White 2018.

Study characteristics
Methods	Design: controlled before‐after study Unit of randomization: none
Participants	Inclusion criteria: children aged < 5 years and women aged 18–49 years within selected households located beyond 5 km from the nearest health facility Exclusion criteria: households and respondents who did not participate or were not available were not replaced
Interventions	Intervention Recruiting and training lay health workers – CHW – to provide iCCM for diarrhoea, malaria, pneumonia (ARI) and malnutrition, including an active case finding approach. iCCM providers were also trained on community engagement, household registration, community mapping and how to conduct household visits, focusing on child health – with the expectation that they would visit every household in their catchment area at least once per month Implementing simplified IMCI‐adapted clinical guidelines for iCCM providers, including an active case finding approach Providing iCCM providers a monthly cash incentive of USD 70 for approximately 20 hours of work per week, additional compensation for training (daily subsistence allowance and travel expenses) Providing iCCM providers with iCCM drugs and equipment Providing iCCM providers and their supervisors with paper and mobile health tools to assist in workflow, help guide clinical decision‐making and collect programmatic data Providing iCCM providers with visual job aids to enable the correct assessment, diagnosis and treatment of children aged < 5 years correctly Providing iCCM providers with supervision (CHW leaders were recruited, trained and paid (USD 220 per month) to provide weekly supervision; and Community Clinical Supervisors were recruited – from nurses, physician assistants and midwives – trained and paid (USD 313 per month) to provide monthly supervision) Comparison Usual facility services in the 3 control districts in Rivercess County: Doedain, population 13,051; Jo River, population 13,900; Timbo, population 19,776. As context the study indicated that gCHV were trained to provide iCCM in both intervention and control districts but actual provision of iCCM by gCHVs was minimal (i.e. careseeking to gCHVs was < 3% at baseline and 0% at endline in both intervention and control districts, see Table 3, page 1257). In terms of health services, the main difference between the intervention and control districts was the intervention described in the study
Outcomes	Objective: to assess whether the programme increased treatment of fever, diarrhoea and ARI compared with a control area during the 1‐year implementation period. Implementation date: August 2015 to July 2016. Location: the study was set in 6 districts of Rivercess County, Liberia. Rivercess County had a population of about 71,000 and was the poorest county in Liberia, with 71.3% of its population within the lowest wealth quintile of the country. Rivercess County also had among the lowest treatment rates for childhood illness and the highest proportion of women describing distance to health facility as a barrier to accessing health care. 3/6 districts were intervention districts (Central C, population 8303; Jowein, population 8921; Yarnee, population 7568) and the remaining 3 districts were control districts. Funding source: Direct Relief and the UBS Optimus Foundation.
Notes
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. Districts were purposefully selected.
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 the paper.
Incomplete outcome data (attrition bias) All outcomes	Unclear risk	Response rates were high: quote: "97.2% in 2015 and 98.4% in 2016 resulting in 455 and 539 surveys, respectively. Within eligible households, 82.2% of listed women participated in 2015 and 84.5% in 2016 (549 and 604 surveys); information about 97.5% of listed children was provided in 2015 and 99.3% in 2016, (340 and 492 surveys). Less than 3% of data items were missing." There was no indication of systematic differences between arms.
Selective reporting (reporting bias)	High risk	Assessing the effect of the intervention on under‐5 mortality was a primary outcome and data were collected. The authors provided the following explanation: quote: "Although we collected data on early childhood mortality rates in both surveys, we were underpowered to detect mortality differences in the timeframe observed." P. 1258.
Baseline outcomes similar	Unclear risk	Risk was unclear. Baseline outcomes were not balanced between intervention and control groups per Table C in Appendix E (online supplementary material). Baseline coverage was higher in the control group for careseeking to an appropriate provider for any illness; careseeking to an appropriate provider for fever; careseeking to an appropriate provider for ARI; and ORT treatment for children with diarrhoea. The authors used a difference‐in‐difference approach adjusted by inverse probability weighting to deal with this type of imbalance; however, the residual risk of bias was unclear.
Baseline characteristics similar	Unclear risk	Risk was unclear. The author's stated, "Overall, the samples were similar (Table 1); however, households in the intervention areas were farther from the nearest health facility than were those in the control areas at both time points. More households in the intervention group were in mining communities and more respondents in the intervention areas completed the survey in English than in the control group. In all groups, IPT weighting produced approximate balance, as seen by decreased standardized differences from the baseline control group. We present full IPT weighting balance diagnostics and an IPT‐weighted version in Appendix C, Table A (available as a supplement to the online version of this article at http://www.ajph.org)." P. 1254. Furthermore, the authors stated, "Our study had several limitations. First, community mapping for the 2015 sampling frame was incomplete, which challenged the comparability of the baseline and follow‐up samples. We used 2 approaches to improve balance between groups and time points: (1) IPT‐weighted modeling and (2) regression adjustment. Results were similar with both approaches … After we applied IPT weights, no covariates had sufficiently different before‐to‐after differences between the intervention and control areas to explain the observed effect on childhood treatment (discussed in Appendix C, available as a supplement to the online version of this article at http://www. ajph.org). However, IPT weighting only corrects shifts in measured confounders, so unmeasured confounders may remain." P. 1257.
Contamination	Low risk	Prior to the study (and through a mechanism not related to the study) a cadre of volunteer lay health workers called gCHVs had been trained on iCCM and deployed to implement it in both the intervention and control districts. The authors stated, "In response to Liberia's poor maternal and child health outcomes, Last Mile Health, a nongovernmental organization, partnered with the Liberia Ministry of Health to implement a CHW programme, which included an iCCM component, in 2 counties in Liberia." (P. 1252). This was the intervention described in the study. The authors indicated that, "This program built upon Liberia's existing "general community health volunteer" programme, which included iCCM but lacked systematic supervision, supply chain systems, and monetary incentives." (P. 1252). These volunteer gCHVs continued to implement iCCM in both the intervention and control districts however implementation was weak, if not negligible, as indicated by the authors in their statement and as evidenced by the results of careseeking at baseline and endline (Table 3, P. 1257). At baseline 2.3% of caregivers in the intervention districts and 2.7% of caregivers in control districts sought treatment from gCHVs. At endline, 2.7% of caregivers in intervention districts and 0% of caregivers in control districts sought treatment from gCHVs in control districts. Since implementation was weak, the effect in terms of coverage negligible, and the fact that gCHVs were in both intervention and control districts, the risk of contamination by the gCHVs is low. The authors also indicated that their study informed the "development of a national‐scale, government‐led program called the National Community Health Assistant (CHA) Program, which uses a cadre of workers called CHAs performing similar duties as the CHWs in this study, which was launched by the Ministry of Health in 2016." (P. 1252). The risk of the CHA contaminating the study is low since it was launched in the areas targeted by the study only after the study was completed.
Other bias	Low risk	No other risks of bias were detected.