Mubiru 2015.

Study characteristics
Methods	Design: controlled before‐after study Unit of randomization: none
Participants	Inclusion criteria: children aged < 5 years, heads of households and caregivers of children aged < 5 years, and women of reproductive age (15–49 years of age) in intervention and comparison districts Exclusion criteria: none reported
Interventions	Intervention Training lay health workers – existing VHT members – to provide iCCM for diarrhoea, malaria and pneumonia (ARI) among children aged 0–59 months Implementing simplified IMCI‐adapted clinical guidelines for iCCM providers (VHT members) Providing lay health workers (VHT members) with incentives, including transport refund and meals during quarterly meetings Implementing referral of children with severe disease to health facilities Providing iCCM providers with iCCM drugs and equipment Providing iCCM providers (VHT members) with supervision; frequency of supervision provided as part of the intervention not reported; however. the study monitored the percent of VHT members who received quarterly supervision; content and approach to supervision not reported Implementing radio spots promoting careseeking Training community leaders to sensitize communities about the work of iCCM providers (VHT members) Comparison Usual facility services
Outcomes	Mortality Under‐5 mortality Coverage of appropriate treatment by an appropriate provider Coverage of appropriate treatment (ACT) for malaria (study took fever as presumed malaria) from an appropriate provider Coverage of appropriate treatment (antibiotics) for pneumonia from an appropriate provider Coverage of appropriate treatment (ORS and zinc) for diarrhoea from an appropriate provider Coverage of careseeking to an 'appropriate provider' of treatment services Coverage of careseeking for treatment services for fever Coverage of careseeking to an appropriate provider of treatment services for fever Coverage of careseeking for fever within 24 hours Coverage of careseeking for treatment services for suspected pneumonia Coverage of careseeking for treatment services for suspected pneumonia Coverage of careseeking for suspected pneumonia within 24 hours Coverage of careseeking for diarrhoea Coverage of careseeking to an appropriate provider of treatment services for diarrhoea
Notes	Objective: to evaluate the effects of iCCM on care seeking behaviour and treatment, 2 years after it has been introduced. Implementation date: July 2010 to December 2012. Location: 3 districts (Masaka, Mpigi and Wakiso) which in 2011 were divided into 8 districts by the government of Uganda (Wakiso, Mpigi, Butambala, Gomba, Masaka, Lwengo, Bukomansimbi and Kalungu). The majority of participants (≥ 67%) lived in rural areas. Funding source: Department of Foreign Affairs Trade and Development Canada through a grant administered by UNICEF.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	High risk	Controlled before‐after study, with no random sequence generation.
Allocation concealment (selection bias)	High risk	Controlled before‐after study, with no allocation concealment.
Blinding of participants and personnel (performance bias) All outcomes	High risk	No blinding of participants and personnel. Lay health workers would have known if they received additional training and this may have biased their performance. Allocation was by village and parents may have known that the health workers at their primary health centre had received additional training and this may have biased their care seeking behaviour or responses to questionnaires, or both.
Blinding of outcome assessment (detection bias) All outcomes	High risk	Blinding of outcome assessors not described in paper.
Incomplete outcome data (attrition bias) All outcomes	Low risk	The number of participating households was increased (from 2080 to 8000) between baseline and endline assessment. The response rate in both assessments were high: 99% (2076/2080) of eligible households participated at baseline and 97% (7734/8000) of eligible households participated at endline.
Selective reporting (reporting bias)	High risk	The outcomes listed in the objective of the paper were presented in the tables. However, grey literature indicates under‐5 mortality was an original objective and that this was collected. The paper substantiated this by indicating a birth history was collected; however, the outcomes on mortality were not reported.
Baseline outcomes similar	High risk	There were some differences in baseline outcomes. Higher prevalence of careseeking for fever, ARI and diarrhoea in the control. Higher % of careseeking within 24 hours (timeliness of careseeking) in the control. Higher % of appropriate treatment for fever and diarrhoea in the control. Higher prevalence of fever, ARI and diarrhoea in the control which may have affected careseeking and treatment.
Baseline characteristics similar	High risk	There were some differences in baseline characteristics. Higher % rural population in control areas. Higher mean household size in control areas. Lower % of "least poor" households based on a household asset index in control areas. Higher % of caregivers with no education in control areas.
Contamination	Low risk	Low risk of contamination due to districts being the unit of analysis and size of districts. VHTs in control areas were not trained on iCCM or provided with commodities for treatment.
Other bias	High risk	6/11 authors had UNICEF affiliations and UNICEF advocates iCCM. The endline survey in the control areas occurred in the dry season whereas the baseline survey for control areas and both the baseline survey and endline survey for the intervention areas were in the rainy season. Ebola may have affected implementation of iCCM, particularly for fever, in the intervention areas.