1. Risk of bias assessment methods for cohort studies.

Bias	Authors' judgement	Support for judgement
Confounding	Low risk Moderate risk Serious risk Critical risk No information	We used the following criteria: Low risk: identified confounders were measured and were balanced across groups (age, sex, destination and duration of travel) Moderate risk: identified confounders were measured and not balanced across groups, or several confounders had not been measured or not reported across groups Serious risk: a critical confounder has been measured and is not balanced across groups
Selection of participants into the study	Low risk Moderate risk Serious risk Critical risk No information	We assessed whether selection into the study was unrelated to intervention or unrelated to outcome, and whether start of intervention and start of follow up coincided for most subjects. Non‐responder bias at the point of selection was considered here for cohort studies. We used the following cut offs for non‐response rate: low risk < 10%, moderate risk 10% to 20%, serious risk > 20%.
Measurement of interventions	Low risk Moderate risk Serious risk Critical risk No information	We used the following criteria: Low risk: the prescription was provided by a travel clinic which also performed the study, and discontinuations were recorded and reported, or all participants were issued with their medication e.g. soldiers or participants were asked to self‐report which medication they took whilst they were taking it. Moderate risk: the prescription was provided by a travel clinic which also performed the study but no information regarding switches and discontinuations was available or patients are asked to self‐report which prophylaxis they took shortly after they finished taking it. Serious risk: Participants were asked to self‐report which prophylaxis they took a long time after they finished taking it.
Departures from intended interventions	Low risk Moderate risk Serious risk Critical risk No information	We assessed whether switches between interventions of interest were available. We assessed whether discontinuations and switches between prophylactic regimens had been recorded and reported.
Missing data	Low risk Moderate risk Serious risk Critical risk No information	We assessed whether outcome data was reasonably complete for most participants. We recorded missing data for included participants e.g. loss to follow up rates and treatment withdrawals.
Measurement of outcomes	Low risk Moderate risk Serious risk Critical risk No information	We assessed whether the outcome measure was objective or subjective. We assessed whether participants or study personnel were blinded to the intervention received. We assessed whether the methods of outcome assessment were comparable across intervention groups.
Selection of the reported result	Low risk Moderate risk Serious risk Critical risk No information	We used the following criteria: Low risk: If the questionnaire was provided in full, or it was clear what was asked within it. Moderate risk: If it is unclear which questions are asked, or information was provided on aggregate. Serious risk: If data captured within the questionnaire was clearly missing.
Other	Low risk Moderate risk Serious risk Critical risk No information	We reported the study sponsor. We classified the analysis of studies sponsored by pharmaceutical companies as independent of the sponsor when it was clearly stated that the sponsor had no input to the trial analysis.

Adapted from Higgins 2011 and ACROBAT‐NSRI tool