Study setting
In Sierra Leone, the campaign was conducted nationwide over 7 days in June 2019, with the primary aim of covering all children aged between 9 months and 15 years with MR, ahead of a switch from measles to MR in the routine program. In addition, OPV was administered to all children under 5, and in half of the country’s districts, Vitamin A and deworming supplements were delivered to children aged 6–59 months and 12–59 months respectively. The campaign was coordinated by the Ministry of Health and Sanitation, with support from the World Health Organization (WHO), United Nations Children's Fund (UNICEF), and other partners. School-based delivery was used exclusively for MR, and other temporary fixed sites were set up at markets, transit points, community centers and other locations. Facilities also delivered the vaccines and supplements during the campaign, and mobile teams were deployed to move around in the communities as well. Most facilities also conducted mop-up activities to reach children who were not vaccinated during the campaign. The post-campaign coverage survey found 93% coverage for MR. The other interventions were not included in the survey, and administrative coverage reported 120% for OPV, and 97% for each of the nutrition interventions.
In Nigeria, the campaigns included in this study were part of a 10-year yellow fever elimination plan. Under this plan, Nigeria has committed to vaccinating at least 80% of the target population (9 months to 44 years of age) by 2026, through a series of immunization campaigns, implemented in a phased manner. This study focused on the yellow fever campaigns conducted in 3 states: Katsina in September 2019, Rivers in February 2020, and Anambra in October 2020. In Anambra, the campaign was implemented alongside a Meningococcal A catch-up campaign targeting 1- to 6-year-olds. The campaigns were coordinated at federal level by the Department of Disease Control and Immunization of the National Primary Health Care Development Agency (NPHCDA), with support from the WHO, UNICEF, and other partners. In each state, local government areas (LGAs) took 10 days to complete the campaign. Campaign implementation was most often managed at ward level, and in some cases by facilities. Temporary posts were the most utilized delivery strategy, but facility-based delivery also took place, and a short mop-up was usually conducted as well. According to post-campaign coverage surveys, the campaigns achieved 83% coverage in Katsina, 82% in Rivers, and 76% (YF) and 96% (MenA) in Anambra.
The campaigns in Sierra Leone and Nigeria relied heavily on financing from Gavi, the Vaccine Alliance (Gavi), which fully funded the vaccines and a large share of the operational cost. This was complemented by funding from both governments. In Sierra Leone, the Global Polio Eradication Initiative also provided support for OPV, and the Canadian government supported the nutrition interventions. Where multiple vaccines and interventions were delivered, delivery was fully integrated, from campaign planning and coordination down to vaccine transport and service delivery.
Study design
We followed an ingredients-based costing approach to capture all financial and economic costs of delivery from a payer perspective, which included the ministries of health and implementing partners. Costs incurred at the government body in charge of the immunization program at the national level, the sites that participated in the campaign, and all relevant administrative levels in between were included. The study covered activities conducted during the campaign, as well as planning, preparatory and wrap-up activities. All resource types were included, such as paid and volunteer labor, per diems, fuel costs, capital costs for vehicles and cold chain equipment, and the cost of vaccines, nutrition supplements, and related supplies. A full list of definitions of costed activities and resource types can be found in Supplemental table 1 and 2.
Ethical considerations
The study was approved by the Health Research Ethics Committee from the University of Nigeria Teaching Hospital and by the Sierra Leone Ethics and Scientific Review Committee. All respondents signed informed consent forms ahead of being interviewed, and were ensured that their personal details would be kept private, and excluded from the final dataset for the analysis.
Sample
The three Nigerian states were purposively selected to represent the north, middle and south of the country. Within these, LGAs and wards were randomly selected. In Sierra Leone, 6 out of the country’s 14 districts were included, with half of these administering only vaccines during campaign, and the other half delivering both vaccines and nutrition interventions. The EPI selected two of the six districts: the riverain district of Bonthe, and the mountainous district of Koinadugu, as they were expected to have incurred higher delivery costs than other districts, and the other four districts were selected randomly. In Sierra Leone, stratified random sampling was used to select a mix of public and private facilities and hospitals in urban and rural areas, while in Nigeria, simple random sampling was applied to select the facilities. The Sample Design Optimizer was utilized to explore efficient sampling designs that minimized the relative error and data collection cost. The final sample included 78 health facilities, and 30 in Sierra Leone, and data were also collected from the 10 relevant local government authorities in Nigeria, and 6 district health offices in Sierra Leone, as well as relevant state and national level offices, and implementing partners.
Data collection & analysis
Data was collected retrospectively between October 2019 and May 2021 using standardized Excel-based questionnaires, tailored to the different administrative levels. Unit prices were obtained at local level, and where not available, national or global level estimates were used instead. This was the case for salary data, fuel prices, replacement prices and useful life measurements, building costs, vaccine prices, and international shipment costs. The cost of volunteer time was estimated at the equivalent of the salary grade of the cadre that best represented their responsibilities, or at minimum wage in the case of day laborers.
Shared costs were allocated to the campaign and to activities based on time spent (for labor and vehicle costs), or the proportion of space used in the cold chain (for cold chain related costs). Capital costs were annualized over their useful life, using replacement prices, and straight-line depreciation. For the economic costs, they were discounted using a 3% rate. Costs were standardized to 2020 currency using the International Monetary Fund (IMF) consumer price index, and subsequently converted into United States dollars (USD) using the World Bank official exchange rates.−
The cost per dose delivered was calculated for each administrative level, weighted by the volume delivered and inverse probability of sampling.− The total delivery cost per dose was obtained by aggregating the unit cost at each level, and the total cost of the campaigns was calculated by multiplying the unit cost by the total number of doses delivered. In the case of Sierra Leone, a national level average was estimated, as well as group estimates for the districts that delivered nutritional supplements, and the group of districts that did not. In the case of Nigeria, we estimated average cost levels for each of the three states. As the campaigns in the three states took place during different time periods, and the national level strategy encompasses a 10-year period, we did not extrapolate the state level findings to a national level average. The datasets from Sierra Leone and Nigeria can be found on Harvard Dataverse., Two-sample t-tests were conducted to test for statistical significance between the various groups (p < 0.05). A bootstrap regression using 1000 replications was also run to generate the confidence intervals (CI).
To estimate the cost savings from integration in Nigeria, we modelled the cost of two standalone campaigns with YF and MenA in Anambra, and compared these with the cost estimated by our bottom-up costing study of the integrated YF-MenA campaign in Anambra. The cost of a YF standalone campaign was estimated based on the practices in Katsina and Rivers, and for the standalone MenA campaign, we leveraged assumptions on MenA campaign practices from other sources. In instances of uncertainty around assumptions, we opted for the scenario that would yield the lowest cost for a standalone campaign. This approach ensured that the findings of our analysis reflect a conservative minimum of the potential savings that could be achieved through integration. The assumptions for the scenarios can be found in Supplemental table 3. To exclude the impact of inflation from the model, we assumed that the two standalone campaigns would take place in the same year.