PubMed, Embase, Scopus, the Cochrane Library, the Center for Reviews and Dissemination Database, the Health Technology Assessment Database, and the China National Knowledge Infrastructure Database were searched to collect economic evaluations. The search encompassed publications from the inception of the databases until June 13, 2024. Our search strategy focused on key terms that corresponding to the predefined main domains of cost-effectiveness analysis, cancer, thrombosis, direct oral anticoagulant. The main search domains were linked using the Boolean operator ‘AND’, and the keywords of the same domain were connected using the Boolean operator ‘OR’. The detailed search strategy and results are shown in Supplementary Table S1.

After the literature search, a two-stage screening process was conducted. First, two researchers independently screened the titles and abstracts of all retrieved studies to exclude obviously irrelevant ones. Then, the full texts of the remaining studies were reviewed to determine their final eligibility based on the predefined inclusion and exclusion criteria.

Inclusion Criteria: (1) Types of studies: Economic evaluations of using DOACs for the prevention or treatment of VTE in cancer patients, including cost-effectiveness analysis, cost-utility analysis, cost–benefit analysis, and cost-minimization analysis. (2) Study Population: Cancer patients with or without VTE. (3) Intervention: Use of DOACs including dabigatran, rivaroxaban, apixaban, or edoxaban. (4) Comparator: Pharmacological therapies and prophylactic measures recommended in guidelines.

Exclusion Criteria: (1) Non-English literature. (2) Exclusion of reviews, theoretical or methodological introductions, and literature on research progress. (3) Exclusion of unpublished or incomplete information such as conference abstracts and letters. (4) Exclusion of duplicate literature. (5) Exclusion of literature that only includes cost analysis and budget impact analysis results.

Two researchers independently extracted literature data and cross-checked their findings. In event of disagreement, they conducted discussions or consultations to reach a consensus. The extracted content includes study characteristics regarding publication (author, year of publication, country), study design (target population, time horizon, comparators, model type, discount rate, outcome measure, perspective, cost type), and study results (cost, outcome, incremental cost-effectiveness ratio (ICER), subgroup analysis, willingness-to-pay threshold (WTP), sensitivity analysis).

The quality of the included studies was evaluated by the Consolidated Health Economic Evaluation Reporting Standards 2022 (CHEERS 2022) checklist (16). It consists of 7 main categories with 28 items: (1) Title, (2) Abstract, (3) Introduction, (4) Methods, (5) Results, (6) Discussion, and (7) Other relevant information. The scoring criteria are as follows: for each item, 1 point was scored for full compliance, 0.5 point for partial compliance, 0 point for not compliance, and items not applicable were not counted. The quality assessment of the literature was independently evaluated by two researchers. In case of discrepancies, they were resolved through discussion or by seeking arbitration from a third researcher.

Due to the differences in healthcare resource consumption across different countries and different perspectives of economic evaluations, there are substantial differences in the characteristics of the study subjects, model assumptions, measurements of costs and outcome parameters, as well as the design of economic models. Therefore, quantitative integration of data is not feasible. Instead, data extraction was analyzed and summarized through descriptions and tables.

A total of 856 studies were identified using the search strategy and 230 duplicates were removed, leaving 626 studies for initial screening. Following further screening of titles and abstracts according to inclusion and exclusion criteria, 550 studies were excluded for not meeting the criteria. Subsequently, 76 studies underwent full-text review, of which 61 studies were excluded for the following reasons: (1) Not a full report (n = 24); (2) Not a comprehensive pharmacoeconomic analysis (n = 24); (3) Intervention or population not of interest (n = 7); (4) Only cost or only efficacy (n = 5); (5) Not in English (n = 1). This review included a total of 15 studies, and the study selection process is presented in a PRISMA flow chart (Figure 1).

The basic characteristics are presented in the Table 1. The studies included in this review were published between 2019 and 2023. Most of the studies were conducted in high-income countries, including the United States (17–24), Spain (25, 26), the Netherlands (27), and Canada (28). Only three studies were conducted in middle-income countries: Brazil (29) and China (30, 31). Among the included studies, eight focused on treatment and secondary prevention of VTE in cancer patients (17, 18, 22, 24, 25, 27, 29, 31), comparing the cost-effectiveness of DOACs (including apixaban, rivaroxaban, and edoxaban) with LMWHs, with one study also comparing with warfarin (29). The remaining focused on primary prevention in cancer patients at risk of VTE (19–21, 23, 26, 28, 30), mainly comparing the cost-effectiveness of DOAC prophylaxis with no thromboprophylaxis and LMWHs prophylaxis. All studies were cost-effectiveness analyses, with the time horizons ranging from 28 days to lifelong. Studies with time horizons exceeding 1 year applied annual discount rates ranging from 1.5 to 5% (18, 20–22, 24, 26–31). All studies utilized decision analysis models for pharmacoeconomic evaluations, employing Markov models or decision tree models, or a combination of short-term decision analysis models and long-term Markov models. Typically included health states were on anticoagulant treatment, off anticoagulant treatment, PE, DVT, intracranial hemorrhage (ICH), non-ICH major bleeding (MB), clinically relevant non-major bleeding, VTE-related death, MB-related death, and death by any cause, etc. Except for one study that did not report the perspective (19), all other studies reported the perspective of analysis, primarily from the public or private healthcare system perspective (20, 21, 24–26, 28–31), followed by the payer perspective (18, 23) and societal perspective (17, 27). One study concurrently used societal and healthcare system perspectives (22). Quality-adjusted life years (QALYs) were used as the outcome measure in all studies, while the ICER was used for comparing the values between treatment strategies.

Du et al. (30), Li et al. (20), Kimpton et al. (28), and Muñoz et al. (26) compared DOAC prophylaxis versus no thromboprophylaxis among cancer patients receiving chemotherapy with intermediate-to-high risk of VTE. These studies included apixaban 2.5 mg twice daily, rivaroxaban 10 mg daily, and edoxaban 60 mg daily as doses for DOAC prophylaxis. All studies utilized data from the AVERT (32) and CASSIN (33) trials, which showed that DOACs significantly reduce the incidence of VTE in cancer patients but may increase the risk of bleeding. The included studies demonstrated that DOACs result in higher QALYs compared to no prophylaxis. However, The results based in Canada (28) and Spain (26) found apixaban prophylaxis was more cost-saving compared to no prophylaxis, contrary to China (30) and the United States (20). When the WTP threshold was set at one time the per-capita gross domestic product (GDP) of China in 2019, apixaban prophylaxis was not cost-effective (30). However, the cost-effectiveness of DOACs for prophylaxis improved in cancer patients with a higher risk of VTE (Khorana score ≥3). In other studies, low-dose DOAC prophylaxis showed cost-effectiveness in patients with a modified Khorana score of 2 or more. Muñoz et al. (26) separately compared the cost-effectiveness of apixaban and rivaroxaban versus no prophylaxis. The result showed that compared to no prophylaxis, apixaban was dominant and rivaroxaban was cost-effective (with additional costs).

Ryan et al. (21) focused on patients with ovarian cancer receiving neoadjuvant chemotherapy, showing that DOAC prophylaxis was effective but more expensive than no prophylaxis, with an ICER exceeding the WTP. They also conducted an exploratory analysis on thromboprophylaxis using low-dose aspirin, which was found to be both more effective and less expensive than no prophylaxis, costing US$531 less per patient and resulting in slightly improved QALYs (0.000246) compared to no prophylaxis.

Glickman et al. (19) evaluated thromboprophylaxis in gynecologic cancer patients, and Bell et al. (23) assessed endometrial cancer patients undergoing minimally invasive hysterectomy. Both studies compared apixaban prophylaxis and enoxaparin prophylaxis, revealing that apixaban prophylaxis was not only less expensive but also more effective. Bell et al. (23) also compared thromboprophylaxis with no prophylaxis, finding that no prophylaxis was superior. Only when the risk of DVT is 14% or higher, apixaban prophylaxis for 28 days would be cost-effective compared with no prophylaxis. When the risk of DVT is 4.8% or higher, a 7-day course of apixaban prophylaxis was found to be cost-effective.

Eight studies focused on the treatment and secondary prevention of VTE in cancer patients who had already experienced VTE (17, 18, 22, 24, 25, 27, 29, 31). All studies compared the cost-effectiveness of DOACs and LMWHs, consistently demonstrating that DOACs were more cost-effective and represented cost-saving strategies. Among them, Lopes et al. (29) also compared the cost-effectiveness of DOACs with warfarin, showing that although warfarin was a more cost-saving strategy, the ICER of DOACs remained below WTP. Three studies (17, 18, 31) conducted subgroup analyses among patients with gastrointestinal malignancies, revealing that although DOACs increased the risk of MB, the lower overall costs resulted in their ICER remaining below the WTP compared to LMWHs.

Two studies compared apixaban, edoxaban, and rivaroxaban in different contexts. Muñoz et al. (25) showed that apixaban had lower costs and better clinical outcomes from a Spanish healthcare perspective. In contrast, Gulati et al. (24) analyzed Veterans Affairs Federal Supply Schedule pricing in the United States and found that although apixaban had lower costs, rivaroxaban was slightly more effective. When incorporating GoodRx data, which reflect commercial pharmacy expenditures, rivaroxaban demonstrated a cost-effectiveness advantage over apixaban.

All included studies conducted sensitivity analyses, with one-way deterministic sensitivity analyses (DSA) in 13 articles (18–25, 27–31), two-way DSA in 4 articles (17, 21, 24, 30), probabilistic sensitivity analyses (PSA) in 13 articles (18, 20–31), and scenario analyses in 8 articles (18, 20–22, 24, 27, 28, 30).

In the context of VTE prevention among cancer patients, the ICER of DOAC prophylaxis versus no prophylaxis is highly sensitive to the risk of VTE, but the results did not reverse (20, 28, 30). In China, the outcomes of DOACs versus no prophylaxis were reversed in populations with a Khorana score ≥3 (30). Muñoz et al. conducted a probability deterministic sensitivity analysis using the incremental net benefit instead of the ICER. The result showed that variations in the probability of cancer mortality with or without prophylaxis made DOACs no longer cost-effective (26). In PSA, there was a higher probability of cost-effectiveness for DOACs in the United States and Canada, at 94 and 99.87% respectively, but the probability is much lower in China and Spain, both <70% (20, 26, 28, 30).

Sensitivity analysis for VTE prevention among ovarian cancer patients receiving neoadjuvant chemotherapy (21) identified drug costs of DOACs, baseline risk of VTE, efficacy of DOACs, and VTE case mortality as factors potentially influencing the ICER of DOACs versus no prophylaxis.

Glickman et al. (19) and Bell et al. (23) showed that changes in all examined parameters did not reverse the results when comparing apixaban prophylaxis with no prophylaxis or enoxaparin prophylaxis under WTP. The results showed that no prophylaxis was favored in 41.1% of trials, apixaban prophylaxis was favored in 33.7% of trials, and enoxaparin prophylaxis was favored in 25.2% of trials (23).

In the comparison of DOACs and LMWHs for treating CAT patients, drug costs were the main driving factors in the United States background (17, 18, 22, 24). Li et al. (18) using a cheaper generic enoxaparin instead of dalteparin would result in the costs of LMWHs being close to those of DOACs, with the ICER below the WTP. Besides drug costs, the relative risk of cancer mortality, the relative utility values, and the probability of non-PE and non-MB death were influential parameters determining the ICER (18, 22). Gulati et al. calculated the costs and effects of apixaban, edoxaban, and rivaroxaban separately, with the sensitivity analysis showing that in real-world scenarios, apixaban would only be the most cost-effective treatment strategy if its monthly cost was below US$530 when the WTP was set at US$50,000 (24).

In other countries, drug costs did not significantly affect the economic outcomes. Parameter changes did not affect the result that DOACs were more cost-effective than LMWHs in Brazil and Spain (25, 29). In the Netherlands, changes in parameters such as the risk of MB, treatment duration, and the risk of VTE recurrence had the greatest impact on incremental costs, with the risks of MB and VTE recurrence significantly affecting incremental QALYs (27). From the perspective of the Chinese healthcare system, the ICER of DOACs and LMWHs was sensitive to the utility and the risk of MB occurrence, which could potentially reverse the economic study results (31).

In all PSA (18, 22, 24, 27), although the ICER might change with the aforementioned parameters, DOACs still had a higher probability of being more cost-effective compared to LMWHs. Lopes et al. (29) also conducted a sensitivity analysis comparing the cost-effectiveness of edoxaban and warfarin, finding that in all simulations, the incremental cost of edoxaban was below the WTP, proving the robustness of the study results.

The results of the included studies and the sensitivity analyses are presented in Table 2.

The percentage of compliance with the items ranged from 72.2 to 88.9%. Details of the quality assessment are presented in the Supplementary Table S2. In the Methods category, less than 50% of the studies complied with the items Health economic analysis plan, Characterizing heterogeneity, Characterizing distributional effects, and Approach to engagement with patients and others affected by the study. Two studies (22, 24) described the health economic analysis plan they used, both based on the recommendations from the Second Panel on Cost-Effectiveness in Health and Medicine (34). Five studies (17, 18, 20, 30, 31) described patient heterogeneity through subgroup analyses within the study, showing how the results varied in specific populations. None of the studies described distributional effects. One study (27) mentioned that it was not suitable to involve patients or the public in the design, conduct, reporting, or dissemination plans, while the remaining reports did not mention whether there was patients or others involvement. Additionally, most reports (17, 18, 20–22, 24, 28, 30, 31) partially met the Setting and Location item under the Methods category, only stating the countries in which the studies were based without mentioning the particular healthcare setting or any other relevant sectors. Under the Results category, apart from de Jong et al. (27), which did not apply to the item Effect of engagement with patients and others affected by the study, none of the other studies mentioned relevant content.