Abstract
Introduction: Anticipated future reductions in donor funding make it important to understand the resource needs for vaccine-preventable disease (VPD) surveillance to allow national programs to include sufficient funds for surveillance activities and facilitate their planning and budgeting. Despite the critical role of VPD surveillance to inform public health decisions, there is limited information about its cost.
To address this gap, costing studies of pre-COVID VPD surveillance were conducted in Nepal in 2018 (fiscal year 2016-17) and Ethiopia in 2022 (fiscal year 2018-19). Based on the experiences from these two evaluations, this presentation provides considerations and recommendations for other countries planning to conduct VPD surveillance costing studies in the context of external donor funding transition. Although there are established methods for costing health service delivery, there are no conventional methods for costing surveillance and similar public health system investments. This presentation highlights novel methods for costing surveillance.
Methods: The methods used for the Nepal and Ethiopia studies were systematically compared by key themes: perspective, costing activities, sampling criteria, extrapolation strategies, data collection modalities, and allocation of shared costs. For each theme, investigators from each study identified methodologic challenges and potential strategies to address these challenges, and recommended practices for future such studies.
Results: The two studies differed in several aspects, including surveillance activity categories and sampling strategies. Costs in Nepal were collected and analyzed by study-specific categories including a subset of surveillance core and support functions, whereas the Ethiopia study used all surveillance support functions from the Global Strategy on Comprehensive VPD Surveillance. Surveillance support functions (e.g., workforce capacity, field logistics and communications) support the core functions (e.g., case investigation). However, in Ethiopia, these support function categories posed some challenges for respondents and stakeholders to understand during cost data collection and analysis. The ease of collecting and using costs by VPD surveillance support function for program budgeting should be explored in future studies.
A mix of random and purposive sampling of surveillance sites was used in both studies. Surveillance sites were selected considering the strata of interest (e.g., rural-urban, ecological zones, health facility type) at each administrative level (e.g., region, zone). In Nepal, sites from all seven provinces and three ecological zones were included, while in Ethiopia, only three of 11 regions were sampled due to budget constraints and security concerns. Results from both studies were extrapolated country-wide using sampling weights and assumptions about the representativeness of purposively sampled units. The Nepal study included a larger proportion of sites (n=48 of 1536, 3.1%) compared to that for Ethiopia (n=47 of 5129, 0.9%). To ensure representative results, if budget allows, those conducting future VPD surveillance cost studies should randomly sample surveillance sites in each administrative level and stratum of interest; sampling strategies should be developed with consideration for the precision needed for the intended use of costing results.
Discussion: The review highlighted potential methodologic tradeoffs in utility and precision of results to assist those conducting and interpreting VPD surveillance costing studies. The recommendations from these evaluations can improve and standardize future studies.