The search strategies yielded different studies, and after removing duplicates, 6277 studies remained. We selected 188 studies that had a high probability of meeting our inclusion criteria for a complete examination (Fig. 1). After completely examining these references, 92 studies met our eligibility criteria and therefore were included in this review.
A total of 96 studies were excluded for the following reasons: no AE was evaluated (n = 38), no control group as placebo or non-comparator for CQ or HCQ (n = 7), non-RCT (n = 4), no report of the outcomes per group studied (n = 6) (Supplementary file), studies are still ongoing (n = 25) (Supplementary file), and unevaluated eligibility criteria (n = 16) (Supplementary file).
Study Characteristics
Out of 92 studies included, eight were on COVID-19 [21–28], 13 studies were on malaria [29–41], 11 on other infectious conditions [42–53], 31 were on rheumatology [54–84], four on dermatologic diseases [85–88], eight on cancer [89–96], five were on metabolic disease [97–101], and the remaining were on other conditions [102–112]. Fifty-seven studies used HCQ (11,274 participants) as intervention and 35 used CQ (6,997 participants). Most studies used placebo as comparator, seven studies were on children [29, 30, 32, 35, 92, 38, 56], two on pregnant women [76, 41], and the others were on adult population. Most studies used daily intervention dose ≥ than 500 mg for CQ and ≥ than 400 mg for HCQ.
In most of the included studies, the participants had chronic conditions, and, consequently, the intervention and follow up were beyond 4 weeks. Meanwhile, the total sample size was higher in studies on acute conditions, with follow up less than 4 weeks. Table 2 (supplementary file) presents descriptive data of all the studies included.
Meta-analysis
Regarding primary outcomes, there is no evidence to support the difference between CQ/HCQ and control group (placebo or non-CQ/HCQ) with regard to the frequency of SAE (OR: 0.98, 95% CI: 0.71–1.36, 25 studies, 11,605 participants, moderate certainty of evidence, Table 1, Fig. 3). Forty-five BAOE studies with 4,503 participants were excluded from this analysis (Fig. 3).
Table 1
Summary of findings according to GRADE approach. CQ/HCQ compared to Placebo or no CQ/HCQ for malarial and non-malarial conditions.
Outcomes | Anticipated absolute effects* (95% CI) | Relative effect (95% CI) | Number of participants (Number of studies) | Certainty of the evidence (GRADE) | Comments |
Risk with Non CQ/HCQ | Risk with CQ/HCQ |
SAE | 12 per 1.000 | 12 per 1.000 (8–16) | OR 0.98 (0.71–1.36) | 11605 (25 RCTs) | ⨁⨁⨁◯ MODERATE a | CQ/HCQ likely does not increase SAE. |
Retinopathy | 18 per 1.000 | 28 per 1.000 (7–105) | OR 1.63 (0.40–6.57) | 344 (5 RCTs) | ⨁◯◯◯ VERY LOW b,c | The evidence is very uncertain about the effect of CQ/HCQ on retinopathy. |
Cardiac complications | 26 per 1.000 | 39 per 1.000 (29–52) | RR 1.48 (1.10–1.98) | 5970 (8 RCTs) | ⨁⨁◯◯ LOW d | CQ/HCQ may result in a slight increase in cardiac complications. |
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI). CI: Confidence interval; OR: Odds ratio; RR: Risk ratio; SAE: Serious adverse events; CQ: Chloroquine; HCQ: Hydroxychloroquine; GRADE: Grading of Recommendations Assessment, Development, and Evaluation |
GRADE levels of evidence High certainty: We are very confident that the true effect lies close to that of the estimate of the effect Moderate certainty: We are moderately confident in the effect estimate: The true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different Low certainty: Our confidence in the effect estimate is limited: The true effect may be substantially different from the estimate of the effect Very low certainty: We have very little confidence in the effect estimate: The true effect is likely to be substantially different from the estimate of effect |
Explanations |
a. The quality of evidence was rated down due to non-inclusion of unpublished data. |
b. According to RoB 2, the studies included were classified as ‘some concerns’ or ‘high risk’ of bias due to the following reasons: missing data, no mention of the method used to measure retinopathy, and no information regarding allocation concealment. |
c. Wide confidence interval. Considering a prevalence of 7.8% of retinopathy in non-diabetic population (Klein,1992), and a 1% of retinopathy risk in the first 5 years of HCQ treatment ( Petri 2020), the minimum sample size required for the detection of this outcome is 4533 (level of significance = 5%; power (1- β) = 80%). The optimal information size criterion was not met, and the quality of evidence was rated down by two levels for imprecision |
d. According to RoB 2, most of the studies included were classified as high risk of bias due to the following reasons: missing data, no mention of the method used to measure cardiac arrhythmia, and no information on if the participants of both groups were subjected to same method and frequency of outcome evaluation |
Regarding the association between CQ/HCQ and the frequency of retinopathy, the evaluation of the risk of bias (most studies did not mention the method used to evaluate this outcome) and imprecision (wide confidence interval, no achievement of optimal information size) did not indicate any clear effect (OR: 1,63, 95% CI -0.4–6.57, 5 studies, 344 participants, very low certainty of evidence, Fig. 4, Table 1). Twenty BAOE studies with 1,559 participants were excluded from this analysis (Fig. 4).
CQ/HCQ may have a small effect on cardiac complications; however, due to missing data and the concern that participants were not subjected to the same method of investigation, the certainty of evidence was low (RR:1.48, 95% CI: 1.1 to 1.98, 8 trials, 5,970 participants, Fig. 5, Table 1). Four BAOE studies with 879 participants were excluded from this analysis (Fig. 3). The complications reported were cardiac arrythmia and prolongation of QTc interval. Two studies reported these complications as SAE (a prolonged QT interval with ventricular arrhythmias, and a case of torsades de pointes) (26)[71]. Two studies were on pharmacokinetic analysis, and all participants performed serial electrocardiograms [40, 37]. Two studies were on COVID-19, and the screening method for the complication was not mentioned. However, one of the studies stated that fewer serial electrocardiograms were performed for patients in the control group than for patients in the HCQ group [22].
For the secondary outcomes, the administration of CQ/HCQ increases the incidence of total AE (RR 1.39, 95% CI: 1.19–1.62, 45 studies, 9,428 participants, supplementary file), nausea/vomiting (RR 1.89 95% CI: 1.48–2.42, 22 studies, 6096 participants, supplementary file), diarrhea (RR 1.64, 95% CI: 1.15–2.36, 19 studies, 5,239 participants, supplementary file), withdrawal due to AE (RR 1.38, 95% CI: 1.11–1.72, 50 studies, 7,760 participants, supplementary file), headache (RR 1.56, 95% CI: 1.11–2.2, 22 studies, 6,131 participants, supplementary file), and dermatological affections (RR 1.61, 95% CI: 1.1–2.33, 16 studies, 3,682 participants, supplementary file). There was no clear evidence to support a difference between the CQ/HCQ and control group with regard to visual and auditory symptoms (RR 1.5, 95% CI: 0.88 to 2.53, 26 studies, 6,6758 participants; RR 1.64, 95% CI: 0.94 to 2.79, 10 studies, 4,687 participants, respectively, supplementary file). Only two studies reported myopathy as AE, and no difference was found between the groups.
For SAE, although more than ten studies were included in the meta-analysis, we could not evaluate publication bias by funnel plot or Egger test because all the studies showed no statistically significant difference between groups (CQ/HCQ versus placebo or non-CQ/HCQ). Nevertheless, the quality of evidence in this domain was rated down due to the non-inclusion of unpublished data.
The subgroup analysis according to type of intervention, patient diagnosis, type of population, daily dosage, and time of follow up did not indicate that CQ/HCQ increased the frequency of SAE (supplementary file). However, for the subgroup analysis of studies (not for subsets of participants), with 3–4 studies in the smallest subgroup, all the effect modification analyses were classified as having low credibility.
In the sensitivity analyses (according to overall risk of bias, placebo or non-CQ/HCQ, sample size), a ‘no true’ CQ/HCQ effect on SAE was observed (supplementary file).