Mitjà 2020b.

Study characteristics
Methods	Open‐label cluster‐randomized trial comparing HCQ with standard care when given to individuals with a history of exposure to SARS‐CoV‐2, for prevention of COVID‐19. Follow‐up was up to day 28, using in‐person visits to the participant's home on days 1 and 14, and telephone interviews on days 3, 7, and 28.
Participants	Setting: community; “screened using the electronic registry of the Epidemiological Surveillance Emergency Service of Catalonia (SUVEC) of the Department of Health. During the COVID‐19 outbreak in Catalonia, a public health ordinance required all patients who tested positive for COVID‐19 in any of the designated diagnostic laboratories to be notified to the SUVEC.” Number of participants: 2525 total: 1225 allocated to HCQ; 1300 allocated to standard care. (Note that baseline characteristics and efficacy outcomes use a modified ITT population as their denominator: 1116 HCQ; 1198 standard care. Adverse events are reported for all randomized participants: 1225 HCQ; 1300 standard care.) Inclusion criteria: “adult individuals ≥ 18 years of age with a recent history of close contact exposure to a PCR confirmed COVID‐19 case (i.e., > 15 minutes within two meters, up to seven days before enrolment) and absence of COVID‐19‐like symptoms on the two weeks preceding enrolment, as either a healthcare worker, a household contact, a nursing home worker or a nursing home resident.” Exclusion criteria: symptoms or signs of COVID‐19 at baseline assessment; “all eligibility criteria are listed in the Supplementary Appendix.” (No appendix was available with the preprint publication.) Age: HCQ arm: mean 48.6 (SD 18.7) years; standard care arm: mean 48.7 (SD 19.3) years. Gender: HCQ arm F:M 813:303; standard care arm F:M 875:323. Types of participant: HCQ arm: 131 (12%) healthcare workers; 302 (27%) household contacts; 550 (49%) nursing home workers; 133 (12%) nursing home residents. Standard care arm: 130 (11%) healthcare workers; 338 (28%) household contacts; 584 (49%) nursing home workers; 160 (13%) nursing home residents. (Note that the denominator for the standard care arm is 1212 rather than 1198.) Definition of development of COVID‐19: "confirmed COVID‐19 episode, defined as symptomatic illness (at least one of the following symptoms: fever, cough, difficulty breathing, myalgia, headache, sore throat, new olfactory and taste disorder(s), or diarrhoea) and a positive SARS‐CoV‐2 RT‐PCR test"; “SARS‐CoV‐2 infection, defined as either the RT‐PCR detection of SARS‐CoV‐2 in a nasopharyngeal specimen or the presence of any of the aforementioned symptoms compatible with COVID‐19”. Comorbidities: cardiovascular disease: HCQ: 130 (11.6%) and standard care: 178 (14.9%); respiratory disease: HCQ: 64 (5.7%) and standard care: 47 (3.9%); metabolic disease: HCQ: 99 (8.9%) and standard care: 94 (7.8%); nervous system disease: HCQ: 170 (15.2%) and standard care: 170 (14.2%).
Interventions	HCQ: 800 mg orally on day 1, followed by 400 mg once daily for 6 days. Total 7 days. Standard care: no treatment. Co‐interventions not reported.
Outcomes	Primary outcome: “confirmed COVID‐19 episode, defined as symptomatic illness (at least one of the following symptoms: fever, cough, difficulty breathing, myalgia, headache, sore throat, new olfactory and taste disorder(s), or diarrhoea) and a positive SARS‐CoV‐2 RT‐PCR test. The primary outcome was assessed in all asymptomatic individuals, irrespective of the PCR result; in a post hoc analysis, we explored the outcome in individuals with positive and negative PCR separately. Time‐to‐event was defined as the number of days from the date of randomization/exposure to the confirmed date of the onset of symptomatic illness.” Secondary efficacy outcomes: “incidence of SARS‐CoV‐2 infection, defined as either the RT‐PCR detection of SARS‐CoV‐2 in a nasopharyngeal specimen or the presence of any of the aforementioned symptoms compatible with COVID‐19” “serological positivity (IgM/IgG) of contacts at day 14” Safety outcomes: “frequency and severity of adverse events (AE), serious AE (SAE), and AE of special interest (e.g., cardiac) up to 28 days from treatment start. Causality was assessed by an external panel of pharmacovigilance consultants.” (Note that this included death and hospitalization.)
Notes	Recruitment: 17 March to 28 April 2020. Sponsor/funding: "mainly supported by the crowdfunding campaign JoEmCorono (https://www.yomecorono.com/) with the contribution of over 72,000 citizens and corporations. The study also received financial support from Laboratorios Rubió, Gebro Pharma, Zurich Seguros, SYNLAB Barcelona, and Generalitat de Catalunya. Laboratorios Rubió also contributed to the study with the required doses of hydroxychloroquine (Dolquine®).” Note that LR and GP are pharmaceutical companies. No mention of their involvement in the study, or lack thereof.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	“Randomization was performed remotely by a member of the study team not involved in participants’ enrollment.” No description of sequence generation method.
Allocation concealment (selection bias)	Unclear risk	“Randomization was performed remotely by a member of the study team not involved in participants’ enrollment… The allocation was revealed to participants after providing written consent on day 1 (baseline).”
Blinding of participants and personnel (performance bias) All outcomes	High risk	High for symptomatic confirmed COVID‐19 (primary outcome) and adverse events. Low for antibody positivity. Open‐label study. Due to symptoms being required to define primary outcome, this would be at high risk of bias due to lack of blinding, as would safety outcomes. Antibody positivity at day 14 would not be influenced by knowledge of group allocation.
Blinding of outcome assessment (detection bias) All outcomes	High risk	High for symptomatic confirmed COVID‐19 (primary outcome); composite symptoms without PCR positivity OR PCR‐positive asymptomatic COVID‐19; and adverse events. Low for antibody positivity and death. No blinding. As above, due to symptoms being required to define primary outcome, this would be at high risk of bias due to lack of blinding, as would safety outcomes. Antibody positivity at day 14/death would not be influenced by knowledge of group allocation.
Incomplete outcome data (attrition bias) All outcomes	Low risk	Low for efficacy outcomes. Unclear for adverse events. Exclusions from "intention‐to‐treat (ITT)" (assessed as modified ITT; primary analysis) were < 10%; reasons were reported, and loss to follow‐up was < 5%. Numbers seemed to be balanced between the 2 treatment arms. Comparison of characteristics between those included vs excluded not presented in preprint. This applies to all efficacy outcomes. There was no imputation for missing data. The safety sample included all randomized participants, so there was low risk of bias for the outcomes of adverse events and death. < 3% of participants either did not receive HCQ in the HCQ arm or started HCQ in the control arm. However, denominators were unclear: 1197 vs 1225 in the intervention arm.
Selective reporting (reporting bias)	High risk	Both of the outcomes currently specified in the trial registry entry (clinicaltrials.gov/ct2/show/NCT04304053) were included in the report. However, disease in contacts of contacts was also specified and is not reported, with no reason provided.
Other bias	High risk	Additional domains for cluster‐RCTs: Recruitment bias: low risk. Appears unlikely, as the rings (clusters) were randomized first, and then the contacts were told their allocation. Baseline imbalance: low risk. No stratified or pair‐matched randomization. Baseline characteristics not disaggregated by cluster. But many clusters, so unlikely to lead to baseline imbalance. Loss of clusters: low risk. No clusters lost. Incorrect analysis: low risk. The analysis accounted for clustering. Comparability with individually randomized trials: high risk. Contamination possible, as this was an open‐label study, and people within clusters may encourage differential adherence to intervention. However, reported adherence was > 95%. This intervention would be expected to work best when given to all contacts of a case rather than some being randomized to the intervention and some randomized to no intervention, which would preclude comparability with an individually randomized trial.