Can Gossip Change Nutrition Behaviour? Results of a Mass Media and Community-Based Intervention Trial in East Java, Indonesia

"...novel theory-driven approaches which employ emotional motivators are capable of having an effect on improving dietary diversity and the regularity of vegetable and fruit intake among children aged 6-24 months."

This study explored whether a behaviour change intervention, Gerakan Rumpi Sehat (the Healthy Gossip Movement), could have an impact on infant and young child feeding practices in peri-urban Indonesia - a place where nutrition knowledge is high but the adoption of healthy nutritional behaviours is poor. Behaviour Centred Design (BCD) is an approach to behaviour change that focuses not on improving knowledge but on identifying and employing behavioural levers - typically, emotional motivators such as affiliation, nurture, and disgust - that enable or inhibit change. Analysis using the BCD theory of change aims to identify where each change is produced and observe how these cause-effect mechanisms operate.

The intervention was designed, delivered, and evaluated according to the 5 stages of the BCD programme development process model:

1. A (Assess) - A framing workshop with key stakeholders identified 3 key target behaviours: (i) improved dietary diversity of complementary food; (ii) the reduction in unhealthy snacks through the promotion of healthy alternatives; and (iii) exclusive breastfeeding until 6 months.

2. B (Build) - Formative research was undertaken using a variety of methods (such as motive mapping, video ethnography, attribute ranking exercises, inventories of personal belongings, and daily scripts) designed to explore the determinants, the context and the drivers of these target behaviours. For example, videos of daily behaviour showed that mothers of toddlers struggled to get children to eat the meals prepared for them. As feeding normally takes place in a public street setting, mothers were embarrassed if their child cried and thus fed them snacks to distract them. Hence, although knowledge of "correct" nutrition was widespread, a variety of social and contextual factors conspired to mean that exclusive breastfeeding was rare, snacking was common, and dietary diversity is low.

3. C (Create) - A creative brief was developed based on the formative research findings. Social pressure from family members and peers was identified as a key driver of all 3 target behaviours; hence, the brief required that the intervention focused on the "affiliation" motive. The creative process was a collaboration between behaviour change theorists, a local creative agency, individuals familiar with the local context, and nutrition experts. (A description of the campaign is given in Table 1 and pictorially in Figure 1.) Emblematic of the GRS campaign was the character of Ibu Rumpi (Mrs. Gossip). The team developed 3 TV adverts (one for each target behaviour) depicting Ibu Rumpi as gossiping about and making judgements on the feeding practices of others before realising, to her (comic) embarrassment, that she was the one who had actually been doing the wrong thing. (One TV advert is below; the others can be seen here.) Each advert intentionally incorporated several key concepts: (i) shock was achieved by an action of hitting the wrong behaviour away (i.e. flicking a bottle into the bin or hitting someone's hand away so they could not provide a bad snack), (ii) the key message of 'Got it wrong, get it right', (iii) an expression of embarrassment by Ibu Rumpi where she slaps her hand to her head and says 'ooalla!', (4) naturalism – the adverts were set in a common streetscape in Sidoarjo which was done so that the audience felt they could relate to the characters.

   By mapping potential channels through which the intended audience engages with their community and environment, the researchers identified informal women's groups called arisans as a potential mode of delivery for community activations. During the arisans, held at convenient community locations, mothers practiced "healthy gossip" through emotional demonstration activities (emo-demos) such that they could support each other to adopt the 3 behaviours. (Click here [PDF] for a full description of each emo-demo). Other elements of the campaign - all of which are described in the paper - included posters, one of which was designed to spark curiosity and get people talking even though no information was given, house-to-house visits, tablet use, integration into Quran recital meetings, social media use (e.g., Facebook groups set up for each community), community noticeboards, and graduation events that were led by village leaders and attended by the family of arisan members. (Graduating mothers were asked to make a pledge to personally maintain the target behaviours and to encourage other mothers in their community to do the same.)

4. D (Deliver) - In each intervention village, 3 facilitators delivered the GRS arisan content and conducted house-to-house visits. The 2 intervention arms comprised almost 129,000 individuals, of which an estimated 4,000 were mothers with children under 2 and 2,000 were pregnant. The 3 TV adverts were each screened every evening on a local TV station (which had a viewer base of only about 30% of the population).

5. E (Evaluate) - The evaluation took the form of a 2-arm cluster randomised trial with a non-randomised control arm. One intervention arm received TV only, while the other received TV plus community activations. The intervention components were delivered over a 3-month period in 12 villages in each arm, each containing an average of 1,300 households. A total of 469 participants were surveyed (163 in intervention arm 1, 154 in intervention arm 2, and 152 in the control arm) during baseline data collection in April 2014; a total of 564 separate participants were sampled at endline (243 in intervention arm 1, 171 in intervention arm 2, and 150 in the control arm) in September 2014, one month after the pilot intervention. There were 2 primary outcomes: dietary diversity of complementary food and the provision of unhealthy snacks to children aged 6-24 months.

   The evaluation suggests that the intervention had some effect on improving dietary diversity and the regularity of vegetable intake among children aged 6-24 months. Dietary diversity scores (DDS) increased by 0.8 points in the arm exposed to TV adverts only (95% confidence interval (CI): 0.4-1.2) and a further 0.2 points in the arm that received both intervention components (95% CI: 0.6-1.4). In both intervention arms, there were increases in the frequency of vegetable and fruit intake. Despite these improvements, the average DDS at endline was still less than the recommended 4 or more food groups per day. The researchers found inconsistent evidence of an effect on unhealthy snacking. Trends indicate that the intervention may have had a positive influence on breastfeeding practice, but this potential effect was not able to be captured within the short 3-month period between baseline and endline surveys.

In reflecting on this process, the researchers note: "This is the first time that the BCD framework has been applied to non-hygiene-related behaviours. It was found to be equally applicable to nutrition, with the ABCDE process generating new insights and new approaches to nutrition promotion. Gossip proved a useful behavioural lever, with beneficiaries identifying with the character of Ibu Rumpi. Consistent with our hypothesis, the findings of this study suggest that mass media campaigns which utilise strong motives rather than health knowledge can be effective in changing behaviour. Complementing TV adverts with community activations is likely to be more effective, but the true value of doing so must be weighed against the logistical challenges of delivering high-fidelity community activations at scale and their cost-effectiveness. Exposure to both campaign components was suboptimal in the pilot intervention and increasing exposure and dose is likely to yield more significant behaviour change. This intervention, based on strong emotional drivers of behaviour change rather than on imparting 'nutritional knowledge', was designed to be replicable at large scale. The results of this pilot intervention have led to the expanded roll-out of a modified version of the programme. This uses a national TV station to screen the adverts and delivers the emo-demo activities through existing government health programmes rather than specialised arisans. A larger trial will evaluate the long-term health outcomes of the intervention."

The limitations section of the paper explores some implications for those seeking to replicate the GRS campaign in other settings. For example, sociodemographic differences were observed between the intervention arms in Sidoarjo district of East Java, and the control sites in Kudus in Central Java, suggesting that these sites were less comparable than the researchers had assumed. Future studies might employ alternative designs for measuring the effect of mass communication - for example, using interrupted time-series data as the counterfactual. Furthermore, the desired coverage of the community activations and dosage of exposure was suboptimal (particularly GRS arisan meetings). This led to the decision to purposively enrol study participants based on exposure status. Therefore, an intention-to-treat analysis regarding the effect of the GRS arisan meetings was not possible. "Capturing the true effect of behaviour change interventions in programme contexts remains challenging given that implementation may not always be optimal and dose of exposure may be low." Finally, there is a growing body of literature that documents the poor validity and reliability of self-reported nutritional assessment, and the principle outcome measures were derived from self-reported data. However, this study used video observation of food intake to explore the relationship between observed and reported behaviour and add insight to the findings. "It is reassuring that in this instance the video observation data generally supported the primary measures."