Terrell 2015.

Methods	Design: cross‐sectional cohort study Study dates: 2012 and 2013 Malaria transmission pattern and local antimalarial drug resistance: "...high risk of malaria (mainly P. falciparum) in Kenya, although the risk is assessed as very low in Nairobi and in the highlands above 2,500 m... widespread resistance to chloroquine" Adverse event monitoring: "...questionnaire‐based, two‐arm cohort study"
Participants	Number enrolled: 2032 completed questionnaires available, 220 failed to indicate which drug they were taking Inclusion criteria: all military personnel on deployment to Kenya who travelled on one of three main body flights on their return to the UK Exclusion criteria: none mentioned Factors influencing drug allocation: "...the choice of drugs considered in this study was limited to mefloquine or doxycycline... participants were free to use another drug should they experience unacceptable adverse effects or where there was an occupational reason" Country of recruitment: UK Country of malaria exposure: Kenya Duration of exposure to malaria: "The majority of participants spent approximately 6 weeks in Kenya with a small number spending a few weeks longer if they filled an administrative role" Type of participants: military
Interventions	Included in review comparisons: 1. Mefloquine* 2. Doxycycline* Not included in review comparisons: 3. Atovaquone‐proguanil* (results not included in the analysis) *dosing regimen not specified
Outcomes	Included in the review : 1. Adverse effects; any 2. Measure of adherence to the drug regimen Outcomes assessed not included in the review: 3. Clinical cases of malaria 4. Impact of adverse effects on self‐reported ability to work
Notes	Funding sources: "The research was not sponsored by any external body" After we submitted the review for peer referee, the author sent us a spreadsheet containing numbers of events relating to a variety of symptoms after the review had been submitted for publication. These data are not included in the review and will require some clarification over how they were collected to allow us to assess risk of bias. This additional information will be considered in future updates.
Risk of bias
Bias	Authors' judgement	Support for judgement
Other bias	Unclear risk	1. Confounding: moderate "Although not formally recorded, each unit can be assumed to be composed of similar populations in terms of number, age, gender, occupation, and general health" 2. Selection of participants into the study: serious "Completion rates were consistently poor throughout the study period with only 150 to 250 questionnaires returned per tranche of around 1,000 troops" 3. Measurement of interventions: low Participants were asked to self‐report which medication they were on while still taking the medication" 4. Departures from intended interventions: moderate "...[participants] were invited to complete the questionnaire for whichever drug they took for the longer period" 5. Missing data: moderate "2,032 completed questionnaires available for analysis of which 10.8% (220) failed to indicate which drug they were taking" 6. Measurement of outcomes: serious The outcome measure was subjective; participants and personnel were not blinded 7. Selection of the reported results: serious "In both arms, some participants indicated that they had experienced an adverse effect but did not report how it had impacted upon their ability to work. They were excluded from the final analysis" Mefloquine: 71 participants, doxycycline: 67 participants 8. Other: low "The research was not sponsored by any external body"