Kennedy 1989.

Study characteristics
Methods	Study design: PCS How were missing data handled? 462/504 HHs remained at follow‐up. Report relied on a longitudinal analysis, the data presented were based primarily on the cohort sample. Randomisation ratio: N/A Recruitment method: NR Sample size justification and outcome used: NR Sampling method: Sony (The South Nyanza Sugar Factory) provided a list of all farmers in the outgrowers' scheme. From this list, a random sample of sugar farmers, weighted by sublocation, was chosen. Once the sample of sugar farmers was chosen, field staff identified the next nearest non‐sugar farmers who met the same selection criteria. Study aim or objective: to evaluate the effects of cash crop production on agricultural production, income and food consumption, and to assess the impact of cash cropping on the health and nutritional status of preschool children and women. Study period: baseline: June 1984 to March 1985; follow‐up: December 1985 to March 1987 Unit of allocation or exposure: HHs
Participants	Baseline characteristics Intervention or exposure group: cash cropping sugarcane Age: NR Place of residence: NR Sex: NR Ethnicity and language: NR Occupation: NR Education: NR SES: mean nominal income per capita: sugar farmers KES 2591 (SD 139), new entrants KES 1956 (SD 42); mean real income per capita: sugar farmers KES 2712 (SD 135), new entrants KES 3070 (SD 38); mean (%) agricultural income used for own consumption: sugar farmers KES 748 (SD 29), new entrants KES 728 (SD 37); mean (%) marketed agricultural income: sugar farmers KES 942 (SD 36), new entrants KES 404 (SD 21) Social capital: NR Nutritional status: HH calorie intake (kilocalories/AE/day): sugar farmers 2689, new entrants 2822; percentage of HHs meeting < 80% of caloric requirements: sugar farmers 30.7, new entrants 17.9; percentage of preschool children with caloric adequacy: sugar farmers 69, new entrants 64; mean WAZ (n) of preschool children: sugar farmers –1.03 (356), new entrants –1.13 (90) Morbidities: NR Concomitant or previous care: NR Control group: no intervention Age: NR Place of residence: NR Sex: NR Ethnicity and language: NR Occupation: NR Education: NR SES: mean nominal income per capita: KES 1924 (SD 231); mean (real) income per capita: 2166 (SD 205); mean (%) agricultural income used for own consumption: KES 822 (SD 43); mean (%) marketed agricultural income: KES 393 (SD 20) Social capital: NR Nutritional status: HH calorie intake (kilocalories/AE/day): 2669; percentage of HHs meeting < 80% of caloric requirements: 30.0; percentage of preschool children with caloric adequacy: 58; mean WAZ (n) of preschool children: –1.17 (556) Morbidities: NR Concomitant or previous care: NR Overall: NR Inclusion criteria: HHs had to have ≥ 1 preschool child; ≥ 1 resident farmer; and own < 20 hectares of land. Exclusion criteria: NR Pretreatment: Sugar farmers have a higher mean nominal income per capita than non‐sugar farmers and higher marketed agricultural income per capita than all other groups. Attrition per relevant group: attrition of the total HHs was 8.3% (42/504) from baseline: no clear attrition present in the sugar farmers subgroup (139 at baseline and 146 at follow‐up); 35.7% (15/42) of new entrants attrited; and 11.3% (26/231) of non‐sugar farmers attrited. Attrition of women was 37.6% (298/793), but 529 new women entered the sample. Attrition of preschool children was 34.9% (409/1171), but 535 new preschool children entered into/were born into the sample. Preschool children from 356 sugar farmer HHs, 90 from new entrant HHs, and 556 from non‐sugar farmer HHs provided anthropometric data at baseline; with 243 sugar farmer HHs, 61 from new entrant HHs and 349 non‐sugar farmer HHs providing data at follow‐up. Description of subgroups measured and reported: 1677 preschool children and 1343 women. Intervention HHs were split into 'sugar farmers' who had completed ≥ 1 harvest at baseline and 'new entrants' who were farming with sugar, but had not completed a harvest (and had consequently not been paid). Total number completed and analysed per relevant group: 146 sugar farmer and 27 new entrant HHs completed the study, along with 205 HHs of non‐sugar farmers. 1024 women and 1297 preschool children completed the study – no per group numbers are available. Total number enrolled per relevant group: 139 sugar farmer and 42 new entrant HHs were enrolled at baseline, as well as 231 HHs of non‐sugar farmers. Total number randomised per relevant group: N/A
Interventions	Intervention characteristics Intervention or exposure group: cash cropping sugarcane Food access intervention category: increase buying power Intervention type: income generation Description: farmers were enrolled into the sugarcane outgrowers' scheme to provide sugarcane to a new factory, South Nyanza Sugar Factory (Sony) according to a contract agreement. Duration of intervention period: ongoing, but study period was 1984–1987 Frequency: continuous intervention, but payment after every harvest (24 months after planting) Number of study contacts: baseline data collection in 1984–1985, and follow‐up in 1986–1987 Providers: Kenyan government Delivery: NR Co‐interventions: NR Resource requirements: NR Economic indicators: NR Control group: no intervention
Outcomes	Proportion of HH budget spent on food; food expenditure for different food groups/items; total per capita food expenditure in previous 7 days Adequacy of dietary intake: percentage of HHs with caloric deficiency; caloric adequacy of preschool children Anthropometry: WAZ; underweight; HAZ; stunted; WHZ; wasting; women weight; adult BMI Morbidity: illness of women and children (all‐cause and diarrhoea)
Identification	Sponsorship source: IFPRI received support as a constituent of the Consultative Group on International Agricultural Research from a number of donors including Australia, Belgium, Canada, the People's Republic of China, the Ford Foundation, France, the Federal Republic of Germany, India, Italy, Japan, the Netherlands, Norway, the Philippines, the Rockefeller Foundation, Switzerland, the UK, the US, and the World Bank. In addition, a number of other governments and institutions contribute funding to special research projects. Country: Kenya Setting: smallholder farm HHs Author's name: Eileen Kennedy Email: eileen.kennedy@tufts.edu Declarations of interest: NR Study or programme name and acronym: smallholder sugarcane outgrowers' scheme Type of record: research report
Notes
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (Selection bias)	High risk	Cohort study and there was no randomisation performed.
Allocation concealment (Selection bias)	High risk	Cohort study and no allocation concealment was done.
Baseline characteristics similar (Selection bias)	Low risk	HHs of sugar farmers had significantly higher (P < 0.05) incomes than non‐sugar farmer HHs, as well as significantly higher (P = 0.05) marketed agricultural income per capita than all other groups. However, the main comparison was between 'new entrants' and 'sugar farmers', between which there were no major differences.
Baseline outcome measurements similar (Selection bias)	Low risk	No baseline outcome measurements were significantly different.
Blinding of participants and personnel (Performance bias)	Low risk	Participants were not blinded, but it was unlikely that the lack of blinding could have resulted in performance bias given the outcomes.
Blinding of outcome assessment (Detection bias)	High risk	Unclear whether outcome assessors were blinded to allocation. Outcomes were subjective/self‐reported and could have been influenced by lack of blinding.
Protection against contamination (Performance bias)	Low risk	NR, but it was unlikely that the control group received the intervention due to the nature of the intervention.
Incomplete outcome data (Attrition bias)	Unclear risk	Attrition was considerable for new entrants and non‐sugar farmers, and no information on differential attrition was provided. It was also difficult to assess the impact of attrition as more HHs were included at follow‐up.
Selective outcome reporting (Reporting bias)	Unclear risk	No protocol available, but all a priori stated outcomes were reported on.
Other bias	Unclear risk	Misclassification bias: unlikely. Measurement bias: high. Some of the outcomes relied on participants having to recall information for several days or weeks.