Cost–effectiveness of emergency care interventions in low and middle-income countries: a systematic review

Abstract Objective To systematically review and appraise the quality of cost–effectiveness analyses of emergency care interventions in low- and middle-income countries. Methods Following the PRISMA guidelines, we systematically searched PubMed®, Scopus, EMBASE®, Cochrane Library and Web of Science for studies published before May 2019. Inclusion criteria were: (i) an original cost–effectiveness analysis of emergency care intervention or intervention package, and (ii) the analysis occurred in a low- and middle-income setting. To identify additional primary studies, we hand searched the reference lists of included studies. We used the Consolidated Health Economic Evaluation Reporting Standards guideline to appraise the quality of included studies. Results Of the 1674 articles we identified, 35 articles met the inclusion criteria. We identified an additional four studies from the reference lists. We excluded many studies for being deemed costing assessments without an effectiveness analysis. Most included studies were single-intervention analyses. Emergency care interventions evaluated by included studies covered prehospital services, provider training, treatment interventions, emergency diagnostic tools and facilities and packages of care. The reporting quality of the studies varied. Conclusion We found large gaps in the evidence surrounding the cost–effectiveness of emergency care interventions in low- and middle-income settings. Given the breadth of interventions currently in practice, many interventions remain unassessed, suggesting the need for future research to aid resource allocation decisions. In particular, packages of multiple interventions and system-level changes represent a priority area for future research.


Introduction
Emergency care is a health systems and service delivery innovation that facilitates early recognition and life-saving interventions for time sensitive acute injuries and illnesses, where a delay of hours may result in avoidable death or disability, or make treatments less effective. 1,2 Frontline providers deliver these interventions across the emergency care system, from scene care to transport to facilities. Conditions addressed by emergency care include trauma, infections, noncommunicable disease and complications of pregnancy. These conditions accounted for nine of the 10 leading causes of death in low-income countries in 2017. 3 For people aged 5-29 years, the most common cause of death is road traffic crashes, causing over 28 million deaths a year, of which most occurring in low-and middle-income countries. 4 Researchers have estimated that over half of deaths in low-and middleincome countries, and up to 2.5 billion disability-adjusted life-years (DALYs) annually, could be addressed through the implementation of effective emergency care. 2 These figures are expected to grow due to factors such as increased use of motor vehicles, increased urbanization and lifestyle changes leading to increases in coronary heart disease. Traumatic injury alone is anticipated to represent a fifth of all ill-health worldwide by 2020. 5,6 Early recognition of acute conditions by the health-care system, and improved access to care, could address much of the ill-health burden and save millions of lives. These facts have been acknowledged by the World Health Assembly Resolution Emergency care systems for universal health coverage: ensuring timely care for the acutely ill and injured. 7 Emergencies occur regardless of whether a health system is prepared to address them. An organized emergency care system can theoretically leverage economies of scope and scale by employing simple low-cost interventions that will save millions of lives. However, little is known about the cost-effectiveness of emergency care interventions in low-and middle-income countries (LMIC), where such interventions may have the greatest impact.
Cost-effectiveness data is essential to inform the resource allocation decisions of policy-makers and regulators. Such data encourages the prioritization of systems and interventions most likely to provide a health and economic benefit. A survey of policy-makers in Brazil, Cuba, Nepal, Norway, and Uganda demonstrated a majority preference for efficiency arguments, such as cost-effectiveness, in formalizing the health priority setting process. 8 Here, we present a systematic review on the cost-effectiveness of emergency care interventions in low-and middleincome countries. Our aim was to characterize the existing knowledge regarding the costs and benefits of delivering emergency care in these settings, to examine the quality of cost-effectiveness analyses and to provide guidance for future research efforts.
Objective To systematically review and appraise the quality of cost-effectiveness analyses of emergency care interventions in low-and middle-income countries. Methods Following the PRISMA guidelines, we systematically searched PubMed®, Scopus, EMBASE®, Cochrane Library and Web of Science for studies published before May 2019. Inclusion criteria were: (i) an original cost-effectiveness analysis of emergency care intervention or intervention package, and (ii) the analysis occurred in a low-and middle-income setting. To identify additional primary studies, we hand searched the reference lists of included studies. We used the Consolidated Health Economic Evaluation Reporting Standards guideline to appraise the quality of included studies. Results Of the 1674 articles we identified, 35 articles met the inclusion criteria. We identified an additional four studies from the reference lists. We excluded many studies for being deemed costing assessments without an effectiveness analysis. Most included studies were single-intervention analyses. Emergency care interventions evaluated by included studies covered prehospital services, provider training, treatment interventions, emergency diagnostic tools and facilities and packages of care. The reporting quality of the studies varied. Conclusion We found large gaps in the evidence surrounding the cost-effectiveness of emergency care interventions in low-and middleincome settings. Given the breadth of interventions currently in practice, many interventions remain unassessed, suggesting the need for future research to aid resource allocation decisions. In particular, packages of multiple interventions and system-level changes represent a priority area for future research.

Search strategy
We systematically reviewed the literature on emergency care interventions in low-and middle-income countries. We searched for peer-reviewed articles published before May 2019, in PubMed®, Scopus, Embase®, Cochrane Library and Web of Science. To capture the heterogeneity of emergency care interventions, we included components of both prehospital and facility-based emergency care system. An example of the search terms used is shown in Box 1 and the full search strategy is available in the data repository. 10 We applied no language restrictions. To limit search results to our context of interest we applied the Cochrane LMIC filter. 9 The study was designed in accordance with the Preferred Reporting Items for Systematic Review and Meta-Analyses guidelines 11 and was prospectively registered with PROSPERO (CRD42018080145).

Selection of studies
We uploaded all identified studies into the software Covidence (Covidence, Melbourne, Australia) for review. Studies were considered for inclusion if they: (i) described a system-wide or individual emergency care intervention; (ii) were implemented in a low-and middleincome country (according to 2018 World Bank classification) or analysed costing data from a low-and middleincome country; and (iii) undertook a full economic evaluation (either a cost-effectiveness analysis, cost-utility analysis or cost-benefit analysis). We defined emergency care interventions as interventions that provide or facilitate the early care of acutely injured and ill patients, whether outside or inside a health-care facility. This definition included early critical care or surgical interventions that commonly take place in an emergency department.
Two reviewers independently assessed titles and abstracts for inclusion of the articles for the full text review. The reviewers only selected studies for full text review if both agreed that the studies met the inclusion criteria. Studies then underwent full text reviews for eligibility by two independent reviewers. Disagreements were resolved by consensus within the study team. To identify additional primary studies, we hand searched the reference lists of included studies.

Quality assessment
Two reviewers appraised the included studies using the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) checklist. 12 To better understand the quality of data, we did a comprehensive scoring of studies, by giving one point for each fulfilled item on the checklist, however, we did not exclude any studies based on quality.

Data abstraction
The following information was extracted from included studies: country; year of publication; intervention; comparator; time horizon; discount rate, study perspective, health outcome, sensitivity analyses and findings. We converted cost results or the cost part of the incremental cost-effectiveness ratios to 2019 Untied States dollars (US$) for comparability. Due to the lack of consensus surrounding the use of cost-effectiveness thresholds, 13,14 we did not apply a global benchmark to each study's results, but left the results to be interpreted within the specific study context.

Description of included studies
Included studies fell within five broad categories: prehospital services; training; treatment interventions; diagnostic tools; and facilities and/or packages of care. These categories are not mutually exclusive, but rather reflect the primary nature of the intervention studied. Table 1 shows a summary of all included articles, interventions assessed and main findings. 10

Prehospital services
Five studies investigated the impact of either introducing a professional ambulance service or improving ambulance response times. Two studies looked specifically at the establishment of ambulance services for obstetric care, 15,23 one at establishing a general emergency medical services system, 16 one for cardiac arrest care, 27 and one modelled the impacts of a decrease in response time. 19 The cost-effectiveness of establishing an ambulance system ranged from US$ 18 in Uganda to US$ 568 642 in Malaysia per life year saved. 23,27 In Thailand, reducing the response time would decrease the yearly national health-care expenditure by US$ 425 million to US$ 850 million for each minute saved. 19 The wide range of results in this category is attributable to significant heterogeneity in the costs, exact interventions provided and impact data used for the analysis.

Provider training
Six studies assessed the impact of training interventions, 25

Treatment interventions
Nineteen studies compared different treatments or patient care pathways. Four studies assessed the treatment of acute bleeding, including two for tranexamic acid injection in trauma patients 22,45 and two studies for recombinant activated factor VII injection in patients with hemophilia. 43

Emergency diagnostic tools
Two studies assessed emergency care diagnostic tools. 44,49 For example, a modelling study showed that using an electrocardiogram for patients with chest pain in India costs US$ 16 per QALY gained. 44

Facilities and packages of care
Four studies assessed the cost-effectiveness of the provision of facility-based emergency care. Three studies focused on the provision of surgical care 17,18,21 while one evaluated a stand-alone diarrhoea treatment centre. 28 The cost-effectiveness of facilities providing surgical care varied from US$ 18 to 265 per DALY averted, 17,18,21 and US$ 4032 per death averted for a diarrhoeal treatment centre. 28 Although most studies analysed actual interventions, three studies used modelling to predict the impact of increased coverage and improved quality of service. 20,29,41 Authors of one paper estimated that the economic benefit would be within the range of U$ 758 billion-786 billion per year globally if the mortality rates in low-and middleincome countries were reduced to the rates in high-income countries. 20 Other findings show that implementing guidelines and improving the standard of care yielded incremental cost-effectiveness ratios between US$ 47 and US$ 474 per DALY. 41

Discussion
We sought to systematically collect and critically appraise the existing literature on the cost-effectiveness of emergency care interventions in low-and middle- income countries. Cost-effectiveness analyses are important for assessing the value for money of emergency care interventions and to allow for prioritization and optimal resource allocation.
Formulating a general conclusion about the wider implication of the findings on the cost-effectiveness of emergency care is problematic, because of the heterogeneity of methods, settings, and presentation of results of the identified studies. For example, few studies used health outcomes that are widely comparable against other disease programmes, such as QALYs or DALYs. This lack, coupled with inconsistent reporting of incremental cost-effectiveness ratios, makes the comparison between the findings of these assessments and other programmes difficult for decision-makers with constrained budgets. Furthermore, some studies fell short of using a contextually-oriented study design and, where possible, empirically derived local inputs. For example, one study used parameters from a high-income setting to estimate cost-effectiveness in a middleincome country, generating results that are difficult to interpret. 27 Overall, we noted that most of the studies were methodologically weak by the quality criteria we applied, failing to provide detailed descriptions of the assumptions taken. Assumptions used to calculate costs and outcomes can greatly influence the final cost estimate and reporting these details can help decisionmakers understand to what level these findings apply to their setting and what level of uncertainty was taken in the review. Furthermore, not reporting the comparator used will hinder readers to understand the context of the results. Only two-thirds of the studies provided an incremental cost-effectiveness ratio, which aid decision-makers by allowing for comparability across interventions and the application of a cost-effectiveness threshold.
Even in the setting of standardized methods and results reporting, there continues to be a lack of expert consensus surrounding the interpretation of cost-effectiveness data outside the original study context that produced it. While the application of gross domestic product-based global thresholds remains a common approach, consideration of willingness-to-pay for health benefits, identification of benchmark interventions, assessment of budgetary-impact, and incorporation of league tables allow for improved contextualization of results and utility for decision-makers. 13,14,54 When comparing the results of our included studied with readily available collated data from other public health interventions, we recommend readers go to the primary literature and ensure context and methodologic comparability.
Another notable finding from our review is that the research focused on single-intervention analyses rather than intervention packages or system changes. An organized emergency care system has the capacity to treat a variety of conditions with a common set of resources, thus gaining efficiencies in per-unit costs by applying economies of scope. Additional positive effects across the health system, such as reducing downstream health-care costs, contributing to public health surveil-lance and preparedness for disasters, can be also achieved by the organization and alignment of emergency care services. More research is needed on the cost-effectiveness of system changes, process improvement and intervention packages. Furthermore, an exclusive health-care perspective was used in most the studies, which may undervalue the broad social impacts and economic burden of lost workforce productivity that can be mitigated with emergency care. This narrow scope of analysis may obscure the broader productivity and economic gains that emergency care interventions provide.
Over 80 studies excluded in the review were costing-only assessments (i.e. no measure of efficacy or benefit was assessed). These studies were often descriptive costing studies of a disease  12 We deemed the overall quality of each item as low if the percentage was below 50%, medium if the percentage was between 50% and 80%, and high if the percentage was above 80%.

Systematic reviews
Cost-effectiveness analyses of emergency care interventions Kalin Werner et al.
entity used to justify spending on preventive measures. For example, the authors of a cost analysis of interpersonal violence in South Africa concluded that the costs of in-hospital care of violence victims warrants investment in primary prevention of these injuries. 55 Costing-only studies were also employed for budgetimpact analysis between two choices, including (i) contrasting expenditures between two health-care settings for a single disease entity 56 and (ii) contrasting expenditures between two patient populations. 57 One study reported an incremental cost-effectiveness ratio calculated by using the reduction in treatment time as a primary outcome rather than a health outcome and therefore was not extracted for final review. 58 When evaluating the research gap, we noticed a sizable discrepancy between the breadth of emergency care interventions in low-and middleincome countries and the amount of published research from these settings. Of the articles we assessed, only 24 out of 137 low-and middle-income countries globally are represented in our findings, indicating a significant gap in research.
Finally, a limitation of our study surrounds the difficulty of labelling emergency care interventions for searchability. Although, we attempted to capture all literature related to emergency care, there may be relevant articles, which were not caught in our search criteria. Unless authors clearly tagged the intervention with terms related to "emergency care," their study may not have been captured by our search. For this reason, we handsearched references of included articles, yielding several additional studies.
Our systematic review demonstrates a relative sparsity of evidence re-garding the cost-effectiveness of emergency care interventions in low-and middle-income countries. Given the breadth of available interventions, numerous potentially low-and high-cost interventions and their impacts remain unevaluated. Our review highlights areas for improvement in the quality of methods and study-design that would facilitate the use of future studies in the decision-making process with regards to the allocation of resources. Overall, the included studies allow us to begin to characterize the literature and establish a research agenda in this area. A primary focus of the future research is the development of cost-effectiveness analyses that evaluate emergency care as a system of integrated care delivery, considering economies of scope and the broader impact of organizing, and aligning health-care provision. ■