THE COSTS OF INTRODUCING A MALARIA VACCINE THROUGH THEEXPANDED PROGRAM ON IMMUNIZATION IN TANZANIA
GUY HUTTON* AND FABRIZIO TEDIOSISwiss Tropical Institute, Basel, Switzerland
Abstract. This report presents an approach to costing the delivery of a malaria vaccine through the expandedprogram on immunization (EPI), and presents the predicted cost per dose delivered and cost per fully immunized child(FIC) in Tanzania, which are key inputs to the cost-effectiveness analysis. The costs included in the analysis are thoserelated to the purchase of the vaccine taking into account the wastage rate; the costs of distributing and storing thevaccine at central, zonal, district, and facility level; those of managing the vaccination program; the costs of delivery atfacility level (including personnel, syringes, safety boxes, and waste management); and those of additional training ofEPI personnel and of social mobilization activities. The average cost per FIC increases almost linearly from US$4.2 perFIC at a vaccine price of US$1 per dose to US$31.2 at vaccine price of US$10 per dose. The marginal cost is approxi-mately 5% less than the average cost. Although the vaccine price still determines most of the total delivery costs, theanalysis shows that other costs are relevant and should be taken into account before marketing the vaccine and planningits inclusion into the EPI.
INTRODUCTION
This article presents the approach to costing the delivery ofa malaria vaccine through the Expanded Program on Immu-nization (EPI), and presents the predicted cost per dose de-livered and cost per fully immunized child (FIC) in Tanzania,which are key inputs to the cost-effectiveness analysis (re-ported in an accompanying paper1).
Cost measurement is one crucial step in presenting cost-effectiveness results, with costs being the numerator in thecost-effectiveness ratio, which gives crucial information onallocative efficiency in terms of the cost per health gain of agiven health intervention.2,3 The cost-effectiveness ratio is es-sentially calculated by dividing the net costs of a health inter-vention by the net health effects.
In conducting a cost study, it is essential to follow appro-priate methods to ensure scientific quality as well as compa-rability with studies of other health interventions. Economicevaluation guidelines have been available since the late 1960sin the days when cost-benefit analysis of development pro-jects was routinely undertaken by Organization of EconomicCooperation and Development government donor agenciesand the World Bank.4,5 In these early guidelines, detailedmethods were presented for cost measurement that were con-sistent with theories of welfare economics.6,7 By the 1980s,economic evaluation guidelines were available for specific ap-plication in the health domain.8–12 These early economicevaluation guidelines for health interventions, as well as laterones,2,3,13–16 have been widely used in the health field, andare commonly referred to as the standard by which economicevaluation studies are judged.
Although in the past it has been recognized that the appli-cation of economic evaluation in the health field was notstandardized and had a lack of guidance,17 the problem is ofa different nature now that there exist an abundance of healtheconomic evaluation guidelines, which propose a variety ofapproaches. While standardization of methods has been at-tempted by several groups in the United Kingdom, theUnited States, and the European Union, EU,14,15,18–20 com-
plete standardization of economic evaluation methods re-mains elusive.
The implication of the different approaches recommendedby these guidelines is that there remains quite some discretionto the analyst in conducting and presenting a cost study.Therefore, this present study endeavors to follow closely thehighest current standards for cost measurement, taking intoaccount the weaknesses inherent in what is essentially a deskstudy.
Previous studies on the costs of adding interventions to theEPI. The delivery of a malaria vaccine through the EPI is anew intervention that has not been implemented or modeledanywhere in the world. Therefore, in conducting a study thatmeasures the hypothetical cost of adding a malaria vaccine tothe EPI, in a first step it is important to identify previous coststudies that have measured costs of adding other interven-tions to the EPI. The main aims of this literature review wereto identify important costs items to give an indication of whatdata are easily available for different costs of the EPI, and toidentify variables, factors and assumptions that need to betaken into account in developing a generalizable cost modeland menu for a malaria vaccine provided within EPI, includ-ing both supply (health system) and demand (population)side variables. In particular, useful information on the mainfeatures of immunization programs, their costs, funding, andperformance was obtained from the World Health Organiza-tion (WHO) website on immunization financing.21
The review found that the costs of introducing a new vac-cine are essentially a function of the cost structure of the EPIand of the particular operational conditions of the program.Among the most important determinants of the incrementalcosts of adding a new vaccine into the EPI are the character-istics of the vaccine itself, the delivery modalities, and thecapacity use of the EPI.22–31 For instance, the studies on theintroduction of hepatitis B vaccine into EPI schedule showedthat at US$1 per dose, approximately 80% of the additionalcosts were due to the vaccine.32–34 The remaining costs weremainly those of supplies, distribution system (mainly coldchain), and social mobilization. However, vaccine deliverycosts vary according to the level of capacity use and volume ofimmunizations given.
Thus, the evidence shows that although the vaccine ac-
* Address correspondence to Guy Hutton, Swiss Tropical Institute,Socinstrasse 57, PO Box, CH-4002, Basel, Switzerland. E-mail:[email protected]
counts for a large part of the incremental cost of adding a newvaccine into the EPI schedule, the immunization program canincur considerable additional costs and these depend heavilyon the operational conditions of the program itself. Based onthese findings, it was justified to gather information on thecurrent status of the EPI program in a malaria-endemic coun-try for the purposes of the cost-effectiveness modeling.Therefore, a method was developed to estimate the incre-mental cost of delivering a potential malaria vaccine throughthe EPI program and it was applied to collect the data neededto calculate vaccine delivery costs in one such country.
Study setting: the EPI in Tanzania. In presenting a cost-effectiveness study based on modeled data, ideally the resultsshould reflect a specific setting. A variety of settings havebeen defined in previous cost-effectiveness modeling studies.For example, the comprehensive study of Goodman and oth-ers35 stratified sub-Saharan African countries by three in-come levels and presented cost-effectiveness simulations foreach of these.
Given the diverse characteristics of the EPI throughoutAfrica, such country stratification was not considered possiblein this present study. Therefore, a single country, Tanzania,was chosen. The EPI in Tanzania was established in 1974 as avertical program and then, as part of the health sector reformsstarted in mid 1990s, it was integrated into the Reproductiveand Child Health Unit in the directorate of Preventive Ser-vices of the Ministry of Health. Immunization services areprovided by 3,544 fixed health facilities in Tanzania, bothpublic and private for profit and non profit. A total of 10% ofthese provide outreach and mobile services.36 The privatesector in Tanzania as a whole provides approximately 40% ofhealth services. From 1996, as a consequence of the decen-tralization reforms, the management of day-to-day immuni-
zation activities at the service provisional point was left to thedistrict and municipal councils.36,37
The recent reforms created a quasi-autonomous drug pro-curement agency, the Medical Store Department (MSD),which is responsible for procurement, storage, and distribu-tion (until district level) of vaccines and related equipment.Other changes introduced in the last few years include gov-ernment financing of procurement of oral polio vaccine, anduse of kerosene in the cold chain, the integration of keroseneand vaccine distribution, supervision and monitoring in thedistrict health system.37
The vaccines provided by the Tanzanian EPI in the year2004 include bacille Calmette-Guérin (BCG) (1 dose), oralpolio vaccine (OPV) (3 doses), diphtheria, pertussis, and teta-nus–hepatitis B virus (DPT-HBV) (3 doses), tetanus toxoid (5doses), Measles (1 dose), and vitamin A (3 doses). In 2002,the immunization coverage at national level was 88% forBCG, 91% for OPV, 89% for DPT-HBV, 89% for measles,and 86% for tetanus toxoid to pregnant women. The drop outrate for DPT1–DPT3 was 6%. However, it should be notedthat the vaccine coverage rate varies widely between EPIproviders.38
A considerable effort was made in recent years to improvethe effectiveness of the program. In 2001, EPI introducedauto-destructing syringes in place of sterilizable needles andsyringes, and incinerators were constructed in all district hos-pitals. Incinerators are not available at the health center anddispensary levels.36 The vaccine wastage rate has been de-creasing in recent years, and in 2002 it was approximately 5%in most regions. However, there is a wide variability amongdistricts in wastage rates, with some of them reporting wast-age rates of up to 16%.38
Table 1 shows the cost structure of EPI in Tanzania for two
TABLE 1Cost (U.S. dollars) structure of Expanded Program on Immunization in Tanzania, fiscal years 2000–2001 and 2001–2002*
Optional information (4)Shared personnel costs 479,256 479,256Long-term training 0 51,250
Total costs (1) + (2) + (3) + (4) 15,350,248 17,111,172* Source: Ministry of Health, Tanzania.36 IEC � information, education, and communication; SIA � supplementary immunization activity.
HUTTON AND TEDIOSI120
financial years (2000/2001 and 2001/2002). In 2000/2001, theEPI budget in Tanzania was US$11.6 million for routine im-munization services and US$3.3 million for supplementaryimmunization services. The program-specific spending onroutine immunization service equated to approximatelyUS$10.6 per DPT3 vaccinated child or US$0.33 per capita.36
In 2001/2002, the budget on routine immunization increasedto US$13.6 million (an increase of 17% from 2000/2001) dueto new vaccine introduction and an increase in other expen-ditures for the program. Total expenditures in 2001/2002 wereclose to US$18 million.
Table 1 also shows the cost profile of EPI program in Tan-zania for the last two years available. Recurrent costs accountfor 67% of total cost and non-recurrent costs for 22% (mainlycold chain equipment). Vaccines and injection supplies ac-count for most of recurrent costs. However, note that the EPIuses and shares certain resources of the national health sys-tem, and the costs of these resources are not included in thesefigures. Examples include general health service personneland managers, health facility buildings, and utilities paid forby the health system (e.g., electricity). Therefore, the costsshown in Table 1 do not represent the full costs of providingthe EPI. In terms of EPI financing, the Tanzanian govern-ment increased its allocation for the program from 2000/2001to 2001/2002 when it funded approximately 44% of overallfinancing for immunization, as shown in Figure 1. The gov-ernment pays mainly for injection supplies, salaries, transportand other recurrent costs, vehicles and cold chain equipment,and some vaccines. Donors pay for vaccines, injection sup-
plies, training, monitoring and surveillance, vehicles, and coldchain and other capital equipment. Donors also fund supple-mentary immunization activities. However, according to theFinancial Sustainability Plan, the Government of Tanzania isprogrammed to take over the funding of all vaccines in thenext few years.
Tanzania’s EPI has faced several problems during the lastfew years, which are also relevant to consider when makingpolicy recommendations based on the findings of cost-effectiveness analysis. The main constraints to a further de-velopment of the immunization program, as perceived by EPImanagers, include inadequate funds; problems in storage ofcertain vaccines; delays in disbursement of funds, from bothgovernment and development partners; delays in the procure-ment process; inadequate refrigerators and spare cold chaincapacity in some districts; shortage of cold chain equipment;and lack of adequate and qualified health staff, especially atfacility level.
Study aims. The aim of the present costing study is to mea-sure the incremental costs of adding a hypothetical malariavaccine to the EPI schedule to enable estimation of cost-effectiveness of such a vaccine.
METHODS
Study perspective and choice of costs presented. After de-fining the study aims, the next step in a cost study is to choosewhich costs to include. The costs included in cost-effectiveness analysis depend first on the perspective of theanalysis, whether it be the health care system, the patients, orsociety. In this study, the cost-effectiveness analysis is per-formed according to a societal perspective, which includes thehealth system, the patient, and other groups affected by theintervention (such as the larger community). Given this per-spective, the costs included and measured must be relevant tothe objectives of the study. In this study, which examines theincremental health impact of a potential new intervention, thecost of interest is the incremental cost associated with theintervention to achieve the health effect. Given the range ofinformation needs of decision makers in the health sector,two types of incremental cost have been selected for mea-surement: marginal cost and average cost.
Marginal cost. The marginal cost consists of the additionalcosts that would be incurred when introducing a malaria vac-cine into the EPI schedule, based on new resources thatwould need to be used in the delivery of the intervention. Thisinformation is most relevant for a decision maker who has tomake resource allocation decisions, based on the immediateresource impact of an intervention. Therefore, for example,when spare capacity in the health system exists, the use of thatspare capacity is not included in the marginal cost analysis.However, when full capacity has been reached and new re-sources are needed, these are included in the marginal costanalysis.
Average cost. The average cost includes all those costs in-volved in delivering a health intervention, whether they areused specially for a new intervention, whether resources areshifted away from other activities, or whether spare capacityis used. Average costing involves sharing the costs of existingcapacity among all the interventions benefiting from thoseresources. The usefulness of presenting full economic costthrough this analysis is that it enables comparison of inter-
FIGURE 1. Expanded Program on Immunization (EPI) fundingsources, financial years 2000/2001 and 2001/2002. GAVI � GlobalAlliance for Vaccines and Immunization.
COSTS OF INTRODUCING A MALARIA VACCINE 121
vention efficiency in the long-term, where all resources can(hypothetically) be redeployed in alternative uses. Therefore,average costs are useful for cost-effectiveness analyses forlong-term planning decisions.
In both marginal and average analyses, all types of cost areincluded where necessary. In this analysis, for policy makingreasons a distinction is made between non-recurrent (capital)cost items (defined as resources that are not wholly used upwithin a one-year period) and recurrent cost items (defined asitems that are used up during a year).
Algorithm for calculating vaccine delivery cost. The spe-cific characteristics of a malaria vaccine are still unknown.This analysis is based on a hypothetical vaccine that must bestored between 2°C and 8°C, has a commercial package simi-lar to that of DPT-HBV vaccine, and requires three doses tofully immunize a child delivered at the same time as the DPT-HBV.
The vaccine delivery cost per dose (Vd) is estimated ac-cording to the formula
Vd = Pd + Dd + Sd + Md + Ed + Td + Zd (1)
where, Pd is the purchase cost per dose, Dd is the distributioncost per dose, Sd is the storage cost per dose, Md is the man-agement cost per dose, Ed is the delivery cost per dose, Td isthe training cost per dose, and Zd is the social mobilizationcost per dose.
The variables in equation 1 are covered in detail in thisreport. All variables are calculated, where relevant, underboth marginal cost and average cost scenarios. The cost perfully immunized child (FIC) with the vaccine is computed bymultiplying Vd by three. However, the average cost per FIC ismarginally greater than three times the cost per dose becauseof the dropout of infants after the first dose. To estimate thetotal number of doses required per year in Tanzania, it isassumed that the coverage rate would be the same as that forthree doses of DPT-HBV in 2003, which was 89%, with adropout rate of 6% from the first to the third dose.
Net vaccine purchase cost. In the cost-effectiveness analy-sis, different price hypotheses are used that range fromUS$1.00 to US$10 per dose. No base case is presented be-cause it may become misleading in presenting results. Instead,cost results are presented under a number of vaccine priceassumptions: US$1, US$2, US$4, US$6, US$8, and US$10.Import duties are not included because these are not an eco-nomic cost but a transfer payment.
The contribution of freight costs to the price at which thecountry receives the vaccines (i.e., including carriage, insur-ance, and freight [CIF]) essentially depends on the originalprice of the vaccine, the packed volume of the vaccines, andthe mode of transport. For DPT-HBV, the contribution offreight to the CIF price is reported in the Tanzanian MedicalStores Department (MSD) documents and in the Global Al-liance for Vaccines and Immunization (GAVI) Financial Sus-tainability Plan. However, in this analysis the price assump-tions include freight costs to the port of entry.
To estimate the total vaccine cost per dose delivered it isassumed that 5% of vaccine is wasted. The purchase costs perdose of the vaccine (Pd) is computed as
Pd = Vp�1 + Qv� (2)
where Vp is the vaccine price and Qv is the wastage rate.Storage and distribution costs. The cold chain system of
EPI in Tanzania includes five operational levels eachequipped with cold chain equipment as follows:39 1 centralvaccine store at the MSD in Dar es Salaam, 8 zonal vaccinestores at the MSD, 15 regional vaccine stores, 116 districtvaccine stores, and 3,544 health facilities (dispensaries, healthcenters, hospitals). All health facilities conduct immunizationactivities and are equipped either with a small absorptionrefrigerator and freezer operating on either kerosene or elec-tricity, or with liquid propane gas or a solar powered refrig-erator.
However, the storage and distribution system is continu-ously undergoing changes and the MSD is restructuring thestorage and distribution policy and is negotiating a new finan-cial agreement with the Ministry of Health. Currently a newmalaria vaccine would be distributed from the central MSD inDar es Salaam to the eight zonal stores and from these di-rectly to districts. The distribution from districts to the healthfacilities providing immunization services is under the directresponsibility of EPI.
The new agreement between MSD and EPI includes a tariffscheme for storage of products at central and zonal level andfor distribution from central stores to zonal stores and thenfrom zonal stores to districts. The tariffs are as follows. Forstorage of cold items at the national store, Tshs 300,000/m3
(US$286) is charged per month. For distribution of cold chainitems to all MSD zones (other than Dar South), Tshs 625,000/m3 (US$595) is charged. For distribution from any MSD zonalstore to all districts, Tshs 145,000/m3 (US$138) is charged.
These new tariffs are the most reliable information cur-rently available on the current and future cost of storage anddistribution of vaccines. Therefore, to estimate the incremen-tal cost of storing and distributing the vaccine the new tariffscheme is used, thus assuming that the MSD will distributevaccine first to zonal stores and then to districts.
To estimate the incremental cost of storage and distributionof the vaccine, the package volume requirement for transpor-tation and cold chain storage at national, zonal, and servicedelivery levels was estimated on the basis of the WHO guide-lines for estimating costs of introducing new vaccines into thenational immunization system.40 These guidelines provide amethod to estimate the total volume package required forstorage and distribution of a vaccine, taking into accountwastage rates, cold chain, and transport grossing factors. Thenew tariffs defined by the MSD for storage and distributionwere then applied to the volume package estimated for thevaccine. The estimated package volume per year was 901 m3
for storage and 684 m3 for the service delivery level. Theestimated package volume for transport per year was3,543 m3.
The MSD distributes vaccines from the central store to theeight zonal stores and from these to district stores every threemonths (four times a year). Every three months there is thusa need to store at national, zonal, and district levels one-fourth (i.e., 3 months’ worth) of the estimated package vol-ume. It is assumed that the MSD is able to distribute thevaccines to zonal and district stores within one month fromwhen it receives them, and thus every three months it has tostore the estimated volume package for a maximum of onemonth at national and zonal levels.
Districts receive the vaccines every three months and dis-tribute them to health facilities monthly. It is assumed that
HUTTON AND TEDIOSI122
every three months districts have to store the estimated pack-age volume of vaccine for a maximum of one month.
The cost of storage per vaccine dose at national level iscomputed according to the formula
Sd =SmNmFy
Nvy(3)
where Sd is the cost of storage per vaccine dose at nationallevel, Sm is the storage cost/tariff per m3 per month, Nm is thenumber of months of storage (over one year), Fy is the vol-ume package per year for storage at all levels (per m3), andNvy is the number of doses of the vaccine per year.
The storage cost at zonal and district level is assumed to bethe same as that at central level. The cost of storage at facilitylevel is computed with the same formula but the volume pack-age required, which is estimated according to WHO guide-lines, is different. This is due to the different grossing factorsthat adjust for the different presumed percentage use of coldchain capacity at different MSD levels. For the national andprovincial level, the grossing factor is 2.9, whereas for districtlevel the grossing factor is 2.2.
The cost of distribution per dose of vaccine from central tozonal stores is computed according to the formula
Dzd =DzmFt�
Nvy(4)
where � =NvT − NvD
NvT
and Dzd is the distribution cost from central to zonal storesper dose of vaccine, Dzm is the distribution cost/tariff fromcentral to zonal stores per m3, Ft is the estimated volumepackage for transport of the vaccine, NvT is the total numberof vaccine doses delivered in Tanzania, and NvD is the totalnumber of vaccine doses delivered in Dar es Salaam.
The adjustment factor � is included to account for the factthat the distribution from central to zonal store for the regionof Dar es Salaam should not be included. This is because theMSD does not charge the EPI for vaccine supplies sent to DarSouth because of its proximity to the MSD.
The cost of distribution from zonal to district stores perdose (Ddd) is computed according to the formula
Ddd =DdmFt
Nvy(5)
where Ddm is the distribution cost/tariff from zonal to districtstores per m3. The cost of distribution of the vaccine fromdistrict to the health facilities, which is under the direct re-sponsibility of the EPI, is assumed to be the same as that fromzonal to district stores. This might overestimate distributioncosts because the distance between district vaccine stores andhealth facilities is normally shorter than that between zonalstores and district vaccine stores. Conversely, it might notoverestimate these costs because of (dis)economies of scale ofdistributing less quantity of vaccine to different facilities.However, in the absence of more detailed data to confirmwhich cost determinant predominates, the same cost per cubicmeter is assumed.
The costs of cold chain storage are mainly capital costsbecause cold storage mainly consists of cold rooms and re-frigerators that last longer than one year. Only personnel and
electricity or fuel for refrigerators are recurrent costs andthese account for a marginal part of storage costs. In theabsence of detailed breakdown from the MSD, it is assumedthat capital costs account for 80% of cold chain costs andrecurrent costs the remaining 20%.
Distribution costs are both recurrent (e.g. fuel for vehicles,fares for air transport, personnel) and capital (e.g. cold boxes,costs of vehicles). Compared with the capital-recurrent break-down of cold chain storage, fuel costs in distributing vaccinesaccount for a considerable proportion of total distributioncosts. Therefore, it is assumed that 50% of distribution costsare capital and 50% are recurrent costs.
Management costs. A wide range of personnel are involvedin delivering a new vaccine, including managers, surveillancestaff, community health workers, nurses, and doctors. Thepersonnel involved in the EPI are distributed throughout allthe levels of the health care system, i.e., national, regional,district, and health facilities.
The introduction of a new vaccine in the EPI will requireadditional management costs at all levels of the EPI system. Itis thus assumed that all personnel of the EPI at national (ex-cluding the EPI manager) and regional levels, the DistrictMedical Officer, the District Reproductive and Child HealthCoordinators, the Medical Officers, and the Medical RecordsOfficers would have to allocate 10% of their working timedevoted to the EPI. These management costs are includedonly in the average analysis because it is uncertain whethernew personnel will be used by the EPI to manage the malariavaccine.
Vaccine delivery costs. The costs at the point of deliveryinclude the recurrent costs of personnel involved in the EPI atfacility level, syringes, and of safety boxes, and the capital costof waste management (other than safety boxes).
Personnel. According to interviews held with the EPI atthe national level, the employees believe that introducing anew vaccine into EPI would not require additional personnelat facility level. The main justification given for this opinion isthat EPI personnel at facility level are now integrated into theReproductive and Child Health Unit and normally they donot dedicate 100% of their time to EPI.
Based on observations of selected health facilities in Tan-zania, it became apparent that there are several ways of or-ganizing an EPI session. In Dar es Salaam infants can getvaccinated on any working day (five) of the week, while inMtwara Region in southern Tanzania vaccination availabilityvaries from selected days every three months for outreach toremote villages to two days per week for health centers.
In terms of the spare capacity of the staff to administer anew vaccine, vaccinators interviewed generally concurredthat a new vaccine requiring five extra minutes per child couldbe accommodated without needing additional staff. However,during the health facility visits the issue of lack of skilledpersonnel was often raised, thus suggesting that staff are oftenunder pressure from the volume of clients. Therefore, theimpression is that the EPI staff could probably accommodatea new vaccine using the current capacity but this may lowerthe quality of services as a whole. Thus, to maintain the mini-mum quality of services, vaccination staff will need to bestrengthened in numbers, targeting those facilities that arealready close to their limit in terms of proportion of workingtime already used.
In the marginal analysis only the incremental cost of vac-
COSTS OF INTRODUCING A MALARIA VACCINE 123
cinators is included, while in the average analysis both thecosts of vaccinators and that of other personnel at district andother facility level staff are included.
In the average analysis it is assumed that personnel of thedistricts other than vaccinators, i.e., medical assistants, healthofficers, and nurses, would have to increase by 10% theirworking time spent on vaccination with the malaria vaccine.
The incremental cost per dose for these personnel (Ild) iscomputed as follows:
Ild =WyPitPmvNp
Nvy(6)
where Wy is the annual gross wage, Pit is the % of staff work-ing time for immunization, Pmv is the percentage increase inthe EPI working time spent on malaria vaccine, and Np is thenumber of personnel.
Data for the annual gross wage, the percentage of workingtime for immunization, and the number of personnel comefrom the Ministry of Health.36 In the marginal analysis it isassumed that these personnel have enough spare capacity toaccommodate the increase in working time required by thenew vaccine and thus the incremental cost is zero.
A cost per dose of vaccine for vaccinators is estimated as-suming an administration time of 7 minutes, and the cost perworking minute of vaccinators is computed assuming 230working days per year and 6 productive hours per day.
The cost per dose is computed as
Ld =Wy
MyAt (7)
where Ld is the cost of vaccinators per dose, Wy is the annualgross wage, My is the total number of working minutes peryear, and At is the vaccine administration time. The personnelvaccine delivery cost is thus computed as average analysis �Ld + Ild and marginal analysis � Ld.
Syringes. For both marginal and average analyses, the sy-ringe incremental cost per dose (Gd) is calculated as
Gd = �NgdGi + NrdGr� �1 + Qs� (8)
where Ngd is the number of injection syringes per dose, Gi isthe unit cost of injection syringes (freight included), Nrd is thenumber of reconstitution syringes per dose, Gy is the unit costof reconstitution syringes (distribution included), and Qs isthe syringe wastage rate.
The syringe wastage rate is 10% as suggested by the WHOguidelines and confirmed by GAVI documents. The cost ofsyringes used is that of the MSD catalogue for 2004,41 whilethe distribution costs are assumed to be 3% of the cost ofsyringes.
Safety boxes. Safety boxes are present at the place of vac-cination, and after vaccination the used syringes are disposedimmediately into these. The safety boxes incremental cost perdose (Bd) is computed as
Bd =�Ns
Yb�Qb Bi
Nvy(9)
where Ns is the total number of syringes, Yb is the capacity ofsafety boxes, Qb is the wastage factor for safety boxes, and Bi
is the unit cost of safety boxes (distribution included).
The capacity of safety boxes is 100 syringes, and the wast-age rate for safety boxes used is assumed to be 11% (giving afactor of 1.11) as reported in the GAVI annual progress re-port.41 The unit cost of safety boxes comes from MSD 2004catalogue (http://www.msd.or.tz/) and includes 3% of the dis-tribution cost.
Waste management incremental cost per dose. The capitalresources required for effective waste management shouldinclude the capital cost of incinerators and any buildings re-quired to house them. Recurrent costs include those associ-ated with incinerator fuel and maintenance, training, and sala-ries of staff. However, in Tanzania only hospitals have incin-erators while in health centers and dispensaries the wastemanagement practice is to throw the safety boxes into a deephole and fire them with kerosene. This was confirmed in allthe facilities visited by the study team.
The waste management of the malaria vaccine should bethe same as that for other vaccines currently deliveredthrough the EPI, and it is unlikely that the EPI would intro-duce different waste management practices because of a newvaccine. The cost of waste management was thus consideredto be zero (except for the safety box cost, as consideredabove).
Training. The EPI personnel must be trained for the ad-ministration of the new vaccine. The training on the newvaccine can either be limited to the period just before orduring its introduction or can continue in successive years. Inthis analysis it is assumed that the training is limited to theintroduction period and has duration of effect of five years,which enables an annual value for training cost to be com-puted.
Training of health workers in Tanzania can be organized atzonal, district, and facility levels. When it is organized at thezonal and district levels the health workers get a per diem tocover the cost of being outside the health facility. It is as-sumed that in the first year of vaccine introduction five daysof training are provided at the zonal and district levels, andfour days are provided at the health facilities. An ingredientapproach is used to estimate the cost of training at each level.
The training at the zonal level is assumed to be organized asa five-day workshop in each of the eight zonal training cen-ters, with two trainers per workshop, and attended by per-sonnel at the regional and district levels (i.e., not staff fromhealth facilities), comprising one regional cold chain officer,one district cold chain officer, one medical records officer,one district reproductive and child health coordinator, andone regional reproductive and child health coordinator.
The daily cost per trainers is assumed to be US$30, theoverheads cost of each premises used for the training (one perzone) US$40, the per diem of personnel US$30, the transportcost per workshop per person US$4, stationary cost per per-son per day US$1, and tea and coffee per day US$ 2 perperson.41 The training at district level is assumed to be at-tended by one person per health facility providing immuni-zation services, and led by one trainer per district. The dailycost per trainer is assumed to be US$30, the overhead costeach premise (one per district) used for the training US$10,the per diem of personnel US$5, the transport per meetingUS$2 per person, and stationary and tea and coffee costs thesame as at the zonal level.36
The training cost per dose of vaccine at the zonal (Tz) andat district levels (Td) is computed according to the following
HUTTON AND TEDIOSI124
formula, assuming the training has duration of effect of fiveyears:
where Ntr is the number of days of training, NtrL is the num-ber of trainers, CL is the cost per trainer per day, Na is thenumber of person attending training, Ca is the per diem oftraining attendants, Cst is the cost of stationary per person perday, Ctea is the cost of tea and coffee per person per day, Nloc
is the number of premises used for training, Cloc is the over-heads cost of premises used for training per day, and Ctra isthe cost of transport per person per workshop. To estimatethe annual equivalent cost, equation 10 can be divided by 5.
The training at the facility level is supposed to be attendedby all personnel at the facility level, except those that werealready involved in training at the district level. A trainer perhealth facility providing immunization services is assumed tobe used, and the daily cost per day of training is US$ 20 plusUS$ 20 of transport cost per trainer (assuming that trainerstravel only once to each facility during the four days of train-ing).
The total training cost at facility level (Tf) is computedaccording to the formula
Tf =Ntr�NtrLCL� + Nf Ctra
Nvy(11)
where Nf is the number of facilities providing immunizationservices. The total cost of training is computed as the sum oftraining cost at the zonal, district, and facility levels:
T = Tz + Td + Tf (12)
Social mobilization. Advocacy and social mobilization ef-forts are crucial for ensuring the successful introduction of anew vaccine. The introduction of the new vaccine should be
followed by an increase in the social mobilization efforts. It isassumed that in the first years after the introduction of thevaccine into the EPI a substantial number of social mobiliza-tion activities will be organized. In the marginal analysis it isassumed that the budget for these social mobilization activi-ties would be approximately equal to the current expenditureon social mobilization (Table 1) and it is thus estimated asapproximately US$ 300,000 per year. This scale of social mo-bilization is warranted by the fact that messages will need toinform the population about the characteristics of the vaccine,and the importance of continuing other preventive and cura-tive strategies.
In the average analysis social mobilization costs are as-sumed to be US$ 450,000 per year. Thus, the addition ofUS$150,000 is assumed to account for the time dedicated tosocial mobilization efforts by the personnel already used bythe health care system. Although this amount may seem rela-tively small, the relatively low unit labor cost in Tanzaniaexplains why there is not a sharp increase in costs.
RESULTS
Target population. In Tanzania in 2003, the number of livebirths was 1,438,000, and 1,289,000 infants survived the firstyear.42 Assuming the same coverage rate reported for DPT-HBV vaccine, the total number of vaccine doses per year is
TABLE 2Target population
Population/vaccine doses % No. Source
Target population 100 1,438,000 42Target first dose 95 1,366,100 36Target second dose 92 1,322,960 36Target third dose 89 1,279,820 36Total number of doses 3,968,880
TABLE 3Summary of costs at different vaccine prices per dose*
Option and cost inclusion Recurrent/non-recurrent
Vaccine price assumption per dose
US$ 1 US$2 US$4 US$6 US$8 US$10
Average cost Recurrent cost per dose 1.41 2.41 4.41 6.41 8.41 10.41Non-recurrent cost per dose 0.07 0.07 0.07 0.07 0.07 0.07Total cost per dose 1.48 2.48 4.48 6.48 8.48 10.48Total cost per FIC 4.43 7.43 13.43 19.43 25.43 31.43
Marginal cost Recurrent cost per dose 1.35 2.35 4.35 6.35 8.35 10.35Non-recurrent cost per dose 0.07 0.07 0.07 0.07 0.07 0.07Total cost per dose 1.41 2.41 4.41 6.41 8.41 10.41Total cost per FIC 4.24 7.24 13.24 19.24 25.24 31.24
* FIC � fully immunized child.
FIGURE 2. Average cost per fully immunized child.
COSTS OF INTRODUCING A MALARIA VACCINE 125
estimated, assuming the same coverage rate reported forDTP-HBV vaccine, to be close to four million (Table 2).
Cost per FIC. Figure 2 shows the average cost per FIC ateach vaccine price per dose assessed. The average cost perFIC increases almost linearly from US$4.2 per FIC at a vac-cine price of US$1 per dose to US$31.2 at vaccine price ofUS$10 per dose.
Table 3 shows more detailed cost data indicating that mar-ginal costs are not considerably less than average costs. Themarginal cost per dose is 6–7 US cents less than the averagecost, thus making it less than US$0.2 per FIC. This differenceis less than the 5% difference between average and marginalcosts.
Table 3 also shows the contribution of recurrent and non-recurrent costs to total cost. As the vaccine price increases,the non-recurrent contribution does not change, thus giving aconsiderably greater weight to recurrent costs at higherprices.
Total cost to the EPI. Figure 3 shows the marginal costs tothe EPI at each vaccine price per dose assessed. The marginalcosts increase almost linearly from US$5.7 million at a vaccine
price of US$1 per dose to US$45.3 million at a vaccine priceof US$10 per dose. In this case, the marginal cost is morerelevant because this is the additional cost that the EPI islikely to finance, in addition to the current resources availableand the annual budget. However, the average cost is not con-siderably greater, but is somewhere in the order ofUS$250,000 more for all vaccine price scenarios less than 5%more than the marginal cost.
Data on the proportion of total costs made up of recurrentand capital costs are provided in Appendix 1, with a distribu-tion that is shown to vary between vaccine price scenarios. Ata vaccine price of US$1 per dose, the capital costs are ap-proximately 4% of the total cost, but this proportion de-creases as the vaccine price increases.
In comparison to the current budget and expenditure pat-terns of the EPI, these total costs represent a considerableimpact on the budget if the malaria vaccine were included inthe vaccination schedule (Table 1 and Appendix 1). The bud-get for 2001/2002 of US$17 million is more than three timesthe cost scenario modeled at a vaccine price of US$1 per dose.At higher vaccine prices, the costs of the malaria vaccinebecome greater than the current EPI budget, increasing tothree times the current budget at a vaccine price of US$10 perdose.
Components of cost. Figure 4 shows the percentage contri-bution of different cost components to total cost at each vac-cine price per dose (US$1, US$2, US$4, US$6, US$8, andUS$10). It demonstrates the change in the contribution ofdifferent cost components (described in the Methods) at dif-ferent vaccine prices. The conclusion is that most cost com-ponents become more insignificant as vaccine price increases.For example, all cost components except vaccine price and
FIGURE 3. Marginal costs to the Expanded Program on Immuni-zation of the malaria vaccine.
FIGURE 4. Contribution of cost components at different vaccineprices.
FIGURE 5. Vaccine delivery cost.
HUTTON AND TEDIOSI126
storage and distribution contribute less than 10% to the totalcost at vaccine prices greater than US$6 per dose.
Cold chain storage and distribution costs are shown in Ap-pendix 1, and account for most of the incremental cost of thevaccine, apart from the vaccine price itself. Storage cost perdose is approximately US$0.03, while the cost of distributingthe vaccine is US$0.08, 66% of which is accounted for by thedistribution of the vaccine from the central to zonal levels.Storage and distribution costs are expected to be the same forboth the average and marginal analysis.
Management costs are included only in the average analysisand are the same for both options. The management costswere insignificant at US$ 0.0023 per dose. All of these arerecurrent costs.
Vaccine delivery costs are shown in Appendix 1 and con-tribute US$0.22 to the average cost and US$0.20 to the mar-ginal cost. All these costs are recurrent costs. Figure 5 showsthe contribution of personnel, syringes, and safety boxes dia-grammatically. The training cost per dose is US$0.03 in bothtypes of analysis, most of which are recurrent costs. The Ap-pendix shows the breakdown by resource input, with the maincontributor (50%) being the cost of trainers. The social mo-bilization cost is US$0.11 per dose in the average cost analysisand US$0.08 in the marginal cost analysis. The entire socialmobilization cost is recurrent.
DISCUSSION
In this report, the delivery cost of a hypothetical malariavaccine was estimated on the basis of the information cur-rently available on the likely characteristics of the vaccineitself and on the EPI in Tanzania. The major strength of thisanalysis is that it is based on information on the current fea-tures of the EPI in Tanzania, uses the best available data oncosts and functioning of EPI, and draws on qualitative data
collected from experts from within the Tanzanian health sys-tem and observations of health facilities by the study team.The cost of delivery is estimated assuming different vaccineprice hypothesis from US$1 per dose up to US$10.
The costs included in the analysis are those related to pur-chase of the vaccine, taking into account the wastage rate;costs of distributing and storing the vaccine at the central,zonal, district, and facility levels; costs of managing the vac-cination program; costs of delivery at the facility level (in-cluding personnel, syringes, safety boxes, and waste manage-ment); and costs of additional training of EPI personnel andof social mobilization activities.
Although the vaccine price still determines most of thetotal delivery costs, the analysis shows that other costs arerelevant and should be taken into account before marketingthe vaccine and planning its inclusion into the EPI. This isparticularly important because new vaccines are likely to havebigger volume packages than that used in this analysis.
The vaccine delivery cost, even when the vaccine price isexcluded, is relatively high and would require additional re-sources to be allocated to the EPI. At a vaccine price of US$1per dose, the total annual cost to the EPI would be more than35% of the current budget.36 When the vaccine price in-creases to US$4 per dose, the total annual cost would increaseto more than US$ 19 million, which is slightly more the annualEPI budget in 2002.
It is thus important to bear in mind that for the vaccine tobe delivered through the EPI some investments are requiredin strengthening the program. In particular, the storage ca-pacity at the central, zonal, district and facility levels wouldneed to be reinforced.
Received September 18, 2005. Accepted for publication November25, 2005.
Acknowledgments: We thank Dan Anderegg for editorial assistance,and the members of the Technical Advisory Group (Michael Alpers,Paul Coleman, David Evans, Brian Greenwood, Carol Levin, KevinMarsh, F. Ellis McKenzie, Mark Miller, and Brian Sharp), the ProjectManagement Team of the Program for Appropriate Technology inHealth (PATH) Malaria Vaccine Initiative, and GlaxoSmithKlineBiologicals S.A for their assistance.
Financial support: The mathematical modeling study was supportedby the PATH Malaria Vaccine Initiative and GlaxoSmithKline Bio-logicals S.A.
Disclaimer: Publication of this report and the contents hereof do notnecessarily reflect the endorsement, opinion, or viewpoints of thePATH Malaria Vaccine Initiative or GlaxoSmithKline BiologicalsS.A.
Authors’ address: Guy Hutton and Fabrizio Tediosi, Swiss TropicalInstitute, Socinstrasse 57, PO Box, CH-4002, Basel, Switzerland,Telephone: 41-61-284-8127, Fax: 41-61-284-8103, E-mails: [email protected] and [email protected].
Reprint requests: Guy Hutton, Swiss Tropical Institute, Socinstrasse57, PO Box, CH-4002, Basel, Switzerland.
REFERENCES
1. Tediosi F, Maire N, Smith T, Ross A, Tanner M, Hutton G, 2006.Predicting the cost-effectiveness of introducing a pre-erythrocytic malaria vaccine into the expanded program ofimmunization schedule in Tanzania. Am J Trop Med Hyg 75(Suppl 2): 131–143.
TABLE 5Vaccine delivery costs
Cost item
Cost per dose administered (US$)
Average Marginal
Vaccine delivery 0.22 0.20Recurrent
Personnel at facility 0.08 0.06Syringes 0.12 0.12Safety boxes 0.03 0.03
Non-recurrent 0 0
TABLE 6Training costs
Cost item Cost per dose administered (US$)
Training (per year) 0.03Recurrent
Trainers 0.015Per diem 0.007Stationary 0.001Tea and coffee 0.001
Non-recurrentPremises 0.000Transport 0.004
COSTS OF INTRODUCING A MALARIA VACCINE 127
2. Drummond M, O’Brien BJ, Stoddart GL, Torrance G, 1997.Methods for the Economic Evaluation of Health Care Pro-grammes. Second edition. New York: Oxford University Press.
3. Gold MR, Gold SR, Weinstein MC, 1996. Cost-Effectiveness inHealth and Medicine. Oxford, United Kingdom: Oxford Uni-versity Press.
4. Little I, Mirrlees J, 1969. Manual of Industrial Project Analysis inDeveloping Countries. Paris: Organization of Economic Coop-eration and Development.
5. United Nations Industrial Development Organization, 1972.Guidelines for Project Evaluation. New York: United Nations.
6. Dasgupta P, 1970. An analysis of two approaches to projectevaluation in developing countries. Industrialization and Pro-ductivity Bulletin. New York: United Nations Industrial De-velopment Organization.
7. Sugden R, Williams A, 1978. Principles of Practical Cost-BenefitAnalysis. Oxford, United Kingdom: Oxford University Press.
8. Drummond MF, Stoddart GL, Torrance GW, 1987. Methods forthe Economic Evaluation of Health Care Programmes. Firstedition. Oxford, United Kingdom: Oxford University Press.
9. Levin HM, 1983. Cost-Effectiveness - A Primer. London: SagePublications.
10. Drummond M, 1980. Principles of Economic Appraisal in HealthCare. Oxford, United Kingdom: Oxford University Press.
11. World Health Organization, 1988. Estimating Costs for Cost-Effectiveness Analysis: Guidelines for Managers of DiarrhoealDisease Control Programmes. Geneva: World Health Organi-zation. WHO/CDD/SER/88.3.
12. Luce BR, Elixhauser A, 1990. Estimating costs in the economicevaluation of medical technologies. Int J Tech Assess HealthCare 6: 57–75.
13. Creese A, Parker D, 1994. Cost Analysis in Primary Health Care:A Training Manual for Programme Managers. Geneva: WorldHealth Organization.
14. Drummond MF, Jefferson TO, 1996. Guidelines for authors andpeer reviewers of economic sumissions to the BMJ. The BMJEconomic Evaluation Working Party. BMJ 313: 275–283.
15. Tan-Torres Edejer T, Baltussen R, Adam T, Hutubessy R,Acharya A, Evans D, Murray C, 2003. Making Choices inHealth: WHO Guide to Cost-Effectiveness Analysis. WorldHealth Organization, Geneva.
16. Baladi JF, 1996. A Guidance Process for the Costing Process.Ottawa, Ontario, Canada: Canadian Coordinating Office forHealth Technology Assessment.
17. Udvarhelyi IS, Colditz GA, Rai A, Epstein AM, 1992. Cost-effectiveness and cost-benefit analyses in the medical litera-ture. Are the methods being used correctly? Ann Intern Med116: 238–244.
18. Drummond M, Brandt A, Luce B, Rovira J, 1993. Standardisingmethodologies for economic evaluation in health care. Int JTech Assess Health Care 9: 26–36.
19. Rovira J, 1994. Standardising economic appraisal of health tech-nology in the European Community. Soc Sci Med 38: 1675–1678.
20. Weinstein M, Siegel J, Gold M, Kamlet M, Russell L, 1996. Rec-ommendations of the panel of cost-effectiveness in medicine.JAMA 276: 1253–1341.
2 1 . W H O , 2 0 0 4 . A v a i l a b e f r o m h t t p : / / w w w . w h o . i n t /immunization_financing/en.
22. Parent du Chatelet I, Gessner BD, da Silva A, 2001. Comparisonof cost-effectiveness of preventive and reactive mass immuni-zation campaigns against meningococcal meningitis in west Af-rica: a theoretical modeling analysis. Vaccine 19: 3420–3431.
23. Brenzel L, Claquin P, 1994. Immunization programs and theircosts. Soc Sci Med 39: 527–536.
24. Monath TP, Nasidi A, 1993. Should yellow fever vaccine be in-cluded in the expanded program of immunization in Africa? Acost-effectiveness analysis for Nigeria. Am J Trop Med Hyg 48:274–299.
25. Feilden R, 1990. Estimating vaccine costs for EPI (ExpandedProgramme on Immunization) cost-effectiveness analysis. Int JHealth Plann Manage 5: 221–226.
26. Brenzel LE, 1990. The Cost of EPI: A Review of Cost and Cost-Effectiveness Studies (1978–1987). Arlington, VA: REACHproject.
27. Shepard DS, Sanoh L, Coffi E, 1986. Cost-effectiveness of theexpanded programme on immunization in the Ivory Coast: apreliminary assessment. Soc Sci Med 22: 369–377.
28. Shepard DS, Robertson RL, Cameron CS, Saturno P, Pollack M,Manceau J, Martinez P, Meissner P, Perrone J, 1989. Cost-effectiveness of routine and campaign vaccination strategies inEcuador. Bull World Health Organ 67: 649–662.
29. Creese AL, 1986. Cost-effectiveness and cost-benefit analyses ofimmunization programs in developing countries. Jelliffe D, Jel-liffe EFP, eds. Advances in International Maternal and ChildHealth. Volume 6. New York: Oxford University Press, 901–927.
30. Martin JF, 1984. Consequences of the introduction of the newinactivated poliovirus vaccine into the Expanded Programmeon Immunization. Rev Infect Dis 6 (Suppl 2): 480–482.
31. Creese AL, Henderson RH, 1980. Cost-benefit analysis and im-munization programmes in developing countries. Bull WorldHealth Organ 58: 491–497.
32. Griffiths UK, Hutton G, das Dores Pascoal E, 2005. Cost-effectiveness of introducing hepatitis B vaccine into the infantimmunization schedule in Mozambique. Health Policy Plann20: 50–59.
33. Edmunds WJ, Dejene A, Mekonnen Y, Haile M, Alemnu W,Nokes DJ, 2000. The cost of integrating hepatitis B virus vac-cine into national immunization programmes: a case studyfrom Addis Ababa. Health Policy Plann 15: 408–416.
34. Hall AJ, Robertson RL, Crivelli PE, Lowe Y, Inskip H, Snow SK,Whittle H, 1993. Cost-effectiveness of hepatitis B vaccine inThe Gambia. Trans R Soc Trop Med Hyg 87: 333–336.
35. Goodman C, Coleman PG, Mills A, 2001. The cost-effectivenessof antenatal malaria prevention in sub-Saharan Africa. Am JTrop Med Hyg 64: 45–56.
36. Ministry of Health, 2003. GAVI National Immunization ProgramFinancial Sustainability Plan 2003. Dar es Salaam: Ministry ofHealth, Tanzania.
37. Semali IAJ, 2003. Understanding Stakeholders’ Roles in HealthSector Reform Process in Tanzania: The Case of Decentralizingthe Immunization Program. PhD Thesis. University of Basel,Basel, Switzerland.
38. Ministry of Health, 2003. Report of EPI Annual Evaluation Meet-ing. March 3–7, 2003. Iringa, Tanazania: Department of Pre-ventive Services, Ministry of Health.
39. Ministry of Health, 2004. Expanded Program on ImmunizationFive Years National Strategic Plan 2000-2004. Dar es Salaam,Tanzania, Ministry of Health.
40. World Health Organization, 2002. Guidelines for EstimatingCosts of Introducing New Vaccines into the National Immuni-zation System. Geneva: World Health Organization. WHO/V&B/02.11.
41. GAVI, 2002. Annual Progress Report to the Global Alliance forVaccines and Immunization and the Vaccine Fund by the Gov-ernment of the United Republic of Tanzania. Available fromhttp://www vaccinealliance org/home/index php.
42. United Nations Population Division, 2003. World PopulationProspects UN 2003. New York: United Nations.
HUTTON AND TEDIOSI128
APPENDIX 1
TABLE 1Total cost of introducing the vaccine in the Expanded Program on Immunization (EPI) at US$1 per dose*
Incremental cost per dose administered Total cost for EPI per year
