Abstract
Background
Inappropriate antibiotic use increases selective pressure, contributing to antimicrobial resistance. Point-of-care rapid diagnostic tests (RDTs) would be instrumental to better target antibiotic prescriptions, but widespread implementation of diagnostics for improved management of febrile illnesses is limited.
Objective
Our study aims to contribute to evidence-based guidance to inform policymakers on investment decisions regarding interventions that foster more appropriate antibiotic prescriptions, as well as to address the evidence gap on the potential clinical and economic impact of RDTs on antibiotic prescription.
Methods
A country-based cost-effectiveness model was developed for Burkina Faso, Ghana and Uganda. The decision tree model simulated seven test strategies for patients with febrile illness to assess the effect of different RDT combinations on antibiotic prescription rate (APR), costs and clinical outcomes. The incremental cost-effectiveness ratio (ICER) was expressed as the incremental cost per percentage point (ppt) reduction in APR.
Results
For Burkina Faso and Uganda, testing all patients with a malaria RDT was dominant compared to standard-of-care (SoC) (which included malaria testing). Expanding the test panel with a C-reactive protein (CRP) test resulted in an ICER of $ 0.03 and $ 0.08 per ppt reduction in APR for Burkina Faso and Uganda, respectively. For Ghana, the pairwise comparison with SoC—including malaria and complete blood count testing—indicates that both testing with malaria RDT only and malaria RDT + CRP are dominant.
Conclusion
The use of RDTs for patients with febrile illness could effectively reduce APR at minimal additional costs, provided diagnostic algorithms are adhered to. Complementing SoC with CRP testing may increase clinicians’ confidence in prescribing decisions and is a favourable strategy.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
Complementing Standard-of-Care with C-reactive protein testing for patients with febrile illness in Africa could effectively reduce the antibiotic prescription rate (APR) at minimal additional costs. |
The incremental cost per consultation for 1 percentage point reduction in APR provides an insightful alternative for the standard health-economic outcomes such as cost per quality-adjusted life-year (QALY) or per disability-adjusted life-year (DALY). |
1 Introduction
Antimicrobial resistance (AMR) is a major global health threat, which increasingly undermines our ability to effectively control and treat infectious diseases [1]. The consequences of AMR can be severe, as it may increase the frequency, duration, morbidity and mortality rates of infectious diseases [1,2,3]. If AMR is allowed to spread unchecked, it will be detrimental in health and economic terms for societies worldwide [4], but especially for low- and middle-income countries (LMICs), where an additional 28 million individuals may be forced into extreme poverty due to increased out-of-pocket spending on life-saving treatment associated with AMR [5, 6]. Over-prescription by clinicians and the availability of over-the-counter (OTC) antibiotics are important causes of inappropriate antibiotic consumption, thereby enhancing the selective pressure on microorganisms, which increases the growth of resistant microorganisms (i.e., selective pressure) that is contributing to AMR [2, 4, 7]. Hence, antibiotic stewardship—use of the right drug, at the right dosage, for the right indication—is considered a critical measure to rationalize antibiotic prescription, reduce unnecessary consumption, and contain AMR [8].
The availability and use of affordable, appropriate diagnostics, especially point-of-care (POC) rapid diagnostic tests (RDTs), would be instrumental to better target antibiotic prescriptions and inform patients on the cause of their symptoms, which potentially could reduce over-the-counter antibiotic use [4, 9, 10]. However, in practice, widespread implementation of diagnostics for improved infectious disease management is limited for various reasons. Firstly, only a few POC RDTs are tailored to LMICs—for example, in terms of prevalent pathogens and suitability to tropical conditions—and affordable [11, 12]. Secondly, available POC RDTs often have suboptimal specificity and sensitivity [13]. Finally, there is a lack of evidence on the clinical impact and cost-effectiveness of RDTs, especially in relation to the long-term effects of AMR, which extends beyond the current-day patients’ quality of life [12, 14, 15]. Generic outcome measures such as the quality-adjusted life-year (QALY) or disability-adjusted life-year (DALY) are not consistently reported in health economic research papers on diagnostics for infectious diseases [14] and the valuation of health outcomes (i.e., QALYs or DALYs) extending beyond the cohort receiving treatment is challenging with prevailing methodologies employed in health economic analyses [16].
There is an urgent need for an improved understanding of the full cost and effects resulting from the implementation of diagnostic strategies that can inform future research and investment decisions by policymakers. The AMR Diagnostics Use Accelerator trial aimed to provide evidence on the usefulness of adopting a package of social and clinical interventions—consisting of POC RDTs, a clinical diagnostic algorithm for physicians, and targeted information for patients—on the clinical outcomes of patients presenting with acute fever to outpatient clinics in LMICs, as well as on antibiotic prescription [17]. The current study uses the results of the trial [18] to conduct a cost-effectiveness analysis to quantify the expected cost, clinical outcomes, and impact on antibiotic prescription and consumption resulting from the implementation of different (hypothetical) test strategies, compared to the current standard-of-care (SoC). Our study aims to contribute to evidence-based guidance of policymakers to inform investment decisions on interventions that foster more appropriate antibiotic prescriptions as well as to address the evidence gap on the potential clinical and economic impact of RDTs on antibiotic prescription.
2 Materials and Methods
2.1 Accelerator Trial
Results from the Accelerator trial (NCT04081051), which have been published separately [18,19,20,21], were used to develop the country-based cost-effectiveness models applied to the three African countries where the trial took place—Burkina Faso, Ghana and Uganda. The intervention in the trial consisted of country-specific diagnostic algorithms, which were used by healthcare workers to determine whether to prescribe an antibiotic or not and are explained in detail elsewhere [19,20,21]. In short, the diagnostic algorithm included pathogen-specific and non-pathogen-specific POC RDTs including group A Streptococcus test, influenza test, malaria RDT, respiratory syncytial virus test, Streptococcus pneumoniae test, typhoid fever test, C-reactive protein (CRP) test, urine dipstick test and complete blood count (CBC) (see Fig. 1). The dengue test was added to the panel for Burkina Faso. Furthermore, communication messages about adherence to prescriptions were communicated by healthcare workers to study participants in the intervention arm. Clinical outcomes and antibiotic prescriptions were compared between the intervention arm and the control arm that followed diagnostic and treatment procedures routinely used in each country. As part of the trial, participants in both the intervention and control arms were followed for 7 days after the first visit to assess their health status and adherence to antibiotic prescription. For the analysis, the trial population was stratified in subcategories based on age and presenting signs and symptoms being respiratory or non-respiratory (see Online Supplementary Material (OSM) Appendix Table A1).
Fever clinical diagnostic algorithm, from Olliaro et al. [13]. CBC complete blood count, CRP C-reactive protein, GAS group A Streptococcus, RSV respiratory syncytial virus, WBC white blood cell count, WHO World Health Organization. *Diagnostic panel depending on local endemicity. †Choice of tests at the discretion of local health practitioners. ‡Unless a concomitant bacterial pathogen identified. §Start treatment followed by culture if needed. ¶And neutrophils > 75% if WBC ≥ 11.000 and/or neutrophils >75% if WBC < 11.000
2.2 Test Strategies
The model simulated seven different test strategies. The first two strategies include the “standard-of-care” (SoC) and “trial-based” strategies, which were modelled alongside the Accelerator trial, based on the outcomes of the control and intervention arm, respectively. Five additional hypothetical test strategies were modelled to assess the maximum effect of different RDT combinations (pathogen-specific and non-pathogen-specific) on the antibiotic prescription rate, costs and clinical outcomes. The five hypothetical strategies were kept similar across the three countries, and include (i) malaria RDT only (country policy in line with World Health Organization (WHO) recommendations [22]); (ii) malaria RDT followed by a CRP test for malaria-negative patients; (iii) malaria RDT followed by a complete blood count (CBC) test for malaria-negative patients; (iv) malaria RDT followed by a CBC and CRP test for malaria-negative patients; (v) malaria RDT followed by the full panel used in the intervention arm for malaria-negative patients. The hypothetical strategies (i–iv) were defined based on clinical relevance as indicated by the principle investigators of the Accelerator trial in the three countries, feasibility for implementation in clinical practice considering the cost of the individual RDTs, the number of RDTs that could be performed within a limited timeframe considering that one malaria RDT (+ CBC for Ghana) constitutes current SoC, and trial data availability per RDT (e.g., a substantial proportion of RDTs were conducted for a limited number of patients, which constrains generalisability). Strategy (v) was selected to assess the hypothetical effect of the trial intervention.
2.3 Model Structure
A decision tree model was developed using R statistical software version 4.2.2 [23]. The general model structure was the same for all compared strategies (Fig. 2). The hypothetical cohort of 1,000 patients entering the model was differentiated clinically based on the presenting signs and symptoms being respiratory or non-respiratory. Subsequently, the result of the malaria RDT informed whether patients received treatment for malaria or continued for additional diagnostic tests and further clinical judgement. Based on additional diagnostic testing—altering per strategy—and clinical judgement, patients were prescribed treatment with or without antibiotics. If no additional RDTs were done, the decision was solely based on clinical judgement. Further branching was the result of the patient adhering to the treatment prescribed. In case antibiotics were prescribed, adherence was defined as completing the treatment as prescribed. For patients who were not prescribed antibiotics, adherence was defined as not using antibiotics, as was recommended by the clinician. Each branch ended with a favourable outcome (i.e., without fever and feeling better at day 7) or unfavourable outcome (i.e., with fever and/or feeling the same or worse at day 7), which were primary endpoints of the trial [13]. The unfavourable outcome was further split into an unfavourable outcome with or without a serious adverse event (SAE). An SAE was defined as any event resulting in a hospital visit.
The test panel in scope of each strategy informed the chance nodes of malaria RDT (positive or negative) and antibiotics prescription (True of False). For the “SoC” strategy, the decision nodes for malaria diagnosis and antibiotic prescription were informed by the RDT results and clinical judgements, or clinical judgement only in case no RDTs were performed. For the “trial-based” strategy, the chance nodes for malaria diagnosis and antibiotic prescription were based on results of parallel RDT testing (i.e., performed at the same time) and clinical judgement. As a result of parallel testing, a positive malaria RDT could coincide with a positive test result on any of the other tests leading to a decision to prescribe antibiotics in case of a positive malaria test. For the five hypothetical strategies, the malaria RDT was done first and informed the decision on malaria treatment in case of a positive test result. In case of a negative malaria test, further testing and/or clinical judgement informed the decision on antibiotic prescription. Due to the sequential approach of testing, the assumption was made that no antibiotics were prescribed in case of a positive malaria RDT for the hypothetical strategies.
A time horizon of 7 days was applied, consistent with the interval patients visited the clinic within the trial [17]. Considering the short time horizon, costs and effects were not discounted. Individuals were excluded from the analysis in case one of the following were missing: presenting signs and symptoms being respiratory or non-respiratory, malaria test results, decision on antibiotic prescription, adherence or non-adherence to prescription, and the outcome being favourable or unfavourable.
2.4 Probabilities
All probabilities within the decision tree were based on the trial data. A distinction was made between dependent and independent probabilities, whereby dependent probabilities were branch-specific and were influenced by previous chance nodes, while independent probabilities were the same across branches. The probability of the presenting signs and symptoms being respiratory or non-respiratory was considered independent of the arm (intervention or control). Dependent variables included the probability of testing positive for malaria, probability of prescribing antibiotics, probability of adherence, the probability of a favourable outcome, and the probability of an SAE. The probability of prescribing antibiotics differed per strategy and was based on the clinical diagnostic algorithm that was informed by the test results of the respective strategy (see OSM Appendix Table A2 for a full overview of probabilities per country per strategy).
2.5 Costs
We took a healthcare perspective, therefore including only direct medical costs for the healthcare system. Costs for labour for healthcare personnel, tests and test analyzers are provided in Table 1. Half-year depreciation costs were considered for training and test analyzers with a 1-year and 5-year total depreciation time, respectively. An overview of the cost for therapeutics can be found in the OSM Appendix Table A3. Average drug costs and diagnostic test costs per consultation were estimated based on trial data. For the unfavourable outcome without SAE, additional costs were modelled as additional therapeutics. For the unfavourable outcome with an SAE, additional costs included cost for additional therapeutics and cost of hospitalization (assuming patients were discharged after two inpatient days [21]). All costs in local currency units were inflated using local inflation rates [24] and converted using official Purchasing Power Parity (PPP) conversion rates to reflect the cost in 2021 International dollars ($), thereby eliminating price level differences across the three countries [25, 26].
2.6 Outcomes
The primary outcomes of the current analysis are the cost per consultation, antibiotic prescription rate (APR), and the incremental cost-effectiveness ratio (ICER) reflected as incremental cost per consultation for 1 percentage point (ppt) reduction in APR. The APR reflects the proportion of consultations in which antibiotics were prescribed (in line with WHO prescribing indicators [33]) and was the primary outcome measure of the Accelerator trial [13]. A reduction in the APR results in a reduction in the selective pressure that contributes to AMR [2, 4, 7]. Although it is uncertain what the exact relation is between APR and AMR, APR is considered an important surrogate outcome to quantify the effect of the test strategies [34]. The incremental difference in costs and APR between strategies results in the ICER, indicating the additional costs per consultation to reduce the APR with 1 ppt. No willingness-to-pay (WTP) threshold exists for the cost per ppt reduction in APR, hence no conclusions can be drawn with respect to the cost-effectiveness of the strategies. Instead, the analysis will enable a relative comparison of the cost-effectiveness between strategies for the three countries.
Secondary outcomes of the current analysis include the number of antibiotic regimens consumed, the number of Defined Daily Doses (DDDs) and antibiotic days-of-therapy. The number of antibiotic regimens consumed includes the total number of antibiotic regimens prescribed and the OTC regimens consumed (which were not prescribed by the healthcare professional). The DDD is a validated method to standardize the number of doses consumed and is developed by the WHO [35] (see OSM Appendix Table A4 for DDD per antibiotic). Since the DDD is impacted by weight- and age-based dosing, days-of-therapy was also included. Days-of-therapy is defined as the days of antibiotic therapy administered to a patient, independent of the dose [36]. Both the DDD and the days-of-therapy include the antibiotics prescribed and the antibiotics bought OTC. Only systemic (i.e., oral and injectable) antibiotics were included in the current analysis. The secondary antibiotic-related outcome measures are included to facilitate the utilization of the study results by future research.
In addition to the antibiotic-related outcome measures, DALYs were included to assess the short-term effect of the different interventions on clinical outcomes. DALYs were considered for the unfavourable outcome leaves of the decision tree. In case of an unfavourable outcome without SAE (i.e., the patient felt worse or the same at day 7), it was assumed that patients needed an additional 7-day cycle to get better, which is similar to new patients presenting at the clinic. To account for the additional disease burden, a disability weight of 0.001 DALY (0.051 DALY/52 weeks) was applied to these patients, which represented the disability weight for an acute, moderate episode of infectious diseases lasting 7 days [32]. In case of an unfavourable outcome with SAE (i.e., the patient had to visit the hospital), a disability weight of 0.0026 DALY (0.133 DALY/52 weeks) was applied, which represented the disability weight for an acute, severe episode of infectious disease lasting for 7 days [32].
2.7 Sensitivity and Scenario Analyses
In addition to the base-case analysis, we conducted a probabilistic sensitivity analysis (PSA) by fitting Beta (for probabilities), Gamma (for effects and costs) and Normal (for time) distributions to the variables based on trial data. If trial data were not available, parameters were varied by minus 20% and plus 20% of the mean, which was considered sufficient to assess the impact of parameter uncertainty on the outcomes. The results of 10,000 simulations are presented by a Bayesian 95% credible interval (CrI). The CrI can be interpreted as a 95% probability that the true (unknown) effect would lie within the interval, given the evidence provided [37].
Additional scenario analyses were conducted to assess the impact of the assumption that no antibiotics were prescribed for malaria-positive cases. This assumption can be considered as an optimistic base-case scenario given that the trial results indicated that antibiotics were prescribed for malaria-positive cases. For the scenario analysis, the effect on the overall APR of different levels of antibiotic prescription for malaria-positive cases was assessed, considering a range of 0–60%.
Furthermore, a scenario analysis was conducted to assess the effect of implementing the strategies for patients with presenting signs and symptoms being respiratory or non-respiratory separately. Therefore, the model was run separately for patients with the presenting signs and symptoms being respiratory (respiratory scenario) and for patients with the presenting signs and symptoms being non-respiratory (non-respiratory scenario).
2.8 Assumptions
Individual patients could have made the decision not to adhere to the prescription of the healthcare professional by buying additional antibiotics OTC. The number of patients who bought OTC antibiotics that were not prescribed by the healthcare professional was based on trial data. Since the details on the costs, DDD and days-of-therapy per OTC antibiotic were not captured in the trial, these numbers were based on the average numbers resulting from the antibiotics prescribed during the trial—stratified per country and age group. Furthermore, the assumption was made that individuals below 5 years of age received liquid medicine unless the medication was not available in liquid form. To model the five hypothetical strategies, the following additional assumptions had to be made: the test and prescription algorithm was strictly followed by physicians; the test sequence started with a malaria test, only in case of a negative malaria test subsequent tests were done, which is a simplification of clinical practice; no antibiotics were prescribed in case of a positive malaria test; the adherence probabilities in the hypothetical strategies were assumed to be equal to the probabilities in the “trial-based” strategy; for the hypothetical strategies, undetected group A Streptococcus, Streptococcus pneumoniae and typhoid fever result in an unfavourable outcome in case the patient was not treated with antibiotics; otherwise reduced antibiotic prescription in the hypothetical strategies did not result in additional negative impact on the quality of life of patients beyond the results of the trial. Finally, it was assumed that all tests were performed at the same clinical encounter.
3 Results
3.1 Context
Based on the results of the intervention arm of the AMR Diagnostics Use Accelerator trial, Table 2 shows the positivity rates, segregated per country per test. For the CRP and CBC test, the number of tests per sub-scale are reported. Table 2 indicates a high positivity rate for malaria and a relatively high percentage of patients with a CRP result > 80 for both Burkina Faso and Uganda. The results for Ghana are less distinct and demonstrate relatively low positivity rates across all tests. The malaria RDT was performed most frequently as part of SoC in Burkina Faso (100%) and Uganda (80.9%), in Ghana both malaria RDT (92.7%) and CBC tests (97.7%) were regularly performed as part of SoC (see OSM Appendix Table A5).
3.2 Outcomes
The total cost per consultation varies between strategies and between countries. The cost categories with the highest impact on the total cost are the test costs, labour cost and therapeutic costs (see Fig. 3). Due to a low number of unfavourable outcomes and low OTC consumption across the three countries, the associated costs only marginally impact the total costs. Similarly, the equipment costs and training costs constitute a fraction of the total cost per consultation. Regarding the five hypothetical strategies, the total cost of testing is dependent on the number of positive malaria tests as in these hypothetical scenarios, subsequent tests were only conducted if the initial malaria test yielded a negative test result. Consequently, the variations in test costs between the strategies in Burkina Faso are small as 646 patients out of the 856 tested positive for malaria (75.5%) (see Table 2). In contrast, the cost differences between strategies are more distinct for Uganda and Ghana due to the lower positivity rate for malaria. In Uganda, 500 out of 1,189 (42.1%) patients tested positive for malaria, while in Ghana 84 out of 738 (11.4%) patients had a positive malaria test. Labour and therapeutics costs are relatively constant across strategies making the cost of tests the main driver of the cost differences between strategies (see OSM Appendix Table A6 for a detailed overview).
The effect of the different strategies on APR varied per country (Fig. 4). Considering the 95% CrI, the “trial-based” strategy resulted in a significant decrease in the APR in Burkina Faso compared to the “SoC” strategy. However, in Ghana and Uganda, the APR in the “trial-based” strategy was not significantly different from the “SoC” strategy given the 95% CrI overlap (see Fig. 4). For the alternative strategies, a significant reduction in APR was observed compared to the “SoC” strategy for all three countries, except for the “malaria + CBC” strategy in Ghana. This could be explained by the fact that malaria and CBC testing was frequently performed in the SoC arm in Ghana. The comparison between the “malaria + all” strategy and the “trial-based” strategy provides insight into the effect the trial intervention could have in theory and the actual effect of the trial intervention, respectively. Across all three countries, there is a substantial difference between these two strategies. A full overview of the impact of the different strategies on the number of antibiotics prescribed, DDD, days-of-therapy and DALY is provided in OSM Appendix Table A7. The impact of the different strategies on DALYs was negligible, which can be explained by the minor differences in the number of unfavourable outcomes between strategies and the low disability weight considered for an unfavourable outcome.
In Table 3, an overview of the incremental cost per consultation for a 1 ppt reduction in APR is provided. The ICER is calculated in a pairwise comparison with the “SoC” strategy and fully incremental. For Burkina Faso and Uganda, the “malaria” strategy resulted in a substantial reduction in APR and a marginal reduction in cost (non-significant) compared to the “SoC” strategy, making the “malaria” strategy dominant (i.e., less costly and results in better health outcomes than the comparator). Expanding the test panel to the “malaria + CRP” strategy could result in a further reduction in the APR, with a pairwise ICER of $ 0.03 and $ 0.08 for Burkina Faso and Uganda, respectively. For Ghana, the pairwise comparison with the “SoC” strategy—which includes both malaria and CBC testing—indicates that both the “malaria” strategy and “malaria + CRP” strategy result in a reduction in costs and a reduction in the APR, making these two strategies dominant versus the “SoC” strategy. When comparing the “SoC” strategy and the “malaria + CBC” strategy for Ghana, minor differences are found, which indicates that the full potential effect of the “SoC” strategy on APR is realized.
3.3 Probabilistic Sensitivity Analysis
The results of the PSA are graphically presented in the cost-effectiveness planes (Fig. 5) with the “SoC” strategy as the comparator. The results of the PSA are in line with the deterministic outcomes and visualize the uncertainty of the deterministic outcomes that should be considered when interpreting the results of the current study. The spread of the scatterplots across countries and strategies is relatively low, which indicates that the results of the model can be considered as robust.
A cost-effectiveness acceptability curve (CEAC) was constructed to present the probability of each of the strategies being cost-effective at different levels of a hypothetical WTP, with a maximum WTP of $ 20 per ppt reduction in APR, which represents an extreme and likely an improbable value. Nonetheless, this extreme value allows for the evaluation and understanding of the dynamics between the different strategies over a wide range of WTP thresholds. Across all three countries, the “malaria” strategy resulted in the highest probability of cost-effectiveness for a WTP ranging from $ 0 to $ 0.40 for Burkina Faso, $ 0 to $ 0.41 for Ghana, and $ 0 to $ 0.22 for Uganda. In case of a higher WTP, the “malaria + CRP” strategy becomes the strategy with the highest probability of being cost-effective across all three countries. With a further increase of the WTP, the “malaria + CRP + CBC” strategy becomes the strategy with the highest probability of being cost-effective for Burkina Faso and Uganda (see Fig. 6). While the CEAC results offer insights in the relative cost-effectiveness of the strategies, they also highlight the need for establishing a WTP threshold per ppt reduction in APR to translate these findings into actionable policy recommendations.
3.4 Scenario Analysis
The first scenario examines the impact of varying percentages (ranging from 0% to 60%) of malaria-positive patients receiving antibiotics. This range was considered to be representative of the percentages of malaria-positive cases receiving antibiotics in the “SoC” and “trial-based” strategies, which were as follows: 48% and 41.6% for Burkina Faso, 17.9% and 20.2% for Ghana, and 33.4% and 62.0% for Uganda, respectively. Across all three countries and all strategies, except for the “malaria + CBC” strategy in Ghana, there was a persistent reduction in APR compared to the “SoC” strategy over the entire range of 0% to 60% (see OSM Appendix Fig. A1). This finding suggests that the strategies’ effect on APR is not solely due to the potential reduction in APR resulting from not prescribing antibiotics to malaria-positive cases. It indicates that there is also an effect on APR in the malaria-negative group due to the additional tests performed or educational intervention introduced in the Accelerator trial, which further supports the finding of the main analysis that the test strategies could effectively reduce APR.
In the second scenario, the potential difference in the effect for patients with presenting signs and symptoms being respiratory (respiratory scenario) versus presenting signs and symptoms being non-respiratory (non-respiratory scenario) was assessed. The results indicate a higher APR across all strategies for the respiratory patient group in both Burkina Faso and Ghana. For Burkina Faso, the strategies resulted in a greater ppt reduction in APR compared to the “SoC” strategy in the respiratory patient population. On the contrary, for Ghana and Uganda, the strategies tended to be more effective in the non-respiratory patient population compared to the respiratory patient population (see OSM Appendix Table A8).
4 Discussion
A health economic analysis was conducted to quantify the expected cost, clinical outcomes, and impact on antibiotic prescription resulting from the implementation of a diagnostic strategy in Burkina Faso, Ghana and Uganda. The “trial-based” strategy resulted in a significant decrease in APR compared to the “SoC” strategy in Burkina Faso, whereas no significant decrease was found in Ghana and Uganda. Alternative diagnostic strategies, modelled as hypothetical cases, indicated that a significant reduction in the APR could be realized by the introduction of a diagnostic test strategy in all three countries. The “malaria” strategy resulted in cost savings but was also part of SoC across all three countries and produced a less pronounced reduction in APR than other strategies. Initial testing with malaria and subsequent testing with a CRP test (for malaria-negative patients) resulted in a pairwise ICER ranging from $ 0.03 to $ 0.08 per consultation for 1 ppt reduction in APR versus the “SoC” strategy for Burkina Faso and Uganda, respectively, and was dominant (i.e., less costly and results in better health outcomes) versus the “SoC” strategy in Ghana.
The results of the hypothetical strategies indicate the maximum effect the tests could have on APR given the assumption that the clinical test algorithm would be strictly followed. As a result, the findings are likely an overestimation of the actual result since in reality multiple factors influence the decision to prescribe antibiotics besides the test results, such as fear of adverse outcomes, lack of trust in POC test outcomes, patient expectations, prescriber’s awareness of AMR, and potential financial burden on patients due to test costs [12, 38]. Indeed, the observed difference in antibiotic prescriptions between the “trial-based” strategy and the “malaria + all” strategy indicate a substantial difference between the actual and hypothetical strategies. The two strategies apply the same package of diagnostic tests, but the hypothetical “malaria + all” strategy resulted in a substantially larger ppt reduction in APR versus the “trial-based” strategy despite the training and education provided as part of the “trial-based” strategy. Based on the results of the scenario analysis, this difference can only partially be explained by the way malaria-positive patients would be treated in the hypothetical strategies and indicates the challenge of changing the status quo of presumptive antibiotic prescription by clinicians. Hence, the results of the current analysis should be considered optimistic if the implementation of the test strategy within clinical practice neglects the broader context of the decision to prescribe antibiotics. A systems approach is required to reduce inappropriate antibiotic prescriptions with diagnostic tests as a valuable tool to rationalize antibiotic prescription rather than the silver bullet.
The primary outcome of the present study is reported in the incremental cost per consultation for 1 ppt reduction in APR, which provides an insightful alternative for the standard health-economic outcomes such as incremental cost per DALY averted or cost per QALY gained. However, cost per 1 ppt reduction in APR is not a widely accepted standard outcome measure yet for interventions targeting antibiotic prescriptions [14, 39, 40]. No studies on antimicrobial stewardship programs in Ghana, Uganda or Burkina Faso have been identified that report the results in cost per ppt reduction in APR. Studies that do report the cost per ppt reduction in APR (recalculated to International $) range from $ 0.07 for an antibiotic stewardship program among primary-care doctors in China [41] to $ 1.59 for a combined primary-care intervention of general practitioner communication training and CRP testing for patients with respiratory tract infections in Europe [42] and $ 3.56 for CRP testing in a primary-care setting for patients with acute exacerbations of chronic obstructive pulmonary disease in England and Wales [43]. Since a WTP for a ppt reduction in APR is lacking, no conclusions can be drawn with respect to the absolute cost-effectiveness of these interventions. However, by reporting the results of the interventions in cost per ppt reduction in APR, comparisons can be made between interventions between countries. Not unexpectedly, interventions produce lower ICERs in LMICs (the present study and the study of Zhang et al. [41]) than high-income countries. However, AMR is a global threat that must be addressed concomitantly in all countries in ways that are proportionate to the wealth and purchasing power of their government and societies.
4.1 Strengths and Limitations
The strength of the current study lies in the relative comparisons made across a range of test strategies and different countries. The relative comparisons provide valuable insights with respect to the most favourable test combination, which can support policy-makers in their decision on the allocation of scarce resources for maximum health benefit. Additionally, the between-country comparisons allowed us to identify parameters influencing cost-effectiveness (i.e., percentage of positive malaria cases), which is a strength of the study. The current analysis also comes with several limitations. To include hypothetical diagnostic strategies, we assumed a perfect diagnostic strategy implementation with healthcare providers strictly following the diagnostic algorithm in their decision as to whether to prescribe antibiotics or not. This is a simplification of reality and most certainly resulted in an overestimation of the effect of the strategies, but it does provide valuable insight in the potential of the alternative diagnostic strategies. Another limitation is related to the assumption that a reduction in antibiotic prescription does not negatively impact quality of life. Although this assumption is supported by previous studies [44, 45], it neglects malaria and bacteraemia co-infection, which could have resulted in negative health outcomes and requires antibiotic treatment [46, 47].
5 Conclusion
The introduction of a diagnostic test panel for patients with acute febrile illness could result in a substantial reduction in the APR with a malaria test as the dominant strategy. Further reduction of the APR at minimal cost could be realized by the combination of a malaria test and a CRP test. The implementation of diagnostic tests for more appropriate antibiotic prescriptions is effective, but requires a WTP per ppt reduction in the APR to decide on the most cost-effective strategy, and should be part of an integrated stewardship program to realize its full potential.
References
Murray CJ, Ikuta KS, Sharara F, et al. Global burden of bacterial antimicrobial resistance in 2019: a systematic analysis. Lancet. 2022. https://doi.org/10.1016/S0140-6736(21)02724-0. (published online First: 19 January 2022).
Michael CA, Dominey-Howes D, Labbate M. The antimicrobial resistance crisis: causes, consequences, and management. Front Public Health. 2014;2:145.
Dadgostar P. Antimicrobial resistance: implications and costs. Infect Drug Resist. 2019;12:3903–10.
O’Neill J. Tackling drug-resistant infections globally: final report and recommendations. Government of the United Kingdom. 2016. https://apo.org.au/node/63983. Accessed 23 July 2021.
Selvaraj S, Farooqui HH, Karan A. Quantifying the financial burden of households’ out-of-pocket payments on medicines in India: a repeated cross-sectional analysis of National Sample Survey data, 1994–2014. BMJ Open. 2018;8: e018020.
Alsan M, Schoemaker L, Eggleston K, et al. Out-of-pocket health expenditures and antimicrobial resistance in low- and middle-income countries. Lancet Infect Dis. 2015;15:1203–10.
Prestinaci F, Pezzotti P, Pantosti A. Antimicrobial resistance: a global multifaceted phenomenon. Pathog Glob Health. 2015;109:309–18.
World Health Organization. Global action plan on antimicrobial resistance. 2015. https://www.who.int/publications-detail-redirect/9789241509763. Accessed 12 Oct 2021.
Laxminarayan R, Duse A, Wattal C, et al. Antibiotic resistance—the need for global solutions. Lancet Infect Dis. 2013;13:1057–98.
Kozel TR, Burnham-Marusich AR. Point-of-care testing for infectious diseases: past, present, and future. J Clin Microbiol. 2017;55:2313–20.
Yager P, Domingo GJ, Gerdes J. Point-of-care diagnostics for global health. Annu Rev Biomed Eng. 2008;10:107–44.
Pai NP, Vadnais C, Denkinger C, et al. Point-of-care testing for infectious diseases: diversity, complexity, and barriers in low- and middle-income countries. PLoS Med. 2012;9: e1001306.
Olliaro P, Nkeramahame J, Salami O, et al. Advancing access to diagnostic tools essential for universal health coverage and antimicrobial resistance prevention: an overview of trials in Sub-Saharan Africa. Clin Infect Dis. 2023;77:S125–33.
van der Pol S, Rojas García P, Postma MJ, et al. Economic analyses of respiratory tract infection diagnostics: a systematic review. Pharmacoeconomics. 2021;39:1411–27.
Verbakel JY, Turner PJ, Thompson MJ, et al. Common evidence gaps in point-of-care diagnostic test evaluation: a review of horizon scan reports. BMJ Open. 2017;7: e015760.
Holmes EAF, Hughes DA. Challenges for economic evaluation of health care strategies to contain antimicrobial resistance. Antibiotics. 2019;8:166. https://doi.org/10.3390/antibiotics8040166.
Salami O, Horgan P, Moore CE, et al. Impact of a package of diagnostic tools, clinical algorithm, and training and communication on outpatient acute fever case management in low- and middle-income countries: protocol for a randomized controlled trial. Trials. 2020;21:974.
Olliaro P, Nkeramahame J, Horgan P, et al. Synthesis and meta-analysis of 3 randomized trials conducted in Burkina Faso, Ghana, and Uganda comparing the effects of point-of-care tests and diagnostic algorithms versus routine care on antibiotic prescriptions and clinical outcomes in ambulatory patients < 18 years of age with acute febrile illness. Clin Infect Dis. 2023;77:S199-205.
Kiemde F, Valia D, Kabore B, et al. A Randomized trial to assess the impact of a package of diagnostic tools and diagnostic algorithm on antibiotic prescriptions for the management of febrile illnesses among children and adolescents in primary health facilities in Burkina Faso. Clin Infect Dis. 2023;77:S134–44.
Kapisi J, Sserwanga A, Kitutu FE, et al. Impact of the introduction of a package of diagnostic tools, diagnostic algorithm, and training and communication on outpatient acute fever case management at 3 diverse sites in Uganda: results of a randomized controlled trial. Clin Infect Dis. 2023;77:S156–70.
Adjei A, Kukula V, Narh CT, et al. Impact of point-of-care rapid diagnostic tests on antibiotic prescription among patients aged < 18 years in primary healthcare settings in 2 peri-urban districts in Ghana: randomized controlled trial results. Clin Infect Dis. 2023;77:S145–55.
World Health Organization. WHO guidelines for malaria. 2023. https://iris.who.int/bitstream/handle/10665/366432/WHO-UCN-GMP-2023.01-eng.pdf?sequence=1. Accessed 5 Oct 2023.
R Core Team. R: a language and environment for statistical computing. https://www.r-project.org/. Accessed 19 Oct 2021.
Inflation, consumer prices (annual %)—Uganda, Ghana, Burkina Faso | Data. https://data.worldbank.org/indicator/FP.CPI.TOTL.ZG?end=2020&locations=UG-GH-BF&start=1999. Accessed 13 Apr 2022.
Fundamentals of Purchasing Power Parities. https://thedocs.worldbank.org/en/doc/332341517441011666-0050022018/original/PPPbrochure2017webformatrev.pdf. Accessed 21 Oct 2021.
PPP conversion factor, GDP (LCU per international $)—Uganda, Ghana, Burkina Faso | Data. https://data.worldbank.org/indicator/PA.NUS.PPP?locations=UG-GH-BF. Accessed 21 Apr 2023.
Boyer S, Nishimwe ML, Sagaon-Teyssier L, et al. Cost-effectiveness of three alternative boosted protease inhibitor-based second-line regimens in HIV-infected patients in West and Central Africa. Pharmacoecon Open. 2020;4:45–60.
Aboagye AQQ, Degboe ANK, Obuobi AAD. Estimating the cost of healthcare delivery in three hospitals in Southern Ghana. Ghana Med J. 2010;44:83–92.
Ccess A, Ottlenecks B, Osts C, et al. Assessing facility capacity, costs of care, and patient perspectives. B OTTLENECKS. 77.
Shillcutt S, Morel C, Goodman C, Coleman P, Bell D, Whitty CJ, Mills A. Cost-effectiveness of malaria diagnostic methods in sub-Saharan Africa in an era of combination therapy. Bull World Health Org. 2008;86:101–10.
Lancet Laboratory—Welcome to Lancet. 2022. https://www.lancet.co.za/. Accessed 22 June 2023.
Salomon JA, Haagsma JA, Davis A, et al. Disability weights for the global burden of disease 2013 study. Lancet Glob Health. 2015;3:e712–23.
How to investigate drug use in health facilities: selected drug use indicators. https://www.who.int/publications-detail-redirect/who-dap-93.1. Accessed 7 Mar 2024.
The opportunity of point-of-care diagnostics in general practice: modelling the effects on antimicrobial resistance—PMC. https://www-ncbi-nlm-nih-gov.proxy-ub.rug.nl/pmc/articles/PMC9243781/. Accessed 12 Mar 2024.
WHO Collaborating Centre for Drug Statistics Methodology. Guidelines for ATC classification and DDD assignment 2021. 2021. https://www.whocc.no/filearchive/publications/2021_guidelines_web.pdf. Accessed 18 Mar 2021.
Polk RE, Fox C, Mahoney A, et al. Measurement of adult antibacterial drug use in 130 US hospitals: comparison of defined daily dose and days of therapy. Clin Infect Dis. 2007;44:664–70.
Hespanhol L, Vallio CS, Costa LM, et al. Understanding and interpreting confidence and credible intervals around effect estimates. Braz J Phys Ther. 2019;23:290–301.
Thompson W, Tonkin-Crine S, Pavitt SH, et al. Factors associated with antibiotic prescribing for adults with acute conditions: an umbrella review across primary care and a systematic review focusing on primary dental care. J Antimicrob Chemother. 2019;74:2139–52.
van Dorst PWM, van der Pol S, Salami O, et al. Evaluations of training and education interventions for improved infectious disease management in low-income and middle-income countries: a systematic literature review. BMJ Open. 2022;12: e053832.
D’hulster E, De Burghgraeve T, Luyten J, et al. Cost-effectiveness of point-of-care interventions to tackle inappropriate prescribing of antibiotics in high- and middle-income countries: a systematic review. J Antimicrob Chemother. 2023;78:893–912.
Zhang Z, Dawkins B, Hicks JP, et al. Cost-effectiveness analysis of a multi-dimensional intervention to reduce inappropriate antibiotic prescribing for children with upper respiratory tract infections in China. Trop Med Int Health. 2018;23:1092–100. https://doi.org/10.1111/tmi.13132.
Oppong R, Smith RD, Little P, et al. Cost-effectiveness of internet-based training for primary care clinicians on antibiotic prescribing for acute respiratory tract infections in Europe. J Antimicrob Chemother. 2018;73:3189–98.
Francis NA, Gillespie D, White P, et al. C-reactive protein point-of-care testing for safely reducing antibiotics for acute exacerbations of chronic obstructive pulmonary disease: the PACE RCT. Health Technol Assess. 2020;24:1–108.
Martínez-González NA, Keizer E, Plate A, et al. Point-of-care C-reactive protein testing to reduce antibiotic prescribing for respiratory tract infections in primary care: systematic review and meta-analysis of randomised controlled trials. Antibiotics. 2020;9:610. https://doi.org/10.3390/antibiotics9090610.
Little P, Hobbs FDR, Moore M, et al. Clinical score and rapid antigen detection test to guide antibiotic use for sore throats: randomised controlled trial of PRISM (primary care streptococcal management). BMJ. 2013;347: f5806.
Organization WH. Guidelines for the treatment of malaria. 3rd ed. Geneva: World Health Organization; 2015.
Wilairatana P, Mala W, Masangkay FR, et al. The prevalence of malaria and bacteremia co-infections among febrile patients: a systematic review and meta-analysis. Trop Med Infect Dis. 2022;7:243. https://doi.org/10.3390/tropicalmed7090243.
Acknowledgements
The authors would like to thank Sarah Girdwood (FIND) and Kyra Grantz (FIND) for editorial help. The authors are grateful to the investigators who were pivotal to data collection, with a special thanks to Rita Baiden, Alex Adjei, Vida Kukula (all from Dodowa Health Research Centre, Accra, Ghana), Halidou Tinto, Francois Kiemde, Adelaide Compaore, Daniel Valea (all from IRSS-Clinical Research Unit of Nanoro, Nanoro, Burkina Faso), James Kapisi (Infectious Diseases Research Collaboration, Kampala, Uganda), Heidi Hopkins (London School of Hygiene & Tropical Medicine, London, UK), David Kaawa-Mafigiri and Deborah Ekusai (both from School of Social Sciences, Makerere University, Kampala, Uganda).
Funding
This study was funded by FIND. The AMR Diagnostics Use Accelerator Trial was funded by the Swiss Agency for Development and Cooperation (SDC), the Federal Ministry of Economic Cooperation and Development (BMZ), and the UK DFID, now Foreign Commonwealth and Development Office (FCDO).
Author information
Authors and Affiliations
Contributions
All authors contributed to the study conception and design. Material preparation, data collection and analysis were performed by Pim W.M. van Dorst with support from Simon van der Pol and Thea van Asselt. The first draft of the manuscript was written by Pim W.M. van Dorst and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.
Corresponding author
Ethics declarations
Potential conflicts of interest
Pim W.M. van Dorst, Piero Olliaro, Sabine Dittrich, Juvenal Nkeramahame, Simon van der Pol and Antoinette D.I. van Asselt have no reported conflicts of interest. Prof. Maarten Postma holds of stocks of Pharmacoeconomics Advice Groningen (Netherlands) and Health-Ecore (Zeist, Netherlands). Prof. Cornelis Boersma holds stocks of Health-Ecore (Zeist, Netherlands). All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the article have been disclosed.
Ethics statement
The AMR Diagnostics Use Accelerator trial was registered with the US National Library of Medicine Clinical Trials database (identifier: NCT04081051). The overarching protocol was reviewed and approved by the Human Research Ethics Committee of the University of Oxford, United Kingdom. In addition, country-specific protocols were reviewed and approved by relevant regulatory authorities and national and/or institutional ethics committees in all participating countries.
Consent to participate
Written informed consent was provided by all participating adults and official caregivers of participating children.
Consent for publication
The study participants and official caregivers of participating children consented to their data being used for publication.
Data availability
The data that support the findings of this study were provided by FIND, who received funding from the UK government for this study. Restrictions apply to the availability of these data and so are not publicly available. Data are available from the authors upon reasonable request and with permission of FIND.
Code availability
The R model is available from the authors upon reasonable request and with permission of FIND.
Supplementary Information
Below is the link to the electronic supplementary material.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License, which permits any non-commercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc/4.0/.
About this article
Cite this article
van Dorst, P.W.M., van der Pol, S., Olliaro, P. et al. Cost-Effectiveness of Test-and-Treat Strategies to Reduce the Antibiotic Prescription Rate for Acute Febrile Illness in Primary Healthcare Clinics in Africa. Appl Health Econ Health Policy (2024). https://doi.org/10.1007/s40258-024-00889-x
Accepted:
Published:
DOI: https://doi.org/10.1007/s40258-024-00889-x