We collected implementation cost data between January 2014 and March 2017 as part of a process evaluation of the Mozambique HPV demonstration project. The specific methods we utilized were: document review, participatory observation, and key informant interviews (KIIs).
Implementation Costing Approach
Our study was performed from the payer perspective, in this case the Government of Mozambique through the Ministry of Health(32). The micro-costing approach followed the three steps of identification, measurement and valuation(33). The resource items necessary for HPV vaccination were identified using the principle of the production function of a health care program(34). We supplemented primary data with data from WHO guidelines for estimating costs of introducing new vaccines into NIPs(35) and guidelines for HPV scale up (36), as well as published literature on HPV vaccine costing(37)(38).
Costs were divided into direct medical and direct non-medical costs(39). Resources were identified and measured using data from HPV demonstration project documents and the comprehensive multiyear plan (cMYP) of the Mozambique NIP(40). These data were supplemented with participatory observations and KIIs conducted as part of the demonstration’s project process evaluation. Gross costing methods were used to analyze data obtained in bundles from the Mozambique Ministries of Health (MOH) and Education for salaries, travel allowances, fuel and other activities implemented by the NIP (37).
Delivery strategy for the HPV vaccine demonstration project
The HPV vaccine demonstration project was conducted in three districts: Manhiça, Manica and Mocímboa da Praia. The districts were selected to reflect the three socioeconomically diverse regions of the country: south, central and northern (Figure 1).
Figure 1: Mozambique HPV vaccine demonstration project sites
The government received Gavi funding for only Manhiça district, and consequently more vaccine-related activities were performed in this district relative to the other two.
The demonstration project was school-based. Vaccines were delivered over one week with subsequent community outreach visits in the days after the school-based vaccinations. In order to simplify the identification and vaccination of adolescent girls, the 10-year age group was selected because they were more likely to be found in “a single class” at schools, enabling targeting to avoid the disruption that would occur if multiple classes were targeted. In the first year of the project, each eligible girl received three doses of the bivalent Cervarix™ vaccine, administered one each in May, June, and November in order to fulfill the WHO recommended schedule of 0, 1, and 6 months. In the second year, following revised WHO guidelines, the number of doses was reduced to two, scheduled at 0 and 6 months. These were administered in June and November to a new group of eligible 10-year-old girls.
Teams composed of a health worker and an auxiliary staff member from all health facilities in the district made visits to up to two community spots (in addition to schools) during the course of the five weekdays of the vaccination week. The teams spent about six hours in each school and six hours in each community outreach spot. In each school, teams identified a teacher to arrange a vaccination venue, register the girls prior to the vaccination day, and organize them in queues ready for vaccine administration by the health workers. The same teacher observed the girls for any symptoms of adverse effects after immunization (AEFI) for a period of up to 48 hours. Overall supervision was provided by two provincial and two district supervisors for each of the three demonstration districts. Additionally, two supervisors from the national level visited each district during the vaccination week to provide oversight and technical support.
Preparatory activities that were undertaken prior to vaccination included vaccine procurement, social mobilization, training, and development and revision of monitoring tools. Vaccines and injection materials were procured and distributed through the existing national immunization cold chain system from the national warehouse in the national capital Maputo to health facilities in the three demonstration districts. Social mobilization messages were developed, piloted and then finalized. Different types of information, education and communication (IEC) materials were developed and produced, including brochures, and radio and television media spots. Additionally, community leaders were engaged to provide community talks in all districts and community activists made door to door visits in only the one district that received Gavi funding. A call-in educational service was started to enable community members to call in and ten telephone operators were trained and provided services during the demonstration project. The three districts each conducted a launching ceremony in which costs were incurred for promotional material items such as banners, T-shirts and caps. Training was conducted in a cascade manner for the health workers and Ministry of Education staff. Monitoring tools that were developed and printed included tally sheets, registration books and vaccination cards. A post introduction evaluation was carried out in Manhiça district where the Ministry of Health received financial support from Gavi.
Base-case cost assumptions
Given that HPV vaccines were delivered to schools and communities where eligible girls were located, recipients and their caregivers did not incur any costs to seek vaccination services. As such patient costs were not considered.
The direct medical costs comprised procurement of vaccines and vaccine injection supplies, vaccine distribution, cold chain costs, social mobilization, personnel, outreach visits, and monitoring and evaluation (M&E). The total annual vaccine cost included: (1) the cost per vaccine dose ($5.70 (including shipping costs)), (2) vaccine coverage for years one and two (number of girls vaccinated), (3) number of doses per girl (three in year 1 and two in year 2), (4) the observed wastage rate of 5%, and (5) a buffer stock of 25%. Vaccines were procured at the beginning of the two-year demonstration period. Vaccine injection supplies which were procured together with the vaccines included syringes and safety boxes. Syringe quantities were assumed to equal the vaccine doses while each five-liter safety box can hold 100 syringes. The syringe wastage rate used was the standard 10% utilized by the Mozambique NIP.
Vaccine distribution costs were related to air and road transport. Cold chain costs included storage costs calculated for the period that the HPV vaccines were stored prior to their utilization. Vaccines arrived in- country six months prior to the vaccination date and required storage until the last dose had been administered in the second year. The cold chain volume in liters per month was estimated and multiplied by the unit cost (cost per liter). We took into account the decrease in vaccine volume over the months, as well as the amounts of time the vaccines spent at national, provincial, district and health facility storage facilities. HPV vaccines did not require a different distribution timeline from other NIP vaccines. Their distribution was integrated into the routine monthly mechanism existing at all health system levels. For social mobilization, we obtained unit costs from project documents to compute the total costs for training and incentivizing community leaders and volunteers. Units costs were used for the calculations of all IEC materials while lump sum quantities were used for television and radio spots development, production and broadcasting.
Personnel costs included training costs and time health workers spent on outreach visits calculated in hours spent at schools and community outreach spots multiplied by wages per hour. Outreach visit costs comprised of health worker per diems and fuel. For monitoring and evaluation, we considered national, provincial and district supervisor per diems, air tickets for national level supervisors and fuel costs. The post introduction evaluation costs included per diem and fuel costs. Lump sum costs were utilized for the development and printing of monitoring tools. All fuel cost calculations are based on kilometers travelled and amount of fuel required in liters. Training costs comprised of travel per diems, training materials and venue costs and were based on actual values incurred during the demonstration project. Education workers’ costs included training costs and time spent during vaccination activities and post vaccination observations for adverse effects which were calculated in hours and multiplied by the wage cost per hour.
Direct non-medical costs were composed of overhead administrative costs. We estimated these by taking the overall annual provincial and district immunization program overhead costs and dividing them by the 52 weeks in a year. We then calculated costs needed during the five weeks that HPV vaccine was administered plus another six weeks for preparatory activities (two weeks prior to the initial dose and one week for each of the remaining four doses that were administered throughout the demonstration project).
Estimation of the cost of a fully immunized girl (FIG) and extrapolation to a one-year national program
In order to estimate the cost per FIG, we added up all two-year total costs in all the cost categories at all levels of the Ministries of Health and Education. We then divided these by the total number of girls who completed the dosing schedule of three and two vaccine doses in years one and two respectively. We calculated credibility intervals (CIs) around all estimated FIG costs (41) by determining upper and lower bound values for the model input parameters through an assumption of 20-30% positive and negative deviation from the baseline cost values, with adjustments to ensure value ranges were plausible. Subsequently maximum likelihood was used to obtain a gamma distribution for each parameter and then 10,000 Monte-Carlo simulations were run to generate possible random costs. The resulting cumulative totals were depicted in histograms and the 2.5 and 97.5 statistical percentiles were utilized to set the confidence intervals endpoints on the graphs.
We extrapolated the cost per FIG to a one-year program targeting all 10-year-old girls in Mozambique. For this projection, new FIG costs were estimated using model input parameters that excluded the first year third dose costs, to accommodate the current two-dose HPV vaccine schedule recommended by WHO. We also made estimates for a one-year program targeting all 10-year-old girls with just one dose because of the growing body of evidence for efficacy and effectiveness of only one-dose HPV vaccination creating the possibility of future dose reduction(42)(43)(44)). The target 10 year old girl population was calculated from the Mozambique 2007 census data from which the 2017 projected population was selected(45). Values were categorized by province to simulate Mozambique’s immunization program costs calculation approach. No province straddles more than one of the three regions of the country and each HPV vaccine demonstration project district’s socioeconomic profile is similar to those of districts contained in the provinces found in each region.
All costs were collected in Mozambican Meticais and converted to US dollars using 2014 official exchange rates(46). Analyses were conducted in Microsoft Excel 2017 (Microsoft, Redmond, WA, USA) and R software version 3.5.3 2019 (Vienna, Austria)(47). We used the SQUIRE checklist when writing our report(48).