PRISMA guideline was used for our systematic review.12
3.1. Criteria for considering studies for this review
3.1.1. Types of studies
Studies focusing on mortality, intubation and mechanical ventilation, and cardiac adverse events among patients taking HCQ and AZT compared to patients taking HCQ alone were included.
3.1.2. Types of participants
We included patients diagnosed with COVID-19 who received either HCQ+AZT or HCQ alone.
3.1.3. Types of interventions
Our treatment arm consists of patients taking HCQ+AZT and HCQ alone as a control arm. Patients in both arms received standard of care.
3.1.4. Types of outcome measures
The mortality, intubation and mechanical ventilation, and cardiac adverse effects among the treatment and control group that occurred during treatment were outcomes of interest.
3.1.5. Outcomes
We compared deaths between treatment and control arm, cardiac adverse effects like QT prolongation or ventricular arrhythmia, and intubation and mechanical ventilation requirements between treatment and control arm.
3.2. Search methods for identification of studies
Pubmed, Cochrane Library, Medline, Clinicaltrials.gov, Google Scholar, and WHO clinical trial registry were accessed by our reviewers (PB and DBS) who independently searched and evaluated the quality of the studies from January 1 to June 2, 2020. We filtered the studies using COVIDENCE and extracted data for quantitative and qualitative synthesis. Any potential conflict was solved taking the final opinion of another reviewer (SK). Another reviewer (ER) assessed the risk of bias and cross-checked all the selected studies.
3.2.1. Electronic searches
We have documented the detailed search strategy in supplementary file no. 1
3.3. Data collection and analysis
We extracted the data for quantitative synthesis through COVIDENCE and did the analysis using RevMan 5.3. Assessment of heterogeneity was done using the I-squared (I2) test. We used random/fixed effect for pooling of selected studies
3.3.1. Selection of studies
We have included randomized controlled trials (RCTs), retrospective observational studies, and case series that have a treatment arm of patients taking HCQ+AZT and a control arm of patients taking HCQ alone in addition to supportive care. We excluded the recently retracted paper by Mehra et al which had reported increased cardiac adverse effects among patients with HCQ and macrolide.13 We excluded studies in which the control arm consisted of the standard of care alone without the use of HCQ. Different articles like reviews, retracted papers, in-vitro studies, editorials, letters to editors, protocols, commentaries, viewpoints, and studies done in the pediatric population were excluded.
3.3.2. Data extraction and management
We evaluated the quality of the studies thoroughly and took into account only the outcomes that were of our interest.
3.3.3. Assessment of risk of bias in included studies
We used the Cochrane ROB tool for analysis of our RCTs shown in figure 1. We used the NHLBI (National Heart, Lung, and Blood Institute) quality assessment tool to assess the risk of bias in our retrospective studies, case series, and cohort studies (Tables 1,2). We used Revman 5.3 for creating a summary of biases for RCTs using the Cochrane tool.
Figure 1: Risk of bias assessment
Table 1: NHLBI quality assessment tool for observational cohort and cross-sectional studies
Table 2: NHLBI quality assessment tool for case series
Supplementary table 1 and 2
3.3.4. Assessment of heterogeneity
The I2 test was used for the assessment of heterogeneity. We interpreted the I2 test done based on the Cochrane Handbook for Systematic Reviews of Interventions as follows:-
i) 0% to 40%: might not be important
ii) 30% to 60%: may represent moderate heterogeneity
iii) 50% to 90%: may represent substantial heterogeneity
iv) 75% to 100%: considerable heterogeneity.
“The importance of the observed value of I2 depends on (i) the magnitude and direction of effect and (ii) the strength of evidence for heterogeneity (e.g. P-value from the chi-squared test, or a confidence interval for I2).”
3.3.5. Assessment of reporting biases
Reporting bias was checked by prefixed reporting of the outcome.
3.3.6. Data synthesis
Statistical analysis was performed using RevMan 5.3 software. Risk Ratio (RR)/ Odds Ratio (OR) was used for outcome estimation whenever appropriate with 95% Confident Interval (CI). The fixed/random-effects model was used according to heterogeneities.
3.3.7. Subgroup analysis and investigation of heterogeneity
We used the random effect model in cases of heterogeneity.
3.3.8. Sensitivity analysis
We used the inverse variance method to assess the effect on the results and running the analysis again to see for sensitivity analysis.