Cost-effectiveness analysis of rapid diagnostic tests for G6PD deficiency in patients with Plasmodium vivax malaria in the Brazilian Amazon

The use of primaquine (PQ) for radical treatment of Plasmodium vivax in carriers of G6PD deficiency (G6PDd) constitutes the main factor associated with severe haemolysis, hospitalization and death in G6PDd individuals in Latin America, especially in males; cases of severe adverse events in females induced by PQ, have not been reported in the Brazilian Amazon [1, 2]. In Brazil, PQ is administered to prevent relapses of P. vivax malaria and to reduce its transmission [3]. However, this radical cure treatment of P. vivax malaria with PQ represents a risk for G6PDd carriers, since the Brazilian malaria national control programme does not recommend a routine G6PDd diagnosis before treatment starts. More than 99 % of the 178,600 new Brazilian cases of malaria in 2013 were reported in the Amazon, of which 81 % were caused by P. vivax [4, 5]. The estimates are approximately 6000 hospitalizations per year for individuals with G6PDd as a result of using PQ, creating a substantial burden, of an estimated costing of US$ 4,858,108.87 per year, for the Brazilian Public Health Service (Sistema Único de Saúde/SUS) [6]. In this context, detecting G6PDd before PQ is prescribed could contribute to improving treatment safety in people with G6PDd, preventing harm done to the efforts to control and eliminate malaria. In Brazilian Amazon, the prevalence of G6PDd in males was estimated at 4.5 % [7]. The identification and implementation of a rapid, efficient and feasible diagnostic method in field conditions is a priority [ 8‐11].

The gold standard G6PDd diagnostic test is a quantitative spectrophotometric assay that requires laboratory setting, complex equipment and specialized staff. In less sophisticated settings, the fluorescent spot test has been widely recommended because it is easy to execute. However, this test also requires specialized equipment, such as an ultraviolet bulb, a water bath and a micropipette. Thus, neither of these two tests can be considered appropriate for field conditions in most of the malarial areas in the Amazon region [8, 12‐14], and they are not used in routine services.

Two RDTs have been evaluated to detect G6PDd in field conditions under different situations: BinaxNOW^® G6PD (BX-G6PD) and CareStart™ G6PD (CS-G6PD). Both are qualitative chromatographic tests that are easily read visually, require minimal training and results can be obtained within 10 min [12, 15, 16].

BX-G6PD presents operational limitations related to temperature, as it provides reliable results only when performed at 18–25 °C range, and the sample required is venous blood [15, 17]. A recently reported multicentre study that included the city of Manaus found a sensitivity of 54.6 % and a specificity of 100 % for this test in malaria patients using a cut-off value corresponding to 60 % of normal activity [16]. The authors mentioned unpublished data for a cut-off of 40 % of normal activity the sensitivity and specificity were, respectively, 66.6 and 100 %.

CS-G6PD requires capillary blood samples collected from sterile finger pricks and results are not affected by high temperatures during its conservation and execution [12]. When this test was used for patients with malaria in Manaus, sensitivity was 45.45 % and specificity 98.71 %, for a cut-off value of 40 % of normal enzyme activity (Brito et al, Personal communication, 2015). In the general population of villages in Cambodia, this cut-off value the test showed a sensitivity and specificity of 93.3 and 97.4 % respectively [18].

The absence of cost-effectiveness analysis comparing the use of RDTs for G6PDd diagnosis with the status quo, where testing before PQ prescription is not performed, is an important gap that makes it difficult to recommend these tests [13, 14, 19]. The absence of a G6PDd diagnosis also represents an obstacle for the future use of tafenoquine, a new promising 8-aminoquinoline that has shown high efficacy in radical cure of P. vivax malaria when administered as a single dose [6, 19, 20]. Therefore, the present cost-effectiveness study was carried out with the objective of estimating the incremental cost-effectiveness ratio (ICER) of introducing a RDT to detect G6PDd in male patients carrying P. vivax in the Brazilian Amazon, in comparison with the routine programme (RP) of the Brazil’s Malaria Control Programme.

A cost-effectiveness analysis (CEA) was done from the perspective of the Brazilian Public Health Service (SUS) for 2013. The study area covers the Brazilian Amazon, composed of 775 municipalities distributed in nine States (Acre, Amapá, Amazonas, Maranhão, Mato Grosso, Tocantins, Rondônia, Roraima and Pará), where the population is approximately 21 million, of whom 10 million live in 34 municipalities with more than 100,000 inhabitants [21]. In the region, 427 municipalities registered cases of P. vivax malaria in 2013 [4].

The research project was submitted to the Ethics Committee of the Faculty of Health Sciences at the University of Brasilia and approved on May 14th 2013 (case number 276,522).

The strategy based on the use of CS-G6PD in 100 % of the population was more cost-effective when compared to RP and the strategy that combines BX-G6PD and RP (BX-G6PD strategy). The BX-G6PD strategy was extendedly dominated (the ICER of BX-G6PD compared with RP was higher than the ICER of CS-G6PD compared with RP) and the strategy CS-G6PD was the most cost-effective for the outcome “adequately diagnosed case”. For “hospitalization avoided”, the RP and BX-G6PD strategies were dominated by the CS-G6PD strategy (Table 3).

The univariate sensitivity analyses for the outcome “adequately diagnosed case” showed that variations in the epidemiological and cost parameters did not have an important impact on the ICERs, whose greatest variation was only US$ 0.45. For “hospitalization avoided” the CS-G6PD strategy remained dominant in relation to the BX-G6PD and RP strategies, as evidenced by negative ICERs, supporting a greater effectiveness and lower cost for the CS-G6PD strategy. Exceptions occurred only in the sensitivity analysis of the variables sensitivity of CS-G6PD and prevalence of G6PDd, with positive ICERs (Fig. 2).

Two acceptability curves were generated for the PSA (Fig. 3), for the outcomes “adequately diagnosed case” and “hospitalization avoided” respectively. For both curves, the strategy based on CS-G6PD presented a high probability of being cost-effective, at 10 % of the limit considered in the analysis.

Additionally, ICERs were estimated considering the highest (18.6 %) and the lowest (4.3 %) proportion of malaria in individuals submitted to the thick blood smear test among the States evaluated. The results demonstrate that the CS-G6PD strategy continues to be the most cost-effective one for appropriate diagnosis of G6PDd, presenting for both proportions the same ICER as that identified in the base-case when compared to the RP. For the outcome “hospitalization avoided”, the CS-G6PD strategy continued to dominate the others. The second analysis of additional sensitivity simulated the amplification of the use of BX-G6PD for the municipalities with more than 50,000 inhabitants (where 40.9 % of the cases of P. vivax malaria in men occurred). Again, the results did not cause significant modifications in the ICERs, and the CS-G6PD strategy continued to be the most cost-effective for the outcome “adequately diagnosed case” and was dominant for “hospitalization avoided”.

Although malaria continues to be one of the main public health problems in the world, there have been some successful results from control efforts in recent years, leading some endemic regions to consider the possibility of malaria eliminating [24]. Tatem et al. showed that the Americas present great potential for elimination of Plasmodium falciparum, as well as operational conditions for controlling P. vivax. In this context, for the regional efforts directed to eliminating malaria to be successful, the strategies directed to combating P. vivax have taken on an even greater importance [9, 25].

The elimination of P. vivax malaria will only be possible with the effective use of an 8-aminoquinoline for the radical treatment of latent P. vivax infections [13, 19]. However, as the use of this type of anti-malarials threatens the safety of G6PDd individuals, previous diagnosis of this condition in patients with P. vivax malaria needs discussion [13, 19], and it is vital that cost-effective diagnostic strategies are identified [1, 6, 11].

The RDTs considered in the study have different accuracies and integrate different strategies. The cost associated with performing each test (training and salary of health workers, test kits and supplies) was also different: US$ 9.81 for BX-G6PD and US$ 4.14 for CS-G6PD.

A hypothetical 2013 cohort of 1,188,523 males presenting with fever and requesting diagnosis for malaria would be administered 87,928 RDT CS-G6PD in males with P. vivax (RDT in 100 % of the municipalities) or 16,179 RDT BX-G6PD in males with P. vivax, based on combined use with RP (RDT in 18.4 % of the municipalities), which could prevent, respectively, 1808 hospitalizations (a cost of US$ 511,522) and 488 hospitalizations (US$ 137,974).

The analysis assumed the values of sensitivity and specificity obtained in studies carried out with patients diagnosed with malaria (Table 2). These estimate a reduction in the accuracy of the RDTs evaluated when compared to the accuracy estimated in studies with the general population. This may occur due to the greater number of new erythrocytes produced as a result of the acute haemolysis provoked by malaria, since new erythrocytes present more G6PD enzymatic activity when compared to mature cells, which may lead to false negative results [16, 26].

A recent study that evaluated the new generation of CS-G6PD demonstrated an improvement in its sensitivity for the general population in Cambodia [18], which could also occur in individuals with malaria, making the strategy even more cost-effective. Corroborating these promising results, another study in an area with high incidence of malaria in Africa recently estimated 100 % sensitivity in the general male population [27].

When the uncertainties concerning epidemiological and cost parameters were tested by means of the univariate analysis of sensitivity, the performance of the strategy based on CS-G6PD was not affected in an important way. The lowest variations in the sensitivity parameter of CS-G6PD and the prevalence parameter of G6PDd altered the dominance relationship of the CS-G6PD strategy for the outcome “hospitalization avoided”, but the strategy remained cost-effective.

The acceptability curves indicate that the probability of the strategy based on CS-G6PD being cost-effective is of approximately 100 %, for both outcomes, at 10 % of the limit of willingness to pay. For the second outcome, when the willingness to pay is US$ 0.00, the probability is over 90 %, demonstrating that the stochastic and simultaneous variation of the parameters for epidemiology and costs used indicates a high probability of the CS-G6PD strategy being cost-effective, despite the uncertainties inherent in the analytical model.

Decision-makers are being faced with increasingly difficult decisions when they consider the addition of new technologies to national malaria control program [28]. Aiming to help in decision-making, various economic studies have examined aspects related to primary prevention, diagnosis and treatment of malaria, with emphasis on vaccines [29, 30], mosquito nets impregnated with insecticides [31, 32], mass screening followed by treatment [33], use of RDTs to diagnose the infection [34, 35] and the use of schemes for chemo-prophylaxis and treatment [36‐38]. Although there are no published studies on cost-effectiveness for G6PDd diagnosis in connection with malaria, recent publications have discussed the urgency of evaluating the efficiency of RDTs used to detect G6PDd so as to guarantee safety in the treatment of these patients [11, 13, 14].

This study presents some limitations, such as the fact that some epidemiological and cost parameters arise from one Amazon state arguing that results can be extrapolated to Brazilian Amazon. However, in the analysis of sensitivity, which includes plausible variations for the region studied, there were no important changes in the ICERs. The prevalence of G6PDd, the costs of hospital food and the salaries of the professionals involved in the RDT were generalized from data on the State of Amazonas, and the proportions of hospitalizations (94.29 %) were calculated from a cohort of G6PDd carriers who were monitored at FMT-HVD [39]. This represents the situation at a center of excellence, differing from the context in small towns of the Brazilian Amazon. Thus, the proportion of hospitalizations may have been overestimated, since in these places there is limited access to health services with moderate to high technological density [6].

Within this scope, the present study estimated a cost-effectiveness ratio that is favorable to the use of RDTs for G6PDd in the Brazilian Amazon, in particular CS-G6PD. Decision-makers in the National Programme of Prevention and Control of Malaria in the region should consider implementation of such RDT strategies for G6PDd diagnosis. The incorporation of RDTs will make the use of 8-aminoquinolines safer and may represent an increase in the effectiveness of P. vivax malaria control in Brazil.

The strategy based on CS-G6PD is cost-effective in diagnosing G6PDd and avoiding hospitalization. These results may help in future decision-making related to the use of CS-G6PD in the Brazilian Amazon when considering a safer P. vivax malaria treatment in the region for G6PDd individuals.