What actually saves children at the doorstep

In Bankass, Mali, the under-5 mortality rate fell from about 150 to about 55 deaths per 1000 live births, a drop of around 60%, over the course of a trial of community health workers (CHWs) ¹. This drop happened in both arms of the trial, after user fees were removed, CHWs were professionalised and health facilities were upgraded everywhere ¹.

It is a before-and-after comparison, not a randomised effect. But it points to a clear lesson. The foundation of a community health system seems to matter more than how often CHWs knock on doors.

The burden

Nearly 5 million children still die each year. In 2024, an estimated 4.9 million children died before their fifth birthday ⁹. The global under-5 mortality rate was 37.4 deaths per 1000 live births ⁹. Of these deaths, 2.3 million (47%) were newborns ⁹.

Most deaths are from causes that CHWs can treat or prevent. WHO lists the leading causes as preterm birth complications, birth asphyxia and trauma, pneumonia, diarrhoea and malaria, all of which are largely preventable or treatable ⁹. After the first month of life, infectious diseases dominate, and malaria alone causes 17% of deaths ¹⁰. The March 2026 UN report also counted more than 100,000 direct deaths from severe acute malnutrition among children aged 1–59 months ¹⁰.

Where the deaths are. Sub-Saharan Africa had 2.8 million under-5 deaths in 2024 (58%) and Southern Asia 1.2 million (25%) ⁹. Sub-Saharan Africa's rate, 71.6 deaths per 1000 live births, is nearly 14 times higher than in Europe and Northern America ⁹. Children born in fragile and conflict-affected settings are nearly three times more likely to die before age 5 than children elsewhere ¹⁰.

Access is the gap. In Mali in 2018, the under-5 mortality rate was 101 per 1000 live births, and some regions were as high as 152 ¹. In Zambia, care is not sought for roughly 40% of children under 5 with fever ⁴. Before the Madagascar trial described below, fewer than 25% of people with fever sought care or got a malaria test ⁵. Distance, user fees and missing health workers and supplies all keep children from timely care ¹.

The trend. Under-5 deaths have fallen by more than half since 2000 ¹⁰. But since 2015 the pace has slowed by more than 60% ¹⁰. The total has stayed at about 4.8–4.9 million a year since 2022 ⁹. UNICEF warns that declining aid is putting pressure on maternal and child health programmes ¹⁰.

How it works

What a CHW does. A CHW is a trained member of the community who gives basic care close to home. Under integrated community case management (iCCM), CHWs diagnose and treat common killers of young children close to home, and refer children with danger signs to a health centre ¹.

In the Mali trial, for example, CHWs treated malaria, diarrhoea and pneumonia in children aged 2–59 months, and moderate acute malnutrition in children aged 6–59 months ¹. They measured arm circumference and referred children in the "red zone" to the health centre ¹. They had a written tool listing danger signs that needed immediate referral, such as difficulty breathing or convulsions ¹.

Fixed-site versus proactive care. There are two ways to organise this work:

• Fixed-site (passive) care. The CHW works from a fixed place in the village. Families bring sick children to the CHW.
• Proactive care (ProCCM). The CHW goes door to door on a schedule, looking for sick people. In Mali the goal was at least two visits per household per month ¹. In Zambia it was weekly ⁴.

The idea behind proactive care is simple: find sick children earlier, before they get worse.

What "professional" CHWs means. In the Mali trial, all CHWs in both arms were trained, paid a salary, supervised monthly and regularly supplied ¹. They worked about 4 hours a day, 6 days a week, and each served about 700 people, in line with Mali's national strategy ¹. Their salary was 40,000 CFA a month (about US$71 in August 2018), the same as the Malian health ministry pays government CHWs ¹.

What the evidence shows

2013–2021: access improved more consistently than survival

Ghana's Newhints trial, published in 2013, randomised 98 zones to pregnancy and newborn home visits or usual care. Families sought facility care for 77% versus 55% of sick newborns, but neonatal mortality did not significantly differ (risk ratio 0.92, 95% CI 0.75–1.12) ¹⁹. Better care-seeking was measurable; a survival gain in that trial was not established.

In Ethiopia, a 2016 cluster trial in 31 districts found that training and supplying community workers did not generate enough demand to raise treatment coverage. Mortality declined similarly in both arms ¹⁸. This is an important counterpoint to the later national success story: a country-level trend cannot establish the effect of a particular programme expansion.

An influential 2018 interrupted time-series study in periurban Mali followed a package of proactive visits, doorstep treatment, supervision, fee removal and stronger clinics. It reported under-five mortality of 7 per 1,000 in 2015 ¹⁶. Without a contemporaneous comparison group, the decline could not identify the package's causal effect or the contribution of home visits. The later Bankass trial tests the added value of proactive visits more directly.

A 2021 Cochrane review of seven iCCM studies found that the approach probably increased care-seeking to an appropriate provider (risk ratio 1.68, 95% CI 1.24–2.27), while mortality evidence was low or very low certainty ¹⁷. The review includes some of the earlier programmes discussed here, so these are not independent confirmations of a survival effect.

Mali: home visits and child survival (Liu et al., 2024)

Who and where. Muso ran the ProCCM trial with the University of California San Francisco, the Malaria Research and Training Centre at the University of Sciences, Techniques and Technologies of Bamako (USTTB), and Mali's Ministry of Health and Social Development ¹. It took place in seven health centre areas of Bankass district, in central Mali ¹. The trial was funded by USAID's Development Innovation Lab, the CRI Foundation, Grand Challenges Canada and the Johnson & Johnson Foundation ¹.

Design. It was a cluster randomised trial in 137 village clusters (NCT02694055). From February 2017 to January 2020 it enrolled 31,761 children, with 52,970 person-years of follow-up ¹. The intervention arm had CHWs who made proactive home visits. The control arm had CHWs who gave care at a fixed site ¹.

Both arms got the same package in January 2017, before the trial started. User fees were removed, and primary health centres got more staff, training, equipment and better buildings ¹.

Primary outcome. Under-5 mortality did not differ between arms. The intention-to-treat incidence rate ratio was 1.02 (95% CI 0.88–1.19) ¹. The per-protocol estimate was 1.01 (0.87–1.18) ¹. During the trial, 909 children died in the intervention arm and 827 in the control arm ¹.

What happened over time. Under-5 mortality fell from 142.8 to 56.7 per 1000 live births in the home-visit arm, and from 154.3 to 54.9 per 1000 in the fixed-site arm ¹. The authors note that this brought the area below almost all other regions of Mali ¹.

Caveats. Only 25.0% of children in the intervention arm lived in households that got the target of two or more visits in the month before a survey ¹. Armed conflict rose in the area from 2018, and six clusters were lost to follow-up ¹. Mortality before the trial was measured by asking women to recall births and deaths over three years, which may be affected by recall bias ¹.

Mali: the maternal care bonus (Ogbuoji et al., 2024)

Who. Duke University and Muso analysed costs within the same trial (69 intervention and 68 control village clusters) ².

What it measured. Antenatal care (ANC) use was a secondary outcome of the trial. The team looked at five ANC measures, such as at least four ANC contacts and ANC started in the first trimester ².

Results. Proactive home visits were cost-saving for every ANC measure ². From the CHW programme's view, each extra ANC outcome saved between US$21.39 and US$79.20. From the view of the full ANC programme, including facility care, the saving was US$1.70 to US$6.30 per outcome ². A CHW contact for ANC cost US$4.03 in the home-visit arm and US$14.51 in the fixed-site arm ².

Caveats. The study measured use of ANC, not health outcomes for mothers or babies ².

Mali: home visits and malaria (Thwing et al., 2026)

Who and where. The US CDC and President's Malaria Initiative and Muso measured malaria within the same Bankass trial ³. The study area covered more than 100,000 people, and it became a conflict zone during the study ³.

Design and outcome. At the end of the 3-year trial, the team tested people of all ages for malaria parasites with rapid tests ³. There was no difference between arms in parasite prevalence at any age, in fever in the past 2 weeks, in care-seeking, or in access to malaria testing for children under 5 ³.

Delivery. CHWs in the intervention arm reached the goal of two visits per month in fewer than half of households ³. At the end of the trial, 39% of people in the intervention arm lived in households with two or more visits in the previous month ³. Over the trial, CHWs recorded 219,410 patient assessments and did 107,428 rapid tests ³.

The authors' reading. With well-supported, supplied, supervised and paid CHWs, fixed-site and door-to-door care gave comparable access to malaria treatment ³.

Zambia: weekly visits where routine care is strong (Rutagwera et al., 2025)

Who and where. PATH and Zambia's National Malaria Elimination Centre ran the trial in Chadiza District, Eastern Province, from April 2021 to May 2023 (NCT04839900) ⁴. The Ministry of Health had already scaled up community malaria case management for all ages over the previous decade ⁴.

Design. It was a two-arm cluster randomised trial. CHWs in the intervention arm visited every household weekly to screen, test and treat for malaria. Routine care at CHW homes and health facilities continued in both arms ⁴. The intervention arm enrolled 26,189 people and the control arm 24,275 ⁴.

Primary outcomes. Malaria prevalence fell from 19.7% to 16.0% in the control arm and from 18.7% to 13.7% in the intervention arm. The difference in change was not significant (adjusted RR 0.97, 95% CI 0.77–1.23) ⁴. Monthly malaria incidence declined slightly faster in the proactive clusters (adjusted IRR 0.98, 95% Bayesian credible interval 0.96–0.99) ⁴. That is a 2% greater decline each month ⁴.

Delivery. CHWs made 272,872 household visits and did 17,835 malaria tests. Of these, 26% were positive, and 99% of positive cases were treated ⁴.

Caveats. Nearly everyone lived within 5 km of a malaria care provider, and routine care was already strong ⁴.

Madagascar: fever care tripled (Garchitorena et al., 2024)

Who and where. Pivot, IRD and the Institut Pasteur de Madagascar ran a cluster randomised trial in 30 health centres in Farafangana District (PACTR202001907367187) ⁵.

Design. Health centres and their CHWs were randomised to malaria case management for all ages or for children under 5 only ⁵. Both arms got CHW training, community messages about free malaria care, monthly supervision and a stronger supply chain ⁵. About 1600 households were surveyed at baseline (2019) and endline (2021) ⁵.

Results. Care-seeking for fever and malaria diagnosis nearly tripled in both arms, from under 25% to over 60%, mostly through CHWs ⁵. The all-ages arm did more rapid tests in children aged 6–13 (RR 1.65, 95% CI 1.45–1.87), with larger gains for people living further from a health centre ⁵.

Caveats. The extra gains for older children were significant only in health system data, not in the surveys ⁵. The authors conclude that a stronger supply chain can bring large gains even without expanding the age range ⁵.

Treating malnutrition at the village level

Niger. In Mayahi district, Action Against Hunger compared CHW-led treatment of severe acute malnutrition with standard treatment at health centres, in a non-randomised trial ⁷. Recovery was 77.2% vs 72.1%, and the cost per child recovered was US$107.22 vs US$116.52 ⁷. The cost of a round trip for families fell from US$3.48 to US$1.30 per visit when treated at a health post ⁷.

Southern Mali. In three districts of Kayes region, Action Against Hunger and the Ministry of Health's Nutrition Directorate added malnutrition treatment to iCCM at 120 health centres and 169 CHW sites ⁶. The cost per child cured was US$303.90–324.90 across three levels of supervision ⁶.

Ethiopia: a national model (Drown et al., 2024)

Research by Exemplars in Global Health with the Ethiopian Ministry of Health describes how the Health Extension Program carried much of Ethiopia's fall in child deaths from 2000 to 2015 ⁸. It was the platform for the PCV-10 vaccine (76% coverage by 2014/15) and for iCCM, and ORS coverage for diarrhoea doubled over the period ⁸. Health extension workers made services easier to reach for rural and pastoralist communities, who make up more than 80% of the population ⁸.

Results at a glance

LimitsWhat didn't work (or the limits)

Extra home visits did not reduce child deaths in Mali. Under-5 mortality did not differ between the home-visit and fixed-site arms (IRR 1.02, 95% CI 0.88–1.19) ¹.

The Mali decline is a before-and-after comparison, not a trial result. This is the most important caveat in this article. The package (free care, paid CHWs, better health centres) was given to both arms at the same time, in January 2017 ¹. There was no comparison group without it. So the trial cannot measure how much of the roughly 60% fall in deaths the package caused ¹. Other changes over the same years could also explain part of it. The authors say only that system-strengthening "may have contributed" ¹. Only a study with a comparison group that did not get the package could measure its effect directly.

Home visits often did not happen. In Mali, only 25.0% of children in the intervention arm met the target of two visits a month ¹. At endline, only 39% of people in the intervention arm lived in households that had two or more visits in the past month ³. The trials tested an offer of home visits, not a full dose of them.

Home visits did not reduce malaria in Mali. There was no difference in parasite prevalence, fever, care-seeking or access to testing ³.

Zambia showed the same pattern. The change in malaria prevalence did not differ, and the gain in incidence was small (2% a month) ⁴. Routine community case management was already strong ⁴.

Conflict disrupted the work. In Bankass, armed conflict rose from 2018 and six clusters were lost ¹. In Gao, insecurity also limited access for a malnutrition trial ¹¹.

Supervision level and cost. In southern Mali, costs per child cured for the three levels of CHW supervision overlapped, so the study could not show which level is best value ⁶.

Evidence gaps. The 2021 Cochrane review found uncertain mortality effects ¹⁷. Ethiopia’s national case study ⁸ and its earlier cluster trial ¹⁸ answer different questions. The Liberian evaluation described below is still at the protocol stage ¹².

Cost and scale

What a professional CHW costs. In the Mali trial, CHWs were paid 40,000 CFA a month, about US$71 ¹. The share of CHW costs (salary, training, supervision and kits) allocated to antenatal care was US$110.94 per CHW-year ². All CHWs in both arms got the same pay and support ².

What care at the doorstep costs. An antenatal contact by a CHW cost US$4.03 to US$14.51, against US$32.27 for a first ANC visit at a facility ². For malnutrition, a child cured by CHW-led treatment cost US$107.22 in Niger ⁷, and US$303.90–324.90 in southern Mali when all programme costs were counted ⁶.

Families save too. In Niger, bringing treatment closer cut the cost of each trip for families by more than half ⁷.

How it scales. Ethiopia's Health Extension Program shows that a national, government-run cadre can reach rural and pastoralist communities at scale ⁸. In Zambia, nearly a quarter of children who got antimalarials in 2018 got them from a CHW ⁴.

Who pays. Only a minority of CHWs worldwide are formally recognised as health workers and receive a salary ¹³. Liberia is one country that changed this. Its National Community Health Assistant programme professionalised CHWs nationwide ¹³. A 2024 case study found that the government overcame pushback from external funders on paying CHWs, through strong political commitment and consensus across government departments ¹³. In the Mali trial, the ministry's salary scale set the pay, while donors funded the trial ¹.

What to watch

Liberia: did the national programme cut child deaths? A study protocol published in 2024 plans to compare under-5 mortality before and after the National Community Health Assistant programme was rolled out in Grand Bassa, from 2018 to 2022 ¹². It will use a household survey and the staggered roll-out across districts ¹². It will be the first estimate of the programme's effect on child mortality ¹². We have not found published results.

Sierra Leone: a short burst of home visits (NCT06395207). PATH, Tulane University and Sustainable Health Systems Sierra Leone designed a three-arm cluster randomised trial in hard-to-reach areas ¹⁴. It tests two months of proactive visits at the start of the malaria season, on top of better-supplied routine care, against routine care alone ¹⁴. The primary outcome is same-day or next-day care-seeking for children with fever. The registry lists 9000 participants, with primary completion in March 2025, but its status is marked as unknown ¹⁴.

Kenya: community health promoters and malnutrition (NCT06441305). IFPRI, UNICEF and Kenyatta University are testing whether community health promoters with smartphones, family MUAC screening and follow-up of defaulters can raise treatment coverage for severe wasting in Turkana County ¹⁵. Primary completion is listed for July 2026 ¹⁵.

Open questions. Which parts of the Mali package mattered most: free care, paid CHWs, supplies, or better health centres? The authors say further analyses of the trial data will try to answer this ¹. The Mali malaria team also calls for more research on which elements drive the gains ³.

What it means

The three trials of proactive home visits do not show that CHWs fail. They show that where CHWs are already paid, trained, supplied and supervised, and where care is free, adding more frequent visits brings little extra benefit for child survival ¹³⁴. The Mali results suggest that the foundation itself may bring large gains, although this was not tested in a randomised way ¹.

For ministries of health. The order of priorities is clear. Pay and supervise CHWs, keep medicines and tests in stock, and remove user fees ¹³. Madagascar shows that a working supply chain can triple access to fever care ⁵. Ethiopia shows that a national CHW cadre can carry vaccines, case management and ORS to every village ⁸. Liberia shows that a government can put CHWs on the payroll, even when funders push back ¹³.

For funders. Fund the backbone first: salaries, supervision and supply chains. Proactive visits still have value for antenatal care, where they save money ². And CHWs can take on more, such as treating malnutrition, at a known cost ⁶⁷.

For programme managers. If you add home visits, measure whether they happen. In Mali, most households did not get the planned visits ¹³. Track stock-outs and CHW pay as closely as you track visits.

The backbone comes first. The doorstep visit adds value on top of it.