Factors influencing the sustainability of homestead vegetable production intervention in Rufiji, Tanzania: A cross-sectional mixed methods study

**Background:** Homestead vegetable gardening is promoted as a nutrition-sensitive intervention to combat micronutrient deficiencies in low-resource settings. However, evidence on the sustainability of such donor-funded programs after external support ends is limited. This study assessed factors influencing the sustainability of the Homestead Agriculture and Nutrition (HANU) intervention in Rufiji, Tanzania, one year after the trial phased out.

**Methods:** A community-based cross-sectional mixed-methods study was conducted from April to June 2021. Quantitative data were collected from 247 randomly selected women (98.8% of the target 250) who had participated in the HANU intervention arm. The mean age of participants was 36 years (SD 8.3); 85% were married/cohabiting; 35.2% had no formal education; mean household size was 7.1 (SD 2.6). Structured questionnaires were administered via tablets using ODKCollect. Qualitative data came from four focus group discussions (FGDs) with women (9–12 participants each)—two with active gardeners and two with non-gardeners—and four in-depth interviews (IDIs) with two extension workers, one community health worker, and one agriculture district officer. Multiple logistic regression was used for quantitative analysis; thematic analysis was conducted using NVivo 12.

**Key Results:** Only 20.24% (50/247) of households sustained homestead vegetable production one year after the intervention ended. In multivariable logistic regression, shortage of seeds was significantly associated with lower odds of sustaining gardening (aOR=0.65; 95%CI=0.46–0.93; p=0.018), while not experiencing a shortage of manure/fertilizers was significantly associated with higher odds (aOR=1.62; 95%CI=1.04–2.46; p=0.031). Other factors—water cost, destruction by free animals, pests/diseases, lack of capital, poor soils, weeds, lack of extension services, limited family labor, household size, age, marital status, and occupation—were not statistically significant in the multivariable model. In FGDs and IDIs, participants consistently reported high water costs, destruction by free-grazing animals (goats, chickens, ducks), pests and diseases, poor soil fertility, shortage of seeds, and lack of capital as barriers. Most active gardeners grew traditional, locally available vegetables (cassava, sweet potato, moringa, pumpkin); fewer than 30% grew introduced varieties (amaranth, tomato, African eggplant, Chinese cabbage).

**Clinical Implications:** The very low sustainability rate (20%) underscores that even well-designed nutrition-sensitive agriculture interventions may fail to produce lasting behavior change without explicit sustainability planning. The significant role of seed and fertilizer access suggests that interventions should incorporate strategies for ongoing, affordable input supply—such as local seed multiplication, compost training, and market linkages—before external support ends. The qualitative findings highlight that water costs, animal destruction, and pest/disease pressure are pervasive structural barriers that require community-level solutions (e.g., bylaws for animal control, affordable fencing, water pricing policies). The shift toward drought-tolerant, pest-resistant traditional vegetables indicates that participants adapted pragmatically, which could inform future intervention design. Policymakers and funders should integrate sustainability plans—including government support, community ownership, and affordable input systems—from the outset of donor-funded nutrition-sensitive agriculture programs.