Costs of seasonal influenza vaccination in South Africa

Abstract Background Influenza accounts for a substantial number of deaths and hospitalisations annually in South Africa. To address this disease burden, the South African National Department of Health introduced a trivalent inactivated influenza vaccination programme in 2010. Methods We adapted and populated the WHO Seasonal Influenza Immunization Costing Tool (WHO SIICT) with country‐specific data to estimate the cost of the influenza vaccination programme in South Africa. Data were obtained through key‐informant interviews at different levels of the health system and through a review of existing secondary data sources. Costs were estimated from a public provider perspective and expressed in 2018 prices. We conducted scenario analyses to assess the impact of different levels of programme expansion and the use of quadrivalent vaccines on total programme costs. Results Total financial and economic costs were estimated at approximately USD 2.93 million and USD 7.91 million, respectively, while financial and economic cost per person immunised was estimated at USD 3.29 and USD 8.88, respectively. Expanding the programme by 5% and 10% increased economic cost per person immunised to USD 9.36 and USD 9.52 in the two scenarios, respectively. Finally, replacing trivalent inactivated influenza vaccine (TIV) with quadrivalent vaccine increased financial and economic costs to USD 4.89 and USD 10.48 per person immunised, respectively. Conclusion We adapted the WHO SIICT and provide estimates of the total costs of the seasonal influenza vaccination programme in South Africa. These estimates provide a basis for planning future programme expansion and may serve as inputs for cost‐effectiveness analyses of seasonal influenza vaccination programmes.

immunised to USD 9.36 and USD 9.52 in the two scenarios, respectively. Finally, replacing trivalent inactivated influenza vaccine (TIV) with quadrivalent vaccine increased financial and economic costs to USD 4.89 and USD 10.48 per person immunised, respectively.

Conclusion:
We adapted the WHO SIICT and provide estimates of the total costs of the seasonal influenza vaccination programme in South Africa. These estimates provide a basis for planning future programme expansion and may serve as inputs for cost-effectiveness analyses of seasonal influenza vaccination programmes. hospitalisations and 11,000 deaths annually, 2 with a high economic burden of approximately $270 million. 3 The health and economic burden of seasonal influenza is exacerbated by the high prevalence of comorbidities in South Africa, such as HIV, tuberculosis and other non-communicable chronic diseases, which are associated with increased risk of severe influenza. 4,5 Influenza vaccination can prevent influenza-associated illness, including among individuals at risk of developing severe complications of the disease. 6,7 The World Health Organization (WHO) recommends that individuals at risk of severe influenza-associated illness be considered for vaccination as part of seasonal influenza vaccination programmes. 8 In 2010, the South African National Department of Health (NDoH) introduced the trivalent inactivated influenza vaccine (TIV) into the public health system free of charge. 9 Annually, the campaign targets selected individuals at risk of severe influenza-associated illness for vaccination. As of 2017, this includes those aged ≥65 years, pregnant women, HIV-infected individuals and individuals aged over 6 months with other underlying medical conditions (UMC), including tuberculosis, heart disease and lung disease. 9,10 While all individuals in these target groups are eligible for vaccination as part of the vaccination campaign, pregnant women and HIV-infected adults are prioritised by the campaign. 9 An overview of the influenza vaccination programme in South Africa can be found in the Supporting Information.
Few persons in targeted groups are vaccinated each year. For example, only 5% of the total number of doses required to cover the prioritised groups were procured for use in the public health sector in 2018 (approximately 900,000 doses). 11,12 This large gap in available doses may be explained by limited economic evidence, particularly in LMIC, to support the case for increased investment in seasonal influenza vaccination, in spite of growing evidence on the benefits of seasonal influenza vaccines in high-risk groups. [13][14][15] To close these gaps, the WHO has developed a standardised mechanism to collect both national and global influenza burden data; to better determine the economic burden among risk groups; and to determine the costs and cost-effectiveness of influenza vaccination in low resource settings. 16,17 South Africa has used this mechanism to good effect in estimating the disease and economic burden of influenza as well as the cost-effectiveness of the influenza vaccination programme. 2,3,18 An assessment of programme delivery and vaccine procurement costs would be useful to identify opportunities for improved efficiency, as well as to plan for potential expansion of the seasonal influenza vaccination programme. 19,20 Furthermore, although a decline in confirmed influenza cases during the COVID-19 pandemic has been reported globally (including in South Africa) due to public health measures, this may result in future compensatory increases in influenza infections. 21 This makes it even more important to understand the costs of the influenza vaccination programme to ensure effective planning and preparedness for a potential influenza epidemic.
This study estimates the financial and economic costs of the 2018 influenza vaccination programme as the base case, as well as the costs of two expansion scenarios where coverage is increased by an absolute 5% and 10% from 2018 coverage rates. The perspective taken is that of the public provider. Currently, TIV is provided as part of the annual influenza vaccination campaign in South Africa. However, there is ongoing global discussion around replacing TIV with the more effective, but more expensive, quadrivalent inactivated influenza vaccine (QIV). 22,23 In order to inform future discussions by policy makers, we also estimate the costs of the programme under a fourth scenario where TIV is replaced with QIV. 23

| METHODS
To support decision making and implementation efforts in countries, WHO has developed the 'Influenza Vaccine Economic Value Chain' guidance package for influenza vaccination. 16,17 This package includes a Microsoft Excel-based planning and costing tool to estimate influenza vaccination costs for the different influenza risk groups-The WHO Seasonal Influenza Immunization Costing Tool (WHO SIICT). 16,17 Further details on the tool can be found in the Supporting Information.
We adapted the WHO SIICT to estimate the cost of South Africa's seasonal influenza vaccination strategy. Using data obtained from primary and secondary sources in South Africa as well as some pre-populated tool inputs, we estimated total  We used a structured questionnaire designed to collect information on resource-use quantities, based on the cost categories described in the WHO SIICT (Table S1). Data were collected from different levels of the health care system, including the national, provincial, district/municipal and facility levels. We purposively selected Tshwane District within Gauteng Province as this district has a range of urban, peri-urban and rural facilities that deliver influenza vaccination, in addition to its convenient location for collection of data within study time and resource constraints. The Supporting Information provides a detailed description of types of resource-use data collected (Table S2) and the designation of the key informants recruited into the study.

| Secondary data collection
Data for other cost inputs were extracted from several secondary sources, which can be found in Table S3. Unit price of the vaccine was obtained from the 2018 Master Procurement Catalogue compiled by the Affordable Medicines Division at the NDoH. 25 Personnel salaries were obtained from the 2018 South African Department of Public Service and Administration (DPSA) salary scales, 26 2018 NDoH and Gauteng Department of Health Annual Reports 27,28 and from the literature. 29 To estimate travel costs for planning, training and distribution activities, a rate of South African Rands (ZAR) 3.61 per kilometre (km) was applied, in accordance with South African Revenue Services guidance. 30 Unit prices of consumables (mainly cotton swabs) were obtained from existing published literature. 31 The pre-populated volume of each vaccine in centimetres cubed (cm 3 ) 32 was multiplied by the number of vaccines procured to calculate total cold chain volume used by the programme. This total volume was multiplied by the cost per cm 3 of the cold chain, obtained from the Programme for Appropriate Technology in Health (PATH) vaccine regional distribution centre cost assessment of 2011, 33 to estimate the cost of cold chain volume used by the programme. This cost was adjusted to 2018 ZAR using the historical exchange rate and the Statistics South Africa Consumer Price Index. 34,35 To account for wastage, a vaccine wastage rate of 5% was assumed, in line with median wastage rate for vaccines in single-dose vials in LMIC. 36,37 However, costs of disposal of used syringes were deemed negligible due to the small size of the influenza vaccination campaign relative to the Expanded Programme for Immunisation (EPI) programme and were therefore excluded from the analysis.
Finally, coverage estimates for 2018 were obtained from the Post Vaccination Evaluation, conducted by the NDoH, 11 while population estimates of each risk group were obtained from a study on influenza in risk groups in South Africa. 12

| Base-case analysis
Total cost of the programme was calculated by multiplying the quanti-  11 Costs were assumed to be constant across the risk groups, as the activities and cost ingredients involved in administering the vaccine at the PHC facility were the same across groups.

| Scenario analyses
We modelled two expansion scenarios to assess total costs of expanding current coverage of the influenza vaccination programme in South Africa, first by an absolute increase of 5% and second, by an absolute increase of 10% within each risk group. These scenarios were considered by steering committee members to be realistic scale- Additional costs introduced in the expansion scenario include costs of expanding cold chain capacity to accommodate additional vaccines. Given that the EPI cold chain system in South Africa is currently operating at full capacity, any additional cold chain capacity required would require capital infrastructure investments. Given space constraints within health facilities, we assumed that capital investments in cold chain would occur only at the district level. Therefore, in both expansion scenarios, we modelled an increase in regional pharmacy cold chain capacity and a corresponding increase in the frequency of vaccine distribution from regional pharmacies to health facilities. Capital costs of procuring additional refrigerators were annualised using a discount rate of 5% and a 10-year useful life. 42,43 Finally, we modelled a scenario in which TIV was replaced with the more expensive QIV in the current vaccination programme, using the same coverage rate as for the base-case scenario. In the absence of a QIV cost per dose, we assumed that the cost of QIV was 150% the cost of TIV. 44,45 3 | RESULTS

| Financial costs: Base-case analysis
Estimates of the financial costs of the influenza vaccination programme in 2018 are presented in Table 2. Costs were calculated in ZAR and converted to USD using a rate of ZAR 13.24 to USD 1.00. 34 In 2018, approximately 900,000 doses of influenza vaccines were available in the public health system, covering approximately 5% of individuals in the risk groups, resulting in a total financial programme cost of approximately ZAR 38.8 million (USD 2.93 million) and a cost per person immunised of ZAR 43.61 (USD 3.29) ( Table 2; Panel A).
The largest financial cost driver was procurement of the vaccines, contributing approximately 99% of total financial cost (Table 3). Distribution, training and microplanning also contributed to financial cost but to a much lesser degree (Table 3).

| Financial costs: Scenario analyses
Total financial cost increased to ZAR 81.3 million (USD 6.14 million) if current coverage within each risk group was increased by 5% and T A B L E 1 Seasonal influenza vaccine coverage rates for the high-risk target groups in South Africa, 2018

| Economic costs: Base-case analysis
Estimates of economic costs of the programme are presented in

| Economic costs: Scenario analyses
Overall, total economic costs increased to ZAR 230 million (USD 17.4 million) and ZAR 356 million (USD 26.9 million) when the current programme was expanded by 5% and 10%, respectively. Similarly, T A B L E 2 Costs of the seasonal influenza vaccination programme as implemented in South Africa in 2018  Table 2). In addition to procurement costs, economic costs associated with immunisation activities were observed to increase with each expansion scenario (Table 4). This was largely due to the facility fee applied to account for non-opportunistic vaccinations in these scenarios. Replacing TIV with QIV also resulted in an increase in total economic costs and cost per person immunised ( where economic costs may be a better reflection of the impact of the intervention on the health system than financial costs.
Although cost estimates across settings are generally not comparable due to differences in health system costs, we observed similar patterns in a study estimating the cost of immunising health workers in Albania. 48 Vaccine procurement cost was the highest financial cost driver of the Albanian programme. 48 Another costing study applied similar methods in Malawi to prospectively estimate the costs of introducing a maternal influenza immunisation programme, and found microplanning, training and social mobilisation to be the largest financial cost drivers, 49 contrary to our findings. This is largely due to the assumption that influenza vaccines were donated and were thus considered an economic cost input, rather than a financial one. 49 This shows the importance of local data, as the costs of an influenza vaccination programme might vary depending on country-specific contextual factors.

| Limitations
We collected data from only one province (out of nine), one district (out of 52) and six facilities (out of approximately 3500). We assumed that the programme in Gauteng Province was representative of the programme across all nine provinces. Although each province operates a facility-based programme, the number of facilities in each district, the number of patients served at each facility and management of the programme are likely to differ across provinces. The extent to which these variations may have biased our estimated costs is unclear and future studies may expand data collection to a more representative sample of provinces, districts and facilities in South Africa.

| CONCLUSION
This study demonstrates the successful use of the WHO SIICT costing tool to estimate financial and economic costs associated with the current influenza vaccination programme in South Africa. Furthermore, it provides evidence to decision makers wishing to understand the potential costs of both expanding the current influenza vaccination efforts and replacing TIV with QIV. As next steps, we suggest the use of cost estimates reported in this study as integral inputs to (i) a budget impact analysis, which considers potential savings from averted influenza-associated illnesses and (ii) an assessment of the cost-effectiveness of QIV compared to TIV.