Protocol for the cost-consequence and equity impact analyses of a cluster randomised controlled trial comparing three variants of a nutrition-sensitive agricultural extension intervention to improve maternal and child dietary diversity and nutritional status in rural Odisha, India (UPAVAN trial)

The UPAVAN trial is a four-arm, cluster randomised controlled trial, implemented in Odisha, India. The trial aims to test the nutritional and agricultural impacts of: (1) a participatory agriculture extension platform of women’s groups viewing videos about NSA practices, and follow-up home visits to encourage the adoption of the practices presented in the videos; (2) a modified version of NSA intervention that incorporates videos on nutrition-specific behaviour change and (3) another modified version integrating a participatory learning and action (PLA) approach of collective identification, prioritisation, and community action to solve local problems. In addition to these three interventions, health system strengthening (HSS) activities, including 2 days of training in maternal, infant and young child nutrition for frontline nutrition and health workers, will be provided in all study areas, including the control arm.

A detailed description of the UPAVAN trial and its interventions are presented in the trial protocol [13]. This paper aims to describe the methodology for the economic evaluation of the trial.

While there is some evidence for the positive impact of NSA programmes on maternal and child undernutrition [11], evidence on their cost-effectiveness is scarce. This constrains any investment case and limits our understanding of the feasibility of scaling up these interventions [10, 11]. Among those studies reporting economic evidence, nearly all are modelling exercises and the majority focus on biofortification programs [14, 15], with only one focussed on home garden programs [16]. Furthermore, although participatory interventions with women’s groups using a PLA approach have been shown to be highly cost-effective in improving maternal and newborn health [17], the evidence of their effectiveness and cost-effectiveness for improving dietary intake [18, 19] and nutrition outcomes [20, 21] is mixed. There is also no evidence, or very limited evidence, on the cost or cost-effectiveness of multisectoral nutrition-sensitive interventions, beyond biofortification and home gardens, in particular using digital platforms. The only economic evidence available for the use of digital platforms is the costing study conducted by Khurana et al. [22], reporting the cost of piloting one of the nutrition-specific intervention components of the UPAVAN intervention.

One of the main reasons for the scarcity of cost and cost-effectiveness evidence on NSA interventions in general, might be related to their multisectoral nature. This makes collecting and interpreting cost (and outcome) data complex, time-consuming and expensive [11]. Moreover, cost-effectiveness evaluations of NSA interventions face another challenge that stems from their multisectoral nature. These analyses focus only on the health outcome either using composite measures of mortality and morbidity, such as Disability Adjusted Life Years (DALYs) or natural units of outcome, such as cases of stunting averted. These measures do not capture the full range of benefits and outcomes, including non-health benefits, generated by NSA interventions; such as food security, women’s empowerment, knowledge and increased income from agriculture production among others [23].

The economic evaluation of the UPAVAN interventions will thus take the form of a cost-consequence analysis (CCA). This approach, to some extent, can address the challenges mentioned above. In CCA, all relevant costs and outcomes of the interventions are listed, in a tabular format, without aggregating into ratios, allowing policymakers to compare the incremental costs with the incremental consequences of the different interventions. CCA is recommended for complex multisectoral public health interventions, such as NSA interventions, that have multiple health and non-health effects, which may be difficult to measure in a common unit [24, 25]. Considering that these interventions are relevant to a wide range of stakeholders, using this approach allows different stakeholders to differentiate where costs and benefits might be accrued in their own sector, e.g. health care, agriculture, and social welfare.

The economic evaluation of the UPAVAN trial will consist of a cost-consequence analysis of the three UPAVAN interventions from a societal perspective.

The specific objectives of the UPAVAN economic evaluation are:

In addition to the above objectives, an equity impact analysis of the UPAVAN interventions will be carried out to evaluate how costs and consequences of the interventions are distributed among different sub-groups within the target population.

The UPAVAN trial is set in four administrative blocks (Patna, Keonjhar Sadar, Harichandanpur and Ghatgaon blocks) in Keonjhar district, Odisha. The total population of Keonjhar district is estimated at 1.8 million, which is 86% rural and agrarian [26]. Keonjhar is a tribal district with around 57% of its population comprised of SCs and STs (12% and 45%, respectively) [26]. It has one of the highest undernutrition rates among women and children in India. The prevalence of underweight and anaemia among women in Keonjhar is 30% and 40%, respectively [9]. Approximately 45% of children aged under 5 years are stunted and 19% are wasted [9]. Among children of 6–23 months of age, less than 10% are fed a minimum acceptable diet [9].

All women aged 15 to 49 years residing permanently (i.e. lived there for more than 6 months) in the 148 trial clusters are eligible to participate in the interventions (i.e. ‘intervention participants’). The trial will evaluate impact on outcomes in ‘trial participants’, i.e. children under 2 years of age, their mothers or female primary caregivers, and their households. The UPAVAN trial will evaluate impacts on outcomes for ‘trial participants’ through baseline and endline repeat cross-sectional surveys.

One hundred and forty-eight clusters were randomly allocated to the four trial arms to give 37 clusters per arm. A cluster is defined as one or two villages and surrounding hamlets, with a minimum population size of 800 people. Three arms will receive different packages of intervention activities, while the fourth arm acts as a control. In all four arms, basic health system strengthening activities will be implemented. The trial will run from December 2015 to May 2020 (53 months), which includes 15 months of trial set-up and intervention development and 32 months of intervention implementation, starting from April 2017 to October 2019.

Three UPAVAN interventions build on Digital Green’s participatory video-based approach [27]. Building on existing community institutions and public systems, the Digital Green approach is a method of delivering agriculture extension using low-cost participatory videos with women’s and farmers’ groups. The approach comprises four steps:

The outcomes of the interventions will be assessed through repeat cross-sectional household surveys at baseline (2016–2017) and endline (2019–2020). The outcomes of the interventions will be assessed relative to the control arm and not relative to each other. Analysis of the trial outcomes will be carried out on groups as randomised (i.e. intention-to-treat) at the individual level, adjusted for clustering. Since the trial outcomes will be evaluated using repeat cross-sectional surveys, loss-to-follow-up of specific individuals will not be an issue. Moreover, survey completion rates are expected to be high in this context, so any participants with missing data will be excluded from primary outcome analysis [13].

The UPAVAN trial has two primary outcomes: (1) child dietary diversity, measured as proportion of children 6–23 months of age consuming four or more out of seven food groups per day, based on a 24-h dietary recall, and (2) mean maternal body mass index (kg/m²) of non-pregnant, non-postpartum mothers or female primary caregivers of children 0–23 months of age.

The secondary outcomes of the trial are: (1) maternal dietary diversity, measured as proportion of mothers or female primary caregivers consuming five or more out of 10 food groups per day, based on a 24-h dietary recall and (2) child wasting, measured as proportion of children with a weight-for-height z-score < − 2 standard deviations (SD).

In addition to the primary and secondary outcomes, the trial has a number of other health and non-health outcomes including maternal and child low mid-upper arm circumference, maternal and child haemoglobin (Hb) concentrations, infant and young child feeding practices, women’s decision-making, women’s time use, gender parity, household economic status and food security and household agricultural production. A complete list of the trial outcomes and their corresponding indicators is listed in Table 2 of the trial protocol [13].

The economic evaluation of the UPAVAN interventions will be in the form of a cost-consequence analysis, conducted from a societal perspective [30, 31]. This includes measuring all costs incurred by the implementing agency (programme-provider costs), community frontline workers and other health care providers, and users, i.e. intervention participants and their households. All associated and measurable outcomes or benefits associated with the intervention over the intervention time horizon will also be enumerated.

Table 1 provides an overview of the resource use and cost measures used. The proposed methods for measuring programme costs, community frontline workers and other health care providers’ costs and participants/household costs are described in more detail in the following sections.

Programme costs are those incurred by the programme provider, i.e. VARRAT, and other partner organisations mentioned earlier. A combination of different costing approaches including activity-based costing [32], expenditure approach and ingredients approach [33] will be employed to estimate direct (financial) and indirect (opportunity) costs of designing and implementing the UPAVAN interventions. In brief, main intervention components and major activities under each component will be identified and defined as cost centres and the costs (mainly extracted through the expenditure records or estimated in case of opportunity costs) will be allocated to these cost centres.

UPAVAN interventions may increase demand for services provided by community frontline health and nutrition workers, i.e. AWW, ASHA and auxiliary nurse midwife (ANM), and other health care providers, such as malnutrition treatment centres or nutrition rehabilitation centres and primary health centres in the project area. In addition, frontline workers and other providers will have incurred a time cost of their direct involvement in the production of videos, attending women groups and attending health systems strengthening activities (Table 1).

UPAVAN interventions may affect the expenditure of study participants and their households in a number of ways. Impact on participants and their families may include changes in care-seeking behaviour and associated costs, and changes in household food and non-food consumption patterns and spending (mainly due to direct and indirect costs of adopting agriculture or nutrition-promotion practices). It also includes the time cost of participation in group meetings, follow-up home visits, and the actions taken by the groups. The interventions and behaviours promoted may also affect the allocation of time within a household.

As mentioned earlier, the economic evaluation of the UPAVAN interventions will take the form of a cost-consequence analysis, tabulating all relevant costs and outcomes of the interventions without aggregating them into a ratio. An incremental analysis of the costs and consequences of the alternative interventions will be conducted; comparing additional costs and outcomes generated by AGRI, AGRI-NUT and AGRI-NUT+PLA with the control or status quo. In addition, the costs will be disaggregated by programme provider, health care providers and participants.

All estimated costs will be presented in 2019 prices in both Indian Rupees and International Dollars (INT$), adjusted for inflation using the Indian Consumer Price Index and converted to 2019 INT$ using the Purchasing Power Parity conversion factor for India. Moreover, the costs will be discounted at 3%, the standard discount rate recommended by WHO-CHOICE [36] and the Gates/iDSI Reference Case for Economic Evaluation [37]. A series of univariate and multivariate sensitivity analyses will assess effect of the main cost drivers on the results. The study design, analytical methods and reporting of the findings will follow the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) Statement [38].

The possibility of conducting a cost-benefit analysis and modeling state-wide and national scale up of the UPAVAN interventions will be explored.

An equity impact analysis of the interventions will be also carried out to evaluate how costs and consequences of the interventions are distributed among different sub-groups within the target population. This will be done through sub-group analyses of all trial outcomes and individual-level costs (e.g. cost of seeking care, opportunity costs of involving in the intervention, etc.) based on selected socio-economic dimensions of the target population, such as education, household wealth and tribal and caste status.

This is protocol version number 1, finalised on 30 November 2018. The intervention implementation started in March 2017 and will end in October 2019 (32 months of intervention exposure). Endline impact evaluation survey will start in November 2019 and continue until May 2020. The detailed timing of data collection points, based on the Standard Protocol Items: Recommendations for Interventional Trials (SPIRIT) timeline, can be seen in Fig. 2. The SPIRIT Checklist is also available as Additional file 1.