Factors associated with nutrition intervention adherence: Evidence from a cluster‐randomised controlled trial in Kenya

**Background:** Child stunting affects an estimated 21% (144 million) of children under five globally, with rural areas disproportionately affected. Nutrition-sensitive agricultural programs, which combine agricultural interventions with nutrition-specific and -sensitive strategies, hold promise for reducing stunting, but adherence to multiple novel products is often challenging. This study aimed to identify factors associated with adherence to a bundle of nutrition and WASH interventions in a cluster-randomised controlled trial in rural Kenya.

**Methods:** The trial was conducted in Lugari and Likuyani subcounties of Kakamega County, Kenya, from March 2018 to April 2020. Clusters (villages or parts of villages) were randomly assigned to either a control group receiving agricultural inputs and training only, or an intervention group receiving the same agricultural program plus nutrition-specific interventions (micronutrient powders [MNP], chickens for eggs, seeds for red onions and indigenous greens) and nutrition-sensitive interventions (soap, chlorine, ORS, zinc). Monthly nutrition and WASH trainings accompanied product distribution. Adherence was measured quarterly: MNP sachets were counted from empty sachets stored by caregivers; egg consumption was recalled for the past 7 days; chlorine use was measured by weighing bottles; soap use was assessed by quarters used; and vegetable consumption was based on years crops were consumed. Quantitative data from 984 households were analyzed using bivariate tests (t-tests, ANOVA) and OLS regressions with cluster-robust standard errors. Qualitative data were collected through 120 barrier-focused feedback sessions and 6 focus group discussions (FGDs) with caregivers.

**Key Results:** Adherence rates were fairly high in the first year but declined in the second year. Mean MNP sachets consumed per year dropped from 112.6 (SD 6.12) in year 1 to 84.5 (SD 5.72) in year 2 (target: 120). Mean eggs consumed per week fell from 3.3 (SD 1.64) to 2.2 (SD 1.55) (target: 7). Chlorine use increased from 267.9 ml per quarter (SD 124.45) to 322.7 ml (SD 126.75) (target: 275 ml). Child age and sex did not significantly affect adherence. Caregivers who were not the biological mother (93% grandmothers) had higher overall adherence (p<0.001) and higher chlorine adherence (p<0.001). Older caregivers (≥39 years) also showed higher overall adherence (p<0.001) and chlorine adherence (p=0.019). Caregiver education was associated with MNP and egg adherence but with small absolute differences. Farming as primary occupation was linked to higher overall adherence (p=0.020). Higher number of prior live births (≥3) was associated with higher overall adherence (p=0.003). Household wealth only predicted egg consumption (p=0.029). Training attendance was high (>95% most months), but primary caregiver attendance declined to just under 75% by month 22. Overall training attendance was positively associated with overall adherence (p=0.001) and MNP adherence (p=0.032) in bivariate analysis. In OLS regressions controlling for confounders, overall training attendance remained a significant positive predictor of egg consumption (coefficient 0.08, p=0.005) and chlorine use (coefficient 6.05, p=0.005), but was negatively associated with MNP use (coefficient -0.34, p=0.029). Qualitative findings revealed that training attendance was motivated by improved child health, the interactive nature of sessions, and product distribution. Barriers to MNP use included limited knowledge by other family members, being busy/forgetting, and child taste issues. High chicken mortality limited egg production. Chlorine was preferred over boiling due to ease of use, cost, and availability.

**Clinical Implications:** This study demonstrates that adherence to multipronged nutrition interventions can be sustained through interactive group training combined with product distribution, active engagement of older caregivers (especially grandmothers), and regular reminders. The decline in adherence over time, particularly for MNP and eggs, underscores the need for ongoing support and potentially additional incentives. The 'grandmother effect' suggests that nutrition programs should not exclusively target mothers but also involve extended family members. The positive impact of training attendance on adherence, even after controlling for confounders, supports the value of continued education and social support. These findings can inform the design of future nutrition-sensitive agricultural programs to improve adherence and ultimately child growth outcomes.