funnel plot, time to fever resolution
There is no evidence of small study effects based on visual inspection.Figure S29 Funnel plot, incidence of mechanical ventilation
There is evidence of small study effects based on visual inspection and Egger’s regression test (PEgger=0.04).Figure S30 Funnel plot, incidence of adverse events
There is no evidence of small study effects based on visual inspection and Egger’s regression test (PEgger=0.31).Figure S31 Funnel plot, incidence of severe adverse events
There is no evidence of small study effects based on visual inspection.Figure S32 Funnel plot, incidence of QT prolongations
There is no evidence of small study effects based on visual inspection and Egger’s regression test (PEgger=0.23).

表S1 PRISMA 2020声明检查表[1]
表S2 MEDLINE检索策略
表S3 EMBASE检索策略
表S4 PubMed检索策略
表S5 中国知网（CNKI）检索策略
表S6 万方数据库检索策略
表S7 万方医学网（Wanfang Med Online）检索策略
表S8 中国生物医学文献服务系统（SinoMed）检索策略
表S9 中文科技期刊数据库（CQVIP）检索策略
表S10 纳入研究及受试者特征
表S11 随机对照试验偏倚风险评估工具2.0（Risk of Bias 2, RoB2）
表S12 非随机干预研究偏倚风险评估工具（Risk Of Bias In Non-randomised Studies - of Interventions, ROBINS-I）
图S1 按研究设计分组的亚组分析森林图：严重急性呼吸综合征冠状病毒2（SARS-CoV-2）核酸转阴时间
随机对照研究与非随机研究的合并均数差无显著统计学差异（P=0.64）。
图S2 按补值与未补值研究分组的亚组分析森林图：严重急性呼吸综合征冠状病毒2（SARS-CoV-2）核酸转阴时间
补值研究与未补值研究的合并均数差无显著统计学差异（P=0.05）。
图S3 按治疗方案分组的亚组分析森林图：第7天检测转阴发生率
不同治疗方案亚组间无显著统计学差异（P=0.18）。
图S4 按治疗方案分组的亚组分析森林图：第14天检测转阴发生率
不同治疗方案亚组间无显著统计学差异（P=0.31）。
图S5 按研究设计分组的亚组分析森林图：第7天检测转阴发生率
随机对照研究与非随机研究的合并比值比无显著统计学差异（P=0.08）。
图S6 按研究设计分组的亚组分析森林图：第14天检测转阴发生率
随机对照研究与非随机研究的合并比值比无显著统计学差异（P=0.79）。
图S7 按研究设计分组的亚组分析森林图：住院时长
随机对照研究与非随机研究的合并均数差无显著统计学差异（P=0.98）。
图S8 按偏倚风险等级分组的亚组分析森林图：住院时长
低/中度偏倚风险研究与高度偏倚风险研究的合并均数差无显著统计学差异（P=0.28）。
图S9 按补值与未补值研究分组的亚组分析森林图：住院时长
补值研究与未补值研究的合并均数差无显著统计学差异（P=0.17）。
图S10 按研究设计分组的亚组分析森林图：病死率
随机对照研究与非随机研究的合并比值比无显著统计学差异（P=0.16）。
图S11 按偏倚风险等级分组的亚组分析森林图：病死率
低/中度偏倚风险研究与高度偏倚风险研究的合并比值比无显著统计学差异（P=0.61）。
图S12 按研究设计分组的亚组分析森林图：发热消退时间
随机对照研究与非随机研究的合并均数差无显著统计学差异（P=0.93）。
图S13 按补值与未补值研究分组的亚组分析森林图：发热消退时间
补值研究与未补值研究的合并均数差无显著统计学差异（P=0.38）。
图S14 按研究设计分组的亚组分析森林图：机械通气发生率
随机对照研究与非随机研究的合并比值比无显著统计学差异（P=0.39）。
图S15 按偏倚风险等级分组的亚组分析森林图：机械通气发生率
低/中度偏倚风险研究与高度偏倚风险研究的合并比值比无显著统计学差异（P=0.77）。
图S16 按研究设计分组的亚组分析森林图：不良事件发生率
随机对照研究与非随机研究的合并比值比无显著统计学差异（P=0.88）。
图S17 按偏倚风险等级分组的亚组分析森林图：不良事件发生率
低/中度偏倚风险研究与高度偏倚风险研究的合并比值比无显著统计学差异（P=0.74）。
图S18 按研究设计分组的亚组分析森林图：QT间期延长发生率
随机对照研究与非随机研究的合并比值比无显著统计学差异（P=0.89）。
图S19 按偏倚风险等级分组的亚组分析森林图：QT间期延长发生率
低/中度偏倚风险研究与高度偏倚风险研究的合并比值比无显著统计学差异（P=0.79）。
图S20 氯喹碱基累积剂量与不良事件发生率Meta回归分析气泡图
氯喹碱基累积剂量与治疗效应间存在显著相关性（P=0.04）。
表S13 Meta回归分析P值表
图S21 机械通气发生率漏斗图
通过目视检查及Egger回归检验可发现存在小样本效应（PEgger=0.04）。
图S22 机械通气发生率修剪-填充分析森林图
图S23 严重急性呼吸综合征冠状病毒2（SARS-CoV-2）核酸转阴时间漏斗图
通过目视检查未发现小样本效应。
图S24 第7天检测转阴发生率漏斗图
通过目视检查未发现小样本效应。
图S25 第14天检测转阴发生率漏斗图
通过目视检查未发现小样本效应。
图S26 住院时长漏斗图
通过目视检查及Egger回归检验未发现小样本效应（PEgger=0.26）。
图S27 病死率漏斗图
通过目视检查及Egger回归检验未发现小样本效应（PEgger=0.83）。
图S28 发热消退时间漏斗图
通过目视检查未发现小样本效应。
图S29 机械通气发生率漏斗图
通过目视检查及Egger回归检验可发现存在小样本效应（PEgger=0.04）。
图S30 不良事件发生率漏斗图
通过目视检查及Egger回归检验未发现小样本效应（PEgger=0.31）。
图S31 严重不良事件发生率漏斗图
通过目视检查未发现小样本效应。
图S32 QT间期延长发生率漏斗图
通过目视检查及Egger回归检验未发现小样本效应（PEgger=0.23）。