Health Economic Publications From the Middle East and North Africa Region: A Scoping Review of the Volume and Methods of Research

The Middle East and North Africa (MENA) region comprises diverse countries in terms of geographic and demographic characteristics, economic performance, and standards of living. Over the last decades, in harmony with the United Nation's Millennium Development Goals and Sustainable Development Goals, the health systems and general health status of this region has undergone considerable development.^[1] Along with the recent reforms aiming for fairer and more accountable healthcare systems, value-based health policy and financing became an emerging priority in the region.^[2,3] Despite the increasing importance of economic considerations in the allocation of resources and the delivery of healthcare, recent reviews have reported limited quality and quantity of health economic evaluations in the Gulf region^[4] and Egypt.^[5] Although the institutional background of health technology assessment and health economic evaluation is still under development,^[6–9] a number of pharmacoeconomic guidelines and expert recommendations have already been published in both high- and middle-income countries in the region.^[10–12] However, the health economic studies in MENA as a whole have not yet been systematically reviewed.

Considering the increasing importance of health economic research in the region, our aim was to review the scope of countries, diseases, technologies, and methods, as well as bibliometric properties of health economic publications from MENA up to 2019. Furthermore, we aimed to explore differences between recent publications from the last 5 years (2015–2019) and earlier publications, as well as high- and middle-income countries.

We followed the applicable PRISMA guidelines in reporting our scoping review.^[13,14] No ethics approval was required for this study. We combined search words related to general descriptions, key methods, and main outcomes of health economic analysis. Although many definitions of the MENA region exist, we selected the following 17 countries: Algeria, Bahrain, Egypt, Iraq, Jordan, Kuwait, Lebanon, Libya, Morocco, Oman, Qatar, Palestine, Saudi Arabia (KSA), Syria, Tunisia, the United Arab Emirates ([UAE] including Dubai and Abu-Dhabi), and Yemen. We conducted a search in PubMed up to December 15, 2019, combining the keywords and countries in the Title/Abstract field, using filters for English language and full-text publications. The detailed search strategy is displayed in Supplementary Table S1.

English language full-text journal articles were included if they reported original research on humans, involved the local population from the target countries, reported costs, and involved a full or partial health economic analysis comparing alternative health technologies. For health technologies, we used a broad definition applied by the World Health Organization (WHO).^[17] We also included efficacy studies reporting costs, which are considered as potentially important sources of health economic evidence.^[18] We excluded speculative articles, which referred to economic research goals or conclusions in their title or abstract, but the supporting evidence was not provided in sufficient detail or methodological rigor to support those conclusions according to our judgment (eg, cost-effectiveness of an intervention was claimed without performing analysis or approximate calculations were provided in the discussion section, without stating the goals, describing the methods, or reporting the results of an economic analysis). We also excluded purely methodological studies, studies conducted on non-native populations (eg, army, migrant populations, or hajj pilgrims), reviews, conference proceedings, commentaries or any nonoriginal research articles, studies that did not involve human subjects (eg, infection vector control), or studies that were not economic evaluations (eg, resource use studies without reporting costs or surveys including cost-related questions but not measuring costs) or did not compare two alternative courses of intervention (eg, cost of illness studies). We considered as a comparison of alternatives if a single health technology was evaluated against no action (eg, savings were reported). The entire process of screening citations and judgment about inclusion was done in parallel by two independent researchers (ZZ, OR), and any differences about final results were resolved by discussion until joint agreement was reached. The selection of articles was performed in two stages. First, articles were screened for eligibility based on their title and abstract. After joint selection, potentially eligible full-text articles were evaluated in detail against all eligibility criteria.

A Microsoft Excel (Microsoft, Redmond, WA, USA)spreadsheet was developed for data extraction according to predefined criteria. Data extraction was performed in parallel by two independent researchers (ZZ, OR), all entries were cross-checked, and differences were resolved. From each publication, we extracted the first author, publication year, and the target country. We indicated if the study involved multiple countries within MENA or countries from other regions. Using the World Bank's classification,^[19] countries were grouped according to their economic status as high income, including Bahrain, Kuwait, Oman, Qatar, KSA, and UAE. Due to the fluctuations over the study period between low-income, lower middle-income, and upper middle-income categories, other countries were categorized as middle income. Studies including multiple countries from different income groups or geographies were categorized as mixed. We extracted the involved patient population and assigned the relevant ICD-10 (5th ed.) chapter and diagnosis code.^[20] In case of screening programs, we recorded the disease and not the screening activity. We also recorded the health technology and involved population for each study. Health technologies were categorized as medicine, device (including diagnostics), vaccine, procedure (eg, surgical techniques or other procedures applied on a single patient), and system (including screening, public health programs, health insurance, or innovations concerning the delivery of healthcare in an healthcare institution in a region or in the whole country). We recorded the type of evaluation according to the standard definitions given in Drummond et al^[21] Studies were labeled as full economic evaluations if both health outcomes and costs were involved: cost-utility (CUA), cost-effectiveness (CEA), cost-benefit (CBA), cost-minimization (CMA), and cost-consequence analyses (CCA). In particular, complex programs tabulating multiple outcomes and costs were categorized as CCA, and studies explicitly reporting the equality of outcomes were categorized as CMA. Budget impact analyses (BIA) and cost comparisons were categorized as partial economic evaluations, and we retained a category for efficacy studies reporting costs as well as other evaluations. For full economic evaluations, we recorded the health outcomes: quality-adjusted life years (QALY), disability-adjusted life years (DALY), specified natural outcomes, monetary outcomes for CBA, and various outcomes without detail for CCA. We also recorded the perspective of the analysis including the following categories: society, health system, healthcare institution, patient, and other. We also captured if the type of study was trial based or model based, using the trial-based category despite the involvement of models, if primary research was reported and included in model inputs. When assigning methodological categories, we relied on the reported information in the articles and made our consensus judgment independently from the categories used by the authors. We also extracted the costing year. If the costing year was not specified, for long-term studies we considered the first year of the study time horizon, and in cross-sectional studies, the final year of data collection. We reported “not available” if the manuscript did not refer to any study timeline, regardless of the actual publication date. Furthermore, we recorded the funding source of the study: industry, mixed with industry involvement, nongovernmental organization (NGO), intergovernmental organization (IGO), government, academic institution, no sponsor if explicitly stated and “not available” if the funding information was disclosed. The type and country of the corresponding author's affiliation was also extracted, using the same categories as for the funding source. We also added professional consulting among the corresponding author affiliation categories. Finally, for each publication, we searched for the most relevant journal subject area and ranking for the publication year in the Scimago database.^[22] Articles published before 1999 or journals not indexed in the database were categorized as “not available.” Studies published in 2019 received the latest available rankings from 2018. We grouped journal subjects into the following main groups: medicine, dentistry, public health, health policy, pharmaceutical sciences, and health informatics. We relied solely on the information provided in the publications during the data extraction, not using complementary sources (eg, contacting authors). As additional data extraction, we recorded whether the authors reported adherence to the Consolidated Health Economic Evaluation Reporting Standards (CHEERS),^[23] as well as the main methods of sensitivity analysis (SA) applied in the studies.^[24]

We used descriptive methods for the data analysis in our study. Cross-tabulated variables were analyzed via the Fischer's exact test, and considering the non-normal distribution of data, continuous variables were analyzed via nonparametric methods. The analysis was performed via Stata 14.2 statistical software.^[25]

The detailed PRISMA flowchart is displayed in Figure 1. Our search yielded 2017 hits, providing 1646 citations after using the English language and full-text filters. Screening by the titles and abstracts yielded 219 full-text papers. From those, 105 articles were included in the review. After evaluation against all predefined inclusion and exclusion criteria, full-text articles were excluded for the following reasons, in order (without double-counting if multiple exclusion criteria were met): (1) not original research on humans (n = 13), (2) the full text not English (n = 2), (3) not the local population studied (n = 7), (4) no health technology evaluated (n = 9), (5) the study not a health economic evaluation reporting costs (n = 6), and (7) no alternative technologies compared (n = 61). Furthermore, we observed during the screening that several articles made conclusions about cost-effectiveness without actually performing economic evaluation using a standard methodology (see examples above). We flagged these studies as speculative studies and found 29 of them among the 219 full-text papers evaluated for eligibility. (8) After evaluating against all preceding eligibility criteria, 15 studies were excluded solely for being speculative. (9) Finally, we excluded one duplicate study. From the 61 studies excluded due to lack of a comparator, 49 studies reported costs with proper methodology but did not compare alternatives. The remaining 12 studies were considered as speculative without a comparator. We found two purely methodological studies and six papers that were not original research articles (mainly review papers), which were excluded earlier for other reasons, hence were not listed in the PRISMA flowchart.

The details of study characteristics, including the description of the involved technology and target population as well as key methodological features and journal rank, were summarized in tabular form. For CUA studies in Supplementary Table S2, we indicated if the health outcome was QALY or DALY. For the remaining full economic evaluations in Supplementary Table S3, we provided the method (CEA, CBA, CCA, or CMA) and the health outcome. The other evaluation category was also included in this table. In Supplementary Table S4, we summarized partial economic evaluations, specifying the method (BIA or cost comparison). Finally, in Supplementary Table S5, details of efficacy studies reporting costs were provided.

The first study was published in 1989, which was followed by sporadic publication activity until 2009 (Fig. 2). Over the last 5 years, the number of publications increased rapidly. The majority of studies were published between 2015 and 2019 (n = 60; 57.1%), with peaks in 2015 (n = 15; 14.2%) and 2019 (n = 19; 18.1%). Eighty-nine (84.8%) studies focused on a single country, three (2.9%) studies involved multiple countries from MENA, and 13 (12.4%) studies were international, also involving countries from outside of MENA. The 105 studies reported 145 country-specific results (Fig. 3), mainly from middle-income countries (n = 91; 62.8%). Most studies reported results from Egypt (n = 32; 30.5%), KSA (n = 29; 27.6%), and Jordan (n = 10; 9.5%). Among the 89 single-country studies, the results were similar, with most focusing on Egypt (n = 26; 29.2%), KSA (n = 24; 27.0%), and Jordan (n = 8; 9.0%). We have not found single-country studies from Yemen and Lebanon.

We found studies from all ICD-10 chapters, with the exception of chapter XII, diseases of the skin and subcutaneous tissue (Fig. 4). The most frequent categories were chapter I, infectious diseases (n = 29; 27.6%), chapter IX, circulatory diseases (n = 8; 7.6%), chapter XI, digestive diseases (n = 8; 7.6%), and chapter XIV, genitourinary diseases (n = 8; 7.6%), followed by chapter II, neoplasms (n = 6; 5.7%), chapter XIX, injuries (n = 6; 5.7%), and chapter IV, endocrine diseases (n = 6; 5.7%). The disease area was not specified in 11 studies, where patient populations were characterized by the healthcare setting (n = 6; 5.7%) or the technology used (n = 4; 3.8%). One study assessed insurance schemes among schoolchildren. The most frequent indications were hepatitis C (n = 7; 6.7%), rotavirus enteritis (n = 5; 4.8%), and renal failure (n = 5; 4.8%), as well as diabetes mellitus (n = 4; 3.8%), cervical cancer (n = 3; 2.9%), and cesarean section (n = 3; 2.9%).

Although the main disease categories did not differ significantly between publications from the 1989–2014 and 2015–2019 periods (Fisher's exact test, p = 0.27), we noticed that somewhat more disease areas were covered in recent studies (n = 16) than in earlier ones (n = 13). Apart from diseases covered by a single study, chapter X, respiratory diseases (n = 4) and chapter VI, nervous system diseases (n = 2) were covered only between 2015 and 2019, while chapter XV, pregnancy was covered only between 1989 and 2014 (n = 4). Growth between the recent and early period was notable among studies in chapter I, infectious diseases (n = 17 vs 12) and chapter IX, circulatory diseases (n = 6 vs 2), while the number of studies with unspecified conditions decreased (n = 3 vs 8).

The main disease areas differed significantly between income groups (Fischer's exact test, p = 0.004). The most frequently covered areas were chapter XI, digestive diseases (n = 6; 15.4%), chapter I, infectious diseases (n = 6; 15.4%), and chapter IX, circulatory diseases (n = 5; 12.8%) in high-income countries; chapter I, infectious diseases (n = 20; 37.7%), chapter II, neoplasms (n = 5; 9.4%), chapter IV, genitourinary diseases (n = 4; 7.6%), and chapter XIX, injuries/external causes (n = 4; 7.6%) in middle-income countries; and chapter I, infectious diseases (n = 3; 23.1%), chapter IV, endocrine diseases (n = 3; 23.1%), and chapter IX, circulatory diseases (n = 3; 23.1%) in mixed-country studies. While majority of studies on chapter I, infectious diseases (69.0%) and chapter II neoplasms (83.3%) originated from middle-income countries, the majority of studies on chapter XI, digestive diseases (75.0%), chapter X, respiratory diseases (75.0%), and chapter IX, circulatory diseases (62.5%) focused on high-income countries.

Most studies focused on systems (n = 44; 41.9%), followed by medicines (n = 31; 29.5%), vaccines (n = 12; 11.4%), devices (n = 10; 9.5%), and procedures (n = 8; 7.6%). The main categories of health technology did not differ between recent and earlier studies (Fischer's exact test, p = 0.70) nor between income groups (Fischer's exact test, p = 0.73).

Within the systems category, we identified studies of infection control (n = 7; 6.7%) screening (n = 7; 6.7%), financial coverage/access (n = 6; 5.7%), hospital technology/management (n = 6; 5.7%), healthcare delivery (n = 5; 4.8%), drug policy (n = 4; 3.8%), as well as public health programs (n = 4; 3.8%), eHealth solutions (n = 3; 2.9%), and treatment strategies (n = 2; 1.9%). Within the medicines category, biologic drugs were most frequently studied (n = 7; 6.7%), followed by antibiotics (n = 4; 3.8%) and insulin (n = 3; 2.9%). Among vaccines, rotavirus vaccine (n = 5; 4.8%), human papillomavirus (HPV) vaccine (n = 3; 2.9%), and pneumococcus vaccine (n = 2; 1.9%), and among procedures, appendectomy (n = 2; 1.9%) and cesarean section (n = 2; 1.9%), were studied most frequently.

Details are depicted in Figure 5. Half of the studies (n = 52; 49.5%) were full economic evaluations. The applied method was CUA, CEA, CCA, CMA, and CBA in 29.5% (n = 31), 14.3% (n = 15), 3.8% (n = 4), 1.0% (n = 1), and 1.0% (n = 1) of studies, respectively. We found 29 (27.6%) partial economic evaluations involving 10 (9.5%) BIA studies and 19 (18.1%) cost comparisons. Also, we identified 22 (21.0%) efficacy studies reporting costs and three (2.9%) other methods: a multicriteria decision analysis (MCDA) study for the selection drugs on a formulary list,^[26] an econometric modeling of waterborne-disease incidence,^[27] and a modeling study of catastrophic costs associated with tuberculosis.^[28] The use of main methods (full, partial, and other evaluations and efficacy studies reporting costs) was significantly different between recent and early studies (Fischer's exact test, p = 0.04) and between country income groups (Fischer's exact test, p = 0.04). We observed an increased number of full economic evaluations in the 2015–2019 period (n = 35; 58.3%) compared to 1989–2013 (n = 17; 37.8%), mainly driven by the increase of QALY-based CUA studies from four (8.9%) to 19 (29.2%). Also, we observed a shift toward a greater proportion of BIA studies within partial economic evaluations from 20.0% (n = 3) to 46.2% (n = 6), although the overall number of partial evaluations decreased from 15 in the 1989–2014 period to 13 in the 2014–2015 period. While the proportion of full economic evaluations (n = 29; 54.7%) was greater in middle-income compared to high-income countries (n = 13; 33.3%), the proportion of QALY-based CUA studies was greater among high-income (n = 8; 61.5%) compared to middle-income countries (n = 11; 37.9%). DALY-based CUA studies were applied only in middle-income countries and mixed-country studies. The method in mixed-country studies was predominantly full economic evaluation (n = 10; 76.9%), using QALYs (n = 4; 30.8%), DALYs (n = 3; 23.1%), and natural outcomes (n = 3; 23.1%) in similar proportions. Among CEA studies, life years gained ([LYG] n = 5; 33.3%) and deaths avoided (n = 2; 13.3%) were the most frequently applied natural outcomes.

Health system perspective was applied in most studies (n = 55; 52.4%) followed by healthcare institution (n = 28; 26.7%), societal (n = 9; 8.6%), patient (n = 8; 7.6%), and other (n = 5; 4.8%) perspectives. Among the nine studies adopting a societal perspective, five were QALY-based CUA (4.8%), two were DALY-based CUA (1.9%), and two were CEA (1.9%) studies. The distribution of perspectives did not differ significantly between recent and early studies (Fischer's exact test, p = 0.27). However, we observed growth mainly in the number of studies applying the health system perspective (from n = 20; 44.4% to n = 35; 58.3%) and the patient perspective (from n = 2; 4.4% to n = 5; 10.0%) between the 1989–2014 and 2015–2019 periods. While studies in high-income countries applied equally the health system (n = 16; 41.0%) and healthcare institution perspectives (n = 16; 41.0%), health system perspective dominated in middle-income (n = 31; 58.5%) and mixed-country (n = 8; 61.5%) studies, followed by the healthcare institution perspective in middle-income country (n = 12; 22.6%) and societal perspective (n = 2; 15.4%) in mixed-county studies.

The studies were trial based in 56.2% (n = 59), model based in 41.9% (n = 44), and econometric in 1.9% (n = 2) of the sample. The study type did not differ between recent and early studies (Fischer's exact test, p = 0.28), but the difference was significant between country income groups (Fischer's exact test, p < 0.001). While all but one international mixed-country studies were model based (n = 12; 93.3%), studies were mainly trial based in high-income (n = 27; 69.2%) and middle-income countries (n = 31; 58.5%).

Five studies reported adherence to the CHEERS guidelines. SA was performed in 47 (44.8%) studies. The applied method was one-way SA in 38 (36.2%) studies, probabilistic SA in 14 studies (13.3%), and multiway SA in 12 studies (11.4%). The multiway SA was a scenario analysis in eight studies (7.6%), two-way SA in three studies (2.8%), and extreme value analysis in one study. In 14 (13.3%) studies, more than one SA method was conducted.

Overall, we observed a great variation in the applied methodology among the studies. When considering the type of study, the method of economic evaluation, the health outcome, and the costing perspective, we identified 34 different categories. The most frequent methods were efficacy studies reporting costs from a healthcare institution's perspective (n = 17; 16.2%), model-based CUA studies using QALYs and reporting costs from a health system perspective (n = 9; 8.6%), as well as trial-based cost-comparison studies (n = 8; 7.6%) and model-based BIA studies (n = 8; 7.6%) using a health system perspective.

The average time lag between the costing year and publication year was 2.6 years (SD 1.6; range 0–7). Although the mean time lag decreased from 2.9 (SD 1.6) to 2.4 (SD 1.6) years 1989–2014 and 2014–2015, the difference was not significant (Mann-Whitney U test with ties, p = 0.11). The mean time lag of studies from high-income, middle-income, and mixed-country studies was 2.4 (SD 1.6), 2.7 (SD1.6), and 3.3 (SD 1.8) years, respectively. The difference was not significant (Kruskal-Wallis test with ties, p = 0.14).

The funding source was not indicated in 31.4% (n = 33) of the studies, funding was stated as none in 22.9% (n = 24), and the sponsor was industry in 17.1% (n = 18), government in 12.4% (n = 13), IGO in 5.7% (n = 6), academic institution in 4.8% (n = 5), NGO in 4.8% (n = 5), and mixed without industry in 1.0% (n = 1) of the studies. The difference in terms of funding source was significant between recent and early studies (Fischer's exact test, p = 0.003). The proportion of undisclosed funding decreased from 42.2% (n = 19) to 23.3% (n = 14), while “funding source: none” statements increased from 11.1% (n = 5) to 31.7% (n = 19) between 1989–2014 and 2015–2019. While government-funded studies became less frequent (from n = 9; 20.0% to n = 4; 6.7%), the number of industry-funded and NGO-funded studies increased from 11.1% (n = 5) to 21.7% (n = 13) and from 0.0% (n = 0) to 8.3% (n = 5), respectively. The funding source did not differ significantly by income group (Fischer's exact test, p = 0.06). However, the proportion of industry funding was greatest among mixed-country studies (n = 4; 3.8%) and lowest among studies of middle-income countries (n = 7; 13.2%). The funding source was stated least frequently in studies of high-income countries (n = 22; 56.4%), and most frequently in mixed-country studies (n = 12; 92.3%).

The affiliation of the corresponding author was an academic institution, healthcare institution, government, industry, and professional consulting in 59 (56.2%), 18 (17.1%), 17 (16.2%), 4 (3.8%), and 2 (1.9%) studies, respectively. There were more corresponding authors from healthcare institutions (n = 14; 23.3% vs n = 4; 8.9%) and fewer from governments (n = 5; 8.3% vs n = 12; 26.7%) in 2015–2019 compared to 1989–2014. The difference was significant (Fischer's exact test, p = 0.016). The affiliation of corresponding authors was also different by income groups (Fischer's exact test, p = 0.001). In high-income countries, corresponding authors were affiliated to academic institutions, healthcare institutions, and governments in 41.0% (n = 16), 35.9% (n = 14), and 12.8% (n = 5) of studies; in middle-income countries in 62.3% (n = 33), 3.8% (n = 2), and 22.6% (n = 12) of studies; and in mixed-country studies in 76.9% (n = 10), 15.4% (n = 2), and 0.0% (n = 0), respectively.

The institution of the corresponding author was from MENA, Europe, United States, Asia, and Australia in 61.9% (n = 65), 18.1% (n = 19), 16.2% (n = 17), 1.9% (n = 2), and 1.9% (n = 2) of studies, respectively. From the MENA region, the affiliated institution of corresponding author was most frequently from KSA (n = 22; 33.9%), Egypt (n = 15; 23.1%), UAE (n = 6; 9.2%), and Jordan (n = 7.7%). We did not find corresponding authors from Algeria, Lebanon, Libya, or Syria. Although we did not find differences between recent and early studies in terms of the country of the corresponding author's institution (Fischer's exact test, p = 0.52), the difference was marked between income groups (Fischer's exact test, p < 0.001). While in 94.9% of studies from high-income countries, the corresponding author was from a MENA country, in the case of middle-income countries, the corresponding author was from MENA, Europe, and the United States in 52.8% (n = 28), 22.6% (n = 12), and 22.6% (n = 12), respectively, and in mixed-country studies from MENA, Europe, and the United States in 0.0% (n = 0), 46.2% (n = 6), and 30.8% (n = 4) of studies, respectively.

The journal subject area was medicine for most studies (n = 54; 51.4%), followed by health policy (n = 18; 17.1%), public health (n = 15; 14.3%), pharmaceutical sciences (n = 12; 11.4%), dentistry (n = 2; 1.9%), and health informatics (n = 1; 1.0%). The journal subject area was not available in the Scimago database for three studies. The journal subject area did not differ between early and recent studies (Fischer's exact test, p = 0.73) and country income groups (Fischer's exact test, p = 0.08). The Scimago journal rank was Q1 in 49 (46.7%), Q2 in 25 (23.8%), Q3 in 20 (19.1%), Q4 in 1 (1.0%), and not available in 10 (9.5%) publications (Fig. 6). The journal rank distribution did not differ significantly between early and recent studies (Fischer's exact test, p = 0.06). However, the proportion of Q1 publications was as high as 92.3% (n = 12) in mixed-country studies, while it was 43.4% in studies from middle-income and 35.9% (n = 14) in studies from high-income countries. The difference by income group was significant (Fischer's exact test, p = 0.03) The proportion of Q1 publications among full economic evaluations, partial economic evaluations, efficacy, and cost studies and other evaluations was 59.6% (n = 31), 39.3% (n = 11), 22.7% (n = 5), and 66.7% (n = 2). Q3 publications were most frequent among studies of partial economic evaluations (n = 7; 25.0%) and efficacy studies reporting costs (n = 7; 31.8%), suggesting that full evaluations had greater chance for publication in leading journals.

According to our best knowledge, our scoping review is the most comprehensive account of health economic publications from the MENA region. We analyzed the target countries, diseases, technologies, and methods, as well as bibliometric properties, of 105 health economic studies published in MENA up to 2019. We found a rapid growth in the number of publications since 2015, with most studies published over the last 5 years, suggesting that the importance of economic evaluations is increasing in the region. Including multicountry studies, we found 145 country-specific results, out of which two thirds were from middle-income countries. As single countries, Egypt and KSA were involved most frequently. The distribution of health economic publications among the countries in our study was similar to the magnitude of overall scientific activity as reported by UNESCO, led by KSA and Egypt and followed by Tunisia, Algeria, and Morocco, with the notable exception of Jordan, which ranked third in our study, while ranked only eighth within MENA in terms of overall research output.^[29]

Infectious diseases followed by circulatory, digestive, and genitourinary diseases were the most frequently studied areas, and systems including infection control, screening, and various complex programs concerning the financing and delivery of healthcare were the most frequently studied technologies. Although, in addition to ischemic heart disease and stroke, back pain and depression were among the top five causes of disease burden in the region in 2010,^[30] musculoskeletal and mental diseases were, respectively, at only 12th and 10th place among the studied disease areas, suggesting that health economic evaluations followed only partially the public health priorities of the region.

Half of the studies were full economic evaluations, and a health system perspective was adopted most frequently. The number of QALY-based CUA studies increased rapidly over the last 5 years, which was probably driven by the availability of locally adapted health utility measurement tools, such as the EQ-5D.^[31,32] However, the lack local value sets for major generic health-related quality of life instruments (EQ-5D, SF-6D, and HUI)^[33–35] may impede the progress of state-of-the-art health economic research in the region. Furthermore, CUA studies using societal perspective were scarce, which is the recommended approach for economic evaluation by the WHO.^[36] The paucity of studies taking a societal perspective may be explained by the lack of data. PubMed searches for common terms concerning the societal impact of diseases, such as “productivity loss” or “informal care,” provided six and eight citations from the 17 countries, respectively. The high share of trial-based studies also suggests that primary data collection was frequently necessary for performing economic evaluations. However, the majority of studies were performed without funding, which may be a barrier to collecting the breadth of data that reflects all relevant costs and consequences of interventions. We observed a small and decreasing number of studies funded by governments and a modest participation of the industry as a sponsor. Although the reporting of funding improved considerably over time, a quarter of the studies did not report the funding source over the last 5 years.

The corresponding author was affiliated to an institution from the region in the majority of studies and in nearly all studies from high-income countries. Half of the studies were published in Scimago Q1 journals, suggesting that considerable professional expertise is available locally in many countries.

Although our study involved most countries and the largest number of studies from MENA, health economic studies from the region have previously been reviewed in high-quality systematic reviews. Al-Aqeel et al^[37] identified 15 health economic studies in KSA up to May 2011, Eljilany et al^[4] reviewed 49 studies from countries of the Gulf Cooperation Council (Bahrain, KSA, Kuwait, Oman, Qatar, and UAE) until the end of 2017, and Farid et al^[5] reviewed 15 studies from Egypt published between 2002 and 2017. While we included only studies that compared alternatives, previous reviews also included studies describing a single technology. Farid et al^[5] excluded studies that compared multiple countries, while we excluded studies involving nonlocal populations. Although the eligibility criteria were different, the comparison with previous reviews revealed that potentially eligible articles were not identified during our search, which is a limitation of our study. The three reviews from the region included a total of 78 publications, out of which 51 were not included in our study. Out of them, eight were identified by our search but were not selected during our screening process. A secondary check revealed that two of those studies would be eligible for our review. Fifteen studies were identified during our screening but did not meet our inclusion criteria. Twenty-six studies were not identified by our search. Nevertheless, 27 studies overlapped, and 23 studies were included in our review, which were potentially eligible yet not included in previous reviews. Overall, from 535, 3074, and 1950 citations, Al-Aqeel et al,^[37] Eljilany et al,^[4] and Farid et al^[5] identified 15 (2.8%), 49 (1.6%), and 15 (0.7%) eligible studies, to which our review could potentially add 4, 11, and 12 more. Our study yielded 105 eligible articles from 1646 citations (6.4%), while it missed at least 26 studies, suggesting that the sensitivity of our search strategy needs to be further developed. Nevertheless, search strategies with reduced sensitivity and greater precision may be acceptable for scoping reviews.^[15] Although the PubMed database has been incomplete in searching for health economic evaluations,^[38] out of the 105 included and 51 missing articles, only one was published in a journal that was not indexed in PubMed. Taken together, the completeness of our study was similar to previous reviews from the region. The strength of our study is that we applied consistent methodology across the region, allowing the representation of health economic studies using the same eligibility criteria across 17 high- and middle-income countries during the entire study period. Similar reviews covering a full scope of health economic evaluations across an entire geographic region are scarce. Pitt et al^[39] performed a global bibliometric analysis of full health economic evaluations published between January 2012 and May 2014, including 2844 studies. In the studied period, the MENA region provided 2% of the global publication output. Augustovski et al^[40] reviewed 72 full economic evaluations published between 1980 and 2004, involving countries of the Latin American and Caribbean region. Valencia-Mendoza et al^[41] included health economic evaluations of selected public health interventions proposed by the Mesoamerican Health Initiative. The review included 92 economic evaluations that were published between 2000 and 2009 involving DALYs or deaths averted as outcomes. A systematic review by Mandrik et al^[42] evaluated the transferability of 34 full health economic evaluations published between 2008 and 2013 on countries of Central and Eastern Europe and the former Soviet Union (CEE/FSU). Other geographically oriented systematic reviews of health economic evaluations generally focused on a single country or a single disease area or technology involving multiple countries.^[39,43]

Although our review is unique in terms of its breadth of geographic and methodological coverage, it has several limitations, which are highlighted as follows. In a way similar to other reviews from MENA or other regions,^{[4,5,37,39,42]} we limited our search to peer-reviewed journals; therefore, potential sources of health economic publications, such as conference posters,^[44] were not covered by our study. Two regional reviews searched for the gray literature,^[40,41,45] and one searched the citation lists of past publications,^[41] which were omitted from our search strategy. Furthermore, in a way similar to three other reviews from MENA and the CEE/FSU,^[5,37,42] we excluded articles in which the full text was not available in English, which was not reported among exclusion criteria in other regional reviews.^[4,40,41] However, due to the last 10 years' efforts to improve the international scientific impact of the region,^[29] focusing on English language publications most probably provided a good representation of the health economic research activity in MENA. This assumption was supported by the finding of Pitt et al,^[39] who did not report the use of Arabic or French publications among health economic studies published over a 27-month period worldwide in languages other than English.

We observed a rapid growth of publications over the last 5 years in MENA, providing input to national, regional, or institutional level value-based health policy and financing decisions. The studies showed a great heterogeneity in the terms of disease areas, technologies, and applied methods, which made the comparison and interpretation of results very difficult. Although CUA studies became prevalent, the societal perspective has been adapted only occasionally. For the future progress of health economic research in the region, in addition to the development of the institutional background, valid and standardized local cost and outcome data should be available to a greater extent.

Supplemental material is available with the article online.