Health Policy and Planning Advance Access published online on November 17, 2007
Health Policy and Planning, doi:10.1093/heapol/czm041
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Cost and cost-effectiveness of nationwide school-based helminth control in Uganda: intra-country variation and effects of scaling-up
1 Department of Infectious and Tropical Diseases, London School of Hygiene and Tropical Medicine, London, UK.
2 Vector Control Division, Ministry of Health, Kampala, Uganda.
3 Schistosomiasis Control Initiative, Department of Infectious Disease Epidemiology, Imperial College, London, UK.
4 Department of Economics, City University, London, UK.
* Corresponding author. Department of Infectious and Tropical Disease, London School of Hygiene and Tropical Medicine, Keppel Street, London WC1E 7HT, UK. Tel: +44 (0) 207 927 2614. Fax: +44 (0) 207 927 2918. E-mail: simon.brooker{at}lshtm.ac.uk
 |
Abstract |
|---|
 Top Abstract IntroductionDescription of the control...Data and methodsResultsDiscussionConclusionAppendix: Resources, quantities...References |
|---|
Estimates of cost and cost-effectiveness are typically based on a limited number of small-scale studies with no investigation of the existence of economies to scale or intra-country variation in cost and cost-effectiveness. This information gap hinders the efficient allocation of health care resources and the ability to generalize estimates to other settings. The current study investigates the intra-country variation in the cost and cost-effectiveness of nationwide school-based treatment of helminth (worm) infection in Uganda. Programme cost data were collected through semi-structured interviews with district officials and from accounting records in six of the 23 intervention districts. Both financial and economic costs were assessed. Costs were estimated on the basis of cost in US$ per schoolchild treated, and an incremental cost-effectiveness ratio (cost in US$ per case of anaemia averted) was used to evaluate programme cost-effectiveness. Sensitivity analysis was performed to assess the effect of discount rate and drug price. The overall economic cost per child treated in the six districts was US$0.54 and the cost-effectiveness was US$3.19 per case of anaemia averted. Analysis indicated that estimates of both cost and cost-effectiveness differ markedly with the total number of children who received treatment, indicating economies of scale. There was also substantial variation between districts in the cost per individual treated (US$0.41–0.91) and cost per anaemia case averted (US$1.70–9.51). Independent variables were shown to be statistically associated with both sets of estimates. This study highlights the potential bias in transferring data across settings without understanding the nature of observed variations.
Key Words: Cost analysis, cost-effectiveness, economic evaluation, variation, scaling up, helminth control, Uganda
KEY MESSAGES
|
|
Introduction |
|---|
|
TopAbstractIntroductionDescription of the control...Data and methodsResultsDiscussionConclusionAppendix: Resources, quantities...References |
|---|
Cost-effectiveness analysis has become a principal tool to evaluate health interventions, guiding health policy in both developed (McDaid et al. 2003
) and developing countries (World Bank 1993; Jamison et al. 2006). Estimates of cost-effectiveness are typically taken from a single study or a few small-scale studies in different countries (Walker and Fox-Rushby 2000), with no attempt to review the possible variation in estimates. However, because both intervention costs and effectiveness differ among locations, a single estimate of cost-effectiveness is unlikely to be universally applicable (Musgrove and Fox-Rushby 2006). More probable is that costs and cost-effectiveness will vary, even within a single country. For instance, intra-country variation in costs has been demonstrated in the delivery of routine immunization in Peru (Walker et al. 2004), antenatal care in Cuba and Thailand (Hutton et al. 2004), a bednet distribution programme in Malawi (Stevens et al. 2005) and a lymphatic filariasis elimination programme in Egypt (Ramzy et al. 2005). Variations in average costs may arise in the short run from differences in the relative costs of inputs, differences in technical efficiency, or, in the long run, from factors associated with economies of scale (Folland et al. 2004). Differences may also reflect variation in respect to the demography and epidemiology of disease, availability of health care resources and system of health care delivery (Drummond and Pang 2001). Understanding how and why costs vary can help in assessing the degree to which cost and cost-effectiveness estimates can be reliably extrapolated across different settings, and can also enable health planners and policy makers to discern what drives costs and to plan future budgets (Drummond et al. 1992; O’Brien 1997; Bryan and Brown 1998; Spath et al. 1999; Drummond and Pang 2001; Walker et al. 2004). This understanding is particularly important for global health programmes which implement a common health package in a range of settings. For example, a number of initiatives are now underway which seek to control a number of tropical diseases, including those caused by parasitic helminth (worm) infections (Albonico et al. 2006; Boatin and Richards 2006; Fenwick et al. 2006; Ottesen 2006).
The staff of these initiatives, together with national programme staff, also need information on how costs change as the programmes are gradually scaled-up. In economics, changes in the level of output may change average costs; as output increases, average costs either remain constant (constant returns to scale), decrease (economies of scale) or increase (diseconomies of scale) (Folland et al. 2004). Many studies assume constant returns to scale, and take average costs per recipient and multiply them by projected output levels (e.g. Fenwick et al. 2005; Brady et al. 2006). In practice, however, available studies demonstrate that average costs vary at different levels of output (Over 1998; Mansley et al. 2002; Valdmanis et al. 2003; Elbasha and Messonnier 2004).
There is a clear need for empirical evidence to better understand variations in cost and cost-effectiveness, particularly in the context of large-scale control programmes. This paper assesses the variation in costs and cost-effectiveness of a nationwide helminth control programme, and the effect of scaling-up on costs. The specific aims are to: (1) investigate the intra-country variation in the cost and cost-effectiveness of a national school-based schistosomiasis and soil-transmitted helminth (STH) control programme in Uganda, (2) determine the effects of scaling-up on costs and cost-effectiveness, and (3) identify the main determinants of average costs.
|
Description of the control programme |
|---|
|
TopAbstractIntroductionDescription of the control...Data and methodsResultsDiscussionConclusionAppendix: Resources, quantities...References |
|---|
In 2003, the Ugandan Ministry of Health (MoH) launched its national schistosomiasis and STH control programme (Kabatereine et al. 2006a
,b). Implemented vertically through the Vector Control Division (VCD) in Kampala, the programme provides anthelmintic (deworming) treatment to schools and communities at risk of morbidity due to helminth infection. In brief, the programme comprises the following activities: community sensitization, training of teachers and community drug distributors (CDDs), and school-based delivery of two anthelmintic drugs. Mass treatment with praziquantel to treat schistosomiasis and with albendazole to treat soil-transmitted helminths was given to all schools and communities in targeted areas. Treatment in schools is carried out by teachers and in communities by CDDs. The programme manager and VCD headquarters staff have overall responsibility for the programme and regularly visit districts to monitor progress. Implementation of the programme at the district level is undertaken by District Vector Control Officers (DVCOs) and district health teams.
To help create awareness and political engagement, a series of national workshops were held in Kampala between 2001 and 2005 (two in 2001, two in 2002 and one each in 2004 and 2005). The implementation of control began with a pilot phase from April to October 2003 targeting 400 000 people, with one sub-county selected for mass treatment in each of the 18 most affected districts (Kabatereine et al. 2006a). In 2004 the number of sub-counties targeted in each of the 18 districts was increased, and in 2005 the programme was expanded to include 23 districts, targeting 2.3 million people (Kabatereine et al. 2006b). In each district, training workshops provided teachers and CDDs with a basic understanding of schistosomiasis and STH, and of how to complete record forms and to administer tablets. The design of training and number of participants varied between districts. Health education messages were delivered through posters, booklets and audio and film media. All information, education and communication (IEC) material was translated into various local languages.
Imported drugs were cleared at Entebbe airport by the Uganda National Medical Stores, who transported them to VCD headquarters. Drugs and IEC material were either transported to the districts by VCD or collected by the districts during routine visits to Kampala. Drug registration and treatment included compiling school enrolment data and community census information to determine the target population and drug needs. The number of tablets provided to each school was calculated on the basis of treatment registers completed by head teachers and CCDs. The drugs were delivered to each school by the DVCOs and were received by the head teacher. Tablets were then administered by teachers on a specified day in all schools under the supervision of the head teachers and community health workers. In communities, treatment was provided by CCDs. Praziquantel (25 mg/kg) was administered to individuals on the basis of height, using locally made height poles, and every individual was given a single dose of albendazole (400 mg). All unused tablets were recovered by DVCOs who also compiled a report of activities.
|
Data and methods |
|---|
|
TopAbstractIntroductionDescription of the control...Data and methodsResultsDiscussionConclusionAppendix: Resources, quantities...References |
|---|
Only costs associated with school-based treatment are considered here because of the global focus of helminth control on the school-age child (Bundy et al. 2006
) and the availability of detailed effectiveness data for schoolchildren (Kabatereine et al. 2007). Cost analysis
Cost data were collected retrospectively from the VCD team in Kampala and from six of the 23 intervention districts (Figure 1). Districts were chosen to reflect differences in disease transmission (Kabatereine et al. 2004) and in socio-economic and health service infrastructure. Data collection was carried out between February and June 2006. A semi-structured questionnaire was drafted and was revised and amended during joint discussions with MoH officials. Data were collected by interviews with district officials using the final questionnaire and by consultation of the programme accounting system in Kampala. Documentation related to expenditure had been checked by each district accountant for accountability and cross-checked by the research team for accuracy.
|
The perspective adopted in the evaluation was that of the government, rather than society, since the costs of accessing treatment were negligible as children were treated in their own schools. Both financial and economic costs were estimated. Financial costs represent cash expenditure paid for the implementation of the intervention on an annual basis. Economic costs include the opportunity cost of using existing Ministry of Health staff and school teachers as well as annuitized capital costs, and represent the true cost of any intervention. Opportunity costs for staff were calculated from salary costs, based on Ugandan civil service pay scales for 2005. Capital costs were annuitized over the useful life of each item using a discount rate of 3%, consistent with the recommendations of the World Bank (1993
). Such annuitization enables an equivalent annual cost to be estimated and reflects the value-in-use of capital items, rather than reflecting when the item was purchased. The assumed useful life of buildings was 30 years, vehicles 7.5 years, motorcycles 4 years and computers 3 years. Vehicle running costs also included maintenance and insurance. The purchase, freight and insurance of drugs was paid in foreign currency (US$). All other costs were paid in Uganda Shillings (USh) and converted to US dollars using official exchange rates, based on average yearly exchange rate: 1 US$ = 1777 USh in 2003, 1807 USh in 2004 and 1844 USh in 2005 (http://www.oanda.com/convert/classic). Monetary costs were adjusted for inflation over time using the Gross Domestic Product (GDP) implicit price deflator (http://ifs.apdi.net/imf/logon.aspx) and expressed in 2000 prices. Details on the resources employed, their unit costs and quantities consumed are provided in the appendix. All costs directly related to research activities were excluded. The cost data are organized into six main cost centres: (1) programme running costs; (2) community awareness; (3) training; (4) imported drugs; (5) drug registration and distribution; and (6) IEC material. The different cost components of the intervention were identified using an ingredients approach, considering both the number of units and the prices of units in local currency (Ugandan Shillings). The unit cost data were combined with numbers treated to calculate, on a district-by-district basis, the average cost per child treated. The relationship between the cost per child treated and the percentage of overall costs due to different cost centres and other independent demographic and geographic variables was assessed using a non-parametric Spearman rank correlation.
Effectiveness
Evidence of the programme effectiveness was measured in terms of anaemia cases averted. Epidemiological data were collected prospectively through longitudinal surveys conducted in 30 schools between 2003 and 2005. The details of the sampling strategy, survey design and procedures are provided elsewhere (Brooker et al. 2004; Kabatereine et al. 2007). Population-based measures of programme impact included parasitological and haematological data which were collected from randomly selected schoolchildren who were followed up over 3 years. Anaemia is defined as haemoglobin concentration (Hb) < 110 g/L. The current analysis focuses on those districts where cost data were collected, thereby excluding effectiveness data from Arua, Bugiri and Mayunde districts. The number of cases of anaemia averted was calculated by multiplying the absolute difference in proportion of anaemia cases averted between 2003 and 2005 by the total number of children treated. This was calculated on a district-by-district basis, as well as, overall, assuming the mean difference in proportion of anaemia cases averted among districts.
Cost-effectiveness analysis
The counterfactual is defined as ‘do-nothing’. This is justified on the basis that prior to the current control programme, no efforts were made to control helminth infection in the country, with only passive detection of cases in health centres and presumptive treatment, although in practice, anthelmintic drugs were rarely available. Cost-effectiveness is defined in terms of the cost per case of anaemia averted, and cost-effectiveness ratios are based on annual economic costs.
Sensitivity analysis
Sensitivity analysis allows for uncertainty within the economic evaluation. It shows how responsive the result is to changes in key economic parameters but also gives an indication of the robustness of the estimate to changes in unknown variables. Sensitivity analysis was undertaken to investigate the effect on the results of varying the discount rate (reduced to 1% and increased to 10%), the prices of the drugs (reduced by 10% and 20% to reflect the use of cheaper drugs in the future) and effectiveness of treatment in reducing the proportion of anaemia cases (reduced by 33% and 50% to reflect differences in the impact of treatment on anaemia in different transmission settings). One-way scenario sensitivity analysis was carried out to assess the impact of key variables on estimates of the cost per anaemia case averted.
|
Results |
|---|
|
TopAbstractIntroductionDescription of the control...Data and methodsResultsDiscussionConclusionAppendix: Resources, quantities...References |
|---|
Total financial and economic costs
The total financial cost of the intervention in the six districts was estimated at US$161 312. The financial costs per district ranged from US$18 015 in Masindi district to US$33 809 in Hoima district. The economic cost of the intervention was calculated by valuing staff time and annuitizing capital costs to provide an equivalent annual cost. The economic costs of the intervention in each district are summarized by the major cost centres in Table 1. The total economic cost was estimated at US$218 303: ranging from US$25 624 in Masindi district to US$44 958 in Hoima district. In each district, the largest individual cost item was the purchase of drugs, ranging from 23.6% of total costs in Masindi district in 2003 to 52.1% in Moyo district in 2005. Community sensitization activities and IEC materials were the next largest items (Table 1).
|
Costs per children treated
The overall financial cost per child treated in the six districts was US$0.39. The total economic cost per child treated in the six districts was US$0.54, which includes the imputed value of labour as well as annuitized capital costs. Considerable variation in the economic costs per child treated existed between districts and between years, ranging from US$0.41 to US$0.91 (Table 2). The economic delivery cost per child treated (which excludes drug cost) also varied considerably: US$0.19–0.69. The cost per child treated is highly sensitive to the total number of children treated (Figure 2). Increasing the number of children treated can significantly decrease the cost per child treated (Figure 2a; Spearman's rho: –0.93, P < 0.001), suggestive of economies of scale. Similar economies of scale were observed in the delivery cost per child treated (Figure 2b; Spearman's rho: –0.93, P < 0.001).
|
|
In order to investigate possible causes of observed variation in costs, the relationship between delivery cost per child treated and the percentage of overall costs due to different cost centres was investigated. Cost per child treated was significantly associated with the percentage of overall costs due to sensitization and awareness (Spearman's rho: 0.769, P = 0.0002). The majority of the costs involved here are per diem (allowances) rates paid to district officials, which ranged from US$4.95–15.44, although the correlation between allowance rates and cost per child treated was non-significant (Spearman's rho: 0.19, P = 0.444). Differences in demographic and geographic factors, including distance of each district from Kampala, geographical area and population density of the district, and differences in epidemiological factors, such as baseline intensity of infection and reduction in infection following treatment, were not significantly associated with costs per child treated.
Cost-effectiveness
Among the 1455 children monitored for the 3-year period in the six districts, the percentage of children anaemic, defined as Hb<110 g/L, fell from 35.2% in 2003 to 18.5% in 2005, following three rounds of treatment. This translates to a 52.5% reduction in the proportion of anaemia cases within the study population. Table 3 reports the proportion of anaemia cases averted over the 3 year period by district. Overall, 0.4 million children were treated at an estimated cost of US$3.19 per case of anaemia averted. Cost-effectiveness ranged from US$1.70 in Moyo district to US$9.51 in Masindi district. Cost-effectiveness decreased with increasing cost per child treated (Figure 3a; Spearman's rho: 0.940.19, P = 0.005) and increased with increasing difference in the proportion of anaemia averted as a result of the intervention (Figure 3b; Spearman's rho: 1.0, P < 0.0001). This suggests that neither costs nor effectiveness are constant and therefore cost-effectiveness varies between districts. Figure 3c indicates a negative association between cost-effectiveness and the number of children receiving treatment (Spearman's rho: –0.828, P = 0.04), suggesting that there are increasing returns to scale in cost-effectiveness with respect to the target population.
|
|
Sensitivity analysis
The sensitivity of cost-effectiveness (cost per anaemia case averted) to variation in key parameters was explored (Table 4). Varying the discount rate made little difference to the estimate of cost-effectiveness. Reducing the prices of drugs by 10% and 20% reduced the cost per anaemia case averted to US$3.07 and US$2.94, respectively.
|
|
Discussion |
|---|
|
TopAbstractIntroductionDescription of the control...Data and methodsResultsDiscussionConclusionAppendix: Resources, quantities...References |
|---|
The cost of school-based control of helminth infection has been widely documented in a number of pilot programmes (Guyatt et al. 1993
, 1994; Holland et al. 1996; Partnership for Child Development 1998, 1999; Mascie-Taylor et al. 1999; Guyatt 2003). Few studies, however, have looked at costs of school-based control under nationwide programmatic conditions (Sinuon et al. 2005; Gabrielli et al. 2006). This current study is the first to document both the costs and cost-effectiveness of a national school-based control programme involving mass treatment for schistosomiasis using praziquantel and for intestinal nematodes using albendazole. The overall economic cost per child treated in the six districts was US$0.54, the overall financial cost per child treated was US$0.39, and the cost-effectiveness was US$3.19 per case of anaemia averted.
These estimates fall below the range of estimates from the experience of the Partnership for Child Development in Africa, where the financial cost per child treated with praziquantel and albendazole was estimated to be US$1.22 and US$0.24, respectively, in Ghana and US$0.79 and US$0.23, respectively, in Tanzania (Partnership for Child Development 1998, 1999). The related economic costs were US$2.94 and US$0.27 in Ghana and US$1.32 and US$0.26 in Tanzania. The programmes in Ghana and Tanzania included prior screening of urinary schistosomiasis at the school level using a questionnaire about symptoms of urinary schistosomiasis, administered by teachers, which added to overall costs. Such an approach is not applicable for intestinal schistosomiasis—the species endemic throughout Uganda—because of the non-specific nature of its symptoms, and therefore mass treatment was provided to all schools in target sub-counties in this study. Furthermore, both the praziquantel and albendazole used in the Ghana and Tanzania programmes were proprietary and not generic products and therefore cost more than in the Uganda programme, where the drugs used were generic products (costing US$0.20 compared with US$0.68 in Ghana and Tanzania). However, our estimates include start-up costs and central running costs, which were excluded in the Ghana and Tanzania estimates. In Burkina Faso, a crude macro-costing of overall costs of a combined school-based and community-based national control programme estimated the financial cost per child of providing praziquantel and albendazole to be US$0.32 (Gabrielli et al. 2006), although this is likely to be an underestimate because of the methodology adopted.
Regarding cost-effectiveness, Guyatt et al. (2001) estimated that the cost per anaemia (Hb < 110 g/L) case prevented over 15 months as part of the Tanzania programme could be US$7.43 using the existing school system to deliver anthelmintics. This higher estimate is due to the higher costs mentioned above and because the intervention only reduced anaemia by 25% in Tanzania. In a study on the island of Zanzibar, Stoltzfus et al. (1998) estimated that the cost per moderate to severe anaemia case (Hb < 90 g/L) averted over 1 year for thrice-yearly mebendazole treatment was US$3.57, increasing to US$16.30 for a case of severe anaemia averted (<70 g/L).
Our study showed that estimates of cost and cost-effectiveness differ markedly with the total number of children treated. Specifically, average costs per child treated ranged from US$0.91 at an output of 7161 children treated to US$0.41 at an output of 37 032 children treated. Over the same output range, delivery costs ranged from US$0.69 to US$0.19. It is also shown that cost-effectiveness increases with increasing output. Various reasons might explain the occurrence of these economies of scale. First, a number of the costs are fixed, and therefore increasing output reduces average fixed costs per child treated. Second, there is increasing ease, through better organization, learning-by-doing and more efficient processes, in implementation as the programme expands (Elbasha and Messonnier 2004). It is possible that further expansion of the programme into more remote areas may entail diseconomies of scale due to greater transport costs and stretched administrative structures and human resources (Johns and Torres 2005). Economies of scale have previously been documented for cancer detection programmes in the USA (Mansley et al. 2002), mass polio immunization campaigns in China (Zhang et al. 1998), vaccination sites in Bangladesh (Valdmanis et al. 2003), a national insecticide-treated net programme in Malawi (Stevens et al. 2005) and a shopkeeper training programme for improving malaria home management in Kenya (Goodman et al. 2006). In the Malawian bednet programme, the scale efficiency savings were mostly related to lowering product or procurement costs (Stevens et al. 2005). Together with these studies, our findings confirm the assertion of Jacobs and Baladi (1996) that assuming constant returns to scale is unlikely to be reliable.
This study also highlights the substantial variation between districts in the cost per individual treated with praziquantel and albendazole and in cost-effectiveness. We found that the cost per schoolchild treated was lowest (US$0.41) in Moyo district and highest (US$0.91) in Masindi district. Cost-effectiveness ranged from US$1.70–9.51 among districts. Because the same costing methods were used in each district, we can exclude methodological inconsistencies as a major source of variation. The results represent a first initial analysis of why costs and cost-effectiveness vary within a country. In economic terms, differences in costs may reflect underlying differences in the underlying production and cost frontiers and in the technical efficiency in delivering the intervention (Folland et al. 2004). We found that the percentage of total costs attributed to community sensitization differed across districts and was statistically associated with the delivery cost per child treated. Differences in these costs were predominantly due to the higher number of participants, especially supervisors from the district, included in the sensitization, and their allowances and salary costs. Because district officials are paid an allowance for such supervision, there is an incentive for some district officials to increase the amount of supervision, possibly leading to inefficiency.
We did not observe that costs varied according to the epidemiology, geography or demography of the district. Hutton et al. (2004) found that the major determinants of the costs of antenatal care in Cuba and Thailand were staffing patterns and productivity, where productivity was assessed using data envelopment analysis (DEA) (Charnes et al. 1995). This analytical approach was employed by Valdmanis et al. (2003) in evaluating vaccination sites in Bangladesh where they identified levels of output which were inefficient in terms of both technical and scale efficiency. Unfortunately, the small number of implementation units (districts) included in the present study precluded the use of DEA and so it was not possible to identify the optimal average costs and scale of operation to maximise technical efficiency. The current study needs to be repeated using larger sample sizes to quantitatively investigate the existence of technical inefficiencies. Further investigation of why costs and effects vary within different settings and between countries would also allow some judgement to be made about the relative impact of independent variables on programme costs and cost-effectiveness in a range of settings, and the extent to which cost estimates can and cannot be generalized to other settings. To guide this empirical work there is a related requirement to develop a common analytical framework for assessing cost variation (Hutton et al. 2004).
There are several qualifications in the present analysis which justify attention. First, although we have inferred that the reduction in the prevalence of anaemia was due to the intervention, one could argue that external factors may be responsible for the observed changes. Cluster-randomized trials are the accepted gold standard for evaluating health interventions delivered at the community level (Kirkwood et al. 1997). In the Uganda programme, however, it was not possible to study control cohorts of children because it was felt that a randomized controlled design, a so-called probability design (Habicht et al. 1999), would not bear relevance to the operational reality of the national programme and would be politically difficult to implement and ethically inappropriate. As a result, there is an opportunity for chance and bias to contribute to the differences observed compared with randomized controlled trials. However, robust statistical analysis indicated that improvements in haemoglobin were largest for children who harboured the heaviest infections at baseline and that observed changes in infection patterns were in accordance with predictions arising from independently validated mathematical models of transmission dynamics (Kabatereine et al. 2007).
A second limitation is that the number of anaemia cases averted is an intermediate health outcome, which does not translate into a universally comparable health outcome measure such as deaths or disability-adjusted life years (DALYs). However, the basis for converting observed changes in patterns of helminth infection and nutrition into DALYs remains controversial (King et al. 2005; Hotez et al. 2006). To date, only one theoretical study has compared the cost-effectiveness of helminth control in relation to other programmes on the basis of DALYs (Warren et al. 1993), and this included a number of assumptions which have been subsequently questioned (Evans and Guyatt 1995). Estimation of alternative outcome measures such as quality-adjusted life years (QALYs) remains problematic in a sub-Saharan African setting, where individuals suffer multiple health insults and are typically unable to distinguish between conditions (Kirigia 1998; Nyandieka et al. 2002). A second alternative outcome measure is the proportion of individuals harbouring a heavy infection (Guyatt et al. 1994), since morbidity is associated with prevalence of heavy infection. WHO (2002) provides definitions of heavy infection based on the intensity of infection as assessed by faecal egg counts. However, these units are specific to individual helminth species, making the definition of a single, multiple-species threshold impossible. Comparison of the cost-effectiveness of school-based helminth control in relation to other public health interventions requires a more universal unit. The advantage of measuring cost-effectiveness in terms of anaemia is that it is an easily assessed outcome, which has been used to evaluate a number of tropical disease interventions (Stoltzfus et al. 1998; Guyatt et al. 2001; Wiseman et al. 2003; Baltussen et al. 2004). It is recognized, however, that the use of anaemia may miss the more subtle health benefits of deworming, such as improved growth and education (King et al. 2005; Hotez et al. 2006).
There are a number of policy implications arising from this study. First, the analysis presents costs that are likely to be representative of a full-scale national programme and suggests that the programme is affordable. In particular, the cost estimates support the conclusions of earlier studies (PCD 1998, 1999; Guyatt 2003) which suggest that regular school-based delivery of simple and safe health interventions is a relatively low cost approach.
Second, the existence of intra-country variation in costs and variable returns to scale clearly indicates that comparison of costs and cost-effectiveness across programme settings and time periods could be misleading unless the effect of differences in input prices and output are taken into account. This is especially important in relation to forecasting costs and cost-effectiveness (Mansley et al. 2002). Many estimates of hypothetical public health programmes assume average cost will remain constant in relation to the population served (Fenwick et al. 2005; Brady et al. 2006). This assumption is, as indicated here, invalid and could lead to inaccurate cost projections. A further implication of the existence of economies of scale identified by Mansley et al. (2002) relates to comparing different interventions. Given that the cost-effectiveness is dependent on output, it is valid only to compare different interventions with similar outputs, or undertake some form of analytical adjustment to empirical estimates (Elbasha and Messonnier 2004).
Third, the results indicate that substantial variation in intervention costs exists within a single national programme. As such, it is important to carefully consider which costs can be reliably extrapolated across different programmes. Further empirical studies, coupled with the development of modelling techniques, can inform future extrapolations. Such studies can also identify potential cost savings and technical efficiencies, and thereby inform policy decisions and promote long-term sustainability of national programmes.
Finally, there is recent interest in the possibility of simultaneously treating a number of parasitic diseases as part of an integrated control package (Hotez et al. 2006; Lammie et al. 2006). Adding more treatments to the current programme may yield economies of scope resulting in lower average costs (Folland et al. 2004). However, this may also cause diseconomies of scope (increasing average costs), whereby adding more treatments overloads capacity and the current treatment is delivered less efficiently (Johns and Torres 2005). This aspect deserves critical attention as integrated programmes are rolled out.
|
Conclusion |
|---|
|
TopAbstractIntroductionDescription of the control...Data and methodsResultsDiscussionConclusionAppendix: Resources, quantities...References |
|---|
Economic evaluation has become a key criterion relevant for priority setting in health and in planning health care interventions. The current analysis is the first to document both the cost and cost-effectiveness of national school-based helminth control and the first to document the intra-variation in both costs and cost-effectiveness. We report the existence of economies of scale and intra-country variation in costs and in cost-effectiveness, and present an initial analysis of the causes of observed variation. The findings highlight the potential bias in transferring data across settings without understanding the nature of observed variations. Failure to do so will ultimately hinder the efficient allocation of health care resources. However, the consistency in the findings suggests that it may be possible to adjust for such variation in future analysis and the challenge remains to develop an analytical framework for understanding and assessing the extent and causes of cost variation. More evidence is clearly necessary on the cost-effectiveness of nationwide control under a range of programmatic conditions and on the underlying causes of variation in cost and cost-effectiveness.
|
Appendix: Resources, quantities and unit costs |
|---|
|
TopAbstractIntroductionDescription of the control...Data and methodsResultsDiscussionConclusionAppendix: Resources, quantities...References |
|---|
The resources employed, the quantities consumed and unit costs are described below under the different cost centres. Many of the costs were divided equally between the school-based programme and the community-based programme, although a cost analysis is only presented for school-based treatment (see Table A1).
|
Programme running costs
The programme has a main office at the VCD headquarters in Kampala, which incurred expenses such as telephone, stationary, computers and vehicles. The proportion of time staff at VCD in Kampala devoted to the programme was estimated. The additional resources used in making the intervention available were also estimated, including new capital costs such as vehicles, building space and equipment.
The financial cost of a new building in Kampala used to store drugs and other equipment and to house some of the programme officers was estimated on the basis of the cost of constructing and furnishing the building (US$87 000), annuitized using an estimated useful life of 30 years. An estimated 10% to cover annual utilities was included. The building is shared with the onchocerciasis and filariasis control programmes and therefore it was assumed that 45% of the building costs were attributable to the current programme. Between 2002 and 2005, five vehicles were purchased by the programme at a total cost of US$134 183. The costs of these vehicles were annuitized over a useful life of 7.5 years and it was assumed that 70% of costs were attributable to programme activities. Annual expenditure on services and repairs was assumed to be 10% of annualized capital costs. Allowances of drivers and staff from the headquarters visiting districts were included in district-level cost estimates. Fuel costs to each district were calculated using MoH guidelines for distance from Kampala to specific district capitals. A computer and fax machine were also purchased at a total cost of US$2245, assuming 100% allocation to the programme and a 3-year useful life. Annuitized capital costs were allocated equally across the 23 districts, attributing half the cost to school-based treatment and half to community-based treatment. Each of the districts was provided with a motorcycle at a cost of US$2899 each, assumed to have a useful life of 4 years and estimated to have an annuitized cost of US$633, and split 50:50 between school- and community-based treatment.
Community sensitization
Prior to treatment, a series of meetings were held with community leaders and school committee members. The format of these meetings and the number of participants varied between districts. A mobile film team from MoH headquarters visited each district and showed a film in several communities to raise awareness about schistosomiasis and soil-transmitted helminths. The opportunity cost for using existing district health officials was estimated.
Training
The costs of the national training workshops were US$16 980 in May 2001, US$11 285 in November 2001, US$13 500 in June 2002, US$26 000 in December 2002, US$20 000 in April 2004 and US$22 107 in April 2005. Attended by national staff and district health staff, the workshops provided general information on schistosomiasis and STH and the national programme, as well as training on treatment registration and recording and drug administration. The total cost of these workshops is divided equally among the 26 districts, allocating half the cost to school-based treatment and half to community-based treatment. The opportunity cost for using existing district health officials for the training and the time of teachers was estimated.
In 2003, a training manual was developed by staff from VCD and SCI. The unit cost was US$10.00, and 10 copies were provided to each district in 2003; this cost was shared 50:50 between school-based and community-based delivery. At the district-level, the training of school teachers and community drug distributors included public awareness, drug treatment and treatment monitoring and record keeping. The expenditure for this training included trainee transport and lunch allowance, stationary (typically exercise books, pens, marking tape, permanent markers and flipcharts), district training facilitators per diem, district drivers per diem and fuel. The estimate of costs also included the per diem and fuel costs of national staff from Kampala attending the district-level training. The unit cost and quantities of each cost element varied between districts and had to be estimated separately.
Drug distribution and treatment
Praziquantel tablets were supplied by Shin Pong Pharmaceutical Company (Kyonggi, South Korea) at a unit price of US$0.072 per 600mg tablet. Assuming 2.5 tablets per child, the drug cost per child treated was US$0.18. Albendazole tablets (400mg) were supplied by International Dispensary Association (Amsterdam, Netherlands) at a unit price of US$0.023, including CIF. The Uganda National Medical Stores cleared the imported drugs and transported them to VCD headquarters in Kampala at a cost of 5% of the drug price. Drug distribution and treatment included school, drug delivery, supervision and recording of treatment, and collection of treatment registers and unused drugs. Per diems or allowances (which varied between districts) were paid to MoH staff within a district and to community health workers to perform these activities; however, teachers were not paid. Fuel and stationary costs were also estimated. The initial cost of locally produced registers to record treatment was US$2.53 but this was subsequently reduced to US$1.64. In 2003 and 2004 the cost of locally manufactured height poles was US$1.29; in 2005 modifications of the pole reduced this cost to US$0.34. The mean treatment dose per child and adult was estimated from treatment registers. Based on experience of other programmes (PCD 1999) and local experience, the wastage rate of drugs was assumed to be 1%. Where activities covered both school-based and community-based delivery of treatment, the costs of the activity were shared 50:50 between the two delivery systems.
Production and distribution of IEC material
Health education messages were delivered through posters, booklets, films and radio shows. Information, education and communication (IEC) material included posters, leaflets and question and answer booklets. These were developed in English and then translated into various local languages by the Health Education department of the Ministry of Health at a cost of US$26 000. This cost was again divided equally among the 26 districts. The distribution channel for the IEC material was the same as the drugs. In addition, an 18 minute educational video film and a 5 minute advocacy film was produced locally and shown widely in each district. During the treatment period, which extended from April to July, radio talk shows were aired frequently on appropriate local FM stations encouraging people to take their drugs. The cost of these shows ranged from US$378.45 in Moyo district to US$30.68 in Masindi district. The cost elements under this cost centre were shared in the proportion 50:50 between school-based treatment and community-based treatment.
|
Acknowledgements |
|---|
In each district, we would like to thank the District Directors of Health Services and the District Vector Control Officers who assisted in data collection and provided access to financial records. We would also like to thank Allen Magezi and Jackson Rwaheru who provided invaluable assistance in data collection, Matilda Temperley who helped calculate the economic costs, and Catherine Goodman, Jan Kolaczinski, Kara Hanson and Damian Walker for helpful advice. The Ugandan national control programme is supported by the Schistosomiasis Control Initiative, which receives funding from the Bill and Melinda Gates Foundation. Data collection was supported by a research grant from the Schistosomiasis Research Programme, based at DBL-Institute for Health Research and Development. The first author is supported by a Wellcome Trust Advanced Training Fellowship (073656).
|
References |
|---|
|
TopAbstractIntroductionDescription of the control...Data and methodsResultsDiscussionConclusionAppendix: Resources, quantities...References |
|---|
Albonico M, Montresor A, Crompton DW, Savioli L. Intervention for the control of soil-transmitted helminthiasis in the community. Advances in Parasitology (2006) 61:311–48.[Web of Science][Medline]
Baltussen R, Knai C, Sharan M. Iron fortification and iron supplementation are cost-effective interventions to reduce iron deficiency in four subregions of the world. Journal of Nutrition (2004) 134:2678–84.
Boatin BA, Richards FO. Control of onchocerciasis. Advances in Parasitology (2006) 61:349–94.[Web of Science][Medline]
Brady M, Hooper PJ, Ottensen EA. Projected benefits from integrating NTD programs in sub-Saharan Africa. Trends in Parasitology (2006) 22:285–91.[CrossRef][Web of Science][Medline]
Brooker S, Whawell S, Kabatereine NB, et al. Evaluating the epidemiological impact of national control programmes for helminths. Trends in Parasitology (2004) 20:537–45.[CrossRef][Web of Science][Medline]
Brouwer W, Rutten F, Koopmanschap M. Costing in economic evaluation. In: Economic evaluation in health care: merging theory with practice—McGuire A, Drummond MF, eds. (2001) New York: Oxford University Press. 68–93.
Bryan S, Brown J. Extrapolation of cost-effectiveness information to local settings. Journal of Health Services Research and Policy (1998) 3:108–12.
Bundy DAP, Shaeffer S, Jukes M, et al. School based health and nutrition programs. In: Disease control priorities in developing countries, 2nd edn—Jamison DT, Breman J, Measham AR, et al, eds. (2006) New York: Oxford University Press. 1091–108.
Charnes A, Cooper WW, Lewin AY, Seiford LM. Data envelopment analysis: theory, methodology and applications (1995) Boston, MA: Kluwer Academic Publishers.
Drummond MF, Pang F. Transferability of economic evaluation results. In: Economic evaluation in health care: merging theory with practice—McGuire A, Drummond MF, eds. (2001) New York: Oxford University Press. 256–76.
Drummond MF, Bloom BS, Carrin G, et al. Issues in the cross-national assessment of health technology. International Journal of Technology Assessment in Health Care (1992) 8:671–82.[Medline]
Drummond MF, O’Brien B, Stoddart GL, Torrance GW. Methods for the economic evaluation of health care programmes (1997) Oxford: Oxford University Press.
Elbasha EH, Messonnier ML. Cost-effectiveness analysis and health care resource allocation: decision rules under variable returns to scale. Health Economics (2004) 13:21–35.[CrossRef][Web of Science][Medline]
Evans DB, Guyatt HL. The cost effectiveness of mass drug therapy for intestinal helminths. Pharmacoeconomics (1995) 8:14–22.[Web of Science][Medline]
Fenwick A, Molyneux D, Nantulya V. Achieving the Millennium Development Goals. The Lancet (2005) 365:1029–30.
Fenwick A, Rollinson D, Southgate V. Implementation of human schistosomiasis control: challenges and prospects. Advances in Parasitology (2006) 61:567–622.[Web of Science][Medline]
Folland S, Goodman AC, Stano M. The economics of health and health care, 4th edn (2004) Upper Saddle River, NJ: Pearson.
Gabrielli AF, Toure S, Sellin B, et al. A combined school- and community-based campaign targeting all school-age children of Burkina Faso against schistosomiasis and soil-transmitted helminthiasis: performance, financial costs and implications for sustainability. Acta Tropica (2006) 99:234–42.[CrossRef][Web of Science][Medline]
Goodman CA, Mutemi WM, Baya EK, et al. The cost-effectiveness of improving malaria home management: shopkeeper training in rural Kenya. Health Policy and Planning (2006) 21:275–88.
Guyatt HL. The cost of delivering and sustaining a control programme for schistosomiasis and soil-transmitted helminthiasis. Acta Tropica (2003) 86:267–74.[CrossRef][Web of Science][Medline]
Guyatt HL, Bundy DA, Evans D. A population dynamic approach to the cost-effectiveness analysis of mass anthelmintic treatment: effects of treatment frequency on Ascaris infection. Transactions of the Royal Society of Tropical Medicine and Hygiene (1993) 87:570–5.[CrossRef][Web of Science][Medline]
Guyatt HL, Evans DB, Lengeler C, Tanner M. Controlling schistosomiasis: the cost-effectiveness of alternative treatment strategies. Health Policy and Planning (1994) 9:385–95.
Guyatt HL, Brooker S, Hall A, et al. Evaluation of efficacy of school-based anthelmintic treatments against anaemia in children in the United Republic of Tanzania. Bulletin of the World Health Organization (2001) 79:695–703.[Web of Science][Medline]
Habicht JP, Victora CG, Vaughan JP. Evaluation designs for adequacy, plausibility and probability of public health programme performance and impact. International Journal of Epidemiology (1999) 28:10–18.
Holland CV, O'Shea E, Asaolu SO, et al. A cost-effectiveness analysis of anthelminthic intervention for community control of soil-transmitted helminth infection: levamisole and Ascaris lumbricoides. Journal of Parasitology (1996) 82:527–30.[CrossRef][Medline]
Hotez PJ, Bundy DAP, Beegle K, et al. Helminth Infections. In: Disease control priorities in developing countries, 2nd edn—Jamison DT, Breman J, Measham AR, et al, eds. (2006) New York: Oxford University Press. 467–97.
Hutton G, Fox-Rushby J, Mugford M, et al. Examining within-country variation of maternity costs in the context of a multicountry, multicentre randomized controlled trial. Applied Health Economics and Health Policy (2004) 3:161–70.[CrossRef][Medline]
Jacobs P, Baladi JF. Biases in cost measurement for economic evaluation studies in health care. Health Economics (1996) 5:525–9.[CrossRef][Web of Science][Medline]
Jamison DJ, Breman J, Measham AR, et al, eds. Disease control priorities in developing countries, 2nd edn. New York: Oxford University Press.
Johns B, Tan Torres T. Costs of scaling up health interventions: a systematic review. Health Policy and Planning (2005) 20:1–13.
Kabatereine NB, Brooker S, Tukahebwa EM, et al. Epidemiology and geography of Schistosoma mansoni in Uganda: implications for planning control. Tropical Medicine and International Health (2004) 9:372–80.[CrossRef]
Kabatereine NB, Tukahebwa EM, Kazibwe F, et al. Progress towards country-wide control of schistosomiasis and soil-transmitted helminthiasis in Uganda. Transactions of the Royal Society of Tropical Medicine and Hygiene (2006a) 100:208–15.[CrossRef][Web of Science][Medline]
Kabatereine NB, Fleming FM, Nyandindi U, et al. The control of schistosomiasis and soil-transmitted helminths in East Africa. Trends in Parasitology (2006b) 22:332–9.[CrossRef][Web of Science][Medline]
Kabatereine NB, Brooker S, Koukounari A, et al. Impact of a national schistosomiasis control programme on infection and morbidity in Ugandan schoolchildren. Bulletin of the World Health Organization (2007) 85:91–99.[CrossRef][Web of Science][Medline]
King CH, Dickman K, Tisch DJ. Reassessment of the cost of chronic helmintic infection: a meta-analysis of disability-related outcomes in endemic schistosomiasis. The Lancet (2005) 365:1561–9.
Kirigia JM. Economic evaluation in schistosomiasis: valuation of health states preferences. A research note. Health Economics (1998) 7:551–56.[CrossRef][Web of Science][Medline]
Kirkwood BR, Cousens SN, Victoria CG, de Zoysa I. Issues in the design and interpretation of studies to evaluate the impact of community-based interventions. Tropical Medicine and International Health (1997) 2:1022–9.[CrossRef]
Lammie PJ, Fenwick A, Utzinger J. A blueprint for success: integration of neglected tropical disease control programmes. Trends in Parasitology (2006) 22:313–21.[CrossRef][Web of Science][Medline]
Mansley EC, Dunet DO, May DS, et al. Variation in average costs among federally sponsored state-organized cancer detection programs: economies of scale? Medical Decision Making (2002) 22:S67–79.[CrossRef][Web of Science][Medline]
Mascie-Taylor CG, Alam M, Montanari RM, et al. A study of the cost effectiveness of selective health interventions for the control of intestinal parasites in rural Bangladesh. Journal of Parasitology (1999) 85:6–11.[CrossRef][Medline]
McDaid D, Cookson R, ASTEC Group. Evaluating health care interventions in the European Union. Health Policy (2003) 63:133–9.[CrossRef][Web of Science][Medline]
Montresor A, Zin TT, Padmasiri E, et al. Soil-transmitted helminthiasis in Myanmar and approximate costs for countrywide control. Tropical Medicine and International Health (2004) 9:1012–5.[CrossRef]
Musgrove P, Fox-Rushby J. Cost-effectiveness analysis for priority setting. In: Disease control priorities in developing countries, 2nd edn—Jamison DT, Breman J, Measham AR, et al, eds. (2006) New York: Oxford University Press. 271–85.
Nyandieka LN, Bowden A, Wanjau J, Fox-Rushby JA. Managing a household survey: a practical example from the KENQOL survey. Kenya Quality of Life. Health Policy and Planning (2002) 17:207–12.
O’Brein BJ. A tale of two or more cities: geographic transferability of pharmacoeconomic data. American Journal of Managed Care (1997) 3:S33–9.[Web of Science][Medline]
Ottesen EA. Lymphatic filariasis: treatment, control and elimination. Advances in Parasitology (2006) 61:395–441.[Web of Science][Medline]
Over M. The effect of scale on cost projections for a primary health care program in a developing country. Social Science and Medicine (1998) 22:351–60.[CrossRef]
Partnership for Child Development. Cost of school-based drug delivery in Tanzania. Health Policy and Planning (1998) 13:384–96.
Partnership for Child Development. The cost of large-scale school health programmes which deliver anthelmintics in Ghana and Tanzania. Acta Tropica (1999) 73:183–204.[CrossRef][Web of Science][Medline]
Ramzy RMR, Goldman AS, Kamal HA. Defining the cost of the Egyptian lymphatic filariasis elimination programme. Filaria Journal (2005) 4:7.[CrossRef][Medline]
Sinuon M, Tsuyuoka R, Socheat D, et al. Financial costs of deworming children in all primary schools in Cambodia. Transactions of the Royal Society of Tropical Medicine and Hygiene (2005) 99:664–8.[CrossRef][Web of Science][Medline]
Spath HM, Carrere MO, Fervers B, Philip T. Analysis of the eligibility of published economic evaluations for transfer to a given health care system: methodological approach and application to the French health care system. Health Policy (1999) 49:161–77.[CrossRef][Web of Science][Medline]
Stevens W, Wiseman V, Ortiz J, Chavasse D. The costs and effects of a nationwide insecticide-treated net programme: the case of Malawi. Malaria Journal (2005) 4:22.[CrossRef][Medline]
Stoltzfus R, Albonico M, Chwaya HM, et al. Effects of the Zanzibar school-based deworming program on iron status of children. American Journal of Clinical Nutrition (1998) 68:179–86.[Abstract]
Valdmanis V, Walker D, Fox-Rushby J. Are vaccination sites in Bangladesh scale efficient? International Journal of Technology Assessment in Health Care (2003) 19:692–7.[Web of Science][Medline]
Walker D, Fox-Rushby JA. Economic evaluation of communicable disease interventions in developing countries: a critical review of the published literature. Health Economics (2000) 9:681–98.[CrossRef][Web of Science][Medline]
Walker D, Mosqueira NR, Penny ME, et al. Variation in the costs of delivery routine immunization services in Peru. Bulletin of the World Health Organization (2004) 82:676–82.[Web of Science][Medline]
Warren KS, Bundy DAP, Anderson RM, et al. Helminth infections. In: Disease control priorities in developing countries—Jamison DT, Mosley WH, Measham AR, et al, eds. (1993) Oxford: Oxford University Press. 131–60.
Wiseman V, Hawley WA, ter Kuile FO, et al. The cost-effectiveness of permethrin-treated bed nets in an area of intense malaria transmission in western Kenya. American Journal of Tropical Medicine and Hygiene (2003) 68(4 Suppl.):161–7.
WHO. Prevention and Control of Schistosomiasis and Soil-Transmitted Helminthiasis. In: WHO Technical Series Report 912. (2002) Geneva: World Health Organization.
World Bank. World Development Report 1993: Investing in Health. (1993) Oxford: Oxford University Press.
Zhang J, Yu JJ, Zhang RZ, et al. Costs of polio immunization days in China: implications for mass immunization campaign strategies. International Journal for Health Planning and Management (1998) 13:5–25.[CrossRef]
Accepted for publication 13 August 2007.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||