BACKGROUND: Low- and middle-income countries (LMICs) are scaling up community health worker (CHW) programmes. Research is needed to understand how CHWs can be integrated into, and supported by, health systems and communities, including evaluation of different approaches to delivering CHW services. This thesis synthesises the evidence, quantifies the impact, and evaluates the process of a proactive CHW workflow designed to reduce treatment delays and under-five mortality. METHODS: We first conducted a systematic review of the effects of proactive case-finding home visits by CHWs in LMICs on mortality, morbidity, and access to care for common childhood illnesses. We then evaluated the effects of proactive CHW service delivery at patients’ homes compared to passive CHW service delivery at fixed village sites in a cluster randomised trial in rural Mali. The primary outcome of the trial was mortality among children under five years of age. The main secondary outcomes pertained to children’s health care utilisation, measured at baseline and 12, 24, and 36 months of follow-up. We conducted a mixed method process evaluation alongside the trial, with embedded realist approaches, to evaluate implementation, mechanisms, and context to explain trial results between and across arms. RESULTS: Our systematic review of 14 reports of 11 interventions found that proactive CHW home visits may improve treatment coverage (RR: 1.59–4.64; low certainty evidence) but effects on prompt treatment and under-five mortality were uncertain, due to limitations in study designs, indirect measures of effect, and unexplained heterogeneity. Our trial found that CHW home visits had no effect on under-five mortality compared to site-based delivery by CHWs. After 12 months, sick children had 22% higher odds of health sector treatment within 24 hours of symptom onset in intervention compared to control clusters (95% CIs: 1.06, 1.41), but no difference at 24 or 36 months. Over all three years, we found modest improvements in children’s health sector consultation in intervention compared to control clusters (aOR=1.12; 95% CIs: 0.99, 1.26). In both arms combined, under-five mortality fell from 148.4 to 55.1 deaths per 1000 live births and 4 prompt health sector treatment more than doubled compared to baseline. Our process evaluation showed that user fee removal, professional CHWs, and upgrades to primary clinics—all in both trial arms—enabled providers to offer acceptable, quality services and trial participants to seek prompt care. In this context, proactive home visits improved access via mechanisms that had already been activated. CONCLUSION: Proactive home visits may accelerate access to care, but user fee removal, professional CHWs, and systems strengthening at primary clinics are foundational to achieving UHC and child survival goals.