A literature search was performed using the following databases: PubMed, Embase, the Cochrane Library, and three Chinese databases (China National Knowledge Infrastructure [CNKI], WANFANG DATA, and Chinese Biomedical Literature Database [CBM]). The search time was from the date of establishment of the databases to January 2019. The following search terms were used, “aprepitant,” “emend,” “cost,” “effectiveness,” “utility,” “benefit,” “economic,” “expenses,” and “pharmacoeconomic.”

Articles retrieved from the literature search were independently screened based on the title and abstract by two authors (TTQ and PM). Studies that did not meet the criteria were excluded. After the initial screening, two researchers (TTQ, PM) independently assessed the full texts of eligible citations. The list of included studies was reached by a consensus. Any disagreements were resolved by discussion or by consulting with a senior author (SDZ).

Data extraction was performed using predesigned data extraction tables in Microsoft Excel. For all studies, the following information: authors, published year, country, type of model, perspective, model details (time horizon, discount rate), source of funding, sensitivity or uncertainty analysis, incremental effectiveness and costs, and ICER were extracted.

The quality of the pharmacoeconomic studies was assessed by a 24-item checklist of the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) statement, which was used as a checklist to rate the quality of reporting in the included papers. The CHEERS statement of the International Society for Pharmacoeconomics and Outcomes Research (ISPOR) Health Economic Evaluation Publication Guidelines Good Reporting Practices Task Force is a guideline intended to improve reporting of the economic evaluation (13, 14). The quality of the included studies was evaluated by the answers to the questions, which were “yes” (reported and scored 1) or “no” (not reported and scored 0) or “partly” (partially reported and scored 0.5) or “NA”(not applicable and scored 1). The studies were separated into four quality categories. Those studies that fulfilled 100% of the items were classified as excellent quality; those that fulfilled between 75 and 100% of the items were classified as good quality; those that fulfilled between 50 and 75% of the items were classified as moderate quality; and those that fulfilled ≤50% of the items were classified as low quality (15, 16).

To facilitate the comparison of ICERs, all costs were converted into US dollars via purchasing power parities (PPP) in the year 2019. PPP was defined as the rates of currency conversion that eliminate the differences in price levels between countries. PPP conversion factors were obtained from the Organization for Economic Co-Operation and Development Stat database (17). We converted the original cost estimates to the target currency and price year.

After a thorough search of the databases, we acquired 169 articles, of which 104 were excluded based on the title and abstract screening. A total of 13 published studies (18–30) were selected for final inclusion (Figure 1) after reviewing.

Table 1 describes the general characteristics of the included studies. Of the 13 studies, eight (18, 21–26, 30) conducted both cost-effectiveness analysis (CEA) and cost-utility analysis (CUA), two studies conducted CUA (19, 20), one study conducted CEA (27), one study conducted cost-consequences analysis (CCA) (28), and one study conducted both CUA and budget impact analysis (29).

Ten of the 13 economic studies included data of clinical outcomes from randomized controlled trials. The clinical data of Chanthawong et al. (22) and Cawston et al. (20) came from systematic review and metaanalysis. Nakamura et al. (27) acquired data from a retrospective analysis of direct medical costs of National Hospital Organization Nagoya Medical Center between January 2009 and December 2013.

Of the 13 studies included, eight of used the decision analytical model (18, 21–25, 28, 30), four used the Markov model (19, 20, 26, 29), and one study did not use model analysis (27).

Eleven studies were conducted from a payer perspective, of which eight were performed from a public payer perspective (e.g., National Health Service, National Health Insurance system, and health-care system) (18, 20, 21, 23, 24, 28–30). In contrast, one study used a patient and statutory health insurance perspective (25), and two studies used the perspective of the payer, but did not describe it specifically (19, 26). One study used a societal perspective (22) and one study (27) did not mention it.

The time horizon for two studies (18, 19) was four cycles (21 days in each cycle), for another eight studies (20–25, 28, 30) time horizon was 5 days, for one study (26) it was five cycles (28 days in each cycle), time horizon of one study (29) was at least six cycles (5 days in each cycle) when CUA was conducted and 5 years when budget impacted analyses, and one study (27) did not mention the time horizon.

Among the included studies, the most common comparison was the aprepitant triple regimen (aprepitant+5-HT3RA+glucocorticoid) vs. standard regimen (5-HT3RA+glucocorticoid) with or without placebo (18, 19, 21, 24–26, 28, 30), the other comparison was aprepitant +5HT3RA vs. aprepitant (27), and the comparison of positive comparators such as netupitan + palonosetron (PAL) (NEPA) (20, 29), metoclopramide (23), and olanzapine (22). Most studies reported ICERs as cost-effectiveness evaluation outcomes.

Three of the studies were funded by Merck & Co (21, 23, 25), one was funded by Sanofi-aventis (19), one was funded by Helsinn Healthcare SA (20), one was funded by MSD Italia srl (28), one was funded by Italfarmaco Spa (29), and one was supported by a Health Outcomes Research Starter Grant from the PhRMA Foundation (26). Two studies (18, 24) did not disclose the funding source, although some authors of these studies were employees from pharmaceutical companies, and some authors received funding or honoraria from pharmaceutical companies. The remaining three studies (22, 27, 30) did not mention any funding information at all.

Twelve studies were of good quality based on the CHEERS checklist (18–26, 28–30), one was of moderate quality (27). The results of the quality assessment of the included studies are shown in Table 2.

We found that the included studies did not report several items on the CHEERS checklist. One of the studies (27) did not mention the perspective of the research, and five studies (21, 24, 25, 27, 29) did not state the results of uncertainty analyses, both the items on the CHEERS checklist that should be reported in the abstracts of economic evaluations.

Two studies used the perspective of the payer but did not describe it specifically (19, 26), and one study (27) did not mention the perspective nor did it use the model analysis.

Reasons for the choice of the economic model were not reported in most of the studies. Two studies did not perform a sensitivity analysis (21, 27). Even some studies that achieved good quality ratings did not meet the checklist criterion for characterizing heterogeneity. The roles of the funders play in the identification, design, conduct, and reporting of the analysis were not reported in all the included studies.

We also synthesized the results of quality rating and the funding information (Table 3). However, no relationship between sources of funding and the quality of the included studies was identified in this review.

The economic outcomes of the included studies are summarized in Tables 4–6.

We systematically searched for, assessed, and summarized the available literature on the cost-effectiveness of aprepitant in patients with CINV. The quality assessment of reviewed articles indicated that most articles were of good and moderate quality. Some studies did not report the reasons for choosing the time horizon or discount rate, and no included studies had a subgroup analysis. We found that only one study (22) in this review used a societal perspective; most studies considered a payer perspective. Societal perspective is the gold standard of pharmacoeconomic studies because it incorporates all costs and health outcomes, although other perspectives may be better for some decision-making situations (42, 43).

The role of funders in the identification, design, implementation, and reporting of research is critical to ensure that readers can reliably detect any potential bias. We found that seven of the included studies received funding from pharmaceutical companies. One study (26) was supported by a Health Outcomes Research Starter Grant from the PhRMA Foundation. Two studies (18, 24) did not disclose the funding source, but some authors were employees from pharmaceutical companies and some authors received funding or honoraria from pharmaceutical companies. Three studies (22, 27, 30) did not mention funding information at all. We found no relationship between the funding sources and the quality of the included studies that were identified in this review.

Ten of the 13 economic studies were highly consistent. The aprepitant regimen (APR+5-HT3RA+DEX) is more economical than the standard regimen (5-HT3RA+DXE), APR+5-HT3RA is more economical than 5-HT3RA, and APR+5-HT3RA+DEX is cost-effective for preventing CINV when compared with MET+ 5-HT3RA+DEX. Although aprepitant brings higher drug costs, the costs associated with vomiting management (such as patient management, hospitalization, and costs associated with the use of rescue drugs) are lower in the aprepitant group. Therefore, the aprepitant triple regimen is a cost-saving strategy.

The other three studies were Cawston et al. (20), Restelli et al. (29) on NEPA economics research and Chanthawong et al. (22) on olanzapine. NEPA is recommended for HEC by the NCCN, ASCO, and the MASCC/ ESMO, but as the drug is not yet on the market in China, it is not recommended in the 2014 edition of the Guidelines for the Prevention and Treatment of Vomiting Related to Cancer Therapy (7–10).

Chanthawong et al. (22) showed that compared with triplet antiemetic regimen, switching aprepitant to olanzapine increased QALY and saved costs. The use of olanzapine as part of standard antiemetic regimen is cost-effective for the prevention of CINV in patients receiving HEC in multiple SEA countries. The 2019 version of NCCA recommends two options for olanzapine for HEC, one is in combination with PAL and dexamethasone, and the other is in combination with NK1RA, 5HT3RA, and dexamethasone (7). The 2016 version of ESMO/MASCC recommends that olanzapine seems to be useful in the prophylaxis of delayed nausea [superior to (fos)aprepitant] and equal to (fos)aprepitant in the prevention of acute symptoms. Olanzapine may be considered with a 5-HT3 RA plus dexamethasone, particularly when nausea is an issue, but when using the 10-mg dose, patient sedation may be a concern [MASCC level of confidence: low; MASCC level of consensus: low; ESMO level of evidence II; ESMO grade of recommendation: B] (9). The ASCO considers that olanzapine lacks high-quality efficacy and safety studies, and hence is not recommended (8). There is no recommendation for olanzapine in China guidelines (10), and there are no indications for preventing CINV in the olanzapine instructions. Olanzapine should be used with caution in older people (44). The use of olanzapine for CINV did not reach the consensus of national guidelines. Therefore, even if olanzapine showed economic advantages when compared with aprepitant, the advantages and disadvantages should be considered in clinical decision-making. More studies are needed to analyze olanzapine or NEPA compared with aprepitant.

This study has several strengths. First, this review is the first comprehensive synthesis of the evidence of cost-effectiveness for aprepitant in preventing CINV. Second, this review includes all published cost-effectiveness studies of aprepitant, and adjusts all cost-related values of different time and countries to 2019 dollars for better comparison.

This study also has several limitations. First, because of heterogeneity in the methodology (e.g., different types of economic models, time horizon, and perspective) and data sources (e.g., effectiveness and safety data, and costing data) of economic evaluations, it is impossible to combine the data (45). Since it is difficult to compare different economic evaluations and reach an overall conclusion regarding the results, reasoning and conducting quantitative analysis (metaanalysis) is impossible. Thus, the explicit and precise estimation of the reported indicators was not possible, so this issue should be considered for using and generalizing the results.

Second, we analyzed the results of economic assessments conducted in different countries with different health care systems and reimbursement mechanisms, and most studies did not use real-world data. Methods such as cost-benefit thresholds, budgeting, and reimbursement should be taken into account, and so the interpretation of the results should be cautious.