Ansah 2015.

Study characteristics
Methods	Trial design: cRCT Unit of randomization: community with ≥ 1 chemical shop Number of clusters: 24 (12 per arm) to obtain 80 adults and 114 children per cluster Data collection: seller kept records on test results, medications dispensed, and whether customer was referred. Slide for parasitology reading at laboratory. Length of follow‐up: 17 months Adjustment for clustering: yes
Participants	Target treatment group: adults and children > 6 months Sample size: 4208 (2719 intervention, 2029 comparison) Exclusion criteria: pregnancy, age < 6 months, signs of severe disease, prescription from health facility
Interventions	Staff who received training: chemical sellers Duration of training: 1 day over and above standard 3‐day malaria case management training for both groups Content of training: treating Pfalciparum malaria after positive mRDT with AL, AQAS, or DP, and referring after negative mRDT Supervision: fieldworkers and supervisors provided technical support. Accuracy of records of drugs dispensed validated by random checks of forms and 'mystery clients'. Direct observation of interactions between chemical sellers and customers by checklist on weekly basis for first month and a further week midway through trial Antimalarials free to participants: no, but subsidized through the Affordable Medicines Facility malaria mRDTs free to participants: yes Additional details: chemical sellers can also sell analgesics, antibiotics (co‐trimoxazole), multivitamins/minerals, haematinics, and antacids.
Outcomes	All‐cause mortality and malaria mortality (risk of bias combined), use of antimalarial when microscopy‐negative, appropriate treatment (defined as antimalarial provision to microscopy‐positive participants and no antimalarial provision to microscopy‐negative participants), number receiving an antimalarial
Notes	Control: chemical sellers dispensing medicines without test results (community‐based treatment of suspected malaria by clinical diagnosis) Country: Ghana Setting: rural Malaria endemicity: not stated Study dates: August 2011–January 2013 Study sponsor: the Malaria Capacity Development Consortium of the London School of Hygiene & Tropical Medicine, with funding from the Welcome Trust and the Bill and Melinda Gates Foundation
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Low risk	Used a programme written in R by a statistician not otherwise involved.
Allocation concealment (selection bias)	Unclear risk	Not described.
Blinding of participants and personnel (performance bias) Mortality	Low risk	Not blinded, but mortality unlikely to be affected by knowledge of intervention.
Blinding of participants and personnel (performance bias) Antimalarial prescribing to microscopy‐ or PCR‐negative people, appropriate treatment	Unclear risk	Not blinded, but unclear how this could affect outcome.
Blinding of participants and personnel (performance bias) Number receiving an antimalarial	Unclear risk	Not blinded, but unclear how this could affect outcome.
Blinding of outcome assessment (detection bias) Mortality	Unclear risk	Not blinded, but unclear who assessed mortality.
Blinding of outcome assessment (detection bias) Antimalarial prescribing to microscopy‐ or PCR‐negative people, appropriate treatment	Low risk	Those preparing and reading slides were blind to allocation and mRDT result.
Blinding of outcome assessment (detection bias) Number receiving an antimalarial	Unclear risk	Not blinded, but unclear how this could affect outcome.
Incomplete outcome data (attrition bias) Mortality	Low risk	1 cluster closed in control arm (8%), and some people refused consent in control, through unclear why. 91.3% positives were successfully followed up. Percentage of participant samples analysed was not similar in both arms.
Incomplete outcome data (attrition bias) Antimalarial prescribing to microscopy‐ or PCR‐negative people, appropriate treatment	Low risk	1 cluster closed in control arm (8%), and some people refused consent in control, through unclear why. 91.3% positives were successfully followed up. Percentage of participant samples analysed was not similar in both arms.
Incomplete outcome data (attrition bias) Number receiving an antimalarial	Low risk	1 cluster closed in control arm (8%), and some people refused consent in control, through unclear why. 91.3% positives were successfully followed up. Percentage of participant samples analysed was not similar in both arms.
Selective reporting (reporting bias)	Low risk	Reporting appears as per the statistical analysis plan. Additional outcomes were registered on Clincaltrials.gov, but are not relevant to this review.
Other bias	Unclear risk	Recruitment after randomization, which may have influenced selection, though little baseline imbalance. Arms similar, though control sellers more likely to have formal training in dispensing, range differences for number of customers per cluster, refresher training. Contamination and uptake of mRDTs were not described.