Community Malnutrition Care Transitions

Evidence

Two studies published July 1 examined different transitions away from facility-centered malnutrition care. One tested village-level treatment for children aged 6–59 months in Somaliland. The other piloted home follow-up for Kenyan infants under six months after hospital treatment. Together they ask whether care can remain connected to families, but they measured different endpoints and do not establish a single clinical effect.

SETiPlus was a cluster-randomized trial in rural villages in the Somaliland districts of Hargeisa, Gabiley, and Faraweyne. Researchers selected and randomized 38 villages, then analyzed 37 after one family health worker stopped working before baseline data collection. Eighteen control villages used integrated community case management in which children identified with acute malnutrition were referred to clinics. In 19 intervention villages, family health workers also provided malnutrition treatment locally. A closed household cohort measured coverage at baseline and six months, while 253 worker-identified cases formed an open cohort for treatment outcomes.

The coverage estimate moved in the expected direction but remained uncertain. At endline, enrollment among children diagnosed by mid-upper-arm circumference or edema was 47.4% in intervention villages and 30.8% in controls. The adjusted difference was 19.6 percentage points, with an odds ratio of 2.55, but the 95% confidence interval was wide at 0.37–17.81 and the result was not statistically significant. The study was underpowered because acute malnutrition prevalence was 3% by arm circumference but 13% by weight-for-height z score, leaving fewer eligible cases than expected.

Treatment initiation showed a clearer access signal. All 164 worker-identified intervention cases started treatment, compared with 25 of 89 control cases, a difference of 71.9 percentage points. Recovery among discharged children was similar: 61.8% in the intervention arm and 66.7% in the control arm. Outcome status was unknown for 40% of control cases that started treatment versus 7.3% of intervention cases, so comparisons require caution. Non-response was recorded in 5.3% of intervention discharges versus none in controls, and relapse in 18.1% of intervention recoveries versus none in controls. Probable worker misdiagnosis was high in both arms—77.4% in the intervention group and 67.4% in the control group—linking expanded access to a serious measurement problem.

The Kilifi County study asked a narrower implementation question. It purposively enrolled 20 mother–infant pairs after treatment at one referral hospital. The infants had a median birth weight of 2.065 kilograms, and 12 of 20 were born prematurely. A breastfeeding peer supporter delivered three home visits and three phone calls over seven weeks, with a community health promoter and a support person incorporated into the follow-up network.

The pilot enrolled its target sample, reported 0% dropout, and completed intervention components on time. Fidelity nevertheless declined across later contacts: 100%, 95%, and 80% for the three calls and 100%, 95%, and 85% for the three home visits. Mothers described the intervention as acceptable and reported greater breastfeeding confidence. These are feasibility and acceptability findings, not evidence of improved growth. The paper says effectiveness for post-discharge weight gain is being tested in a trial.

Analysis — Access and accuracy must travel together

The cross-study pattern is operational rather than clinical: moving support closer to families can reduce a break in contact, yet proximity does not itself establish accurate classification or better recovery. SETiPlus made treatment initiation far more common among cases identified by family health workers, while its high probable-misdiagnosis rate, uncertain coverage estimate, and similar recovery proportions showed how decentralized delivery can carry measurement and training weaknesses into the pathway. The Kenyan pilot used a different structure—a breastfeeding peer supporter linked with community health personnel and a family-selected support person—and showed that repeated post-discharge contact could be delivered and accepted. It did not test whether that contact changed weight gain. An emerging, unproven inference is that community malnutrition programs may need to treat diagnostic quality and continuity of support as coupled design problems. Expanding reach without validating identification could scale error; improving identification without maintaining follow-up could leave the transition home unsupported. Because the studies differ in age, country, design, and endpoint, this is a research hypothesis, not evidence that one community model is superior. A stronger test would measure diagnostic agreement, implementation fidelity, sustained recovery, and relapse within the same adequately powered program.

Limitations

These studies cannot be pooled as if they tested the same intervention. SETiPlus was an underpowered cluster trial among children aged 6–59 months in rural Somaliland. Its closed household cohort lost 21% of households across both arms and excluded newly arriving households from the endline coverage sample. Arm circumference and weight-for-height identified very different numbers of children, and several treatment outcomes had small denominators. The high probable-misdiagnosis estimates further complicate interpretation of initiation and relapse, while diagnostic accuracy was not measured for clinic staff.

The Kenyan study was a descriptive pilot from one referral hospital with 20 purposively selected pairs and no comparison group. Although it reported 0% dropout as an implementation metric, the paper also states that two infants died before the end of follow-up; the design cannot estimate a survival effect or attribute those deaths to the intervention. Organization vehicles were used for home visits even though the planned main-trial model involved public transport, leaving real-world logistics partly untested. Feasibility measures and participant reports cannot establish effects on growth, readmission, survival, or longer-term breastfeeding. Results from either study may not transfer to other health systems, age groups, diagnostic protocols, or causes of malnutrition.