Perception and Acceptance of Micronutrient Fortified Bouillon Cubes among Non-Index Household Members in a Randomised Trial in Two Districts of Northern Ghana

Abstract

Background: Micronutrient deficiencies remain a major global health challenge, disproportionately affecting women of reproductive age and pre-school children in low- and middle-income countries (LMICs). In northern Ghana, inadequate nutrient intake may contribute to high rates of iron and iodine deficiencies exceeding 40% among pre-school children. These deficiencies contribute to adverse health outcomes and exacerbate socio-economic disparities. Bouillon cubes, a condiment consumed by over 90% of households in northern Ghana, offer a culturally acceptable and scalable vehicle for delivering essential micronutrients through fortification. Researchers in the Condiment Micronutrient Innovation Trial (CoMIT) project conducted a randomised controlled trial (RCT) to evaluate the impact of household use of multiple micronutrient-fortified bouillon cubes among non-pregnant, non-lactating women of reproductive age (WRA), lactating women 4–18 months postpartum (LW), and pre-school children aged 2–5 years (PSC) in the Kumbungu and Tolon districts of northern Ghana. However, the perceptions and acceptance of these bouillon cubes among household members not enrolled in the RCT (non-index household members) remain unexplored. Objectives: The main objective of this study was to investigate the perceptions and acceptance of micronutrient-fortified bouillon cubes among non-index household members (NIHM) of WRA, LW and PSC participating in the CoMIT project RCT. Specific objectives were to (1) develop and validate a questionnaire to measure perception and acceptance of micronutrient fortified bouillon cubes ; (2) analyse the changes in perception and acceptance during the intervention, identifying key drivers and barriers; and (3) evaluate the associations between the perceptions and acceptance of NIHMs and the haemoglobin (Hb) and spot urinary iodine concentration (UIC) of index participants (WRA and PSC) in the CoMIT RCT and examine household-level factors as effect modifiers. Methods: This study was a longitudinal mixed-methods study that involved NIHMs (≥15 years) from households of index participants enrolled in the CoMIT RCT in the Kumbungu and Tolon districts of northern Ghana. The index participants were WRA (n = 1,028), LW (n = 690), and PSC (n = 690), recruited through door-to-door visits using the random walk method. After consent was obtained for the index participant, one NIHM was randomly selected from the household roster using the Kish Table. Following baseline assessments of index participants, households were randomised to receive biweekly rations of either (a) multiple micronutrient fortified bouillon cubes (containing folic acid, vitamin A, vitamin B12, iodine, iron, and zinc) or (b) control cubes containing iodine only, to be used in household cooking over a 9-month period. Quantitative data were collected from NIHMs at three time points: pre-intervention (T0; n = 1,047), early intervention (T1, within the first two months of the intervention; n = 731), and late intervention (T2, within the final two months of the intervention; n = 796). At T0, background characteristics were assessed using structured instruments. At T1 and T2 NIHMs’ perception and acceptance of the study bouillon cubes were assessed using a 29-item questionnaire, with most items eliciting responses on a 5-point Likert scale. Focus group discussions (FGDs; n = 24) were conducted at mid-intervention to explore health, sociocultural, sensory, and economic factors influencing NIHMs’ perception and acceptance of the study bouillon. Among enrolled WRA and PSC, Hb (g/dL) and spot UIC (µg/L) were measured at baseline and endline. Eighteen items specific to the study bouillon cubes (8 for perception, 10 for acceptance) were identified and validated using exploratory (EFA) and confirmatory (CFA) factor analyses of T1 data. Composite scores were computed as normalised, weighted means of validated items at T1 and T2, with higher scores indicating greater perception or acceptance. Statistical analyses included (a) Bayesian linear mixed-effects models to evaluate longitudinal changes in perception and acceptance and identify key drivers, and (b) Generalized Estimating Equations (GEEs) alongside Generalized Additive Mixed Models (GAMMs) to examine associations between NIHMs’ perceptions or acceptance and index participants’ Hb and spot UIC and household-level factors modifying these associations. Analyses were adjusted for pre-defined covariates (e.g. sex, age, food insecurity, and socio-economic status) that were significant at p ≤ 0.1, as well as baseline Hb and UIC for models examining the association between NIHMs’ perceptions or acceptance and index participants’ Hb and spot UIC. Baseline household-level factors (e.g. household head’s age, socio-economic status, bouillon consumption, household size) were tested as effect modifiers. Focus group discussions (FGDs) data were analysed using Braun and Clarke’s six phase framework, complemented by Latent Dirichlet Allocation to identify thematic patterns in participants’ responses. Results: A total of 1,047 NIHMs (mean age 40.0 ± 14.7 years; 59% female; 74.5% without formal education) provided responses. Exploratory factor analysis identified two constructs, Perception (8 items) and Acceptance (11 items), explaining 56% of variance. Confirmatory factor analysis retained 18 items (Perception=8; Acceptance=10) and showed good model fit (chi-square/df=1.91, root mean square error of approximation [RMSEA]=0.04, comparative fit index [CFI]=0.98, Tucker–Lewis index [TLI]=0.98, and standardised root mean square residual [SRMR]=0.06). Internal consistency was (Cronbach’s alpha = 0.71 for Perception and 0.72 for Acceptance). Between T1 and T2, the mean ± SD perception score for the study bouillon cubes increased significantly from 3.4 ± 1.1 to 4.1 ± 0.6 (adjusted mean difference = 0.74, 95% CrI: [0.63, 0.86]). In contrast, the acceptance score remained consistently high (4.59 ± 0.38 to 4.58 ± 0.35; adjusted mean difference = –0.01, 95% CrI: [–0.06, 0.04]). These patterns were similar across intervention arms. At both time points, more positive perception scores were observed among wives, children, and parents compared to household heads, and among urban compared to rural residents. Perception scores were lower among formal employees than among farmers. Interaction effects showed greater increases in perception scores from T1 to T2 among formal employees and small business owners compared to farmers, and smaller increases among wives, children, and parents compared to household heads. Higher acceptance was associated with being male, owning a small business, and having higher household socio-economic status (SES). Acceptance was lower among urban residents, those with higher baseline bouillon consumption, and mildly food-insecure households, but higher among severely food-insecure households. The FGDs further revealed that perceived health benefits, favourable sensory attributes, and cultural compatibility contributed to the sustained high acceptance of the study bouillon cubes, whereas barriers to perception and acceptance included misconceptions (e.g., fears of long-term male infertility) and occasional changes in food colour. In adjusted GEE models, NIHMs’ perception and acceptance scores at T1 and T2 showed no significant association with endline Hb concentration in PSC or WRA. However, significant effect modification by household factors was observed. In PSC models, perception and acceptance at T2 interacted significantly with the household head’s age and baseline bouillon consumption. The negative interaction with age indicates a weaker association at higher household head ages, while the positive interaction with baseline bouillon consumption suggests a stronger association at higher consumption levels. In WRA models, acceptance at T1 and T2 interacted significantly with household head age (negative and positive directions, respectively), and with SES and household size at T2 (positive and negative directions, respectively). Perception at T2 also interacted with household bouillon consumption and SES (positive direction). Adjusted GAMMs revealed complex associations between NIHMs’ perception and acceptance scores and endline UIC. In PSC, perception at T1 showed a non-linear association with UIC, while acceptance at T1 and T2 showed positive linear associations. In WRA models, both perception and acceptance scores at T1 and T2 were significantly associated with UIC in non-linear patterns. These associations were strongest in male-headed households, those with basic education, and households experiencing food insecurity. Conclusion: This study developed and validated a tool to assess household perceptions and acceptance of fortified bouillon cubes. The findings show that perceptions improved over time and acceptance remained consistently high, suggesting that fortified bouillon cubes are a culturally appropriate and widely accepted fortification vehicle in northern Ghana. While NIHMs’ perception and acceptance were not directly associated with endline haemoglobin levels in PSC and WRA, these relationships were modified by household-level factors. In contrast, both perception and acceptance were significantly associated with endline urinary iodine concentration, with patterns varying according to household characteristics, including the sex of the household head, education level, and food security status. These findings support the need for complementary implementation strategies to enhance the nutritional impact of bouillon cube fortification programmes, particularly among vulnerable households.