Cost-effectiveness analysis of malaria chemoprophylaxis for travellers to West-Africa

This study is a cost-effectiveness analysis based on malaria cases imported from West Africa to Switzerland and was done from the perspective of the Swiss health system. Our work did not require ethics committee clearance because we used available published data on malaria risk and malaria associated costs. No human subjects were recruited for the analysis. West Africa was defined by OECD-definition as Benin, Burkina Faso, Cameroon, Chad, Ivory Cost, Ghana, Guinea, Guinea-Bissau, Gambia, Mali, Mauritania, Niger, Nigeria, Liberia, Senegal, Sierra Leone, Togo [10]. Cape Verde was excluded because it is not a malaria risk area.

A decision tree model (as shown in Figure 1) was developed using TreeAge Pro 2009 (TreeAge Software, Williamstown WA). We estimated the impact of an 80% reimbursement for the cheapest malaria chemoprophylaxis regimen or a voucher for the equivalent sum of money ("80% reimbursement strategy" label in tables). In the model we compared that strategy to the situation of no reimbursement, which is the current situation in Switzerland ("current situation of no reimbursement" label in tables). We calculated the cost for malaria chemoprophylaxis for a two week stay in West Africa (Table 1). So, in the proposed 80% reimbursement strategy every traveller who is prescribed malaria chemoprophylaxis would get a discount of € 17.77 (i.e. 80% of € 22.21 for mefloquine as Mephaquin^®).

Travellers were divided in two theoretical groups: 1) those who didn't take any chemoprophylaxis and 2) those who used chemoprophylaxis. The chemoprophylaxis users were divided between the three recommended regimens mefloquine, atovaquone/proguanil and doxycycline. The decision tree takes adverse events into account. The traveller may experience adverse events (side effects) leading to withdrawal, or may experience adverse events without withdrawal or may not experience adverse events from the malaria chemoprophylaxis. At the end, the probability of contracting malaria is reduced (or not) by the prophylactic effectiveness of the chosen malaria chemoprophylaxis regimen. The probability of contracting malaria for withdrawals was estimated to be equivalent to those travellers who didn't use malaria chemoprophylaxis.

The 80% reimbursement strategy presents the same decision tree as the current situation of no reimbursement but differs on two points: the probability of recourse to malaria chemoprophylaxis and the respective cost of 80% of the price for mefloquine.

The value to compare the two situations was determined by calculating the cost and the health outcomes. The main outcome measure is given as incremental cost-effectiveness ratio (ICER) in terms of incremental cost per averted malaria case. Other outcome measures included the number of malaria cases, deaths and averted malaria cases for the reimbursement strategy, the average cost of malaria and associated chemoprophylaxis and the probability of contracting malaria per journey to West Africa.

The estimates of variables used in the model were probabilities and medical costs stemming from a review of the literature and are summarized in Table 2.

The costs are given in Euro (€) and were converted Swiss Francs (CHF) to Euro (€) in 2009: 1 Swiss Franc (CHF) = 0.660 Euro (€) = 0.575 British Pound (£) = 0.936 US Dollar ($).

The results are presented as incremental cost effectiveness ratio (ICER) in terms of cost per averted malaria case using an 80% reimbursement of the cheapest malaria chemoprophylaxis compared with the current situation where there is no reimbursement for malaria prophylaxis. ICER is calculated by dividing the incremental cost of the reimbursement strategy by the savings gained in the malaria cases. Using data from the Swiss Federal Statistical Office (SFSO) we estimated the annual number of Swiss travellers to West Africa to be 70'000 [23]. The estimation on the annual cost and effects (number of malaria cases) was conducted with the given number of travellers from Switzerland to West Africa. Costs and benefits were not discounted given the evaluation of malaria chemoprophylactic prevention and treatment would occur within a year.

To test for the precision and robustness of the model-results, deterministic sensitivity analyses were performed. For the one way sensitivity analysis, all model parameters were assigned a uniform probability distribution where possible, allowing ± 30% variability around the baseline. The variables on efficacy and probability of drop out, adverse events and no adverse events of malaria chemoprophylactic regimens were varied in their confidence interval. Furthermore, uncertainties around the result of the base-case were assessed in a probabilistic sensitivity analysis by using 10,000 sets of parameter values which are randomly sampled from a beta-distribution reflecting the ranges of variations given in Table 2 [24].

The results are shown in Table 3 for the situation of recourse to malaria chemoprophylaxis of 55%, 68.7% and 82.4%. The absolute results calculated for 70'000 travellers to West Africa are presented in Table 4. The 80% reimbursement strategy leads in all situations to a significantly lower probability of contracting malaria per traveller and also to significantly lower costs per traveller in the situation of 55% and 68.7% recourse than in the current situation of no reimbursement. Only in the 82.4% recourse situation the cost would be higher in the 80% reimbursement strategy. Assuming current use of malaria chemoprophylaxis at 55% the reduction in malaria cases is 47% with an 22% decrease of cost (€ 65.45 to € 50.94per traveller); for a 68.7% recourse the reduction of malaria cases is 45% with a 9% reduction of cost (€ 50.05 to € 45.36per traveller); for a 82.4% recourse the reduction of malaria cases is 38% with 9% higher costs (€ 36.64 to € 39.78 per traveller).

In a scenario with 68.7% usage of chemoprophylaxis, the net cost for the public health system would be € 328'300 and a reduction of 278 malaria cases (from 624 to346), if the 80% reimbursement of the cheapest malaria chemoprophylaxis would be introduced (Table 4).

The main medical effectiveness of the model is given in the number of malaria cases prevented and secondarily by the number of malaria related deaths prevented. The results are summarized in Table 3 and 4.

The 80% reimbursement strategy is dominant (i.e. cost saving and more effective than the current situation) in both situations of 55% and 68.7% recourse to malaria chemoprophylaxis. For the 82.4% recourse situation the incremental cost per additional malaria case prevented is € 2'302 and for prevention of one malaria related death € 191'833.

These results underline the high influence of the current situation on the recourse to malaria chemoprophylaxis in our model.

The one-way sensitivity analysis on the cost of the 80% reimbursement strategy shows the highest sensitivity for the following variables: probability of recourse to malaria chemoprophylaxis, probability of contracting malaria without malaria chemoprophylaxis, cost of mefloquine regimen, the decrease in the number of travellers without malaria chemoprophylaxis in the reimbursement strategy. All other variables have a minor effect on the incremental cost effectiveness ratio. The effects on the incremental cost effectiveness ratio per averted malaria case are presented in Table 5 for these parameters and the ICER of reimbursing atovaquone/proguanil or doxycycline instead of mefloquine are shown.

The probabilistic sensitivity analysis (Figure 2) resulted in a 95% probability of being dominant in the situation of 68.7% recourse to malaria chemoprophylaxis based on data from Table 2.

This study gives a good overview about imported malaria from a high risk area in the form of a model with the all important variables in it. We tried to include as many direct factors as possible in the model while also keeping it as simple as possible. It has to be remembered that we conducted a cost effectiveness analysis with the comparison between the current situation of no reimbursement and a strategy of 80% reimbursement of the cheapest malaria chemoprophylaxis. The model is robust and changes in input data such as underestimations of cost of treating malaria cases or hospitalization rate will not unduly influence the outcome. The model can also easily be adapted to more precise input factors or to other population group with a different situation on imported malaria.

A weakness of the analysis is the lack of exact data on the input factor 'use of or recourse to malaria chemoprophylaxis' and this factor has a major impact on the ICER (incremental cost effectiveness ratio). There were no specific data on chemoprophylaxis use available for Swiss residents. So we used an estimation based on European residents derived from the data of the airport study which was conducted in nine different airports in Europe with passengers residing in Europe and boarding a flight to West Africa [12]. This information is, however, not representative, because targeted destinations and flights were pre-selected for practical reasons, the study participation was on voluntarily basis and unprepared travellers were probably less likely to participate. This meant a bias in favour of those using chemoprophylaxis and a resulting overestimation of recourse to malaria chemoprophylaxis. The three studies from France, which Pistone et al. analyzed for travellers to West Africa, might underestimate the recourse to malaria chemoprophylaxis for Swiss travellers [13‐15]. The proportion of immigrants from West Africa living in France is higher than in Switzerland. Several studies have shown that visiting friends and relatives (VFR) are at increased risk of acquiring malaria [5, 7]. One of the reasons is the lower recourse to malaria chemoprophylaxis.

The incidence of malaria in travellers, who stopped taking malaria chemoprophylaxis (withdrawals), was assumed to be the same as travellers without malaria chemoprophylaxis. But they could still have a limited protection from the taken malaria chemoprophylaxis and it would overestimate the incidence of malaria.

In our model we considered only the three recommended chemoprophylactic agents for West Africa from the Swiss Federal Office of Public Health (FOPH): mefloquine, atovaquone/proguanil and doxycyline. Others, like chloroquine, which has a partial prophylactic effectiveness, were excluded from the model. This could lead to an overestimation of the incidence of malaria cases.

The increase in malaria chemoprophylaxis users for our 80% reimbursement strategy was assumed based on the study of Pistone et al. [11]. We made a conservative assumption and took the same factor (57%) for the increase of malaria chemoprophylaxis-users as Pistone et al. in their 65% reimbursement strategy. This would possibly underestimate the ICER. But it has to be remembered that travellers seeking travel advice in a health clinic still have to pay for the visit by themselves.

The number of travellers to West Africa was estimated using air transportation statistics, but it included only travellers with a flight ticket where the final destination was an airport in West Africa. Travellers flying to a destination in West Africa via secondary routes are not registered. This fact has no influence on the incremental cost and effectiveness as well as the ICER, but they have a main impact on the absolute number of malaria cases, malaria related deaths and their associated cost for the Swiss health system. This makes the validation of our model more difficult. The results of the model with the estimated annual number of travellers of 70'000 overestimated in all three situation of recourse to malaria chemoprophylaxis the number of malaria cases reported at the Swiss Federal Office of Public Health (FOPH). The origin of the malaria acquisition is not known for every reported malaria case and may underestimate the true number of malaria cases from West Africa. Furthermore, the model does not include malaria cases that are treated outside of Switzerland.

The 80% reimbursement strategy of only the cheapest malaria chemoprophylaxis (mefloquine) makes the proportional prices of the three recommended malaria chemoprophylaxis agents very diverse: for a two week stay the price reduction for mefloquine would be 80%, for doxycyline 59% and for atovaquone/proguanil 21%. These percentages change with chosen brand names and travel duration. In Switzerland, broken packs are not dispensed in this situation. This idea of giving a voucher for malaria chemoprophylactic medication equivalent to 80% of the cost of the cheapest anti-malaria regimen is a new and an innovative approach to increase the recourse to prophylactic drugs. Mefloquine has the advantage of wide applicability (including small children, pregnant women, long-term travellers) and is comparable with other malaria chemoprophylaxis in its prophylactic effectiveness and frequency of severe adverse events but is associated with a high incidence of non-serious adverse events, particularly in women, that may impact on well being [9]. Contraindications for mefloquine are: active depression, a recent history of depression, generalized anxiety disorder, psychosis, or schizophrenia or other major psychiatric disorders, or with a history of convulsions. The similar prophylactic efficacy of the three regimens, differing mainly in their contraindications, makes this 80% reimbursement of the cheapest malaria chemoprophylactic agents a reasonable proposition. The realization of the 80% reimbursement strategy of the cheapest malaria chemoprophylaxis, mefloquine, for a journey to West Africa could be made by delivering a voucher for the given travel duration approved by a valid individual ticket for a destination in West Africa.

In the literature, one other cost-effectiveness analysis has been conducted from the perspective of the French health system for travellers to Sub-Saharan Africa and the results showed that the incremental cost of their 65% reimbursement strategy has a positive cost-effectiveness impact [11]. Two older studies analyzed cost minimization in use of malaria chemoprophylaxis in travellers and concluded that malaria chemoprophylaxis is cost-effective from the perspective from the British health care system and also from the German and Swiss systems respectively [25, 26].

In most economic analyses, where two strategies or more are compared, the incremental cost effectiveness ratio, ICER, is given in life years gained or QALY (quality adjusted life years). This study used averted malaria cases as ICER and not QALYs. The National Institute for Health and Clinical Excellence (NICE), is a specific health authority of the National Health Service (NHS) in Britain which gives guidelines for appropriate treatment regimens for different diseases. NICE considers a treatment intervention with a threshold of € 22'962 to € 34'443 (£20'00 to £30'000) per QALY to be cost effective. If our study results were expressed in terms of QALYs, the theoretical calculations are as follow:

The average age at death due to malaria in Switzerland is 51.3 years [4]. The life expectancy in Switzerland is around 83 years [27]. That would lead to 31.7 life years lost. If we assumed that the average quality of these 31.7 life years is 0.8 it would be 25 QALY. The costs of the 80% reimbursement for one malaria related death averted is € 7673 (€ 191'833 divided by 25 QALY), using the study study assumption that 82.4% of travellers to West Africa use chemoprophylaxis, The 80% reimbursement is considered cost effective according to the NICE guidelines.

This study attempts to provide economic data on the costs of malaria prevention and the expenses associated with imported malaria in industrialized countries. This analysis is based on traveller data. Travellers to West Africa are not homogeneous but comprise tourist, business travellers and increasing numbers of "visiting friends and relative" travellers. Currently in Switzerland the main burden of malaria is borne by this last group and future research can possibility focus on this group in economic evaluations as some studies have shown that chemoprophylaxis use in this group falls far short of the 55% recourse to chemoprophylaxis assumed in this analysis [7]. Other countries with large migrant populations can also consider such economic evaluations for malaria and travel-related illness.