|Publication:||Name: Journal of Health Population and Nutrition Publisher: International Centre for Diarrhoeal Disease Research Bangladesh Audience: Academic Format: Magazine/Journal Subject: Health Copyright: COPYRIGHT 2009 International Centre for Diarrhoeal Disease Research Bangladesh ISSN: 1606-0997|
|Issue:||Date: Dec, 2009 Source Volume: 27 Source Issue: 6|
1. Map of the study sites (Fig. A1)
2. Details on parameter assumptions
2.2. Private economic demand
3. Model of economic costs and benefits
[FIGURE 1 OMITTED]
2. Details on parameter assumptions
Estimates of the value-of-statistical-life (VSL) are now available for a number of less-developed countries. We extrapolate data primarily from two recent studies in Delhi, India (1) and Matlab, Bangladesh (Maskery B et al. 2008. An estimate of the economic value parents in rural Bangladesh place on ex ante risk reductions for their children. Presented at 16th Annual Conference of the European Association of Environmental and Resource Economists). Both studies estimated VSLs in the range of US$ 30,000 to US$ 100,000. The study in Bangladesh examined parents' willingness-to-pay (WTP) to reduce their children's risk of death. The Indian study examined the WTP of commuters to reduce risk of death from traffic accidents. Two other studies in India used a labour-hedonic approach (2,3) and found much higher VSLs (US$ 0.37 m-US$ 1.4 m). Given that average incomes are very low in the two slums in Kolkata, we prefer a conservative estimate of US$ 50,000.
Private economic demand
Equations A1-3 and Figure A2 show the exponential demand relationships used in the analysis. As discussed in the text, we also assume that only 80% of the population learns about the programme. For more detail on the methods used for collecting the data and the econometric results, see Whittington et al. (4).
Percentage of children aged 2-4.9 years covered at user-fee (p)=0.73 x exp (-0.14 x p) (Eq. A1)
Percentage of children aged 5-14.9 years covered at user-fee (p)=0.69 x exp(-0.27 x p) (Eq. A2)
Percentage of adults aged over 15 years covered at user-fee (p)=0.62 x exp(-0.28 x p) (Eq. A3)
[FIGURE 2 OMITTED]
Manufacturing technology for the Vi vaccine has been transferred to local producers in many countries. In Viet Nam, locally-produced Vi vaccine is sold to the public sector for around US$ 0.56 per dose in 2007 (Jodar L. Personal communication. 2007), and several private-sector Indian producers have also acquired the technology and have offered prices to the public sector of US$ 0.45 per dose or less for multi-dose vials (DeRoeck D. Personal communication. 2007). We assume that vaccines can be purchased at a price of US$ 0.45 per dose. We add 15% to the price (US$ 0.067 per dose) to cover customs, insurance, and freight based on the average percentage used by UNICEF (5). Furthermore, we assume 10% wastage of vaccines (adding US$ 0.052 after customs, freight, and insurance have been added). The total acquisition cost, including the cost of wastage, customs, freight, and insurance, is US$ 0.57 per dose.
Delivery costs depend on a number of factors and vary widely in the published literature, even for similar programmes (e.g. EPI vaccinations). The reviews of both Fox-Rushby et al. and Pegurri et al. found that most economic evaluations of vaccine lacked transparency in the calculation of costs (6,7). Furthermore, many economic evaluations assume that the vaccination programmes can be readily implemented through the existing health infrastructure (typically the country's EPI vaccination programme) with minimal additional delivery costs (7). This will not be the case with typhoid-vaccination programmes, since Vi has to be provided in campaigns outside the infants' EPI schedule because it has not been proven effective in children aged less than two years. Countries may, however, follow the WHO's recommendation to harmonize Vi vaccination with school-based diphtheria or tetanus vaccination.
The only published estimate of delivery costs for typhoid vaccines is from a cost-benefit analysis for a slum community in India, although it is based on a personal communication about an unpublished study in Viet Nam (8). The authors use a delivery cost estimate of US$ 0.9-1.7 (in 2007). Lauria and Stewart [Lauria D and Stewart J. Vaccination costs. 2007. UNC Department of Environmental Sciences and Engineering (Unpublished manuscript)] reviewed data from 22 vaccine-cost studies in low- and middle-income countries, including unpublished cost-data from several DOMI vaccine trials, and found that the median delivery cost per dose (after removing four outliers) was US$ 0.8 (mean US$ 1.1), with estimates ranging from US$ 0.10 to US$ 5.7 per dose. Because delivery costs in middle-income countries tend to be twice those in low-income countries (9,10), Lauria and Stewart's best judgement is that delivery costs are on the order of US$ 0.5 per dose for a low-income country, such as India.
We assume that delivery costs are captured in a constant marginal cost per vaccinated individual rather than including fixed (i.e. set up) costs. This implies constant returns to scale in vaccination. We assume that the marginal delivery cost per dose is the same for a school-based programme (strategy S and C) as for a community-based vaccination programme (strategy 'CA'). Although it is possible that average delivery costs for school-based programmes may be lower than for community-based programmes (because health staff-time might be used more efficiently and because less social marketing might be needed), we feel that the body of evidence is not strong enough to warrant the use of different delivery costs for our programme options.
We use Lauria and Stewart's estimate of US$ 0.5 per dose for delivery costs. Importantly, we include estimate of time/travel costs for vaccination (commonly ignored in evaluation literature). For each dose, we assume that every vaccine recipient walks 10 minutes to a nearby clinic (no financial transportation costs) where he or she spends 20 minutes waiting to be vaccinated and walks 10 min utes back. We value this time equally for adults and children at one-half of the median hourly wage in our Tiljala sample (US$ 0.18). The economic costs of travelling and waiting to be vaccinated is, therefore, US$ 0.06 per dose (0.67 hours * US$ 0.09/ hour). The final estimate of total social cost is US$ 1.13 per dose.
3. Definitions of economic benefits and costs Coverage
For ease of exposition, we assume that a programme targets a population of size Pop. This could be either the total population of the area, or the population within a specific age-group. We assume that only a fraction h of the population hears about the vaccination programme. Although this fraction would likely be related to the amount of effort (i.e. costs) spent on information and advertising, we have no information to identify this relationship, and we will assume that this fraction is exogenous to the model (we assume this fraction is 80%). The size of the population who hear of the programme is, therefore, h x Pop, which we call N.
We assume that the Government may ask users to share some cost of the programme through a user-fee p. Because Vi immunization requires only one dose, we need not distinguish between fees levied per dose vs per immunization. Vaccine recipients will face other costs in choosing to be vaccinated, both financial costs of travelling to the clinic (e.g. taxi, bus, etc.) and the economic costs of the time spent travelling and waiting in the clinic (11). In practice, these costs will vary among the population, based on their location compared to the nearest vaccination clinic and the queues at clinics$. For simplicity and because Tiljala and Narkeldanga are compact urban slums, we assume here that the total travel/waiting costs is a constant t per dose. The total cost that users face is therefore (p+t).
The proportion of people who choose to be vaccinated will be a decreasing function P of the costs that users face. The total number of people vaccinated ('coverage') is:
Coverage = N x P [p+t] (Eq. A4)
Cases and deaths avoided
The population incidence rate is I In the absence of herd-protection effects, the effectiveness of the vaccine in preventing cases is assumed to be a constant, Eff. This effectiveness is independent of coverage rates, and unvaccinated persons experience no reduction in their chances of contracting the disease, regardless of coverage. The duration of the vaccine's effectiveness (in years) is Dur. The total number of cases avoided in the population is:
Cases avoided = Dur x Eff x P[p+t] x N x I (Eq. A5)
We assume that some fraction of those who fall ill eventually die from the disease. Multiplying this case-fatality rate (CFR) by the number of cases avoided gives the number of deaths avoided:
Deaths avoided = CFR x Dur x Eff x P[p+ t] x N x I (Eq. A6)
Since Dur, Eff, N, and I are constants, the number of cases avoided and deaths avoided are proportional to the coverage rate P(x) (for clarity, we suppress notation of the coverage function P). If coverage is a monotonically-decreasing function of the costs that users face (as economic theory would suggest), the number of cases avoided decreases with increases in either the user-fee and the travel/time costs.
We might be interested in knowing how many cases are expected to continue to occur even in the presence of the vaccination programme. There will continue to be cases in three subsets of the population: (a) those who never heard about the programme, (b) those who heard about the programme but chose not to be vaccinated at cost p+t, and (c) those who chose to be vaccinated but were not protected because the vaccine is not 100% effective (Eq. A7). In the absence of herd protection, the only way to eliminate all cases in the population is to achieve 100% coverage with a 100% effective vaccine.
Remaining cases = (Pop - N) x I + Dur x N x [1-PW] x I + Dur x N x PW x I x (1-Eff) (Eq. A7)
The total cost of the vaccination programme will comprise fixed costs plus variable costs. Rather than trying to estimate fixed costs a priori, which would require information on the number and staffing of outposts (which we do not have), we assume that costs can be subsumed into a single constant marginal cost function. These variable costs are a function of the total number of doses delivered (q), which is, in turn, a function of the total coverage (again, typhoid Vi requires only one dose per immunization):
Total doses delivered = q = N x P[p + t] (Eq. A8)
As described above, we split the variable cost function V(q) into three parts: vaccine-acquisition cost, vaccine-delivery cost, and travel/waiting costs. As discussed above, we assume that the travel/waiting costs are a constant t per dose. We also assume that per-unit acquisition cost is a constant (Acq). This assumption seems reasonable in the context of evaluating a programme of fairly limited size (e.g. one neighbourhood). A city-wide, regional, or (certainly) national immunization programme might have economies of scale in acquisition or manufacturing so that marginal and average costs would vary with the number of vaccines demanded. The variable costs V(q) are:
V(q) = q x [Acq + Deliv + t] (Eq. A9)
Substituting (A8) into (A9) gives the expression for the total costs (Eq.A10). The total costs will be a function of only one varying parameter, the user-fee p.
C(p) = N x P[p+ t] x (Acq+ Deliv + t) (Eq. A10)
We examined several different measures of economic benefits in assessing whether vaccination programmes would pass a social cost-benefit test. The first benefit measure includes the cost-of-illness (COI) avoided by preventing a case of the disease. Cost-of-illness includes both direct and indirect costs, and financial and economic costs. We break down cost-of-illness further into privately-borne COI (PrivCOI) and public-sector COI (PubCOI). COI incurred in the second and third years of the programme is discounted to give a net present value [COI over duration = CO[I.sub.0] + COE/(1+disc)+CO[I.sub.2/ [(1+disc).sup.2] where disc = financial discount rate]. The benefit of a programme using this measure (avoided COI benefits) is simply the COI avoided multiplied by the number of cases avoided:
Avoided COI benefits = (PubCOI+PrivCOI) x [Dur x Eff x P(x) N x I] (Eq. A11)
The second benefit measure adds the value of mortality risk reductions by multiplying the number of deaths avoided by an estimate of the VSL. Adding this mortality-risk-reduction benefit to the COI benefits gives 'COI+VSL benefits'. It is possible that this approach double-counts private COI. For example, if VSL is estimated by directly asking about WTP for a risk-reduction programme (i.e. like the stated preference approach used by Maskery et al.), respondents could be including ex ante private COI in their WTP for the risk reduction.
Avoided COI +VSL benefits = (PubCOI + PrivCOI) x [(1-CFR) x Dur x Eff x P(^) x N x I ] + VSL x [CFR x Dur x Eff x PW x N x I ] (Eq. A11)
The third benefit measure derives from stated preference studies of what households said that they were willing to pay for vaccines. The average WTP per vaccinated person comprises per-capita expenditure (equal to the fee p) plus average per-capita consumer surplus (CS, which collapses to -1/[[beta].sub.p] in our econometric models). We multiply this average WTP measure by the number of people who choose to be immunized at user-fee p and add the public-sector treatment cost savings (which we assume people did not include in their private valuations):
WTP+avoided public COI benefits = (p+CS) x [P(x) x N] + PubCOI x [(1-CFR) x Dur x Eff x P(x) x N x I] (Eq. A12)
Note that since consumer surplus is a constant (1/[[beta].sub.p]), total per-capita WTP benefits increase linearly with the user-fee. Intuitively, as the fee increases, the people remaining in the average (because they still buy a vaccine at the higher fee) are those with higher WTP.
(1.) Bhattacharya S, Alberini A, Cropper M. The value of mortality risk reductions in Delhi, India. J Risk Uncertain 2007;34:21-47.
(2.) Shanmugam KR. Self selection bias in the estimates of compensating differentials for job risks in India. J Risk Uncertain 2001;22:263-75.
(3.) Simon NB, Cropper ML, Alberini A, Arora S. Valuing mortality risk reductions in India: a study of compensating wage differentials. Washington, DC: World Bank, 1999:29.
(4.) Whittington D, Sur D, Cook J, Chatterjee S, Maskery B, Lahiri M et al. Rethinking cholera and typhoid vaccination policies for the poor: private demand in Kolkata, India. World Dev 2008;37:399-409.
(5.) United Nations Children's Fund. Reference vaccine price list. 2001. (http://www.unicef.org/supply, accessed on 1 March 2007).
(6.) Fox-Rushby JA, Kaddar M, Levine R, Brenzel L. The economics of vaccination in low- and middle-income countries. Bull World Health Organ 2004;82:640.
(7.) Cookson ST, Stamboulian D, Demonte J, Quero L, Martinez de Arquiza C, Aleman A et al. A cost-benefit analysis of programmatic use of CVD 103-HgR live oral cholera vaccine in a high-risk population. Int J Epidemiol 1997;26:212-9.
(8.) Poulos C, Bahl R, Whittington D, Bhan MK, Clemens JD, Acosta CJ. A cost-benefit analysis of typhoid fever immunization programs in an Indian urban slum community. J Health Popul Nutr 2004;22:311-21.
(9.) World Bank. Investing in health. Washington DC: World Bank, 1993. 348 p.
(10.) Hinman AR. Economic aspects of vaccines and immunizations. CR Acad Sci/Life Sci 1999;322:989-94.
(11.) Jeuland M, Lucas M, Deen J, Lazaro N, Whittington D. Estimating the private benefits of vaccination against cholera in Beira, Mozambique: a travel cost application. J Health Econ 2009;91:310-22.
(12.) Kim D. Strategy for determining vaccination user fees and locations: a case study in rural China. Chapel Hill, North Carolina: University of North Carolina at Chapel Hill, 2007. (Doctoral dissertation).
$ In addition, waiting-times might decrease with more vaccination clinics and increase with additional expenditure on advertising ceteris paribus. Waiting-times may also decrease with increasing user-fees as the fees reduce demand and may reduce queues (the vaccine provider will, of course, try to match expected demand and staffing levels to minimize unused staff-time). Kim uses more complicated spatial/GIS techniques to model these types of trade-offs in evaluating optimal locations for vaccination clinics (12).
|Gale Copyright:||Copyright 2009 Gale, Cengage Learning. All rights reserved.|