Introduction
Lebanon, an East Mediterranean country, has around 5.3 million residents (Reference Bakhach and Doghman1). Despite being a small country, the Lebanese healthcare system is highly fragmented (2). The multiple public and semi-public payers offer healthcare benefit packages that differ significantly in the included benefits (2), with consequences of horizontal inequity (Reference Tao, Henry, Zou and Zhong3). (Reference Tao, Henry, Zou and Zhong3)This stems partly from the absence of an explicit and systematic economic assessment (4), leading to different perceived economic value of health technologies (Reference Tao, Henry, Zou and Zhong3). There is a growing need for a systematic approach to assess the value of health technologies; this has been recognized by the Lebanese Ministry of Public Health (MoPH) through its commitment to rely on healthcare Economic Evaluations (EEs) to optimize its healthcare investment decisions (5).
Moreover, this commitment comes as Lebanon has been going through an unprecedented and multifaceted crisis since 2019, disrupting the healthcare sector at large by limiting the health system’s dedicated resources (2;Reference Dahham, Kremer and Hiligsmann6). Consequently, Lebanon needs to favor healthcare investment decisions holding good value-for-money: economic evaluations (EEs) serve as a tool for assessing the economic value of different health technologies by comparing their incremental costs and outcomes (Reference Evers, Goossens, de Vet, van Tulder and Ament7).
However, to date, no national explicit health technology assessment framework, requiring the submission of EEs exists (4;5), and few budget holders require the submission of a partial or full EEs (4). Moreover, the submitted EEs often remain unpublished, escaping the possibility of appropriate scrutiny of their methods and results. On the other hand, there is no study yet that identifies published EEs or the methodological quality of published EEs in Lebanon. This is particularly important, especially in the lack of national health economics expertise and the previous absence of national methodological and reporting guidelines (Reference Daccache, Rizk and Dahham8;Reference Gheorghe, Gad, Ismail and Chalkidou9).
While the MoPH has committed to relying on EEs to assess the value of interventions and inform decisions, evidence is needed regarding the quality of the current practice. Noting that EEs inform decision making only when it is of high methodological quality, we aim to systematically review and critically appraise the published EEs in Lebanon to identify current practice and inform future evaluations with the goal of reducing the methodological standardization gap within the Lebanese context.
Methods
Review design
We followed the guidance for economic evidence review and synthesis (Reference Higgins, Thomas and Chandler10). The reporting of the literature searching component followed the Preferred Reporting Items for Systematic reviews and Meta-Analyses literature search extension (PRISMA-S), and that of the systematic review followed the PRISMA statement (Reference Rethlefsen, Kirtley and Waffenschmidt11). Ethical approval was not required.
Study selection criteria
We included original articles; full EEs comparing the costs and consequences of two or more health-related interventions using cost–benefit, cost-effectiveness, or cost-utility analyses, or cost-minimization analyses (CMAs), and studies conducted in Lebanon irrespective of the language and the date of publication. Partial EEs, with the exception of CMAs (budget impact analysis, cost of illness, etc.) and other types of publications (posters, editorials, letters, and abstracts) were excluded.
Search strategy
We searched the NHS Economic Evaluation Database (NHSEED)-database ceased in 2015, MEDLINE via Ovid, Embase via Ovid, and the Cumulative Index to Nursing and Allied Health Literature (CINAHL) via EBSCO, using two concepts: EEs and Lebanon. For the first concept, we employed the validated NHSEED search filter, which has a relatively high sensitivity (0.999) and a relatively fair precision (0.040). For the concept of Lebanon, we conducted a search for the truncated free-text word “Leban*,” and the controlled vocabulary term “Lebanon.” The electronic search (Supplementary Material 1) was validated and executed by an information specialist; we searched each database from inception, and the search was conducted on the 23 October 2021 and updated on 23 July 2025. Bibliographies of included articles and relevant reviews were hand-searched for papers missed by database searches.
Study selection
Records were exported to Endnote 20.1. Duplicates were removed automatically, then manually (WM). Titles and abstracts of the records were screened for eligibility by two authors, independently and in duplicate (WM/RR). The full texts of potentially eligible studies were retrieved and assessed for inclusion by pairs of authors, independently and in duplicate (WM/RR and WM/DB). Included studies were original, English-language, full economic evaluations, and conducted in Lebanon.
Data abstraction
Data were abstracted by pairs of authors, independently and in duplicate (WM/RR and WM/DB), using a data abstraction form developed by the researchers (Supplementary Material 2), detailing the research team composition, shedding light on the local capacity in health economics; the objective of the study and the type of assessed technologies; the methodology, that is, adopted perspective, analytical framework, identification, measurement, and valuation of costs and outcomes; the presentation of the results; sensitivity analyses; and target audience of the evaluation.
Critical appraisal
Critical appraisal was performed by pairs of authors, independently and in duplicate (WM/CD and WM/RK) using the validated and widely used Consensus on Health Economic Criteria (CHEC) for trial-based EEs and the CHEC-Extended for model-based EEs (Reference Evers, Goossens, de Vet, van Tulder and Ament7;Reference Watts and Li12).
Data synthesis
A narrative tabulation and synthesis of the characteristics, methods, and results of the studies was performed, with a set of recommendations based on the methodological gaps identified.
Results
We identified 1109 records, out of which eight studies were included (Figure 1).
PRISMA flow chart of study selection. NHS EED, National health service economic evaluation database; CINAHL, Cumulative index to nursing and allied health literature.
Characteristics of included studies
The studies were published 2012–2024. Six studies included non-Lebanese researchers (Reference El Karak, Rassy, Bassil, Awali, Hanna and Saroufim13–Reference Sharma, Seoud and Kim17), one included only Lebanese researchers (Reference Nasser, Mansour, Abi Nader and Metni18), and one had no Lebanese researchers (Reference Cameron, Mantel-Teeuwisse, Leufkens and Laing19). Five studies entailed model-based EEs (Reference El Karak, Rassy, Bassil, Awali, Hanna and Saroufim13–Reference Matni, Yamout, Koussa, Khamis, Fleifel and Sharifi15;Reference Sharma, Seoud and Kim17;Reference Nasser, Mansour, Abi Nader and Metni18), two were trial-based EEs (Reference Rizk, Hiligsmann, Karavetian and Evers16), and one was a CMA (Reference Cameron, Mantel-Teeuwisse, Leufkens and Laing19).
Methodologies of included studies
Model-based economic evaluations
The perspectives were unclear in two studies, but assumed to be institutional given included costs and outcomes (Reference El Karak, Rassy, Bassil, Awali, Hanna and Saroufim13;Reference Nasser, Mansour, Abi Nader and Metni18), the National Social Security’s (Reference Matni, Yamout, Koussa, Khamis, Fleifel and Sharifi15), and the societal perspective (Reference Kim, Sharma, O’Shea, Sweet, Diaz and Sancho-Garnier14;Reference Sharma, Seoud and Kim17). All studies adopted a lifetime horizon and a bottom-up costing approach. Outcomes, discounting practices and rates, and presentation of results varied. All studies conducted sensitivity analyses: deterministic (Reference Nasser, Mansour, Abi Nader and Metni18), deterministic and probabilistic (Reference El Karak, Rassy, Bassil, Awali, Hanna and Saroufim13;Reference Matni, Yamout, Koussa, Khamis, Fleifel and Sharifi15), scenario and deterministic (Reference Sharma, Seoud and Kim17), and scenario, deterministic, and probabilistic analyses (Reference Kim, Sharma, O’Shea, Sweet, Diaz and Sancho-Garnier14) (Table 1).
Methodological characteristics of the included model-based economic evaluations (n = 5)
Note: CUA, cost-utility analysis; CEA, cost-effectiveness analysis; QALY, quality-adjusted life-year; DALY, disability-adjusted life-year; LYS, life-years saved; NA, not applicable; LBP, Lebanese pounds; GDP, gross domestic product; WTP, willingness-to-pay.
Trial-based economic evaluation
Both trial-based EEs were led by Lebanese researchers (Reference Rizk, Hiligsmann, Karavetian and Evers16;Reference Abi Hana, Abi Ramia, Burchert, Carswell, Cuijpers and Heim20). One study adopted a societal perspective and looked at the cost-effectiveness of nutrition education by dedicated dietitians in patients undergoing hemodialysis; it entailed CEA, with phosphorus as outcome, and CUA, with QALYs as outcome, with a bottom-up costing approach. The time horizon was 1 year, and the WTP thresholds varied between three times the GDP and thresholds from the United Kingdom and United States. Sensitivity analyses and nonparametric bootstrapping were conducted (Reference Rizk, Hiligsmann, Karavetian and Evers16). The other trial-based EEs adopted two perspectives: societal and healthcare. It assessed the cost-effectiveness of an e-mental health solution, and measured disease-specific outcomes rather than QALYs. The time horizon was 5 months (Reference Abi Hana, Abi Ramia, Burchert, Carswell, Cuijpers and Heim20) with a WTP threshold of 220 USD for a disease-specific response, and a threshold of 840 USD for a remission, when adopting a healthcare perspective (Reference Abi Hana, Abi Ramia, Burchert, Carswell, Cuijpers and Heim20;Reference Kroenke, Spitzer and Williams21).
Cost-minimization analysis
This CMA looked at savings from substitution of originator brands by their equivalent lowest-priced generic in low-to-middle-income countries (LMICs), including Lebanon (Reference Cameron, Mantel-Teeuwisse, Leufkens and Laing19). The savings were captured from the perspective of patients and private reimbursement agencies. Savings were calculated over 1 year, using the 2009 sales volumes from IMS Health, Inc.
Findings of included studies
Sharma et al. suggested that an increase in cervical cancer screening to 50 percent while extending the intervals provides greater health benefits, at a reasonable cost, and a more equitable health gains distribution (Reference Sharma, Seoud and Kim17). Kim et al. found that the HPV vaccination was cost-effective compared with no intervention. However, efficiency diminished when the vaccination cost increased (Reference Kim, Sharma, O’Shea, Sweet, Diaz and Sancho-Garnier14). Karak et al. reported oncotype DX, a diagnostic tool, among intermediate risk patients as cost-effective (Reference El Karak, Rassy, Bassil, Awali, Hanna and Saroufim13). Nasser et al. showed early treatment strategy with DAAs as cost-effective (Reference Nasser, Mansour, Abi Nader and Metni18). Matni et al. indicated the dominance of cladribine over alemtuzumab, fingolimod, and natalizumab (Reference Matni, Yamout, Koussa, Khamis, Fleifel and Sharifi15). Rizk et al. concluded that nutrition education is likely to be cost-saving, but unlikely to be cost-effective (Reference Rizk, Hiligsmann, Karavetian and Evers16). Abi Hanna et al. concluded that an e-mental health solution holds an 80 percent chance of being cost-effective for a 220 USD WTP per disease-specific response according to PHQ9 (Reference Abi Hana, Abi Ramia, Burchert, Carswell, Cuijpers and Heim20;Reference Kroenke, Spitzer and Williams21). Cameron et al. estimated substantial savings of 2008 I$4,035,033 from a hypothetical switch from brand originators to the equivalent lowest cost generics (Reference Cameron, Mantel-Teeuwisse, Leufkens and Laing19).
Critical appraisal of included studies
Study population, competing alternatives, and research question
All EEs defined their population clearly. However, whether the study’s population is consistent with the population in Lebanon was not clear unanimously. For example, Matni et al. validated the population baseline characteristics with local experts (Reference Matni, Yamout, Koussa, Khamis, Fleifel and Sharifi15), whereas Karak et al. adjusted some probabilities to reflect recent Lebanese data (Reference El Karak, Rassy, Bassil, Awali, Hanna and Saroufim13). All EEs clearly described the competing alternatives. However, in the EEs by Nasser et al., the comparator regimen was unclear (Reference Nasser, Mansour, Abi Nader and Metni18). Finally, all EEs posed a well-defined research question in an answerable form (Table 2).
Critical appraisal of included studies (n = 8)
Appropriateness of the economic study design
The study design was appropriate to study objectives, except for the trial-based EEs, where the time horizon was not long enough to capture all costs and outcomes (Reference Rizk, Hiligsmann, Karavetian and Evers16;Reference Abi Hana, Abi Ramia, Burchert, Carswell, Cuijpers and Heim20). All EEs considered both costs and outcomes, except in Cameroon et al., assuming similar outcomes across compared technologies (Reference Cameron, Mantel-Teeuwisse, Leufkens and Laing19).
Model structural assumptions and validation methods (evidence underpinning the model, its quality, its uncertainty, and face, performance, and outcomes validity)
The model should ensure a reasonable representation of the real-life case, should perform correctly and consistently, and results generated should be realistic (Reference Hammerschmidt, Goertz and Wagenpfeil22).
Structural assumptions
Four studies relied on preexisting models (Reference El Karak, Rassy, Bassil, Awali, Hanna and Saroufim13–Reference Matni, Yamout, Koussa, Khamis, Fleifel and Sharifi15;Reference Sharma, Seoud and Kim17). One study did not indicate if the model was adopted, but that it followed the 1-year natural disease course (Reference Nasser, Mansour, Abi Nader and Metni18). Information about the model was not presented clearly in all studies. In Kim et al. and Sharma et al., the authors referenced the studies that created and validated the models, and studies that utilized the model (Reference Kim, Sharma, O’Shea, Sweet, Diaz and Sancho-Garnier14;Reference Sharma, Seoud and Kim17): the structural validity of the model could thus be verified by back-tracing. Karak et al. mentioned a study where the model was utilized, but did not reference where the model was first created and validated (Reference El Karak, Rassy, Bassil, Awali, Hanna and Saroufim13;Reference Paulden, Franek and Pham23). Accordingly, back-tracing the original model to ensure structural validity becomes challenging. In contrast, Matni et al. clearly detailed the model (Reference Matni, Yamout, Koussa, Khamis, Fleifel and Sharifi15) and indicated the study from which it was adopted and the source of clinical data that supported the hypothesis underpinning the model (Reference Matni, Yamout, Koussa, Khamis, Fleifel and Sharifi15;Reference Hettle, Harty and Wong24). Nasser et al. mentioned that the model represented the natural course of the disease, without detailing the structural validity of the model; it was unclear whether it was adopted (Reference Nasser, Mansour, Abi Nader and Metni18).
Face validity
The four studies that relied on previously utilized models were representative of the real-life problem (Reference El Karak, Rassy, Bassil, Awali, Hanna and Saroufim13–Reference Matni, Yamout, Koussa, Khamis, Fleifel and Sharifi15;Reference Sharma, Seoud and Kim17). While Nasser et al. mentioned that the model represents the natural course of the disease, earlier stages of the disease were not presented; hence, the comparison of costs and benefits for early versus delayed treatment has two different starting points (Reference Nasser, Mansour, Abi Nader and Metni18). The model did not accurately compare alternative disease courses: a delayed treatment of hepatitis C patients with DAAs might imply an early treatment with lower-cost therapies, including Pegylated interferon, or no therapy. Instead, the Markov model implied four health states, and despite the authors listing annual transition probabilities collected from literature, transitions between health states were not reflected in the model. Moreover, the uncertainty around assumptions was not clearly addressed, and the length of the 1-year model cycle was unjustified. Finally, the model was not tested for structure, performance, and outcomes validity (Reference Nasser, Mansour, Abi Nader and Metni18).
Performance and outcomes validity
All models adopted from previous EEs were tested for performance and outcomes validity in the original studies (Reference El Karak, Rassy, Bassil, Awali, Hanna and Saroufim13–Reference Matni, Yamout, Koussa, Khamis, Fleifel and Sharifi15;Reference Sharma, Seoud and Kim17). Nasser et al. did not detail performance or outcomes validity (Reference Nasser, Mansour, Abi Nader and Metni18).
Time horizon
The CMA estimated the annual savings generated by the switch to generic medicines, relying on 2008 prices, over one budget year (Reference Cameron, Mantel-Teeuwisse, Leufkens and Laing19). The remaining EEs considered lifetime costs and outcomes (Reference El Karak, Rassy, Bassil, Awali, Hanna and Saroufim13–Reference Matni, Yamout, Koussa, Khamis, Fleifel and Sharifi15;Reference Sharma, Seoud and Kim17;Reference Nasser, Mansour, Abi Nader and Metni18), except the trial-based EEs, adopting a 1-year time horizon, which is inappropriate to capture the full extent of outcomes (Reference Rizk, Hiligsmann, Karavetian and Evers16).
Perspective of the studies
Various perspectives were adopted, including the societal perspective (Reference Kim, Sharma, O’Shea, Sweet, Diaz and Sancho-Garnier14;Reference Rizk, Hiligsmann, Karavetian and Evers16;Reference Sharma, Seoud and Kim17). Cameroon et al. adopted the perspective of the private sector, with data deriving from IMS Health Inc. (Reference Cameron, Mantel-Teeuwisse, Leufkens and Laing19); being a multicounty study, no segregation of the private sector was reported. In Karak et al., the adopted “Lebanese perspective” was unclear (Reference El Karak, Rassy, Bassil, Awali, Hanna and Saroufim13). Nasser et al. adopted a “Lebanese third party payers” perspective, without specifying it (Reference Nasser, Mansour, Abi Nader and Metni18). In these two studies, included costs and outcomes suggest an institutional perspective, despite not clearly reporting this. Matni et al. adopted the perspective of the National Social Security Fund (NSSF) (Reference Matni, Yamout, Koussa, Khamis, Fleifel and Sharifi15).
Costs identification, measurement, and valuation
The quality of the costing exercise differed between studies. In Karak et al., identification and measurement of resources in natural units were unclear. Also, the valuation relied on the units’ cost in Lebanon for some of the resources, and in Canada for others (Reference El Karak, Rassy, Bassil, Awali, Hanna and Saroufim13). For the valuation, Karak et al. and Rizk et al. relied on exchange rates and not purchasing power parity to transfer costs between currencies (Reference El Karak, Rassy, Bassil, Awali, Hanna and Saroufim13;Reference Rizk, Hiligsmann, Karavetian and Evers16). Moreover, Nasser et al. did not specify the years in which prices were collected (Reference Nasser, Mansour, Abi Nader and Metni18). Matni et al. conducted the valuation based on the prices paid by the NSSF, with indication of exact years (Reference Nasser, Mansour, Abi Nader and Metni18). Matni et al., Sharma et al., Kim et al., and Rizk et al. detailed the costing exercise (Reference Kim, Sharma, O’Shea, Sweet, Diaz and Sancho-Garnier14–Reference Sharma, Seoud and Kim17). The latter detailed identification, measurement, and valuation steps (Reference Rizk, Hiligsmann, Karavetian and Evers16). Abi Hana et al. specified the cost components of providing the e-mental health solution, and limited the cost of the comparator, enhanced usual care, to the cost of the message delivered to patients as a prompt to seek usual care. The consequent healthcare cost for the intervention and comparator was not included, despite its importance (Reference Abi Hana, Abi Ramia, Burchert, Carswell, Cuijpers and Heim20).
Outcomes identification, measurement, and valuation
Karak et al., Sharma et al., Kim et al., and Matni et al. retrieved outcomes data from clinical trials (Reference El Karak, Rassy, Bassil, Awali, Hanna and Saroufim13–Reference Matni, Yamout, Koussa, Khamis, Fleifel and Sharifi15;Reference Sharma, Seoud and Kim17). Information about the methods used was provided in the referenced studies of the original models and Supplementary Materials. Although Nasser al. mentioned that DDAs have an efficacy rate between 85 and 99 percent, a 90 percent efficacy rate was assumed for early- and late-stage hepatitis C patients treated with DAAs, without a systematic approach for the assumption (Reference Nasser, Mansour, Abi Nader and Metni18). As for Rizk et al., the study relied on effectiveness data drawn from the Nutritional Education for Management of Osteodystrophy trial conducted on Lebanese patients (Reference Rizk, Hiligsmann, Karavetian and Evers16). Abi Hanna et al. relied on disease-specific outcome measures collected in the clinical trials. Researchers defined a score for treatment response and remission (Reference Cuijpers, Heim and Ramia25;Reference Cuijpers, Heim and Ramia26). Finally, Cameron et al. considered that generics and originator brands are interchangeable regarding efficacy and safety, once therapeutic equivalence is proved (Reference Cameron, Mantel-Teeuwisse, Leufkens and Laing19).
Appropriate incremental analysis and discounting of costs and outcomes
The increments in the health benefit and cost were measured in all EEs, except Cameroon et al., where outcomes were considered similar (Reference Cameron, Mantel-Teeuwisse, Leufkens and Laing19). Matni et al. reported the incremental cost-effectiveness ratios despite the cladribine intervention being dominant (Reference Matni, Yamout, Koussa, Khamis, Fleifel and Sharifi15). Noting that EE guidelines did not exist for Lebanon, Nasser et al. did not adopt any discounting (Reference Nasser, Mansour, Abi Nader and Metni18), while four model-based studies discounted costs and outcomes at different rates between 1.5 and 3.5 percent (Reference El Karak, Rassy, Bassil, Awali, Hanna and Saroufim13–Reference Matni, Yamout, Koussa, Khamis, Fleifel and Sharifi15;Reference Sharma, Seoud and Kim17).
Uncertainty and sensitivity analysis
Karak et al. conducted a deterministic sensitivity analysis by lowering the discount rate on costs and outcomes from 5 to 1.5 percent annually. No other uncertainties were addressed (Reference El Karak, Rassy, Bassil, Awali, Hanna and Saroufim13). Sharma et al. conducted a scenario analysis scaling up the coverage to 70 percent, considering a random subset of women from the population, versus the same women screened year over year, throughout their lifetime, in the baseline assumption (Reference Sharma, Seoud and Kim17). Concerning parameter uncertainty, all studies conducted at least deterministic sensitivity analyses, for example, Nasser et al. conducted a sensitivity analysis examining the effect of reducing the dual therapy cost by 25 percent (Reference Nasser, Mansour, Abi Nader and Metni18), while Matni et al. exerted a 50 percent variation on inputs that have a significant impact on the results, including the effect of DMD on disease progression, relapse rate, discontinuation rates, and discounting rates (Reference Matni, Yamout, Koussa, Khamis, Fleifel and Sharifi15). Moreover, Sharma et al. considered different scenarios based on varying cost per vaccinated girl (Reference Sharma, Seoud and Kim17). Three studies conducted probabilistic sensitivity analyses, addressing parameter uncertainty (Reference El Karak, Rassy, Bassil, Awali, Hanna and Saroufim13–Reference Matni, Yamout, Koussa, Khamis, Fleifel and Sharifi15). Abi Hanna et al. relied on multiple imputation with chained equations (MICEs) to address the uncertainty around missing data; no bootstrapping was conducted (Reference Abi Hana, Abi Ramia, Burchert, Carswell, Cuijpers and Heim20).
Conclusions
The authors reached conclusions while acknowledging methodological limitations and uncertainties. For example, Cameroon et al. suggested substantial savings by switching from originator brands to generics, while recognizing the variations between the prices of medicines paid by patients versus payers (Reference Cameron, Mantel-Teeuwisse, Leufkens and Laing19). Karak et al. concluded that Oncotype-DX is cost-effective, despite challenges related to obtaining local cost data (Reference El Karak, Rassy, Bassil, Awali, Hanna and Saroufim13). Rizk et al. acknowledged an uncertainty related to the assumption of attrition cases and baseline differences (Reference Rizk, Hiligsmann, Karavetian and Evers16). Abi Hanna et al. concluded that the intervention is dominant when adopting a societal perspective. However, when adopting a healthcare perspective, it is 80 percent likely to be cost-effective, when the WTP thresholds are 220 USD per response and 840 USD per remission (Reference Abi Hana, Abi Ramia, Burchert, Carswell, Cuijpers and Heim20).
Generalizability of the results
Three studies discussed the generalizability of the results. Cameroon et al. included 17 countries, acknowledging that the prices of the medicines are country-specific and bound to the time of data collection, rendering generalizing the results to other countries impossible (Reference Cameron, Mantel-Teeuwisse, Leufkens and Laing19). Conversely, Kim et al. framed their research to offer “preliminary insights for policy makers” in the extended Middle East and North Africa countries, mentioning that additional factors should be considered by policy makers (Reference Kim, Sharma, O’Shea, Sweet, Diaz and Sancho-Garnier14). Rizk et al. mentioned that the trial underpinning the EEs was conducted following a practical fashion, increasing its generalizability (Reference Rizk, Hiligsmann, Karavetian and Evers16).
Potential conflict of interest and discussion of ethical and distributional issues
Cameroon et al. received partial funding from GlaxoSmithKline (Reference Cameron, Mantel-Teeuwisse, Leufkens and Laing19). Two of the authors of the study by Matni et al. are employees of Merck Serono, and another author is an employee of IQVIA, the firm commissioned by Merck Serono to support the development of the EE (Reference Matni, Yamout, Koussa, Khamis, Fleifel and Sharifi15).
Discussion
This systematic review provides a comprehensive assessment of published full economic evaluations examining the value for money of healthcare technologies in the Lebanese context. We screened thirteen full texts (Reference El Karak, Rassy, Bassil, Awali, Hanna and Saroufim13–Reference Abi Hana, Abi Ramia, Burchert, Carswell, Cuijpers and Heim20;Reference Nasser, Nassif and Mahfouz27–Reference Basbous, Yehya, Salti, Tamim and Nabulsi32), and included eight studies with varying methodologies and quality (Reference El Karak, Rassy, Bassil, Awali, Hanna and Saroufim13–Reference Abi Hana, Abi Ramia, Burchert, Carswell, Cuijpers and Heim20). Among the thirteen screened full texts, three were excluded because although they were titled “cost–benefit analyses,” they only compared the costs, and not the health outcomes of the alternatives (Reference Khneisser, Adib, Assaad, Megarbane and Karam28;Reference Bahr, Bzieh, El Hayek and Adib29;Reference Basbous, Yehya, Salti, Tamim and Nabulsi32). We found few EEs. This relates to the absence of an explicit HTA framework in Lebanon requiring the submission of EEs for reimbursement decision making (4). The first full EEs by a Lebanese researcher were published in 2017, indicating a relatively nascent practice (Reference Rizk, Hiligsmann, Karavetian and Evers16). Moreover, researchers experienced in EEs were mostly non-Lebanese, denoting the limited local research capacity, as formal HTA requiring the submission of EEs as part of a dossier remains limited in LMICs (Reference Daccache, Rizk and Dahham8). EEs co-authored by these experts (Reference Kim, Sharma, O’Shea, Sweet, Diaz and Sancho-Garnier14;Reference Sharma, Seoud and Kim17) met a higher number of methodological criteria.
From a methodological standpoint, substantial heterogeneity was observed, particularly with respect to the choice of discount rates and the application of willingness-to-pay thresholds. Lower methodological quality was predominantly observed in model-based economic evaluations conducted by Lebanese researchers. Across the included studies, recurrent shortcomings included unclear specification of the analytical perspective; deficiencies in the identification, measurement, and valuation of costs and outcomes; inadequate model validation and calibration; limited sensitivity analyses; and inconsistent application of discounting and WTP thresholds.
Of interest, although the choice of the perspective affects the results of an EE, the studies adopted different perspectives. Given the fragmented nature of the healthcare system in Lebanon, specifying the institution of choice is required, as institutions vary greatly regarding benefit packages, the standards of care, and the costs incurred, as no unified purchasing exists (2). Conversely, it would be inefficient to conduct EEs from the perspective of each payer separately. Albeit not always feasible, tailoring the analyses to different payers could better inform the use of health economic information in decision making. Half of the studies adopted a societal perspective (Reference Kim, Sharma, O’Shea, Sweet, Diaz and Sancho-Garnier14;Reference Rizk, Hiligsmann, Karavetian and Evers16;Reference Sharma, Seoud and Kim17;Reference Abi Hana, Abi Ramia, Burchert, Carswell, Cuijpers and Heim20). When not adopting a societal perspective, the narrower perspective was poorly defined (Reference El Karak, Rassy, Bassil, Awali, Hanna and Saroufim13;Reference Nasser, Mansour, Abi Nader and Metni18;Reference Abi Hana, Abi Ramia, Burchert, Carswell, Cuijpers and Heim20). Although a societal perspective is preferred (Reference Byford and Raftery33), a recent systematic review on EE guidelines in LMICs found that only four out of the thirteen included guidelines recommended its use (Reference Daccache, Rizk and Dahham8).
Several recurrent methodological weaknesses were identified in relation to costing practices across included studies. Primarily, cost categories, and sometimes tariffs, were directly retrieved from the primary studies from which the model was adopted (Reference El Karak, Rassy, Bassil, Awali, Hanna and Saroufim13). Noting that the cost per patient could vary considerably based on contextual factors across countries, inappropriate transferability of economic data could lead to misleading results and inefficient investment decisions (Reference Goeree, Burke and O’Reilly34). The transferability of cost data should be completed while separating the units of resources utilized and the respective monetary values per unit (Reference Goeree, Burke and O’Reilly34). We highlighted the suboptimal reporting and variation of the costing approaches in included studies (Reference El Karak, Rassy, Bassil, Awali, Hanna and Saroufim13;Reference Nasser, Mansour, Abi Nader and Metni18), also previously observed in LMICs (Reference Dahham, Rizk, Kremer, Evers and Hiligsmann35). In the model-based studies using local costs, the identification of costs relied mostly on guidelines rather than collecting patient-level information (Reference El Karak, Rassy, Bassil, Awali, Hanna and Saroufim13;Reference Nasser, Mansour, Abi Nader and Metni18). Moreover, as the fragmentation in the Lebanese healthcare system implies varying prices for the same healthcare resource across the multiple payers, conducting scenario analyses applying different price discount scenarios for the comparator(s) could be helpful in generalizing results (Reference Briggs36). Furthermore, as two studies were conducted during a period that witnessed a severe currency depreciation (Reference El Karak, Rassy, Bassil, Awali, Hanna and Saroufim13;Reference Matni, Yamout, Koussa, Khamis, Fleifel and Sharifi15), valuation of costs should be reported in both local currency and purchasing power parity dollars, while specifying exchange rates and data collection dates, where possible (2;Reference Dahham, Kremer and Hiligsmann6).
The outcomes considered in included studies also differed. In the absence of a local valuation set to elicit Lebanese preference-based utilities (37), the utility weights used for QALYs’ calculations were imported from other countries (Reference El Karak, Rassy, Bassil, Awali, Hanna and Saroufim13;Reference Rizk, Hiligsmann, Karavetian and Evers16;Reference Nasser, Mansour, Abi Nader and Metni18). While this was acknowledged only in two studies (Reference Matni, Yamout, Koussa, Khamis, Fleifel and Sharifi15, Reference Rizk, Hiligsmann, Karavetian and Evers16), the others did not justify the choice of the utilities adopted. Adopting utilities that reflect the preferences of other populations could bias the results and favor investment decisions that might not reflect the preferences of Lebanese society (Reference Gerlinger, Bamber and Leverkus38). Furthermore, regarding QALYs and life-years saved, the survival data were derived from the Lebanese population only in Matni et al. (Reference Matni, Yamout, Koussa, Khamis, Fleifel and Sharifi15). The remaining studies extrapolated survival data from foreign populations. Within the existence of life tables for the Lebanese population, this practice is questionable (39).
The validation of the models was not mentioned explicitly in all studies. Researchers developing EEs relying on models should provide clarity about the structure, face, performance, and outcome validity (Reference Odnoletkova, Goderis and Pil40).
Furthermore, although discounting could largely affect the results of an EE (Reference Attema, Brouwer and Claxton41), the practice varied from no discount (Reference Nasser, Mansour, Abi Nader and Metni18) to discounting only costs at 1.5 and 5 percent (Reference El Karak, Rassy, Bassil, Awali, Hanna and Saroufim13) to discounting both costs and outcomes at 3 percent (Reference Kim, Sharma, O’Shea, Sweet, Diaz and Sancho-Garnier14;Reference Sharma, Seoud and Kim17), and 3.5 percent (Reference Matni, Yamout, Koussa, Khamis, Fleifel and Sharifi15). As these studies were developed in the absence of local EE guidelines (Reference Al-Harakeh, Abbas and Hassan42), the variation could be explained by the lack of international consensus about discounting costs and effects (Reference Attema, Brouwer and Claxton41).
The maximum WTP for an extra unit of outcome represents the cost-effectiveness threshold, against which the results of EEs are compared (Reference Grosse43). A central debate in defining this threshold is whether to adopt a “supply-side” approach based on the opportunity cost, or a “demand-side” approach based on the consumption value of health (Reference Grosse43). The topic of WTP threshold has not been explored in Lebanon without an agreed-upon threshold (Reference Daccache, Rizk, Hiligsmann, Evers and Karam44). Consequently, the studies differed in their approaches. Since there is no WTP threshold agreed upon by local policy makers, the claim that an intervention is cost-effective lacks consensus and legitimacy, rendering results potentially less binding and impactful. The latter could be explained by the lack of a formal and legitimate maximum willingness to pay threshold (Reference Daccache, Rizk, Hiligsmann, Evers and Karam44). This could potentially be addressed by interpreting results across different ceiling ratios, relying on the results presented in the cost-effectiveness acceptability curve (Reference Briggs36). Moreover, one study measured disease-specific outcomes and relied on WTPs for the interpretation of the results, rendering the utilization of the results even more complex in the context of Lebanon (Reference Abi Hana, Abi Ramia, Burchert, Carswell, Cuijpers and Heim20).
Knowing that health economics is not an exact science, sensitivity analysis is an integral component. Although all studies conducted sensitivity analyses, some did not address relevant uncertainties. Concluding results based on a point estimate of the ICER, without the relevant sensitivity analyses and cost-effectiveness acceptability curve, could be misleading (Reference Jain, Grabner and Onukwugha45).
Local EE guidelines were recently developed (Reference Daccache, Rizk, Hiligsmann, Evers and Karam44;Reference Daccache, Karam, Evers, Hiligsmann and Rizk46). As this systematic review looks at EEs developed before the emergence of these guidelines, assessing the methodological quality of the future EEs might give a sense of the impact of the guidelines.
Moreover, our study did not rely on a critical appraisal instrument designed to generate an index score such as the Quality of Health Economic Studies (QHES) (Reference Motheral47), as done in the study by Tran et al. in Vietnam, limiting international comparability (Reference Tran, Nong, Maher, Nguyen and Luu48).
Strengths and limitations
We pioneer in exploring the extent and quality of full EEs of health technologies in Lebanon. We searched multiple databases using a sensitive strategy and followed methodological best practices. We conducted a rigorous appraisal using validated and widely used checklists. More importantly, this systematic review responds to a national priority identified by policy makers and stakeholders (5) and could be replicated in other countries where health economics is emerging. Moreover, this study serves as a bridge between theory and practice and invites researchers to further account for the methodological quality of future research. Conversely, while we tried to be as exhaustive as possible, we might have missed relevant studies. Moreover, we did not search the gray literature and could not have access to one potentially eligible study for full-text screening. Finally, the utilized critical appraisal tool does not take into account the methodological heterogeneity driven by the type of healthcare intervention, nor other relevant specificities (Reference Odnoletkova, Goderis and Pil40).
Recommendations for future economic evaluations
HTA and EEs are emerging in Lebanon. Their use to inform decisions is an objective of the Lebanese MoPH (5). Our systematic review identified few published EEs with varying quality. As high-quality EEs are crucial to inform healthcare investment decisions, we recommend strengthening local expertise and increasing adherence to the recently developed local EE guidelines (Reference Grosse43), as detailed in Box 1. Our findings align with those of similar studies in emerging countries, calling for serious attention to the quality of further EEs (Reference Tran, Nong, Maher, Nguyen and Luu48;Reference Teerawattananon, Russell and Mugford49).
Recommendations for future EEs
-
1. Define clearly the adopted perspective, while considering the fragmentation of the Lebanese healthcare system.
-
2. Costs should be identified, measured, and valued systematically and transparently.
-
3. Costs should be reported in both local currency and purchasing power parity dollars, while indicating the exchange rates and data collection date. The whole process should be reported explicitly.
-
4. Scenario analyses with varying comparator(s) prices could improve generalizability in a fragmented healthcare system.
-
5. Model validation: researchers should ensure that the model is underpinned by solid evidence, represents the real-life case at hand, performs without errors, and generates outcomes that are valid.
-
6. Deterministic and probabilistic sensitivity analyses should be conducted, along with an explicit discussion about their impact on the results.
-
7. Results should be carefully interpreted in the absence of a legitimate maximum WTP threshold. Considering multiple maximum WTP thresholds might improve interpretation.
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1017/S0266462326103961.
Author contribution
W.M. and R.R. are equal first authors.
Funding statement
This research received no specific grant from any funding agency, commercial, or not-for-profit sectors.
Competing interests
This publication was based on the end-of-studies project while completing WM’s MSc in Health Economics for Healthcare Professionals at the University of Aberdeen. WM currently assumes the role of health economics and market access lead for the Middle East and North Africa at Boston Scientific.
Open access
