• Diseases
  • African Trypanosomiasis
  • Buruli Ulcer
  • Chagas Disease
  • Cholera
  • Dengue
  • Dracunculiasis
  • Emerging infectious diseases
  • Fascioliasis
  • Leishmaniasis
  • Leprosy
  • Lymphatic Filariasis
  • Malaria
  • Onchocerciasis
  • Schistosomiasis
  • Soil transmitted helminthiases
  • Trachoma
  • Tuberculosis
  • Yaws
  • Zoonoses
• Global health topics
  • Agriculture
  • Biotechnology
  • Climate change
  • Capacity building
  • Community health
  • Conflict
  • Diagnostics
  • Economics
  • Environment
  • Ethics
  • Equity
  • Funding
  • Gender
  • Health policy and systems
  • Human rights
  • Immunology
  • Innovation
  • Operations research
  • Pharmaceutical products
  • Poverty
  • Primary health care
  • Research priorities
  • Research translation
  • Social environment
  • Vaccines
  • Vectors
• News
• Reviews
• Editorial opinions
• Research articles
• Reports
• Virtual journals
• Profiles
• Policy & Strategy
• Funding opportunities
• Networking
• Resources
• Toolkits
  • Knowledge Translation (KT) Toolkit
  • Health Research Systems Analysis (HRSA) Toolkit

Communities of practice

• Disease Reference Groups
• Thematic Reference Groups

A Basis For Financial Decision-Making On Strategies For The Control Of Human African Trypanosomiasis

24 Jan 2008

Source: WHO/TDR

A P M Shaw ¹ , P Cattand ² , P G Coleman ³ , M John ⁴

¹A P Consultants, Upper Cottage, Abbotts Ann, Hampshire, SP11 7BA, UK
²1 rue de l'Hotel Dieu, F-74200 Thonon, France
³CTVM, University of Edinburgh, Easter Bush, Roslin, Midlothian, EH25 9RG, UK
⁴Le Kerrio, 56230, Berric, France

Introduction

This discussion paper seeks to identify the main issues involved in designing cost-effective strategies for the control of human African trypanosomiasis and to highlight areas where more information needs to be compiled or new research initiated.

Some form of control of the disease has been ongoing in most parts of Africa since the beginning of the last century. Thus many schemes have been funded and budgeted for, however the economics of the different approaches and their cost-effectiveness has only been studied sporadically. Thus, whilst most people working on the disease have a very clear idea of what the most cost-effective strategies are in their locality, given their resources and price structure, little has been done to bring together the information thus gained in order to help decision-makers design strategies and prioritise in new areas or under new circumstances.

This paper is the first step in an ongoing exercise to

• identify the main factors influencing the cost-effectiveness of the different approaches to controlling the disease

• review the current literature on this subject

• identify potential research areas and information needs.

For 1999, the burden of human African trypanosomiasis was estimated at 66,000 deaths and 2 million DALY's lost [15]. Although only 45,000 new cases were reported [16], the likely number of people affected is probably ten times greater, thus approaching half a million. Of the 60 million people thought to be ‘at risk’ only 3 – 4 million are covered by some form of disease surveillance.

The control of human African trypanosomiasis is based on combinations of the four different approaches illustrated in Figure. 1.

Figure 1

• treating those human patients diagnosed with the disease;

• trying to improve diagnosis, by some level of surveillance, designed to find patients and ensure that they are treated earlier, hopefully while they are still in the first stage of the disease

• proceeding to more active forms of case-detection, aimed not only at finding and diagnosing patients, but also at reducing the size of the human reservoir of the disease, this applies to the gambiense form;

• trying to reduce the chance of people picking up infections from domestic animals or wildlife, to this end cattle are being routinely treated around new outbreaks in Uganda, to control the rhodesiense form of the disease (personal communication I. Maudlin);

• controlling tsetse fly populations so as to reduce transmission of both forms of the disease

• and finally, as indicated in Fig. 1 by the two boxes marked ‘point of contact’, avoiding areas likely to lead to infection - a strategy that people have used since time immemorial. This applies above all to avoiding contact with the tsetse fly. This strategy has mainly been employed by livestock keepers trying to prevent their cattle becoming infected and either dying or being less productive. However, people have become aware of the dangers of working in certain tsetse-infested thickets, and have avoided these. In the past it has been thought that people could avoid contracting the rhodesiense form of the disease by simply avoiding areas containing wildlife. In the early stages of an epidemic of sleeping sickness, the disease has tended to be found among people whose occupations put them at risk, by bringing them into contact with infected flies, particularly at certain times of the day (when collecting water, washing clothes, or when entering game reserves, as in the case of beekeepers, hunters, park rangers, etc.). Avoidance, whilst not specifically discussed here, is a control strategy of sorts, although in practice it has mainly been used by cattle herders to protect their stock

Finding and treating patients

As cited above, WHO believes that only about 10% of sleeping sickness patients are correctly diagnosed as having the disease, and receive treatment. This proportion may be larger in endemic foci where active surveillance is undertaken, but it could also be a great deal smaller. Typically patients who are detected passively have suffered from symptoms for some time, possibly years, in the case of gambiense sleeping sickness, and have had several attempts at having their symptoms treated and at obtaining a correct diagnosis. Usually, this will have involved several trips to their rural health centre, possibly also to a nearby hospital or treatment centre, a visit to a local healer, and being treated for malaria and other diseases before being diagnosed as having sleeping sickness. Both during trips to the health centre and while the patient is hospitalised, a relative will usually be needed to accompany and look after the patient. As the disease progresses, the patient in search of a diagnosis will have become more and more of a burden on his family, requiring care and being unable to undertake normal activities.

A very rough estimate of the possible cost to patients while trying to obtain a diagnosis was made by the author in Landell Mills (2000), for the situation of rhodesiense in Uganda. This came to US$ 25 per patient, including estimates for the cost of transport to rural health centres and hospitals, treatments for malaria and painkillers, and an estimate of time taken by relatives to accompany and care for the person. For those rhodesiense patients who are never correctly diagnosed, but do receive some treatments, this cost would rise to a minimum of US $50, to include consultations with a local healer, further trips to treatment centres, possibly a short stay in hospital, and care by relatives.

For those correctly diagnosed, the costs of treatment were estimated in WHO [14] and are given in Table 1, on the basis of then current drug costs, treatment regimes in use and estimates of the cost of hospitalisation.

However, these costings will need to be re-evaluated in the light of current plans to make drugs available free of cost from WHO for a specified number of years. The costs of transport and administration will then remain, but for the relevant drugs the cost paid by recipient countries or programmes will then consist only of transport and drug administration, so that, as the table indicates, treatment costs could be reduced by significant percentages. Nevertheless, in economic as against purely financial, terms, the use of these drugs will still represent a resource cost, which should be taken into account. Furthermore, their availability on these terms will alter over time. Taken together, this implies that the average cost for patients passively diagnosed and correctly treated, a high proportion of whom will already be in the second stage of the disease, would be of the order of $ 150 – 300 each.

In order to analyse this aspect further, there is a need to:

• analyse case histories of individuals being treated for the disease in order to determine by what process they obtained a diagnosis, how long this took and how much this cost them and the health services

• and to collate information on currently used treatment regimes and the costs of hospitalisation at trypanosomiasis treatment centres.

Economics of controlling the human reservoir: gambiense form of the disease

Turning next to active surveillance, which has been the mainstay of the programmes to combat the gambiense form of the disease, a set of calculations was undertaken for the WHO [14] expert committee report, which are given in a series of tables in Annex 95 of that report. The discussion below is based on these calculations, which were originally presented in Shaw, John and Cattand [13] and followed the same approach as those prepared for the previous WHO Expert Committee on Human African Trypanosomiases by Shaw and Cattand [13]. In order to produce a coherent set of costings it was necessary to try and use a real situation as a basis, while making some adjustments to produce a scenario in line with generally accepted norms. Thus, the figures and prices used were based mainly on M. John's work in the Moyo District of Uganda [4]. The figures used in 1985 were based on WHO's work in the Daloa area of Côte d'Ivoire.

As a next step the calculations in WHO [14] have been used to create a spreadsheet (Microsoft Excel ©) model. This allows the results to be calculated for any starting prevalence. The population covered, number of units involved in surveillance, sampling intensity, sensitivity and specificity of screening and diagnostic tests, as well as all prices and other costs can also be varied. The results of repeated runs of this model enable the relative cost-effectivity of different sampling strategies at different prevalences to be analysed. These are presented and discussed in the series of graphs that follow.

The original analyses were based on five alternative surveillance strategies. These were the classic mobile teams, fixed-post surveillance and the less widely used innovative techniques of filter paper sampling by trained community animal health workers, either visiting the community or based at rural health centres.

• Fixed-post or passive surveillance, where patients presenting with symptoms that are difficult to diagnose or don't respond to treatments, say for malaria, are eventually referred to a treatment centre and tested for a variety of disease, including trypanosomiasis, and those with the disease are eventually diagnosed. The initial screening test is performed on wet blood.

• Filter paper sampling at rural health centres, where community health workers based at rural health centres receive some training in collecting samples on filter paper and then routinely test any new patients presenting themselves at the health centre for whatever reason.

• Filter paper sampling by community health workers, who have been trained in collecting filter paper samples and then spend 20% of their time collecting samples and following up seropositive individuals. This was based on experience in Uganda [4] and Côte d'Ivoire [5].

• Monovalent mobile teams, the classic surveillance teams; in this scenario the CATT is performed on whole blood, and all the parasitological tests except for lumbar punctures are done by the team in the field. The monovalent teams work only on trypanosomiasis.

• Polyvalent mobile teams, which operate in the same way as monovalent teams, except that only a third of their work consists of screening for trypanosomiasis.

In order to standardise the results, these were calculated for an area with a human population of 100,000 people, containing ten rural health centres, and where 20 community animal health workers were operating. The numbers screened by each strategy were assumed to be as given in Table 2.

The cost calculations covered:

• initial screening using the CATT test

• parasitological confirmation using gland punctures, CTC test, m-AECT and lumbar punctures

• training of staff in specialised techniques needed for diagnosing trypanosomiasis

• administrative overheads

• depreciation (annualised cost) of capital items (vehicles, laboratory equipment, etc.)

• travel allowances and a share of salaries of all staff in proportion to the amount of time they spend on trypanosomiasis control

• running costs for vehicles, specialised equipment and other recurrent costs for each surveillance strategy.

Firstly, the relative performance of the different surveillance strategies in terms of the cost of finding gambiense patients was analysed. The results, in terms of US$ per trypanosomiasis patient found, are given in Figure 2. Obviously the cost of finding a patient declines very rapidly as the prevalence increases, since a higher proportion of those screened are infected. This clearly does not reflect any increase in efficiency, simply that more and more of those screened are infected, so the costs of the operation are averaged out over a larger number of individuals. Fig. 2 is given in three sections, so that the differentials between the costs of each sampling strategy can be seen more clearly.

At very low prevalences (1% or less, see Fig. 2a), such as those encountered in past years in areas where surveillance was reasonably regular, and the disease was considered to be under control, the costs at a 0.05% prevalence are well over US$ 2000 per patient found using rural health centres or mobile teams, and drop to just under US$2000 for community health workers. Passive or fixed-post detection costs only US $50, since there are virtually no overheads. When the prevalence reaches 1%, the cost of passive detection falls to just over US $20 per person, surveillance using community health workers costs just under US $100 per person, and using mobile teams or rural health centre costs between US$ 120 and US$ 140.

Fig. 2a

At medium level prevalences (from 1% to 5%, see Fig. 2b) these costs per patient found continue to fall, and the differentials between strategies narrow further, with the cost of passive detection being US$14, community health workers US$22 and the other three options about US$30.

Fig. 2b

At high level prevalences (over 10%, see Fig. 2c), the cost of patients found passively falls to around US$ 10, for all surveillance strategies at a prevalence of 20%. When the prevalence reaches 50%, the cost of detection becomes very low, around US$ 5 for all surveillance strategies, except passive detection where it remains at about US$10.

Fig. 2c

Given the way in which the figures were calculated, by independently building up the cost of each strategy using local norms and prices, it was surprising how similar the costs for finding patients using different strategies were. Setting aside passive detection, of the four active detection strategies, community animal health workers using filter paper were consistently the most cost-effective. The purpose built mobile team concerned only with trypanosomiasis surveillance was also consistently the most costly.

The almost total convergence of costs at high prevalences is due to the fact that at higher prevalences more than half of all costs are diagnostic costs for initial screening and parasitological examinations (ranging between US $ 3.50 and US$ 2.50 per person). At these prevalences, most individuals are sero-positive and have to be re-examined and the running cost and overheads associated with each strategy are spread over a large number of patients.

In order to further examine the relative effectiveness of the different surveillance strategies, Fig. 3 shows what proportion of the population could be sampled in a year using different approaches. This reflects the assumptions made about the possible workload that each form of active surveillance can tackle. The figures shows the situation if, in the hypothetical area with a human population of 100,000, the number of rural health centres is fixed at 10 and the number of community health workers who can assign a significant proportion of their time to active case detection for sleeping sickness is fixed at 20. Thus it is the inputs by the mobile teams which can be increased as illustrated, from spending half their time in the area to having two teams working there. Based on experience, it was also assumed that the mobile teams were able to undertake further examinations on a higher proportion of CATT+ patients than was possible under the other surveillance strategies based on community health workers. These filter paper based sampling strategies involve a delay before the results are known and individuals testing positive then have to be recalled and taken to a treatment centre for further testing. The mobile teams can undertake many of the parasitological tests themselves, and then transport the remaining individuals to a treatment centre for final confirmation of their disease status.

Fig. 3

Under these conditions, as illustrated in Fig. 3, finding and treating a majority of the patients in the population can only be done using one or more monovalent mobile teams. The other active surveillance strategies could be effective in detecting the presence of the disease or gradually eroding the size of the human reservoir, provided that the incidence is not high.

Turning from the costs per individual found with the disease, to the total investment required for finding trypanosomiasis patients, Fig. 4 illustrates these for the five different surveillance strategies for prevalences of 1% and 10%. As would be expected, the costs largely reflect the proportion of the population screened by each strategy (Fig. 3 and Table 2). Although costs would vary from country to country, and have probably increased somewhat since these figures were published in 1998, they give an idea of the orders of magnitude involved, ranging from US$50,000 to US $ 60,000 for a monovalent mobile team, to a minimal investment for fixed post/passive detection.

Fig. 4

In order to understand how the situation evolves at high prevalences better, details of the calculation for intervention by a mobile team are shown in Fig. 5. The cost of treating the patients found has now been added to produce total costs. These have been broken down into four categories. The cost of surveillance (logistics, share of salaries, etc.) is given as ‘sampling strategy’. Diagnostic tests covers both initial screening and parasitological confirmation. These costs dominate total costs at very low prevalences, once a prevalence of 1% is reached, then all other costs are dwarfed by the cost of treating patients. Given the uncertainty about how to value drug costs, these have been included at their recent commercial cost, but separated from other treatment costs (administration of drugs and hospital care). Thus the graphs can be read so as to exclude the cost of drugs. A notional figure for transport or for their economic cost could be added in subsequent analyses.

Fig. 5a

Fig. 5b

The costs of controlling the disease in the area postulated, with one monovalent mobile team screening 36% of the population in a year, thus range from just over US$ 50,000 where the prevalence is 0.1% to over US$ 4 million at a prevalence of 70%.

This analysis thus examines the ways in which the costs of controlling the human reservoir vary with sampling strategy, sampling intensity and prevalence. The spreadsheet model produced provides a basis for extending this analysis to cover other surveillance protocols, price sets, test sensitivities etc. The figures used, it will be remembered, were based on conditions in Uganda, and were extrapolated from the situation encountered there. It is likely that the costings for the extremely high prevalences need revising upwards, since in the model the labour and time requirements have not been fully adjusted to deal with the work involved in a situation where virtually all individuals test positive to the screening test and need parasitological confirmation.

To update and validate this analysis in different circumstances, what is required is

• collating and examining data from budgets and actual expenditure from a range of surveillance activities in different countries

• finding out what protocols are used in different field situations and what results are obtained, particular in terms of what sequence of tests are used to obtain parasitological confirmation and how many patients are detected by each test at different prevalences.

Economics of controlling the animal reservoir: rhodesiense form of the disease

Turning to the rhodesiense form of the disease, its more acute course means that patients usually present with symptoms shortly after infection, nevertheless, diagnosis is often slow and inaccurate. Control of this form of the disease has relied less on active surveillance, and more on vector control.

Recent research results have added a powerful and cost-effective tool to the armoury of control methods for this form of the disease. The proof that cattle have now become the main reservoir of the disease [3] in South-eastern Uganda, has as its corollary that treating cattle would control the reservoir, and if done at suitable level, would stop transmission to humans. T. b. rhodesiense is not pathogenic to cattle, however the drugs used to control it also kill T. vivax and T. congolense which are pathogenic to cattle and prevalent among the cattle population in the area. Thus treating cattle generates an economic benefit which is independent of the control of the disease in humans.

A preliminary and very approximate economic analysis of the control of rhodesiense disease in South-eastern Uganda was undertaken as part of a DFID-commissioned review of its research work [6]. Although based on extremely approximate assumptions, this highlighted the potentially very favourable situation if it were possible to control the disease by treating cattle. Drug treatments cost US$ 1.75 to 2 per dose, and it is thought that between 5% and 20% of cattle carry cattle pathogenic trypanosomes (personal communication, Paul Coleman). Based on work done on the impact of trypanosomiasis on livestock production, and current milk and cattle prices in Uganda, it was estimated that treating 1000 cattle around a disease focus could yield a benefit of between US$ 500 – US$ 3000, as compared to a cost of $1750 – 2000. Furthermore, if this expenditure on cattle treatment was successful in reducing the incidence in humans, the financial benefit in terms of sleeping sickness treatment cost avoided would probably more than justify the expenditure.

The preliminary analysis of the economics of disease control over the past decade, in southeastern Uganda, considered four categories of costs: research, vector control, medical surveillance and cattle treatment [6]. The benefits were calculated by considering three likely alternative scenarios for what the disease incidence might have been if there had been no control activities. The monetary benefits consisted of costs saved for treating patients, benefits to cattle production and patients' costs incurred while seeking treatment (see Section 2 above). The non-monetary benefits were in terms of DALY's averted. The monetary benefits produced benefit-cost ratios ranging from 1.08 to 1.98. Because the project produced a net financial gain, the cost per DALY was actually negative, ranging from -$1.50 to -$7.50. These results should be treated with caution, as they depend on very rough assumptions, however they do point to the likelihood that this control approach, where it is feasible, could be extremely cost-effective.

Research into cattle as a reservoir of T.b.rhodesiense is ongoing, and will be reported on at the meeting by those directly involved in the work. From the economic point of view, one key issue is what proportion of the cattle population will need to be treated and with what frequency, in order to generate financial benefits which cover the costs of controlling the disease in humans. Another important variable is the ratio of reported to unreported cases of the disease in humans. Obtaining this information depends on the results of epidemiological studies. The economic gains to cattle production need to be more accurately estimated, using data on the prevalence of cattle pathogenic trypanosomes and their impact on livestock productivity.

Vector Control

At this stage of the work, it has not been possible to review existing information about the costs of vector control in detail. Recent years have seen few initiatives to use vector control to deal with human trypanosomiasis. Costs are available from recent work in Uganda, where pyramidal traps were very effective in reducing tsetse density and disease incidence. The costs of the work there (personal communication R. Floto) came to ECU 781,000, excluding technical assistance, and at today's prices this would be about US $ 1.1 million.

Costs of vector control will tend to be very specific to the situation studied, varying with terrain (especially for target and spraying operations), with the type of organisation (especially for targets, traps or screens) and with the type of settlements and rural economy (affecting the availability of labour for maintaining traps and targets and the presence of cattle if these are to be treated with insecticides). Estimates of the cost of vector control operations, such as those cited below, almost invariably only include the marginal costs, that is the extra costs involved in field work, and neglect the considerable overheads, which can double or triple costs per sq km.

• Aerial spraying, using fixed wing aircraft and the sequential aerosol technique, typically involving spraying the area in a cycle of five at carefully timed intervals; apart from the sterile insect technique, this tends to be the most expensive control method, with the great advantage of achieving a very rapid reduction in the fly population. Costs are likely to be well upwards of $500 per sq km.

• Traps, screens or targets - the cost of these operations is far more variable than for aerial spraying, depending the number deployed per sq km or per linear km, and the way in which they are deployed and serviced, cost for target operations probably range around $300 – 400 per sq km, and low-cost trap or screen-based operations could cost as little as $100 – 150 per sq. km.

• Treating cattle with insecticides (not to be confused with treating cattle with trypanocides, as discussed below), where various ‘pour-on’ formulations are used, again the costs are very difficult to estimate, since the pour-on formulations also protect against tick borne diseases, and the number of cattle to be treated per sq km is also a variable as is the number of treatments per year. The cost of pour-on formulations currently ranges around $1.50 – 2.00 for an adult bovine. Insecticide treatment of cattle reduces fly density, but whether or not this would be of use in preventing human trypanosomiasis depends entirely on the local epidemiology of the disease.

A useful series of comparative costings for tsetse control methods for one country, along with details of how they were calculated can be found in Barrett (1997) for the case of Zimbabwe. Current research and information needs are for updating costings for the various strategies, and collating information on the impact that vector control operations undertaken in the past have had on the incidence of sleeping sickness. The extent to which community involvement and inputs, especially of labour, can be sustained over long periods of time also needs to be revisited.

It is difficult, especially for gambiense disease, to separate the effects of controlling the human reservoir from those of vector control, since the two are usually undertaken at the same time. The cost-effectiveness of vector control versus case-finding and treatment was modelled in Shaw (1989). At this time it appeared that case-finding and treatment was the more cost-effective at lower incidences, and when dealing exclusively with a human reservoir. The conclusion, then as now, was that epidemiological models and economic models need to be integrated in order to be of use in decision-making,

DALY's and the cost-effectiveness of sleeping sickness control

Finally, in order to examine the wider economics of controlling sleeping sickness in relation to the control of other diseases, an estimate of the cost per DALY averted is needed. So far there have been very few comprehensive attempts to calculate the actual and potential DALY's lost due to either form of sleeping sickness.

Recent work in Uganda has produced calculations of DALY's for rhodesiense [7,8]. This work will be reported on elsewhere in this workshop, and integrated into more comprehensive economic analyses. The authors point out that the age distribution of trypanosomiasis patients very closely follows that of the active adult population. This means that the disease tends to hit the most economically productive group of society hardest, so that family livelihoods and community prosperity are much affected. This data confirms observations made throughout Africa for both forms of the disease. The authors estimate and that at the time of diagnosis, patients will have been suffering from symptoms of the disease for an average of 61 days, they then require hospitalisation for an average of 34 days and that for patients correctly diagnosed and treated the case fatality rate is 5.3%. For unreported cases, the outcome is assumed to be inevitably fatal. Based on the age distribution of patients, the authors estimate that the number of DALY's for unreported and thus untreated patients is just over 20 years (personal communication Dr. Paul Coleman). This figure was used to underpin the economic analysis discussed in Section 4 above. In this case, because controlling the animal reservoir reduced the rate of tranmission to humans, and it was assumed that for every reported case there was one unreported case, the full twenty years applied to half of the cases averted. This DALY figure thus also means that the cost-effectiveness of rhodesiense sleeping sickness control compares very favourably with those of other high priority health control activities, such as malaria (Goodman et al., 2000) EPI, HIV.

For gambiense, no published DALY figures based on detailed field records were encountered, however ongoing work for Southern Sudan indicates that a similar situation probably exists (personal communication Dr. Anne Moore). An attempt to look at the economics of alternative treatments for second stage gambiense patients used the age-at-death distribution calculated for rhodesiense patients in Uganda [9]. This article also concluded that the standard treatment for second stage patients represented a very attractive cost per DALY averted, ranking with the most cost-effective interventions such as childhood immunisation and blood-screening for HIV.

Figure 6 above takes this discussion a bit further by modelling how the situation could be analysed if more data were available. It takes the cost of treatment and detection of patients using a mobile team at different prevalences, as shown in Figure 2, and divides it by a conservative estimate of the number of DALY's averted. The ‘zero’ baseline figure, which gives the highest cost is based on each patient treated, that is premature death prevented, representing 15 DALY's averted. This figure was selected as a conservative estimate, taking into account the long asymptomatic period for gambiense disease and thus calibrated to be rather lower than the figure for rhodesiense. On this estimate, once the prevalence has reached 1%, the cost per person found and treated falls to US$ 330, and thus the cost per DALY averted falls below the threshold of US$ 25, defined by WHO as ‘very attractive’. At higher prevalences, the cost per DALY averted stabilises at between US$ 10 and US$ 12.

Fig. 6

In order to add another dimension to the analysis, three further lines have been drawn showing the effect if the number of DALY's averted per patient found and treated were increased by 0.5, 1 or 1.5. This analysis needs to be developed once more data has been analysed or collected showing:

• what the actual figure for DALY's averted per patient treated is likely to be,

• how the prevalence evolves from year to year in gambiense foci.

The latter brings the discussion back to epidemiology, since, at low prevalences, this multiplier could be seen as a measure of Ro, the basic reproductive rate of the disease.

There is thus a clear agenda for analysing existing data on the progress of epidemics and for collecting new data, in order to add to the knowledge of the epidemiology of the disease. The low proportion of all cases which are actually recorded has meant that knowledge of the year to year changes in prevalences is often patchy or anecdotal, nevertheless, for foci which have been the subject of more intensive control work over a number of years data does exist. In this context, epidemiological models [10] have an important role to play.

Conclusions

This paper has tried to cover the main issues involved in looking at the economics of controlling both gambiense and rhodesiense sleeping sickness. Some aspects, notably vector control, have only been superficially treated. Nevertheless it is hoped that it provides a sufficient basis for discussion on what the priorities for research and information gathering designed to help planning funding and resource allocation and gauge what returns to expect from these investments.

Returning to the information requirements that were noted at the end of each section, these all fall into one of two categories: economic and epidemiological. Some of the issues and needs identified have been included in the table below, which provides a simple framework for categorising the information required.

From the point of view of those allocating funds within the health sector, the recent emergence of DALY calculations for sleeping sickness has made it very clear that control of this disease represents an extremely cost-effective investment. This is linked to two factors. The first is the inevitably fatal outcome of the disease, which has been discussed above. The second is the focal nature of the disease, which means that although the population at risk is large, the disease is nevertheless location specific, so that control operations can target circumscribed geographical areas where the disease is known to be present.

Comments