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

Background There is growing evidence that home vegetable gardening interventions improve food security and nutrition outcomes at the family level. This study assessed factors influencing the sustainability of homestead vegetable production intervention in Rufiji district, Tanzania, one year after the cessation of external support. Methods This was a cross-sectional study using both quantitative and qualitative data collection methods. A total of 247 randomly selected women from households who participated in the homestead vegetable intervention were interviewed using a structured questionnaire. The study held four focus group discussions with women from households that participated in the intervention, and four In-Depth interviews with two extension workers, one community health worker, and one agriculture district officer. Multiple logistic regression for quantitative data and thematic analysis for qualitative data was conducted. Results About 20.24% (50/247) of households sustained homestead vegetable production for one year after the intervention phased out. Lack of seeds (adjusted OR=1.26: CI=0.39-0.89) and either manure or fertilizers (adjusted OR=1.69: CI =1.08-2.63) were significant factors influencing the sustainability of homestead vegetable production. In the Focus Group discussions (FGDs) and In-Depth Interview (IDIs), all participating women and extension workers reported high cost of water, destruction from free-grazing animals, agriculture pests and diseases, poor soil fertility, shortage of seeds, and lack of capital affected homestead vegetable production sustainability. Conclusion Existing individual, community, and system challenges influence the sustainability of external-funded agriculture and nutrition interventions. The study findings underscore the importance of community authorities, scientists, and policymakers in having a well-thought sustainability plan in all promising external-funded interventions.

Participants in the intervention arm received three main intervention packages: (i) agriculture 1 0 5 inputs such as seeds, watering cans, manure, fertilizers, and training to promote homestead food 1 0 6 production and increase food diversity, (ii) nutrition counselling, including prevention and management 1 0 7 of child malnutrition, and (iii) a health-focused intervention including information on micronutrient 1 0 8 supplementation, integrated management of child illness, and safe water, sanitation and hygiene 1 0 9 practices. The extension and community health workers provided these training and educational  This was a cross-sectional study design using a mixed-methods sequential explanatory data 1 1 5 collection approaches, where quantitative data analysis results led to development of themes for 1 1 6 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 10, 2022. ; https://doi.org/10.1101/2022.05.06.22274693 doi: medRxiv preprint 7 qualitative survey (22). The study was conducted from April to June 2021, one year after the phase-out 1 1 7 of HANU trial. The population for this study was women who previously participated in the Homestead 1 1 8 Agriculture and Nutrition (HANU) trial in 5 randomly selected intervention villages.

9
Participants' source list was acquired from the HANU trial database, where all 500 participants 1 2 0 from five intervention villages were subjected to simple random sampling. A proportion-based 1 2 1 representation of intervention households per village was considered when sampling to obtain a 1 2 2 representative sample size from each village. Household replacement was considered after missing the village was also generated, and a standard orderly sequence was applied to guide selecting households 1 2 5 from the list for replacement. This involved collecting data at one point from 247 randomly selected 1 2 6 households. A simple random sampling technique was used for the quantitative survey to select women 1 2 7 who participated in the HANU trial under the intervention arm. For qualitative survey, purposive sampling was performed to achieve uniform representation of the 1 2 9 four Focus Group Discussions (FGDs) of women from five villages that received the HANU 1 3 0 intervention; two FGDs involved women actively practicing home gardening to date and two FGDs 1 3 1 involved women not practicing home gardening since the end of the HANU trial. We conducted four 1 3 2 In-Depth Interviews (IDIs) with two extension workers, one community health worker, and one 1 3 3 agriculture district officer. The sample size for the quantitative survey was pre-determined by the number of participants in the 1 3 7 trial under the intervention arm (500 women) and the logically feasible time frame of three months. The number of participants in the intervention trial arm who could be enrolled was estimated before (as 250) 1 3 9 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 10, 2022. ; https://doi.org/10.1101/2022.05.06.22274693 doi: medRxiv preprint 8 to be sufficient to determine sustainability status of the intervention, but no formal sample size 1 4 0 calculation was performed. Quantitative data collection was done by trained research assistants using closed-ended structured 1 4 4 questionnaires administered in electronic devices (tablets). The collected data was then uploaded and 1 4 5 stored to Ifakara Health Institute(IHI) server in Dar es salaam, Tanzania for safe data management. Open-ended questions were used to generate information from a total of four FGDs and four IDIs. IDIs explored participants' perceptions of potential strengths, weaknesses, opportunities, and threats that 1 4 9 may influence the sustainability of vegetable gardening in the absence of external financial and 1 5 0 technical support. Each FGD had 9 to 12 participants. FGDs and IDIs were conducted using a topic 1 5 1 guide that had been pilot-tested. Two experienced qualitative researchers conducted FGD and IDI in 1 5 2 Swahili language and in a private place to ensure confidentiality. All the FGDs and IDIs were audio-  Quantitative data were cleaned and analyzed by using STATA version 15. Numerical variables 1 5 7 were summarized using means and standard deviations. Categorical variables were summarized using primary endpoint of the study (sustaining gardening) was assessed by bivariate analysis. To get a deeper understanding of the data on how different variables impact sustainability of 1 6 1 homestead vegetable gardening, logistic regression models were used to estimate the odds ratio (OR) 1 6 2 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 10, 2022. ; https://doi.org/10. 1101/2022 for the association between active practising vegetable gardening status and potential agriculture For FGDs and IDIs, thematic analysis was done with the aid of NVivo 12 pro qualitative analysis 1 6 8 software (23). Transcripts were imported into NVivo, and two experienced social scientists did the coding. The list of codes and main theme frequencies were summarized. Only themes that answered the 1 7 0 specific study objective were identified during the thematic analysis. Furthermore, themes from all five 1 7 1 study villages including FGDs and IDIs were combined after thematic analysis due to the similarity of 1 7 2 their results. Ethical clearance to conduct this study was approved by the Institutional Review Board of Ifakara granted. The Principal Investigator assured that the study was conducted in full compliance to the of Tanzania. Written informed consent was obtained from all study participants. Confidentiality and 1 8 0 anonymity of participants' data were highly maintained. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

8 4
A total of 247 women were recruited and interviewed in the quantitative survey with the mean (sd) 73.7% lived in their own house or residence. The mean family size was 7.1 (2.6) persons (Table 1).  is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 10, 2022. ; https://doi.org/10.1101/2022.05.06.22274693 doi: medRxiv preprint 1 2 About 20.24% (50/247) participants reported actively participating in vegetable gardening one year 1 9 2 after the nutrition-sensitive homestead agriculture intervention had phase-out. All participants with 1 9 3 active vegetable gardens were growing cassava, sweet potato, moringa, and pumpkin, considered 1 9 4 traditional vegetables whose seeds/seedlings are locally available. Less than 30% of participants with 1 9 5 active vegetable gardens were growing amaranth, tomato, African egg-plant, and Chinese-cabbage 1 9 6 which participants were supplied with seeds during the intervention phase (Fig 1). In the multiple logistic regression predictive models for the sustainability of home vegetable 2 0 0 gardens, participants in households who could buy seeds were significantly more likely to have active in households that did not experience a shortage of manure and fertilizers were significantly more likely 2 0 3 to have active home vegetable gardens (adjusted OR =1.69; 95%CI= 1.08 -2.63; p = 0.020) ( Table 2). . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 10, 2022.  CI 95%, 95% confidence interval.

1 0
Adjusted odds ratio for probable confounding effect-adjusted for water cost, destruction by free animals, pests and diseases, 2 1 1 water shortage, lack of capital, poor soils, weeds, shortage of seeds, lack of extension services, cost of agriculture inputs, 2 1 2 limited family labour, lack of time to work, theft, damage by weather, and shortage of manure and fertilizers.
2 1 3 2 1 4 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 10, 2022. ; https://doi.org/10.1101/2022.05.06.22274693 doi: medRxiv preprint 1 4 In FGDs and IDIs, factors mostly reported by participants to influence the sustainability of 2 1 5 homestead vegetable production include garden and vegetable damage caused by free-grazing animals, 2 1 6 cost of water, pests and diseases, shortage of seeds, shortage of manure and fertilizers.

1 7
Majority of the participants were not able to build a strong fence for garden protection from free-2 1 8 grazing poultry and animals such as goats, chickens, and ducks. Due to the high cost of hard and strong 2 1 9 fencing materials, most gardeners built short-term fences using coconut thatches, sesame trees, and old 2 2 0 bed nets. good economic status to buy water, wells, fence or live close to water sources" (Male IDI 2 3 0 Participant).

3 1
All participants reported water for gardening to be a challenge in terms of availability, access and 2 3 2 cost. It was informed that water availability was not a big problem in nearly all study villages because 2 3 3 there are enough public and private water taps and wells. However, the cost of water was a big 2 3 4 constraint as most participants could not afford the cost of water for domestic and home garden uses. "Cost for water is the biggest challenge let me tell you, because you have to buy water for 2 3 6 cooking, drinking, washing clothes, and bathing these are compulsory and requires money; so 2 3 7 you choose not to irrigate vegetables until I get extra money" (Woman FGD participant).

3 8
. CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

(which was not certified by peer review)
The copyright holder for this preprint this version posted May 10, 2022. ; https://doi.org/10.1101/2022.05.06.22274693 doi: medRxiv preprint 1 5 "One of the biggest challenge is the problem of access to water and its costs. Therefore, farmers 2 3 9 had to seek water from those with wells and water pipes where the water was sold at a highest 2 4 0 cost compared to their economic status. Even in the rainy season water was sold but the farmer 2 4 1 did not have to buy water every day as sometimes it rained" (Male IDI Participant).

4 2
Pests and diseases were reported to be a big problem even before the phase-out period of the 2 4 3 homestead agriculture nutrition project in Rufiji, Tanzania considering that participants were 2 4 4 discouraged from using industrial chemical pesticides for safety reasons.  "Despite providing education to farmers to use natural remedies but pests and diseases were 2 5 0 very high so the effectiveness of these natural remedies was limited, for example they used and diseases by about 20% or 30% compared to industrial pesticides" (Male IDI Participant).

5 3
Most of the participants in FGDs and IDIs reported a shortage of seeds as a constraint on the better 2 5 4 sustainability of homestead vegetable gardening. They reported that vegetable seeds were provided 2 5 5 timely and free of charge during the intervention phase as the project incentives. "Before the phase-out of the intervention all seeds and manure or fertilizers were provided for free 2 5 7 but now days I have to buy seeds and fertilizer from shops. Since Seeds and fertilizer are sold at 2 5 8 high price sometimes it is difficult to afford" (Woman FGD participant). Majority of participants reported limited availability and accessibility of manure and fertilizer was 2 6 0 affecting vegetable productivity because of poor soil structure and fertility. 2 6 1 . CC-BY 4.0 International license It is made available under a is the author/funder, who has granted medRxiv a license to display the preprint in perpetuity. (which was not certified by peer review) The copyright holder for this preprint this version posted May 10, 2022. ; https://doi.org/10.1101/2022.05.06.22274693 doi: medRxiv preprint