Study of cost per treatment with ivermectin using CDTI strategy: final report, March 2005

If, , t a Study of Cost Per Treatment with lvermectin using CDTI Strategy Final Report March 2005 PRIMARY INVESTIGATOR Deborah A. McFarland Rollins School of Public Health, Emory University RESEARCH ASSOCIATE and CO-INVESTIGATOR Nicholas Menzies, RSPH CO.INVESTIGATORS Zzkariaou Njoumemi, University of Yaounde, Cameroon Obinna Onwujelrwe, University of Nigeria. Enugu Support provided by The African Programme for Onchocerciasis Control (APOC) Technical Services Agreement n"08/1 I 1/296 1 1. Executive Summary Community Directed Treatment with Ivermectin, or CDTI, is the predominant intervention used by the African Programme for Onchocerciasis Control (APOC) to control Onchocerciasis in meso- and hyper- endemic areas across sub-Saharan Africa. This report documents the process and results of a cost study of CDTI conducted in Cameroon. Nigeria and Uganda in 2004. This cost study sought to quantify the resources devoted to CDTI by all the major in-country partners - communities. national governments. non-govemmental development organizations and APOC - in a purposively selected sample of projects, and describe what form these inputs took and what programmatic activities they were devoted to. The methods used were consistent with the current practice of economic evaluation; in addition an attempt was also made to assess and value in monetary terms the time devoted by communityJevel volunteers, and to set this alongside other contributions to the project so as to give an appreciation of the community's investment in CDTI. The results of this study indicate that the average total cost per treatment of CDTI in the 1 1 projects assessed was US$0.58. If the value of volunteer time is included then the average cost per treatment rises to US$0.74. Definite trends are evident in the break down of these costs across the 1l projects, and there is also significant variation, both in aggregate and disaggregated costs. Through revealing the current structure of CDTI support it is hoped that these results will aid management and sustainability planning for CDTI, both at the project level and at the programme level. O a t t a 2 t2. Acknowledqements The investigators wish to thank the numerous people and organizations involved in this study. Without their generous contribution of time and effort, this study would not have been completed. We wish to thankthe staff of APOC in Ouagadougou, the staff of each of the CDTI projects we visited, the staff of the NGDOs - HKI, the Carter Center Global 2000, SSI, UNICEF and CBM - WHO staff in the three countries in which we conducted the study - Cameroon, Nigeria and Uganda - the staff of the NOCP in each country, national, regional and local government staff, and last but not least, the CDDs and community members in all the communities we had the privilege to visit. We are indebted to the wisdom, experience and dedication of all those we met. Your contributions to the elimination of onchocerciasis as a public health problem are extraordinary. No cost study can capture such intangibles. a I J 1. Executive Summary 2. Acknowledgements 3. Contents 4. Abbreviations 5. Forward 6. Study Design 7. Methods 8. Analysis 9. Results 10. Discussion I 1. Conclusion Perspective Intervention Outcome Measure Cost Year Nature of Costs Disaggregated Costs Expenditure Vs. Budgetary Data - Methodologt Development - Sampling of Projects - Collection of Data Management of Specific Errors and Data Gaps Volunteer Time Valuation ond Interpretation Allocation of Shared Resources Key Outputs CDTI Costs Summed Across All Projects CDTI Costs Broken Down By Project CDTI Costs Vs. Various Explanatory Variables a a 4 4- Abbreviations a t APOC CDD CDTI FLHF GNI JAF LGA MDP MOH NGDO NOCP REMO TCC LINICEF wHo African Programme for Onchocerciasis Control Community-D irected Distributor Community-Directed Treatment with Ivermectin Front Line Health Facility Gross National Income Joint Action Forum Local Government Area Mectizan Donation Programme Ministry of Health Non-Governmental Deve lopment Organization National Onchocerciasis Control Programme Rapid Epidemiological Mapping of Onchocerciasis Technical Consultative Committee (APOC scientific advisory group) United Nations Children's Fund World Health Organization 5 5. Forward The African Programme for Onchocerciasis Control (APOC) was formed in 1995, and is a multinational disease control programme designed to combat Onchocerciasis (River Blindness) in endemic areas across Sub-Saharan Africa. This programme is a partnership between private industry, international health and development institutions, non-governmental development organizations, domestic governments, and affected communities. with successful programme implementation requiring the input of resources from all these partners. The principle modality used for disease control is mass drug administration (MDA) of the microfilaricide Mectizan@ (ivermectin)' to communities that are meso- and hyper-endemic for onchocerciasis. The yearly administration of this drug interrupts the life-cycle of the parasite in infected individuals, and prevents them from experiencing the distressing symptoms and pathology associated with untreated infection. In the long-term it is thought that the regular treatment of entire communities will break the transmission of the parasite and thus eliminate the disease in the local area, contributing to the elimination of onchocerciasis as a disease of public health and socioeconomic importance throughout Africa, and so contribute to improving the welfare of its people2. MDA is achieved via Community- Directed Treatment with Ivermectin (CDTI), a strategy which relies on the close involvement of communities in all phases of design and implementation, with designated community members (Community-Directed Distributors, or CDDs) dispensing the drug to their fellow community members during a distribution period once per year. using a simplified treatment algorithm. A more detailed explanation of the CDTI strategy can be found elsewhere3. The administration of this programme is organized into discrete projects, which usually match the structure of the underlying health system of the country. and thus a particular project will usually be responsible for an entire state, province. or region, although there are exceptions. Each CDTI project receives support from those partners mentioned earlier. as well as from a national level administrative unit, the National Onchocerciasis Control Programme (NOCP), which falls under the authority of the national Ministry of Health. The programme objective of APOC is'to establish, within a period of 12 to 15 years, ffictive and self- sustainable, community-directed ivermectin treatment throughout the endemic areas within the I lvermectin is donatedfree of charge by Merck & Co, Inc, for as long as necessary to all who need it. ' Programme Documentfor Phase II (2002-2007) and the Phasing-out Period (2008-2010) (document JAF 7 8) ll orld Healt h Or ganization, African Pr ogramme for Onchocerciasis C ontrol. 3 Community-Directed Treatment with lvermectine (C DTI) : A practical guide for trainers of community. APOC 1 998 I I 6 geographic scope of the Programme, and, if possible, in selected and isolated foci, to eradicate the vector by us ing environmentally s afe methods.a Thus external financial and administrative support, from the APOC Trust Fund and many international NGDOs, has been provided for the development of these CDTI projects. The intention is that this external funding should be reduced and phased-out once projects have become firmly entrenched within the community and relevant government health services, and the recurrent financial costs are borne by governments, communities, or other willing partners. In practice this has involved two funding cycles of 5 years each, to 85 CDTI projects in 19 different countries, covering communities with an aggregate treatment population of almost 59 million people. These projects were initiated progressively over the years following APOC's inception. Recently there have been moves towards providing a third abbreviated tranche of funding to selected projects in support of sustainability activities, but it is still generally understood that external funding will progressively decline as domestic sources of support (from national, regional and local governments and communities) take increasingly more responsibility for maintaining the projects. The successful transfer of both administrative control and financial responsibility is now seen as crucial for project sustainability. As concluded by a mid-term evaluation conducted in 2000, sustainability is the key issue for Phase 2 of APOC, and the major challenge for the future of Onchocerciasis control, and up to that point there had been inadequate planning to mobilise the resources and set in place the routines whtch will ensure the continuation of the programmet . Another recenttrend can be seen in the enthusiasm of project partners to incorporate other development activities into CDTI, such as Vitamin A distribution or MDA to treat Lymphatic Filariasis, thus taking advantage of the broad distribution network represented by the CDD system. Integration of this nature promises a number of benefits - multiplication of the public health benefits provided by the CDDs, higher project efficiency resulting from economies of scope, and the possibility of attracting new funding to projects. Although a large multi-site study is currently assessing the benefits of integration, there is presently a poor understanding of how different forms of integration will affect the burdens borne by current project partners, and whether the potential benefits are realized in practice. Although this cost study will not look at the transition or recurrent costs of integration, it will provide baseline data against which integration activities can be compared. Quantitative data on the nature of project support will serve to illuminate these pressing questions facing the Programme - what resources are being devoted to CDTI projects? Who are they being provided by? o Programme Documentfor Phase II (2002-2007) and the Phasing-out Period (2008-2010) (document JAF 7.5). llorld Health Organization, African Programme for Onchocerciasis Control s Ransome-Kuti O. et al. Report: h,ternal Mid-term Evaluation. African Programme for Onchocerciasis Control, September 2000 t 7 What are they being used for? This data will serve a number of purposes. At its crudest level, by comparing the total annual resources invested in projects with their total annual treatments it will be possible to estimate the annual cost-per-treatment for Onchocerciasis control. When cross-sectional studies on this subject have been undertaken in the past, the cost-per treatment results they produced often ignored important costs, or did not use standardized protocols. thus hampering comparability across projects. Cost data are also an important resource for project management, particularly to inform decisions in the area of sustainability planning - an understanding of the structure of current project support, and what activities the resources are used for, is an important step in the process of working out what activities will need to continue after APOC support has been phased-out, and where the resources will come from to carry them out. The calculation of objective cost per treatment results also provides a concrete basis for advocacy for future funding. Finally, through comparing the per-treatment costs between different projects it will be possible to investigate those factors that are thought to drive higher project costs, and possibly reveal successful strategies to deal with these problems and maximize project efficiency. It is also anticipated that in the future the methods used in this cost study might be simplified and trimmed down to serve as a financial evaluation tool that projects would be able to apply themselves. and thus provide feedback of financial support across a wider range of projects, whilst minimizing the data collection requirements of the cost study. This financial evaluation will not only provide useful feedback for the administration of each particular project, but also contribute to the pool of information and knowledge used to manage the programme as a whole. a 8 nThis section outlines the general concepts underlying the methods used in this cost study, the specific nature of the costs being included in the study, and how they will be presented to maximize the programmatic relevance of the results. Perspective The perspective defines the limits of assessment in an economic evaluation. It is the answer to the question - "Whose costs do we care about? ". In this study the costs we are concemed with are those borne by the major in-country partners in a CDTI project, namely APOC, the government, the NGDO, and the community. This has a number of important ramifications: The cost of drug development, manufacture, and delivery to the country are excluded. These costs are borne by Merck & Co., Inc., and although not assessed in this study, they are known to be substantial. It is felt that the inclusion of these costs, while giving a more complete estimate of the total resources devoted to a particular CDTI project, would detract from the usefulness of the results, given that it is impossible to influence these costs through better project management, and they play little role in sustainability planning (two processes that this study hopes to inform). All supra-national overheads are excluded. This includes the costs of APOC's intemational-level organization, and the costs incurred by the NGDO's outside the country in question, as well as the costs associated with any meetings or committees devoted to CDTI outside of the country. All these costs are excluded, for the same reason as given above. The costs incurred by the community are included, although these costs are limited to those borne by the Community Directed Distributors and those other community members who help them with their tasks. As will be seen in the cost study results, the bulk of community resources dedicated to any CDTI project will be in the form of donated time, by CDDs and those that help them. The valuation of this volunteer time is not straight-forward, as is dealt with in the analysis section. Figure 1 shows the organizational levels excluded and included in this study a 9 E_ 6E9 E8o!o Eo>. - - ) Fq P.-o.9, oEi 5l--UO.= i c.=.= h E_ tEo Xo;6,> :-cE(, F o =Y oEAIEB 5E().==.8 oE o >E' o o oo o o E'o E i5 xo E' (E !, oE E .E I o -g o G .N (E cn o -o)g, lI- NATIONALINTERNATIONAL g I I I I I I I I I I I I I I I I I I I I I I I I Intervention Another important consideration is to define the exact activities being costed. This analysis aims to determine the costs associated with Community-Directed Treatment with Ivermectin in a small sample of mature, successful CDTI projects. Thus the assessment of costs was conducted so as to: Exclude the costs of clinic-based treatment with lvermectin. Clinic-based treatment is a fundamentally different intervention to CDTI, and for the purposes of comparability and programmatic relevance of results the costs associated with clinic-based treatment were excluded. For the same reason the denominator data used to calculate cost-per-treatment figures (number of treatments in the project area in the cost year) also excludes clinic-based treatments. Exclude the cost of 'add-ons' to CDTI drug distribution (eg Vitamin A distribution). Although there will likely be benefits to be gained from the integration of vertically- structured programs, it is important to restrict analysis to those costs specifically associated with Ivermectin distribution, so the results from different projects can be validly compared. In this way the results of this study can acl as a baseline against which the benefits of integration can be compared. Outcome Measure In this study the costs of each CDTI project are being assessed relative to the outcomes of that project, and these outcomes are being measured in terms of the number of treatments provided via CDTI in the cost year. Using this outcome measure assumes that (a) treatments are the primary output we are concerned about with CDTI projects (in that all project benefits are a result of treatment), (b) treatments have a proven association with health benefits for the treated population and (c) the relationship between treatment and health benefit is relatively equal across different projects. It could be argued that it would be better to assess the health benefits to the treated populations more directly, such as through estimating the death and disability avoided through treatment, and thus we would gain a much more valid estimate of how much good each project is doing. However the complexities of such a methodology and the expense they entail would require a significant expansion of this study. If the three assumptions listed above are accepted as largely valid, the number of treatments is an outcome measure that is objective, easy to measure, and perhaps most importantly, is a variable over which project management can have direct influence. 9 l1 Cost Year Given the possibility of project costs varying substantially over the course of different seasons, the year is the natural unit for measuring costs. In this evaluation we chose not to use a set cost year for all projects, given that different projects complete their financial years at different times, and cost collection would be much more time-consuming if we tried to compile data on a financial year different to that used in the project. Instead. for each project we chose to cost the project financial year that aligned most closely wtth the January - December 2003 calendar year (and for which complete data were available). Within each project we used the same cost year for all sources of funding. since otherwise a long-term shift in the distribution of costs between project partners could bias our results either up or down. The specific cost years used for each project are as follows: Figure 2 Nature of Costs Although up to this point the term 'cost' has been used with little explanation. it is important to clarify what resources are being identified as costs. There are two major distinctions usually identified within the concept of cost: direct vs indirect costs, andfinancial vs economic costs. A discussion of both is important to concretely define what kind of inputs we are counting when we enumerate costs. In this study the direct costs are those costs which are directly attributable to project activities - for example, the per diems paid to project personnel when they go out on evaluation missions. the fuel used by the vehicle that carries them, the repairs it needs when they get back, the paper the evaluation report is written on and the postage to send it to the national office - all of these are c Anambra, Nigeria Oct 2003 - Sep 2003 Kasese, Uganda Jan - Dec 2003 Centre Trois, Cameroon Dec2002 - Nov 2003 Kisoro, Uganda Jan - Dec 2003 Cross Rivers, Nigeria Jul 2002 - Jun 2003 Masindi, Uganda Jan - Dec 2003 Ebonyi, Nigeria Oct 2003 - Sep 2003 Nord, Cameroon Nov 2002 - Oct 2003 Enugu, Nigeria Oct 2003 - Sep 2003 Yobe, Nigeria Jan - Dec 2003 Hoima, Uganda Jan - Dec 2003 t2 r"' a\' I r r.. I _-i :.I:i .Y:!v.lJi ,' ;'l-' - ^r;. . ,l direct costs. The indirect costs are those costs which, although not directly attributable to any project activity, are necessitated by the factthat the project exists. These costs, often refened to as overheads, include such things as cleaning services, utilities charges, the wages ofany security or maintenance staff, and the rental cost or opportunity cost of the building that the project headquarters is housed in (opportunity cost is explained below)6. In this project both direct and indirect costs were included in data collection, so as to give a more complete estimate of the cost of the CDTI project. A more complex situation arises where the CDTI project (or whatever organizational unit is being costed) is housed within a larger facility such as a Ministry of Health, and shares the same services, utilities, etc., as other Ministry of Health departments. In this case there is often no pre- existing allocation of these overheads to the CDTI project, and so a method must be chosen to allocate a reasonable portion of the total overheads costs to the CDTI project, for the purposes of this analysis (specific methods are discussed in the methods section). Financial costs are those costs that relate to cash expenditures on project activities, and all of these should be recorded in a project's financial accounts for the year, or in the financial accounts of a project partner. Although it is important to include these costs in the study, there are also a number of other important costs which need to be included and that are not often reflected in the standard accounts. These non-financial costs represent resources devoted to the CDTI Project that, while not being associated with any direct expenditure, could be put to good use somewhere else were they not being used by the CDTI project - in other words they have an opportunity cost (in that through using them for this project we are forgoing the opportunity of putting them towards the next best other use). The sum of these non-financial costs plus the financial costs represents the total economic cost of the CDTI project, and it is this economic cost that this study is assessing. Non-financial or opportunity costs include such things as: the ' free' use of building space provided by the Ministry of Health or other proj ect partner, the use of capital items such as vehicles or computing equipment, the purchase of which occurred in a earlier cost year, and the time devoted to the project by CDDs and other community members 6 In the economic literature the lerms direct and indirect costs are also used to denote a dffirent, more specific distinction, with indirect costs referring to those costs borne by patients though missed work and leisure time while travelingfor and receiving treatment. This term is closer to the economic cost distinction made later in this text, and is not the meaning intended here. 13 , If one were to try to diligently include all such non-financial costs it would quickly become apparent that this is a task with no clear ending - should one include the cost of food eaten by CDDs while distributing, and some value for the use of the village square in which they conduct the distribution? Or more extremely, should one try and include some value for the use of the country's basic infrastructure, such as roads? In both of these cases the answer is no, for a number of reasons. In this study the assessment of these non-financial costs were restricted to those cost items which were thought to represent a significant cost to the CDTI project a priori, which could reasonably be expected to be put to another use were they not used by the project, and for which a valid cost could be estimated. Methods used for the estimation of these costs are elaborated in later sections. It is important here to remember those costs which have been excluded, such as Merck's costs and other supranational overheads, which have been excluded for the reasons given earlier. A more debatable exclusion is the costs incurred at the first-line health facility level. Although many sustainability evaluations have documented little input from this level, it is clear that costs could be incurred here (and not just funded from elsewhere) that should correctly be counted as part of the study. The decision to exclude these costs is mainly a pragmatic one - given the likelihood that these costs would be minor in comparison with other sources of funding, and that they would be diffrcult to disentangle from general healthcare funding, it was decided to exclude them in order to restrict the amount of field work needed to collect the cost data. so as to fit the budget allocated to this research project. In this study the costing approach taken has been to tailor the level ofaccuracy sought to the size of the cost being measured - if it is a large cost, then any inaccuracy will have an influential effect on the results, and so careful measurement is important. Conversely, any inaccuracy in a minor cost will have little effect on the final results, and so the investigator's time is better spent measuring the large costs well. Disaggregated Costs As well as producing summary results for both the total cost and the per-treatment cost of the selected CDTI projects in the chosen cost-year, this study also aims to present the results with these total costs disaggregated along three programmatically relevant sets of categories: 0 14 ICosts disaggregated by source offunding. As mentioned earlier, this study assessed the costs invested in a CDTI project by all of the major partners, and results will be presented so that the different contributions of each can be seen and compared. The specific sources of funding identified are as follows: o APOC support to the National Onchocerciasis Control Program o National Government support to the National Onchocerciasis Control Program o APOC support to the CDTI project o Regional Government support to the CDTI project o Local Govemment support to the CDTI project o Community support to the CDTI project o NGDO support to the CDTI project The NOCP makes an important contribution to the administration and oversight of each CDTI project, and so a portion of the costs of running the NOCP must be included in the costs for each CDTI project. The allocation system to determine the size of this portion is discussed in the analysis section. A schematic diagram displaying the relationship between the different funding sources is shown below: I 15 Figure 3 Schema identifying the sources of support to each CDTI project v APOC NOCP National Govt Regionail Govt Local GovtT*GBO Cornmunity It is clear that the three-tier system of government we use in this model will not fit all situations (such as where there is no separate level of local government involved between the regional government and the community, or where a national govemment contributes directly to a single national CDTI project). In those instances where the organizational environment of a project differs from that shown in model, then the costing protocol can be adapted to fit. Costs disaggregated by input type. Secondly, the results will be presented with total costs separated into the various types of resources (inputs) used in the project: o Transportation - the recurrent costs associated with vehicular transport. namely fueling, insurance, maintenance, repairs. road taxes or fees, and rental costs if any vehicles are rented. o Consumables - the recurrent costs associated with project activities and general office running, namely office stationary. photocopying. IEC materials, food / refreshments, 76 computer consumables, equipment hire / maintenance I repair, telephone, internet and other communications costs. o Employed Personnel - the recurrent costs associated with paying salaries to employed personnel, including any supplements, emoluments, insurance or other benefits of employment. o Per Diems - the recurrent costs associated with per diems, allowances, DSAs etc. o Overheads - the recurrent indirect costs associated with project administration and maintenance of the work environment, namely utilities charges, building rental or equivalent, cleaning services, security guards, and other ancillary personnel. o Capital Costs (Vehicles) - representing those capital costs associated with vehicles, namely the annuitized cost of bicycles, motorbikes, cars, trucks and or boats used in the project. o Capital Costs (Other) - representing those non-vehicle capital costs associated with project, namely the annuitized cost of such items as computer equipment including expensive software, communications equipment such as faxes, audiovisual equipment such as televisions and projectors, photocopiers, office furniture, and generators. o Volunteer Time - the monetary value of the donated time of CDDs and other community members in implementing CDTI at the community level. C o s t s di s aggr e gat e d by pr o gr amm at i c act iv ity. Finally, the total costs are broken down by programmatic activity. These functional areas were chosen so as to capture the major classes of activity involved in a CDTI project. The categories used were: o Training of Volunteers. This involved the training of CDDs and any other community volunteers, and includes the costs incurred by both the trainers and the trainees. o Other Training. This involved all other training, at whatever level, and as with the last category includes the costs incurred by both trainers and trainees. o Mobilization and Sensitization. This included all promotion, information dissemination and advocacy related to the project. 5 17 o Drug Distribution Chain. This included all activity involved in getting the drugs from the port at which they entered the country to the CDDs. o CDD Drug Collection. This involved the time and resources used by CDDs collecting the drugs and bringing them back to the community. o CDD Drug Distribution. This involved the time and resources used by the CDD distributing the drug in the community. o Census. This involved the time and resources spent by CDDs completing the annual update ofthe register in his or her coverage area. o Monitoring and Evaluation. This involved all supervision, monitoring and evaluation, at all levels, and includes the costs incurred both by the supervisors/evaluators and those who were supervised/evaluatedT. o Reporting to APOC. This involved the preparation and transmission of reports and other required documentation to APOC HQ. o All other Administration. This involved all other reporting, accounting and general office administration. o Other Project Activities. This catch-all category involved all other activities not already mentioned, and, if any entry was made in this column during data collection, a description was entered as to the specific nature of the activities being performed. Expenditure vs Budgetary Data This study used expenditure data rather than budgetary data - how much money was actually spent during the cost year, and what activities it was actually spent on, rather than what was planned. Budgetary data, while often being easier to obtain, can fail to show where money is over or under-spent in any given year, or. more commonly, where funding has been reallocated between different activities and input types within a set budget. One exception to this expenditure rule is with capital items, where costs were included that relate to expenditures made in years prior to the cost year in question, and thus data were collected on these historical expenditures. 3 7 Costs of APOC funded 3 and 5 year sustainability evaluations are not included in M&E costs. 18 7. Methods Methodo logy Development The methods and instruments used to conduct this cost study were developed and refined at an investigators meeting held at APOC Headquarters in Ouagadougou, Burkina Faso, in June 2004. These methods were designed so as to meet the informational needs of APOC and other stakeholders, while adhering to accepted conventions for the conduct ofeconomic evaluation, and minimizing the costs associated with data collection. The data collection and analysis instruments were developed specifically for this project, although their basic form was based on costing instruments designed by the Lymphatic Filariasis Support Center, Atlanta USA. Sampling of Projects This study was restricted to mature, successful projects. Maturity (projects in the 5'h year or more of APOC funding) was seen as important because the results must be relevant to the question of sustainability. Many of the questions surrounding project funding and changes in the contributions from different project partners arise as a project nears the end of its APOC funding period, and it is these questions that this study hopes to inform. The restriction to successful projects (defined as having coverage rates comfortably above the 65%o level) was made to ensure that our cost estimates were not biased upwards by the fact that the projects in question had poor coverage rates. Although a cost study can greatly aid the diagnosis of why a project is failing, this initial study aims to establish a base-line, and determine the range of costs that might be found in a CDTI project when it is functioning well. Although this suggests a certain normative element to the cost estimates resulting from this study (that they will show what a project should cost), it must be remembered that, firstly, selecting successful projects in terms of coverage rates does not imply that these projects are similarly successful in containing costs. Secondly, it must be recognized that there will be many valid reasons for why a project may incur a higher cost-per- treatment than others, such as differences in the underlying infrastructure of the country, or a local culture that is wary and slow to accept the distribution program. Indeed, this study does not include any projects with one of the most commonly recognized sources of increased costs - civil conflict. The decision to exclude this as well as some other possibly relevant determinants of cost was made to fit the time and budgetary constraints of this study, and it is hoped that assessment of the impact of these variables can be performed in later studies. 19 5 From the pool of successful, mature projects. a sample was selected purposively to exhibit variation over a small number of variables thought to be to be important determinants of cost: Project size (as indicated by the total number of treatments in the cost year); Country (reflecting both differences in the underlying health system infrastructure and organization. as well as differences in input prices: NGDO partner; and Loaiasis co-endemicity. Figure 4 lists the projects included in the sample and their basic characteristics. Figure 4 Although it is accepted that project maturity and therapeutic coverage are likely to be influential determinants of costs, these factors can not be used to explain any variation in the cost results since we considered these variables in choosing our original sample. .. l:\ 20-30%Cameroon HKI 163,286Centre Trois Province 271,450Cameroon G2000Nord Province 577,902Anambra State Nigeria G2000 UNICEF 742,264Cross Rivers State Nigeria 464,879Nigeria G2000Ebonyi State G2000 714,296Enugu State Nigeria CBM 338,304Yobe State Nigeria 76,713Uganda SSIHoima District G2000 77,575Kasese District Uganda G2000 16,403Kisoro District Uganda 44,326Uganda SSIMasindi District 20 It should also be noted that the definition of 'project' used in this study is the final functional project that would exist once APOC funding support has been removed - in other words the organizational units of the country's health system. This implies provinces, states and districts in Cameroon, Nigeria and Uganda respectively. Because of this, some current projects (defined as such by APOC) have been separated for the purposes of this study, namely the Anambra/Ebonyi/Enugu CDTI Project in South East Nigeria, and Phase I in Uganda (Hoima, Masindi, Kasese and Kisoro Districts). Collection of Dat* By source offunding: Data collection was organized around the funding sources identified in Figure 3. For each funding source, cost collection involved the review of accountancy records and project documentation, and interview of key personnel (usually project coordinators and accountants). Where possible these data were confirmed through cross-reference with other sources at a later date. Costing for the NOCP in each country involved summing the total contributions (in the cost year) from the two contributors to the NOCP - the National Government and APOC. Where a staff member within the WHO country office is involved in the financial management of CDTI projects, this contribution was included. Having identified the total cost for the NOCP in the cost year, a portion of the cost was allocated to each CDTI project. Methods used for this allocation are outlined in the next section. At the project level, costs were collected individually from each of the funding sources that contributed directly (APOC, Regional Government, Local Government, NGDO, Community) and a separate costing questionnaire was completed for each. Given that it was not feasible to collect costs from every community and local government involved in a given project, a sample of communities and local governments was taken, and the results from the sample used to produce an estimated total cost for that funding source. Given the small sample size used (to reduce the time and logistical complexity involved), local governments and communities were sampled purposively for inclusion in the sample. The following method was used to select the sample: (l) Local governments were selected first. Three local govemments were selected within the project area, and were chosen so as to reflect a range in the variables that were considered a 2t important to determining costs at this level. The variables were total population, dispersion of the settlements within the local government area, distance from the capital city of the region, quality of the roads and ease of travel. andtherapeutic coverage rates. Each of the selected local governments was surveyed to determine their contribution to the CDTI project, and these results were pooled, averaged, and then multiplied by the total number of Iocal governments involved in the project to estimate the total Local Government contribution. (2) l2 communities (4 per local government) were selected for the sample, although sometimes more communities were added during the fieldwork, to ensure at least 24 CDDs were included in the final analysis. This restriction to the 3 local government areas was made so as to simplif, the logistics of data collection. Communities were selected to reflect a range on several important variables while being generally representative of the communities in the entire project. The variables were: community size, dispersion of communities in the area, distance from the local town, number of CDDs, and therapeutic coverage rates. The costs for these communities were pooled. averaged, and multiplied by the total number of participating communities in the project to estimate the total Community contribution. It must be noted that because projects were selected purposively rather than randomly. standard statistical techniques cannot be used to calculate confidence-intervals around the estimated proj ect-level result. By input type: At each funding source, costs were collected according to the input types mentioned in the previous section. The general format followed was: (l) The total gross expenditure on each particular cost item (e.g. one vehicle. or an employed individual) was calculated. (2) The respondent was asked to indicate what portion of that cost item was used for CDTI in the cost year (for example, local government staff may have been involved in a number of health programs, and thus a fraction of their total salary would be allocated to the CDTI project as costs, while for CDTI project staffthat fraction would be 100%). (3) This cost was then converted to US Dollars at the exchange rate that existed at the time the cost was incurred, and inflated to present day (August 2004) US Dollars so that all 22 comparisons between projects were in the same monetary units. This currency conversion was performed at a later stage, after all the crude cost data had been collected. Certain input types involved special treatment: Volunteer time: this category represents the time devoted to the project by CDDs and other community members who helped them. An estimate of this time was made by asking how much time was devoted to the project in the last distribution period (although this may have been slightly different to the cost year, asking about the most recent complete distribution was done to achieve the most accurate responses). Time was measured in full hours, and then this was divided by 8 to give the number of equivalent 8- hour working days devoted to CDTI. Other community members involved in the project were identified by asking the CDD and local village leader, and their time allocations assessed in a similar fashion and summed to find the total for the community. Capital items: this involved costing capital items fu zse during the cost year, and data on purchase cost and year were collected for all such items. Given that the usefulness of capital items is spread overanumberof years, we needto allocate a portion of theirtotal purchase cost to the cost year. If this is not done we would see an artificially low cost estimate, since we are ignoringthe contribution of these capital items (most of which are purchased near the start ofthe project) and the fact that they would eventually need to be replaced. In this study we estimated the average useful life for all capital equipment to be 5 years, and the purchase cost was allocated equally over the 5 years. Use of buildings and other facilities: these cost items, like capital items, are useful across a number of years. Because of the longevity of buildings it is often difficult or impossible to estimate a 'purchase price' incorporating the original construction costs as well as later refurbishments and improvements. In this case it is often more accurate to obtain an equivalent rental price (at the best rate possible), and this method was used to find the annual value of a building or facility. By programmatic activity Respondents were also asked to indicate which programmatic activities each cost item was used for, and how much it was used for each. This information was then used to allocate a percentage of the total cost of the item for each activity. By summing the costs for each particular activity within a input type we were able to generate the costs disaggregated by the three classifications Z5 discussed in the study design section - costs disaggregated by the source of funding, the input type, and the programmatic activity they were used for. Overhead was one input type treated slightly differently. Because, by definition, overheads are not dedicated to any particular programmatic activity, we could not just ask how they were used. In this case we took the employed personnel's allocations across the different programmatic activities, averaged these (with the average weighted according to the total time each staff member spent on the project), and then used this weighted average to allocate the overhead costs to the various activities. Figure 5 (following page) shows a typical page from the costing questionnaire. and shows how each cost was collected and dealt with. This sheet collects costs on employed personnel. The first table collects crude cost data, and for each cost item calculates the portion ofthe cost attributable to CDTI, and then the equivalent value in August 2004 US Dollars. The second table collects data on which programmatic activities each item was used for, and the third table calculates the totals for each activity for that input type. The dark grey cells collect descriptive data, the light grey cells collect cost data, and the cells already containing numbers (0.00 in this case) are not for data entry, but rather contain formulae that calculate the cost results automatically in the Excel version ofthe costing sheet. Av o i danc e of d oub I e -c ounting Throughout the data collection considerable effort was made to prevent double-counting. given that funds were often spent by one party that had actually been provided by another party. A similar situation arises where transfer payments were made between different funding sources (e.g. 'motivation fees' paid by the provincial government to CDDs in the Cameroonian projects). In cases like these care was taken to ensure that costs were assigned to the actual funding source. Allocation of shared overhead costs Where a CDTI project office shared facilities within a larger organization (such as a Ministry of Health building), or where the overheads of a project partner are split between CDTI and another programmes, the overheads were allocated using employee time as a guide. For example, if a particular office had 6 full-time employees. and 3 of them spent half their time on CDTI and half on other activities, while the other 3 did not deal with CDTI at all. then the percentage of overheads attributable to CDTI would be 25Yo. This is calculated by dividing the employee time spent on CDTI (3 employees x 50/100) by the total employee time (6 employees x 100/100). I 24 oq o@ oq oa, oq oo oq o@ 9odtOf= c 'Es JO EN cOoe- o g EERPE s e; o G BsE9S soq o soq o soq o :eoo o oq o oq o oq o oq o oq o soo o E ofi a ; Ea-e o o -Oooo'E o> o-E s c.aEorE oq o oq o oq o oq o c)q o soq oEEEEs oq o oq o oq C' oq o oo c; oi 6€ =< s E.= =E Pclitoote oq o o oq o oq o oq o soo ci o-ED- :a O- ci E EEE s PoEOitooEe ;e EDtr .E .9 o9.s E Ei =ur s EDC .E .9 o9lE E Ei =ut oq o oco oq o oo c; oo ci soq o G -4.G- EEFC tr oo o oq o oq o oq o oq o aEo.6 E= .o6FL oq o oq o oq o oq o c,o c; c o oE.g3;6 5 E,' i5 s tr o oE.g:x6686 i5 oco oq o oq o oq o oo G; soo o uio .ECC Etroo e- .9EEoF =s=6=i $E i oq o -c E€96rEN .N'E E2o6 =o s -tr EE eG agN .NEE2 €8 oq o oIo oc qo oq o soq o =a(, €E oq o be 6.: <F s o:, Igt.E <F oq o oo o oo o oco oq o ;eoq o i€E$:gi& E o(, oEoc o c o o E (E o o o E o ctq o be -o:l F> s o oo 6 c oo o a) E o 6 o o = () Il .1i o oo oo -9 CL E LrJ bc P$ g6F> oq o oq o o o oco oq o soo ci ;; g 5 E SE;6dEt .ll +a6soE o (o llt oI olrlcl olotLIol o-' It o oa E ut ,i o o o o o o tr, sE 3EgE Ev(, rE E> =o U' e o E,o oo o Ec oo oo o tro ooq. g) l! c o o G = eo E t' G q s ooo G o eI Ug c o :i rEooco o<go o!o.E o o. !, so oo o. to E trG trS o(J o G o es Or e. *e5 E'E -(,G'O c<Oeoot6 o o 1O oq. E G o.tr>9';5E laa)6<o>oro cootr =Ooo:trd E oF o() :> o o 0) 'd (L 0)E o o -o oooo lrJ(9 I 6 c o o o o. to -9 CL tr tu ,i tO o o o ootr :Et il dg E=96 t. 6)Eo- oo o oo e, (E o -=.Y.E{E E:rieItnljio> EE9>trrr IEtrz EE .o6FL t rO o f .9II r- a This section details the process of data analysis - how various data collection errors were dealt with, how volunteer time was valued, how shared resources (such as NOCP support) were allocated to particular projects, and how key output statistics were generated. Management of SpeciJic Errors and Data Gaps During data collection the cost records were not often found in an ideal state - sometimes data were not recorded in the same format as used in this study, sometimes key documents and statistics were incomplete or missing. Adjustments and estimations thus had to be made in concert with the experience of those most knowledgeable with the data. Some of the more influential adjustments made were as follows: For the Nigerian NOCP for 2003, the accounts showed significant spending on two items: sustainability workshops, and computer software called 'Health Mapper'. The sum total for these items was US$l13,104. While these are important and worthwhile expenditures. it was decided that they should be excluded, since they do represent future costs. Data on the purchase price of vehicles was occasionally impossible to find. In this case an estimate was made by using the purchase price for a similar vehicle purchased around the same time in the same country (found in a different costing sheet). Similarly, the purchase date of capital items was sometimes missing. In this case the year in which the project began was used as the most reasonable estimate. The sample size for local governments and communities in Ebonyi. Enugu and Anambra States was smaller than intended. For this reason caution must be used when reviewing the cost estimates for the funding sources in these projects, given the greater variability around the cost estimates. Volunteer Time Valuation and Interpretation CDDs and others give their time freely to ensure that drug distribution and other project activities occur, but it does not follow that their time has no value. In the design of this project it was thought crucial to capture this community contribution, and to place it alongside the other sources of resources invested in a project. To capture the value of the community resource of donated time, we had to ascribe a monetary value to volunteer labor. To do so we used values for the average per o I 26 capita Gross National Income (GNI) in 2003 for each of the three countries studied - Cameroon, Nigeria and Uganda (sourced from the World Bank's World Development Indicators). These yearly estimates were used to compute a daily estimate, and this daily estimate was multiplied by the total number of equivalent S-hour days contributed by CDDs and others to find the community contribution. The values used are shown in Figure 6: Figure 6 Cameroon us$640 us$1.75 Nigeria us$320 us$0.88 Uganda us$240 us$0.66 a t Source: World Bank. World Development Indicators 2004 Although valuing the communities contribution is important, it must be understood that this valuation system is somewhat arbitrary - although the World Bank's per capita GNI figures maybe representative of the country in general, it is less obvious that they validly capture the value of the activities an individual would have undertaken were they not a CDD, or how much it would cost to pay someone to do the CDD's job. The correct value to put on a volunteer's time is open to debate, and for this reasons the study results are presented in two ways - with the value of volunteer time excluded and with it included. Allocation of Shared Resoutces A number of costs had to be allocated between projects. These included: - The costs of the NOCP. - The costs of NGDOs that work on more than one CDTI project. The APOC project costs where a project has been disaggregated into its individual states or districts, as in the case of Anambra/Ebonyi/Enugu in Nigeria and Phase I in Uganda. In these cases there are two general methods that could be used for allocating costs. The first method is to allocate the costs equally (for example the costs of the NOCP would be allocated equally to all projects over which it had supervision), while the second method would be to allocate 27 . 1,,.i*lri.,.,iii ;x, i[ .r,..,.y:ixi: qcfri#i$": lfr,':,' J-,r'-,;'.i': the costs proportionally according to some indicator of project size (for example the costs of an NOCP would be allocated to all projects proportional to the number of treatments completed in that cost year). The decision as to how these costs should be allocated is not easy. Some ofthe drivers of costs, such as the costs of reporting and regular evaluation missions, would be the same independent of the size of the project, and so this would suggest equal allocation, while other costs would vary depending on the size of the project, and so this would suggest proportional allocation. It was decided for the purposes of this study that this first class of costs (those that do not vary by project size) are likely to predominate, and for this reason shared costs were allocated equally to the projects underneath them. Sensitivity analysis shows that the choice of allocation strategy has a much greater effect on one project compared to all the others - Kisoro District in Uganda, due to the small number of treatments performed in Kisoro compared to other Ugandan projects. It is felt that the equal allocation method. while causing Kisoro's per-treatment costs to appear large, are reflective of the situation in the field. Key Outputs The key summary statistics for each project are the total annual cost. and the total annual per- treatment cost. These results are presented with volunteer time excluded and included, and also disaggregated by source of funding, input type. and programmatic activity. An average is presented for all of the projects included in the study and also for each of the countries involved, and these average results are presented in the same manner as the individual projects results. Where average results are calculated, the contribution of each project to the average is weighted according to project size (in terms of total treatments in the cost year). These results are then assessed with reference to the key independent variables identified at the outset - project size, country. NGDO partner and Loaiasis co-endemicity. Care must be taken when interpreting these results. As will be seen, per-treatment costs between projects vary substantially, and averages disguise these differences. Also, any difference between the per-treatment costs, while being partially explained by the independent variables shown, will also vary due to factors of the project environment not assessed in this study, and many of these factors will not be under the complete control of project management or the various partners involved. Such factors include poor country infrastructure. civil conflict, and wariness of drug- taking in the community. a t 28 9o s,Q c 'Es ;R oo: @ o @ o @ oo o@ Erna* sI;;E o t *xOO;95 = soq o soq o soq o soq o oq o soo o 5 oZ a EePqe stg 'E .,, E EEE'd';odE s uEEt'd.: ".t oq o oq $EEE.9 oq o oq o oo et o eo oo ct oi 6-E s Ee AE =E oq o o d o o oco soq o r**E€E s Po .=A Etoo&e ;s Po .Eoitoote o oq oo ,{ oq ooo soo ct Oq o o; oqo oq o soco G -o.ag- E5Ftr c oq o oco oIo og o oq o TDtr .E .9 9Eq .E "E =uJ s E"= .E .9 o9c E EE =o oq o aEo.6 E= .o6FL Oco o d o ci q o oo e; o gE= 6Er= i5 s o oE.g =xc685 i5 oo ci oo: oq o o oo c; soo c, -$e,G'CE Etroo P- .9EEoh = o 3x€tt== -GE;"8 g oco -tr EEe(gtEN -N'E E'E €3 s -tr:.e -=-eGON -N'EE'i s8 oq o co o oq e soq o =a =t! sE oq o 0, :/ :E <F s be gE <F oco o oco oq o oq o ;eoq o i€E$:€:E 6 oo o o o o o Ei: o oq o (E o E9 PS g6F> s o o() 6 tr oo o oc o E o o = .: o o 6 o oo oo -9o E UJ IE P.E gEF> oo ci oc oqo oco oq o soo o ;; e rE.EE= =9Zo6iEE .:r6So5 o (o r{ o{ o) E o ah L o o- Eo -9o E uJ ti o oo o o otr ij,E biE5 l!v(JbE E> oo . o E" 0) Go 5 tr oo o o- o o u, o a. u; |E c o q, Ge o 'tr E +i G S o(,(, G d eS Bg o :ioootro o<Au ^65*o oA o o t4 o a. G' lt (E o a)o G q) QI o: e. ts =3oOc<Oroo td o o t4 oq. B o o.tr>9'ts 'r= ot) -o4o>olt 9icoOE 3c)oOo=trdt o o 0, :> o o o) 'd (L o) o c)3 oooo N UJ o q o o o o o- o oa E IJJ an o oo o o otr I 6 oo o o- .E o- E o -=.Y6 8E E!r5 -9>in !2o>OE:t9>trrr EEtE E3 .obFo_ rl) 6) J .9l! (t C !c This section details the process of data analysis - how various data collection errors were dealt with, how volunteer time was valued, how shared resources (such as NOCP support) were allocated to particular projects, and how key output statistics were generated. Management of SpeciJic Errors and Data Gaps During data collection the cost records were not often found in an ideal state - sometimes data were not recorded in the same format as used in this study, sometimes key documents and statistics were incomplete or missing. Adjustments and estimations thus had to be made in concert with the experience of those most knowledgeable with the data. Some of the more influential adjustments made were as follows: For the Nigerian NOCP for 2003, the accounts showed significant spending on two items: sustainability workshops, and computer software called 'Health Mapper'. The sum total for these items was US$113,104. While these are important and worthwhile expenditures, it was decided that they should be excluded, since they do represent future costs. Data on the purchase price of vehicles was occasionally impossible to find. In this case an estimate was made by using the purchase price for a similar vehicle purchased around the same time in the same country (found in a different costing sheet). Similarly, the purchase date of capital items was sometimes missing. In this case the year in which the project began was used as the most reasonable estimate. The sample size for local governments and communities in Ebonyi, Enugu and Anambra States was smaller than intended. For this reason caution must be used when reviewing the cost estimates for the funding sources in these projects, given the greater variability around the cost estimates. Volunteer Time Valuation and Interpretation CDDs and others give their time freely to ensure that drug distribution and other project activities occur, but it does not follow that their time has no value. In the design of this project it was thought crucial to capture this community contribution, and to place it alongside the other sources of resources invested in a project. To capture the value of the community resource of donated time, we had to ascribe a monetary value to volunteer labor. To do so we used values for the average per a I 26 capita Gross National Income (GNI) in 2003 for each of the three countries studied - Cameroon, Nigeria and Uganda (sourced from the World Bank's World Development Indicators). These yearly estimates were used to compute a daily estimate, and this daily estimate was multiplied by the total number of equivalent 8-hour days contributed by CDDs and others to find the community contribution. The values used are shown in Figure 6: Figure 6 Cameroon us$640 us$1.75 Nigeria us$320 us$0.88 Uganda us$240 us$0.66 Source: World Bank. World Development Indicators 2004 Although valuing the communities contribution is important, it must be understood that this valuation system is somewhat arbitrary - although the World Bank's per capita GNI figures maybe representative of the country in general, it is less obvious that they validly capture the value of the activities an individual would have undertaken were they not a CDD, or how much it would cost to pay someone to do the CDD's job. The correct value to put on a volunteer's time is open to debate, and for this reasons the study results are presented in two ways - with the value of volunteer time excluded and with it included. Allocation of Shared Resoutces A number of costs had to be allocated between projects. These included: - The costs of the NOCP. - The costs of NGDOs that work on more than one CDTI project. - The APOC project costs where a project has been disaggregated into its individual states or districts, as in the case of Anambra/Ebonyi/Enugu in Nigeria and Phase I in Uganda. In these cases there are two general methods that could be used for allocating costs. The first method is to allocate the costs equally (for example the costs of the NOCP would be allocated equally to all projects over which it had supervision), while the second method would be to allocate 27 the costs proportionally according to some indicator of project size (for example the costs of an NOCP would be allocated to all projects proportional to the number of treatments completed in that cost year). The decision as to how these costs should be allocated is not easy. Some ofthe drivers of costs, such as the costs of reporting and regular evaluation missions, would be the same independent of the size of the project, and so this would suggest equal allocation, while other costs would vary depending on the size of the project, and so this would suggest proportional allocation. It was decided for the purposes of this study that this first class of costs (those that do not vary by project size) are likely to predominate, and for this reason shared costs were allocated equally to the projects underneath them. Sensitivity analysis shows that the choice of allocation strategy has a much greater effect on one project compared to all the others - Kisoro District in Uganda, due to the small number of treatments performed in Kisoro compared to other Ugandan projects. It is felt that the equal allocation method, while causing Kisoro's per-treatment costs to appear large, are reflective of the situation in the field. Key Outputs The key summary statistics for each project are the total annual cost, and the total annual per- treatment cost. These results are presented with volunteer time excluded and included, and also disaggregated by source of funding, input type, and programmatic activity. An average is presented for all of the projects included in the study and also for each of the countries involved, and these average results are presented in the same manner as the individual projects results. Where average results are calculated, the contribution of each project to the average is weighted according to project size (in terms of total treatments in the cost year). These results are then assessed with reference to the key independent variables identified at the outset - project size, country, NGDO partner and Loaiasis co-endemicity. Care must be taken when interpreting these results. As will be seen, per-treatment costs between projects vary substantially, and averages disguise these differences. Also, any difference between the per-treatment costs, while being partially explained by the independent variables shown, will also vary due to factors of the project environment not assessed in this study, and many of these factors will not be under the complete control of project management or the various partners involved. Such factors include poor country infrastructure, civil conflict, and wariness of drug- taking in the community. ? t 28 a9. Results The results will be presented in the following three sections: I. CDTI Costs summed across all projects included in the study: . Total Cost and Cost Per Treatment . Total Cost Broken down by Source of Funding . Total CDTI Cost Broken down by Input Type . Total CDTI Cost Broken down by Programmatic Activity . National Government Costs Broken down by Input Type and Programmatic Activity . Regional Government Costs Broken down by Input Type and Programmatic Activity . Local Government Costs Broken down by Input Type and Programmatic Activity . APOC Costs Broken down by Input Type and Programmatic Activity . NGDO Costs Broken down by Input Type and Programmatic Activity II. CDTI Costs broken down by project: ' Total Cost and Cost Per Treatment, by Project ! Cost Per Treatment for each CDTI Project, Disaggregated by Source of Funding t Cost Per Treatment for each CDTI Project, Disaggregated by Input Type . Cost Per Treatment for each CDTI Project, Disaggregated by Programmatic Activity IIL CDTI Costs vs. various explanatory variables: . Differences in Cost Per treatment by Country . Differences in Cost Per Treatment by Price Level . Differences in Cost Per Treatment by Project Size . Differences in Cost Per Treatment by NGDO . Differences in Cost Per Treatment by Loaiasis Coendemicity . Is External Funding a Complement or a Substitute for Government Funding at the Project Level? All basic analyses will be shown with the value of volunteer time both included and excluded. All further analyses and graphs will be shown with the value of volunteer time excluded. I. CDTI Costs Summing Across All Projects Total CDTI Costs and Cost Per Treatment Total Project Cost (volunteer time valued) = Total Project Cost (volunteer time not valued) = Per Treatment Cost (volunteer time valued) = Per Treatment Cost (volunteer time not valued) = t I 29 s0.74 $0.58 Total CDTI Cost Broken down by Source of Funding Total CDTI Cost By Source of Funding (volunteer time not valued) lnput Type TotalGost Cost Per Treatment National Government (NOCP) Regional Government Local Government Communities NGDO APOC APOC (NOCP) Breakdown of CDTI Costs by Source for All Projects (volunteer time not valued) National Govt (NOCP) 1.3Yo APOC (NOCP) 1 .8o/o Regional Govt 11.3o/o APOC 26.7% Local Govt 37.3% NGDO 15.3% Communities 6.3% 30 $27,ng I $228,456 T- $0.0 $754,094 I $0.22$127,$5 F$309,111 | $0.09 $539,674 I $0.15 $36,515 I $0.01 Total CDTI Cost Broken down by Input Type Totat CDT! Gost By lnput Type (volunteer time not valued) lnput Type TotalGost Cost Per Treatment Transportation Consumables Employed Personnel Per Diems Overheads Volunteer time Capital Costs (Vehicles) Capital Costs (Other) Breakdown of CDTI Costs by lnput Type for All Projects (volunteer time not valued) Capital Costs (O,theo Transportation 10.7o/o Capital Costs (Vehicles) 9 3Yo O\erheads 3.60/o Consumables 17.20/o Per Diems 11.2o/o Employed Personnel a 3l 46.9o/o $216,448.86 | .06 $349,799.m | $o $950,583.90 | $0.27I $o.oz - I $o.oz $228,252.U $o.oo I $189,036.99 T $21.226.85 T $0.00 .05 $o.oi Total CDTI Cost Broken down by Programmatic Activity Total CDTI Cost By Programmatic Activity (volunteer time not valued) lnput Type Total Cost Per Treatment Training of Volunteers All Other Traininq Mobilization / Sensitization Drug Distribution Chain CDD Druq Collection CDD Drug Distribution Census Monitoring and Evaluation Reportinq to APOC All Other Administration Other Project Activities Breakdown of CDTI Costs by Activity for All Projects (volunteer time not valued) Other P roject Activities 3.2% Training of Volunteers 12.60/o All Other Administration 11.1% Reporting to APOC All OtherTraining 12.SYo3.6% Monitorj ng and Evaluation Mobilization / Sensitization 18.50/" 25.5o/o Census 0.0o/o Drug Distribution Chain 11.0% CDD Drug Distribution 1 .7o/o CDD Drug Collection 0.2o/o ] 5Z $255,330 $0.07 $253, 271 $0.07 $374,f09 s0_1 1 $222,290 $4,200 $0.00 $33,700 $4la $0.00 $516,872 $73,695 [_I $0.02 $225,333 $63,811 .06 National Government CDTI Costs Broken down by Input Type and Programmatic Activity Breakdovvn of National Govemment CDTI CosB by lnput Type for All Projects (volunteer time not valued) Capital Costs (Vehicles) 0.8o/o Capital Costs (Other) 0.7o/o Overheads 16.9o/o Transportation 6.SYo Consumables 25.5o/o Per Diems 18.6% Employed Personnel 31 .iYo a Breakdown of National Government CDTI Costs by Activity for All Projects (volunteer time not valued) Training of Volunteers 1.1o/o Other Project Activities 8.$Yo All Other Training 8.9% Mobilization / Sensitization 1',!.90/o All Other Administration 26.4o/o Drug Distribution Chain 3.3o/o Reporting to APOC 8.2o/o Monitoring and Evaluation 31 .4o/o JJ Regional Government CDTI Costs Broken down by Input Type and Programmatic Activity Breakdown of Regional Government CDTI Costs by lnput Type for All Proiects (volunteer time not valued) Capital Costs (Vehicles) 2.1o/o Capital Costs (Other) 0.4o/o Overheads 10.0o/o Transportation 9.80/o Per Diems 7.60/o Consumables 12.5o/o Employed Personnel 57.7o/o Breakdown of Regional Government CDTI Costs by Activity for All Projects (volunteer time not valued) Other Project Activities 2.5% Training of Volunteers 8.6% All Other Administration 14.3% AII OtherTraining 8.8% Reporting to APOC 5.70h Mobilization / Sensitization 15.00/o Monitoring and Evaluation 27.60 Drug Distribution Chain 17.5% t 34 Local Government CDTI Costs Broken down by Input Type and Programmatic Activity Breakdown of Local Govemment CDTI Costs by lnput Type for All Projects (volunteer time not valued) Capital Costs (Vehicles) 2.lYo Capital Costs (Other) 0.1Yo Overheads 3.1Yo Transportation 6.3% Per Diems 6.1o/o Consumables 1.1Yo Employed Personnel 81 .3o/o a Breakdown of Local Government CDTI Costs by Activity forAll Projects (volunteer time not valued) Other Project Activities 4.60/o All Other Administration Training of Volunteers 21 .40/o7.9o/o Reporting to APOC 0.9o/o Monitoring and Evaluation 21.00/o All Other Training 14.5o/o Drug Distribution Chain 10.6% Mobilization / Sensitization 19.2o/o 35 Breakdown of APOC Costs (NOTF + Project) by lnput Type for All Projects (volunteer time not valued) Capital Costs (Other) 2.8o/o Transportation 16.5o/o Capital Costs (Vehicles) 22.3% Overheads 0.5olo Consumables 20.3o/o Per Diems 21 .90/o Employed Personnel 15.7o/o APOC CDTI Costs (NOCP and Project) Broken down by Input Type and Programmatic Activity Breakdown of APOC Costs (NOTF + Project) by Activity for All Projects (volunteer time not valued) Other Project Activities 2.7% Trarning of Volunteers 10.2% All Other Administration 10.81o Reporting to APOC 5.5% All Other Training 13.60/o Mobilization / Sensitization 11/% Monitoring and Evaluation 34/% Drug Distribution Chain 11 .5o/o a e 36 NGDO CDTI Costs Broken down by Input Type and Programmatic Activity Brcakdown of NGDO Costs by lnput Type for All Projects (volunteer time not valued) Capital Costs (Othe0 0.9% Transportation Capital Costs (Vehicles) 11 .7o/o 7.6% Overheads 6.5o/o Consumables 27 .1o/o Per Diems 10.90h Employed Personnel 35.3o/o a Breakdown of NGDO Costs by Activity for All Projects (volunteer time not valued) Other Project Activities Training of Volunteers 1.8% 4.2o/o All Other Administration 20.3o/o All Other Training 14.Oo/o Reporting to APOC 6.4o/o Drug Distribution Chain 11.5% Monitoring and Evaluatron 28.7o/o 5t Mobilization / Sensitization 13.1o/o II. CDTI Costs Broken Down by Project Total Costs and Cost Per Treatment, by Project Cross Rivers State. Niqeria Total Project Cost (volunteer time valued) Total Project Cost (volunteer time not valued) Per Treatment Cost (volunteer time valued) Per Treatment Cost (volunteer time not valued) Yobe State. Niqeria Total Project Cost (volunteer time valued) = Total Project Cost (volunteer time not valued) = Per Treatment Cost (volunteer time valued) = Per Treatment Cost (volunteer time not valued) = Ebonvi State. Niqeria Total Project Cost (volunteer time valued) = Total Project Cost (volunteer time not valued) = Per Treatment Cost (volunteer time valued) = Per Treatment Cost (volunteer time not valued) = Enuqu State. Niqeria Total Project Cost (volunteer time valued) = Total Project Cost (volunteer time not valued) = Per Treatment Cost (volunteer time valued) = Per Treatment Cost (volunteer time not valued) = Anambra State. Niqeria Total Project Cost (volunteer time valued) = Total Project Cost (volunteer time not valued) = Per Treatment Cost (volunteer time valued) = Per Treatment Cost (volunteer time not valued) = t 38 $412.421 $260.984 $0.35 $290.043 Hoima District. Uqanda Total Project Cost (volunteer time valued) = Total Project Cost (volunteer time not valued) = Per Treatment Cost (volunteer time valued) = Per Treatment Cost (volunteer time not valued) = Masindi District. Uqanda Total Project Cost (volunteer time valued) = Total Project Cost (volunteer time not valued) = Per Treatment Cost (volunteer time valued) = Per Treatment Cost (volunteer time not valued) = Kasese District. Uqanda Total Project Cost (volunteer time valued) = Total Project Cost (volunteer time not valued) = Per Treatment Cost (volunteer time valued) = Per Treatment Cost (volunteer time not valued) = Kisoro District, Usanda Total Project Cost (volunteer time valued) = Total Project Cost (volunteer time not valued) = Per Treatment Cost (volunteer time valued) Per Treatment Cost (volunteer time not valued) Centre Trois nce. Cameroon Total Project Cost (volunteer time valued) = Total Project Cost (volunteer time not valued) = Per Treatment Cost (volunteer time valued) = Per Treatment Cost (volunteer time not valued) = Nord Province. Cameroon Total Project Cost (volunteer time valued) = Total Project Cost (volunteer time not valued) = Per Treatment Cost (volunteer time valued) = Per Treatment Cost (volunteer time not valued) = a 39 $0.76 $0.55 $2.78 $2.33 $336.836 $278.894 $2.06 $1.71 $1.21 $0.86 83 ;21 $0.64 s232.242 <f, oq N@ oq N @. o o o @ lt E' o (E E) o E) E"(E oEa ^0,PE5-oN> ,=o6Z -oO otrfi:O-L _or-ObEoE -O -) o o a oo o E(! o FI o o. !o & p q .s' \J $ 0o oo $ S'\!\NI sU q) q,) qJ \J 4 U oo tt, 63 oN o oo cI a!ooorL .L o o- o u? o @ oq o@ a JJ tr ol! uJ(, d uJ E oz oo(, o GY (! .E o I E" UJ oll uJ o G = o o o oo e o .9,Y g ll E(E ol! o o o t ooIo lcotoJd IIOC Co oz >=oo36Eg 6:=oEZ6:g.nf- .:'6!lE..ooo oorxo90-o- TI!!TII $G -e6.oE =-8C0)9 >oF,. [i e;;E-ed.En6 .E g 6 E H bOO#=axs€EEEZ9:6r'o-a83fi36S38 rItrtrTIItr oq N@ oq N@ o CL ct ;I E' o .E E'I o E') ED lEo= og 60?giN6 =, oa) OE0,r- o-. ^-O*o Ftro=Oo -}, o o o oo o E oo a o o- \) F\' 1 €' qJ $" 06 0os ql L{ !FrI sIBqJ \J \) l\ U\ 4 a ooou?) E3oN o oo o auoo oL .F(#b G oq o @ oq oeJ t ol! UJo t UJ E oz ar, o F o c oo g o6 v oooo(E Y o(! = (g .E o I ED =tr UJ o .o UJ oll o (E ! E .E 2 o .>tr oo o o lcoloJd tIGC c'ls oq (Y) @ oq(\l @ oq N@ oof tooN ^6l*o -1 (J6-. o E (g o FI o o- oq @ o u? o@ oq o@JJ tr ol! ul(9 E UI o o F 6) oo o o6 Y g5 e EE-F E EaE,EEfiE !c.IEEE:i?€=E ;EEE EEEEEg-HgEF E=EsEe.$Es=8IltrtrMtrIltr oooo IEY oG (E .E oI E .cl E .E g, IJJ '; E oll UJ o o o o E o oo o 't o .o t, o G E'o ED ED G o i5 BgNE c) 'il9 (, .=oF'ili o.g x!6s o o o o o o F (! o I o o. tcafo.rd IIOC t q\ Q b6 \ .s' \) !ov 06 OA a Uq,I iH Ut q,) U t) H nl{ q t oz c.)$ 6' ED(! o t oz L o o E oo o o .!2x 0?' LI 6Z'0$ zs'0$ rt'0$ 0z'0$ 6e'0$ ooooqqq9 F990OOOO (sJelloO Sn ,OOZ) uoOnqliluoC luauqEsl rad ,L'0$ 9t'0$ 9Z'0$ ?t'0$ 9Z'0$ 0z'0$ 8r'0$ sl'0$ zr'0$ z0'0$ 0r'0$ 60'0s 80'0$ 80 ez'0$ z t'0$ z !'0$ e! ,0'0$ 0r'0$ oq tz' o!i o or EI IE oo(, z c o Etr o o o E o o o o o. o .Eo '" o o o o E .E P ! o o- q) {\l\I s q) U q\\ d e { + B toq + 3 u t) =..s' qJ aq ,S a t$ FoI I I oo oz I o E o o o T oooo lEY E ,6 .E = L__l tg .E o I o -o E a! tr gt IJJ .; oIt ut sz'0$ (, o 60 si()d !z'0$ & oq o oq o l z0'0$ rz'0$ '.1' I ;j t 1 i L i i .i Ij I I I I I I .t il. I I I I I l , I.r: i I i l .1.. I I i L l l I i l l l t, <f,s oq a o a a oq C oOEoit aDB(\t ;oo@Odoo o E(!6EdFOL o o- orf c,g oN c;o oo c;o(Il o ,9z C C) oz 5E o (LoxoO6e5:E O (I1\vanfEs;E888(5(I)xo(,(LL IT!trIII (Et, G E"f o o o E GI(J t,o E oz o Ei: oo 6 6,(J o U' lt t,o (E E'o El cn(E ,.2o f. oo @ o oo o (E E o o. 90'0$ zs'0s zz'o! -S s' \) \,) !\ NJ\ 4 U aI s) ,qJ\ ao L t o 6 X H v) d t, (n a) o U F U oL'o$ 90'0$ soot 9 r'o$ z9'o$ o$ 90'0$ ra)$ o 'qN@ oq N@ luauleoJl rod lsoc oou?q @o o u? o@ oq ora !,o E oz o .EF L o o 6 o .No C'o oL o- E') '6 (g E o ='= o E ag oLF L o o- o oo 4"" "r\ ,"n "% -{ \ [: ,.% .r"\ \oo o- oo@ oo o_ oot* o oo o^ oo rO oo o- oos oooooo oo o- oo ooo oo(o oooo(f) N Eo E (E EF L o) or O- Nea3 5C) .!!I r95(LFIt e00z u! sluauleoJl oN \) r.rq Uq) \ €' \) B\) t\ h\ h q \J CJ \.) .ts a a Differences in Cost Per Treatment by NGDO Dffirences in Cost Per Treatment by Loaiasis Coendemicity Average Cost Per Treatment, By NGDO $2.00 $1.60 ${.20 $0.80 $0.40 $0.00 CBM Global 2000 HKI SSI UNICEF Cost Per Treatnrent for Centre-Trois (Coendemic Loaiasis) vs. Projects without Loaiasis Coe ndemicity $2.00 $1.60g o o ! st.zo o 5 E so.aot; o o. $0.40 $0.00 Centre Trois Average for All Others ) 46 $1.71 $0.58 $0.38 $0.45 $1.71 $0.s2 F-$ 9o. ?V"oo o% %,,, \ "rr" %.v %,o % \ ,,, 9r- 's/ tgo -( oo o@ 5o o E ooJ + EE a ,. F88EZ<lr ooo(osN ooo@@@ o@ o@ oq @ oc! @ E' .E!, lJ- o(9 E .9 ED ot t, G Eo oJ E' '.i E, =; E' .=!, IJ. o o o. !t (g oo oz .g oo E) .E o ocJnos qcBg uro4 6urpunl luauleaJl-Jad 6\. q) \){ USl \ q,rs 06 \ \ q) !r a -s 4 B qJ 0o \ qJ li ki a The generous use of graphics in the results section is intended to aid the reader in recognizing themes in the cost data. Similarly in this discussion section the intent is to identify trends and significant findings in the cost data - to abstract from the data the general structure of funding to CDTI projects, and how that differs between projects and along the three cost categorizations. It must be noted that the intent is not to assign definitive meaning to the trends and significant findings, nor to infer a causal link in any of the particular correlations that may be seen between CDTI costs and the various independent variables. Although it is tempting to begin making these conclusions, this task should ultimately fall to those most experienced with the context of financial support to CDTI projects - APOC management and the Technical Consultative Committee, NGDO management, and National-level CDTI programme staff in the various APOC countries. This caveat aside, the results reveal the following (with discussion following the order used in the results section above): The average cost per treatment of CDTI was $0.58, taking into account all the costs described in the introductory sections of this report. When volunteer time was valued and incorporated into the calculation this per-treatment cost rose to 50.74. The Local Government bore the greatest costs of any CDTI partner, contributing more than one third of the total resources invested in CDTI activities in the projects sampled. The next largest contributor was APOC Project funding, followed by the NGDO and the Regional Government, with National Government, APOC NOCP funding, and Community contributions being relatively minor. Payments made to individuals (salaries and per diems) represented more that half of all costs, with the salaries of Employed Personnel representing by far the largest cost category. Consumables was the next largest, followed by Per Diems and Transportation costs. Capital Costs and Overheads represented relatively minor costs, although the contribution from Vehicles approached 10%. The most costly programmatic activity was Monitoring and Evaluation, accounting for a quarter of all costs, followed by Mobilization and Sensitization. Training, of both volunteers and others, also accounted for about a quarter of all costs, and the remaining costs were spread over a number of other categories. As might be expected, the community-level 48 10. Discussion activities performed by CDDs incurred very minor financial costs. It is also notable that reporting to APOC consumed less that 4Yo of all resources. When we look at how the costs break down for each particular Source of Funding, it can be seen that an increasingly larger fraction of costs is taken up by Employed Personnel as we progress down the levels of government, with salaries accounting for more than four fifths of funding at the Local Government level. Given Local Government's considerable contribution to total costs, it can be seen that this overwhelming salary allocation at the local government level drives the high proportion that salaries represent in the aggregate cost per treatment results. Salaries are significant in the respective NGDO contribution, however only represent the forth largest input type when we look at the pooled (NOCP and project) funding from APOC, behind Capital Costs, Consumables, and Transportation. APOC and the NGDO are also notable as the predominant suppliers of capital goods to the projects. Looking at the activities supported by each Source of Funding, Monitoring and Evaluation is important in all cases, although more so for APOC, the NGDO and the National Government. There is a surprising amount of similarities between funding source when we look at the other programmatic activities, although Local Governments use relatively few resources on Reporting to APOC, and on Other Administration. When we compare the crude results for individual projects, we can see there is considerable variation. This variation is skewed, with the lowest cost per treatment result - Hoima District in Uganda at $0.33, lying closer to the majority of the cost results than the two top-end outliers -Centre Trois in Cameroon at $1.71 and Kisoro in Uganda at $2.33. The graph of cost per treatment by number of treatments suggests that, with the exception of Kisoro and Centre Trois, the cost per treatment remains relatively constant regardless of the number of treatments. In all cases the inclusion of volunteer time results in a small but still significant increase in the cost per treatment result. When we look at these differences disaggregated by the source of funds, we can see that in general a higher cost per treatment result involves higher contributions across all funding sources although there are a number of deviations from the norm: o Kisoro's high cost per treatment result is dominated by the NGDO contribution, which represents well over $ 1.00 per treatment alone. 49 o Centre Trois's high cost per treatment figure is similarly dominated by APOC funding, about $0.80 per treatment. A similar high contribution from APOC can be seen in Nord Province, the other Cameroonian project. o Ebonyi has a high contribution from Local Government compared to other projects When we look at the project cost per treatment differences disaggregated by input type we see much more uniformity, with most projects mirroring the significant contribution from salaries seen in the aggregate cost per treatment result. Two minor deviations from the norm can be seen in Ebonyi, where salaries dominate costs more than in other projects, and Masindi, where proportionally more is spent on Per Diems than in other projects. Looking at the mix of programmatic activities in different projects, we can see that Monitoring and Evaluation are key factors in the three most costly projects (Kisoro, Centre Trois and Nord). In Kisoro Other Administration also represents a significant cost. Monitoring and Evaluation is generally less significant in the Nigerian projects, with Mobilization and Sensitization generally being of more importance in Anambra, Cross River, Ebonyi, Enugu and Yobe. When we look at the contributions from Government, APOC and the NGDO across projects we can see again the significant contribution made by governments, but we can also see how this shows marked differences between different projects. The largest source of this variation is the NGDO partner, ranging from less than five cents per treatment to more than one dollar per treatment. Looking at the variables we identified a priori as being potentially influential determinants of project costs, we can see that there is a difference between the three countries investigated, Cameroonian projects on average being much more costly than Ugandan or Nigerian projects. It must be noted however that the Cameroonian average only contains two projects, and so the high cost per treatment result for Centre Trois will be influential. In comparison, the high cost per treatment figure of Kisoro will be of much lesser effect, both because of the larger number of projects included from Uganda and also due to Kisoro's small size meaning it will not have a large effect on the pooled results. The apparent lack of Local Govemment contribution in Uganda is an artifact of the data collection processt. , I 50 8 In Ugandawe collected costs at the national and district level but not at the sub-district or sub-county level. oProject size does seem to have an effect on per treatment costs, although there is significant variation at the low end, with per treatment costs showing much less variation in those projects with over 250,000 treatments in the cost year. If there is an effect it is clearest with Kisoro, which is less than half the size of the next smallest project, and twenty times smaller than the average project size. However, the relatively low costs of Hoima and Masindi show that small size, when not in the extreme, is not a barrier to low per treatment costs. The apparent variation in cost per treatment between NGDOs must be contextualized by noting that this represents just one project, Centre Trois, which we will retum to in the next bullet point. All of the other NGDOs have relatively similar cost per treatment results. It is hard to draw conclusions about the one Loaisis Coendemic project, given it is the only project included with those particular circumstances. However it may be the most obvious explanation for Centre Trois's high cost per treatment resulq which is more than three times the average ofthe other projects. Although not a cost determinant as such, it is thought provoking to look at how project-level government support (Local and Regional Government) varies with APOC project funding and NGDO funding. Assessing this link quantitatively, vs find that the correlation coefficient between APOC project funding and Government project funding is high, atr=0.71 (the commensurate correlation for NGDO funding is only r:0.1 1). Although it is not possible to display it graphically, we can control for the effect of other cost determinants statistically, and by controlling for the differences between countries the correlation coefficients rise even more. The fact that the coefficients are both positive and high does not necessarily lead to a particular conclusion, but it is worthy of note that in general external project funding varies with domestic project funding, and that this effect is stronger when country-level effects have been controlled for. In this sense external funding is a complement to domestic funding, not a substitute. a 51 11. Conclusion This study is the first attempt to examine the total cost structure of CDTI projects across countries. It can not be said to be representative of all the APOC CDTI projects but it does offer some perspective on the cost profile of a given CDTI project with respect to source of funding, type of input and uses of funds (activities). The study raises questions about whether it is appropriate to value volunteer time, the inclusion of employed personnel at all govemment levels as a CDTI cost and the sensitivity of the allocation method per activity of the employed personnel. The following graph shows an analysis addressing one of the issues - the effect of excluding the cost of govemment employed personnel from the total costs. crrph 4 : Showing th. Proportlon of Costs Paid by APOC, NGDO, Gov.rnmht .hd Comnlty ior Each Progr.mm.tlc Actlvity, Sunrrilhg Acrcs Atl Projccts, rnd Excludlng EEploycd Parsonhcl ahd Voluht@r Tlm. Other Prcjcct ktivities Al Other &minisrataon Ropoding to POC hnitonng and Evaluataon I POC - NOCP ! FOC - Prcj€ct O NGDO I Gou - National I Go{ - R6gional I Gou - Local I CommunityCDD Drug Distdbution COD Drug collection Drug Distribulion Chain mbilization / Sensnization Al Oth.r Training Training ofVoluntoers o% 10% 2O'/o 30./o 100/n 50% 600/o Percont.g6 of total Cost 7Oo/o AOo/o 90vo 100% The proportion of total costs funded by APOC increases relative to the costs funded by various levels of government and communities for all the activities with the exception of the three activities that take place exclusively at the community level - CDD drug collection, CDD drug distribution, and census taking. Further analysis ofthe data are planned to provide insights into these questions. a q 52 I I