Pharmacoeconomics Evaluations of Oral Anticancer Agents: Systematic Review of Characteristics, Methodological Trends, and Reporting Quality

TAGEDNV A L U E I N H E A L T H R E G I O N A L I S S U E S 1 6 ( 2 0 1 8 ) 1
1 5

Available online at www.sciencedirect.com

ScienceDirect journal homepage: www.elsevier.com/locate/vhri

Pharmacoeconomics Evaluations of Oral Anticancer Agents: Systematic Review of Characteristics, Methodological Trends, and Reporting Quality 1 1,2 1 D1X X Ahmad Al Kadour, MSc D2X X , Wafa D3X X Al Marridi, MSc D4X X , Daoud D5X X Al-Badriyeh, PhD D6X X * 1

College of Pharmacy, Qatar University, Doha, Qatar; 2Pharmacy Department, Sidra Medical and Research Centre, Doha, Qatar

TAGEDPA B S T R A C T

Objectives: To review literature characteristics, describe methodological trends, and assess the reporting quality of the economic evaluations of oral anticancer drugs (OACDs). Methods: The review included comparative economic evaluations of OACDs. The search was conducted via PubMed, Embase, EconLit, and Economic Evaluation Database, and studies till December 2017 were included. Using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses checklist, literature inclusion and data extraction were performed in duplicate by separate investigators. Outcome measures were literature characteristics, gaps and methodological trends, and reporting quality using the Consolidated Health Economic Evaluation Reporting Standards checklist. Data were summarized on the basis of methodological themes of interest. Descriptive statistics and tabulations were used for result presentation. Results: Out of 241 found articles, 21 were included. There is a recent increasing interest in the economics of OACDs, whereby the cost per quality-adjusted life-year, via costutility analysis, is the most used for decision making. Most of the

Introduction The oral chemotherapy landscape has developed over the years. The first approval of oral anticancer drugs (OACDs) by the Food and Drug Administration was in the early 1950s, which rapidly grew in the early 2000s, overtaking the growth of intravenous chemotherapy in the last 5 years [1,2]. It is estimated that more than a quarter of the 400 anticancer drugs now in the development pipeline are planned to be oral agents [3]. Seeing the minimized inconvenience of infusion, including pain, anxiety, and inpatient status, OACDs have been proposed to be associated with improved quality of life (QOL) in patients. The oral administration allows patients to receive their therapy at home, with only follow-ups taking place in a hospital setting. These, together with an increased incidence of cancer, availability of more therapeutic alternatives, and insufficiency of hospital resources, have led practices to move toward the use of OACDs [4,5]. Cancer care requires a significant amount of control, particularly for dosing and its timing. With OACDs, however, much of

studies were from the payer perspective, and the primary sources of data were clinical trials, expert panels, and medical charts. The dominance status (higher effect, lower cost) was a commonly reported outcome. Decision-analytic modeling was used in most of the studies, mostly including Markov modeling. Studies were highly heterogeneous in methodological aspects, and the included studies did not meet most of the reporting quality criteria. Conclusions: High heterogeneity in methods in studies may limit the robustness and transferability of results, potentially misleading decision makers toward wrong decisions on OACDs. The transferability and generalizability of results are further limited by a “less than ideal” adherence to current reporting standards. Keywords: cancer, economic evaluation, methods, oral chemotherapy, systematic review. Ó 2018 Published by Elsevier Inc. on behalf of International Society for Pharmacoeconomics and Outcomes Research (ISPOR).

this control is in the hand of the patients. The lack of coordinated care results in a possible level of errors, nonadherence, and increased adverse events [6]. As a consequence, and despite the advantages, the use of oral chemotherapy has been controversial. Importantly, and within the context of the present research, the economic burden is especially a prominent issue that has an impact on the use and prescribing of OACDs, influencing patient access to the drugs. Oral chemotherapies tend to be costly. As of 2014, in the United States, for example, most of the newly marketed OACDs had a price that exceeded US $10,000/mo [7]. In addition to the financial burden, prescribing OACDs is believed to be partly shifting the economic paradigm of cancer care and medical service from the hospital setting to the community, creating a loss of potential resources for hospitals, which may create a barrier to the widespread use of the drugs and, hence, reduced patient access [8,9]. Also important, and partly a result of the aforementioned burdens, is that it is common for the cost of OACDs not to be covered by government reimbursement because OACDs are not administered in a hospital or a clinical setting,

Conflicts of interest: The authors have indicated that they have no conflicts of interest with regard to the content of this article. * Address correspondence to: Daoud Al-Badriyeh, Pharmacoeconomics and Outcomes Research, College of Pharmacy, Qatar University, Doha 2713, Qatar. E-mail: [email protected] 2212-1099/$36.00 – see front matter Ó 2018 Published by Elsevier Inc. on behalf of International Society for Pharmacoeconomics and Outcomes Research (ISPOR) https://doi.org/10.1016/j.vhri.2018.05.003

2

V A L U E I N H E A L T H R E G I O N A L I S S U E S 1 6 ( 2 0 1 8 ) 1
1 5

even if they are administered on the basis of accepted treatment protocols [10]. Nevertheless, OACDs still have considerable advantages, as discussed earlier, which have the likely potential for downstream cost-savings. Hence, although most of the evaluations of OACDs seem to be focusing on treatment adherence and safety, it is only logical that future research will focus more on the economic implications of OACDs, especially for guiding toward the right national reimbursement plans for oral chemotherapies in settings. It is, therefore, anticipated that there is and will continue to be an increasing interest in conducting economic evaluations of OACDs, as opposed to that of nonoral chemotherapies [11]. Several studies [12
25] have indeed been performed to find that the cost-savings with OACDs as associated with the reduced need to treat adverse events, and with the enhanced patient time and medical resource utilization, exceeded the relative increase in acquisition costs of the drugs as compared with the intravenous formulations. Here, the absence of research technique standardization and poor consistency with established standards, filling knowledge gaps on literature characteristics, and methodological trends will be of functional incentive to researchers in arranging and sorting out their local explorations. In the present study, the objective was to systematically review the literature characteristics, the methodological trends and gaps, and the reporting quality of the economic evaluations of oral chemotherapies in the literature. It is important to note that the clinical, practice, and policy aspects of studies are outside the scope of the present review. Information about such literature characteristics and trends will be of practical value for consideration by researchers in settings that are looking to perform pharmacoeconomics research on OACDs. This information is added to enable a better understanding of the quality of evidence by decision makers as they contrast this against current gaps and quality of reporting in the literature. Results from the present study can also be inceptive to journal editors and commentators in enhancing the quality of distributed research.

Methods The Preferred Reporting Items for Systematic Reviews and MetaAnalyses reporting checklist was followed for the purpose of the present study (see Appendix I in Supplemental Materials found at 10.1016/j.vhri.2018.05.003) [26].

Search Strategy A systematic review of the literature was conducted via the databases PubMed, Embase, EconLit, and the National Health Service Economic Evaluation Database (EED). The search was last updated in July 2017 for PubMed, Embase, and EconLit. The EED, however, ceased its search in December 2014 and did not include publications from later dates. Articles found via PubMed were differentiated by labeling them as “PubMed—indexed for MEDLINE” or “PubMed—in process.” The former refers to articles that exist in MEDLINE, whereas the latter refers to articles that exist only in PubMed. The PubMed search terms were the Medical Subject Heading terms “antineoplastic agents,” “administration, oral,” “neoplasms,” and “cost-benefit analysis,” in addition to the free-text term “oral.” The Embase index terms used were the Emtree terms “oral drug administration,” “antineoplastic agent,” “neoplasm,” “cost-effectiveness analysis,” “cost-benefit analysis,” “cost-utility analysis,” “cost-minimization analysis,” “economic evaluation,” “cost,” and “pharmacoeconomics,” in addition to the free-text terms “oral,” “neoplasms,” “cancer,” “cancers,” and “tumor.” The EED search keywords were similar to those in Embase. The EconLit search

terms were “antineoplastic agent,” “anticancer,” “oral,” “neoplasm,” “cancer,” “tumor,” “cost effectiveness,” “cost utility,” “cost benefit,” and “economic evaluation.” The full PubMed search strategy is given in Appendix II in Supplemental Materials found at 10.1016/j.vhri.2018.05.003.The same was adapted for the other databases. The database search included the gray literature, such as books, dissertations, conferences, working papers, and governmental publications, and was supplemented with a screening of references in the included articles and also a general Internet search using Google and Google Scholar, where free-text searching used the same search terms as in the primary search.

Selection Criteria Inclusion criteria The inclusion criteria were outlined in terms of the PICO (Population, Intervention, Comparison, Outcome) framework:  Population: Cancer-based underlying disease.  Intervention: Study of the use of at least one oral chemotherapy in cancer.  Comparison: Therapy-based comparative studies.  Outcome: Peer-reviewed publications of comparative studies till December 2017 were included. No considerations were made on whether the articles were freely available. No considerations were made on whether the studies were retrospective or prospective. Of interest in the included literature were the characteristics, methodological trends and gaps, and the reporting quality of the economic evaluations.

Exclusion criteria The exclusion criteria included the following:  Non-English language  Nonhuman studies  Noncomparative research (e.g., letters, general reviews, and editorials)  Non
drug-based studies

Data Extraction Screening for initial eligibility via the search terms was done by assessing the titles and abstracts first. Found articles via the database search were further screened for eligibility through a manual analysis of study abstracts. Then, for final inclusion in the study, a follow-up manual screening by reviewing the full text of the initially eligible articles was conducted. This process, in addition to data extraction, was separately performed for conformance by two of the authors. Disagreements were further discussed by the research team as led by one of the authors. Before formal data extraction, and for validation purposes, a random sample of three included articles was independently reviewed by each of the study authors before being discussed to ensure consistency and agreement among all. Extracted data from included full texts were related to study characteristics and methodological features, such as comparators, study objectives, setting and perspective, type of evaluation, research design, types and sources of clinical and economic data, time adjustment, time horizon, limitations, and uncertainty analyses. Descriptive statistics and tabulations were used to present the results.

Quality Assessment Economic evaluations were scrutinized by using the Consolidated Health Economic Evaluation Reporting Standards (CHEERS)

TAGEDNV A L U E I N H E A L T H R E G I O N A L I S S U E S 1 6 ( 2 0 1 8 ) 1
1 5

checklist [27]. Although there are other available checklists for economic evaluations, such as the Quality of Health Economics Studies, the purpose of an evaluation checklist in this review is not to assess the quality of methods or to assess bias to generate evidence. CHEERS is considered the most comprehensive and appropriate as a reporting quality tool. The quality assessment was independently conducted by the different individual researchers, as described earlier.

Results Literature Search A total of 241 articles were found using search databases. After screening and removing duplicates, only 21 studies [12
25,28
34] were eventually included for consideration in this review (Fig. 1). No additional articles were added through Google or through the screening of references of reviews.

Main Characteristics of Economic Evaluations Main features of the economic evaluations are presented in Table 1. The important quality aspects of research reporting that need to be adhered to are presented in Table 2. The first study of all the included studies was published in as recently as 2003 [31]. A total of 12 OACDs, among which capecitabine was the most common, were evaluated across the included studies. A summary of the study comparisons is presented in Table 3. Clinical data in all studies were based on randomized controlled trials (RCTs) as the primary data source. More than 85% of the studies (n = 18) in this review were cost-utility analyses (CUAs), including two studies that also incorporated parallel costeffectiveness analyses (CEAs) [19,20]. Table 4 presents the utility

3

values as in different articles along with the primary source of estimation. Most of the evaluations were from the payer perspective (19 out of 21). In all the studies, the types of costs measured were consistent with the perspectives used, but none of the studies that were from the social perspective included the indirect cost of lost productivity. More than 77% of studies (n = 14) included simulation modeling as part of their analysis, as presented in Table 1. A non-Markovian decision-analytic model was constructed in six of these studies [16
18,20,23,25]. The Markov modeling was used in 13 studies [12
18,21,29
33]. Appendix III in Supplemental Materials found at 10.1016/j.vhri.2018.05.003 demonstrates a simplified decision tree that is used in most of the Markov simulation studies (n = 8) [12
18,21]. In the Markov modeling, all studies were based on published RCT data about the probabilities of health state transitions for overall survival and disease-free survival. In some instances, however, data were obtained from hospital case and safety reports, as was done by Perrocheau et al. [28], or partly from observational cohort studies, as in the study published by Goss et al. [20]. Cost data were collected from various study sources. Most study sources were drug databases, local and national insurance databases including the British National Formulary (n = 16), RCTs and literature sources (n = 8), and expert panels (n = 5). Total gain in life-years, progression-free life-years, and qualityadjusted life-years (QALYs) were also commonly reported outcome measures, as presented in Table 1 (12 out of 21). The important probability of being cost-effective on the basis of willingness-to-pay measures was reported only in five articles [12,21,22,25,31].

Effectiveness and Safety Data Only 10 out of the 21 studies had clear or partial definitions of effectiveness outcomes; in addition, different studies defined

Fig. 1 – Flow diagram of the study selection.

4

Table 1 – Characteristics and methodological aspects of included studies. Study, country

Aller et al. [12], Spain

Comparators,

Analysis type;

Time horizon;

neoplasm

model structure

perspective

Sunitinib vs. SFN and BEV/IFN, mRCC

CUA; Markov structure; 6-wk cycle

10 y; third-party payer

Primary outcome

QALYs, PFLYs, LYs

perspective

Definition of

Source of primary

effectiveness

clinical data

PFS* Progression

Death

Pazopanib vs. sunitinib, mRCC

CUA; decisionanalytic model

5 y; Canadian health care system perspective

QALYs

PFS Progression Death

Carlson et al. [23], United States

Erlotinib vs. docetaxel and pemetrexed, NSCLC Capecitabine vs. IV 5-FU/LV, CRC

CUA; decisionanalytic model

2 y; US health care payer perspective

QALYs

CUA; Markov structure; cycle

Lifetime; NHS and societal perspectives

LMs QALMs

PFS OS (not defined in the article) RFS OS

Cassidy et al. [18], United Kingdom

Chen et al. [24], China

Imatinib vs. IFN, CML

Di Costanzo et al. [16],

Capecitabine vs. IV

Italy

5-FU/LV, CRC

Ghatnekar et al. [29], Sweden

Dasatinib vs. highdose imatinib, CML

Glen and Cassidy [17], United Kingdom

Capecitabine vs. IV 5-FU/LV, CRC

Goss et al. [20], United

Lenalidomide/BSC

States

without EPO vs. BSC with EPO, myelodysplastic syndromes

length not defined in the article CUA; modeling information not defined in the article CUA; Markov structure; cycle length not defined in the article CUA; Markov structure; 1-mo cycle CUA; Markov structure; cycle length not defined in the article CEA and CUA; decision-analytic model

QALYs

3, 4, 5, and 10 y and

QALMs

RFS

lifetime; Italian NHS perspective Lifetime; societal perspective

LYs QALYs

Lifetime; NHS and societal perspectives

QALMs

1-y; US health care

Transfusion-

payer perspective

independent timex QALYs

OS (not defined in the article) CP AP BPz RFS OS (not defined in the article) Reduced transfusion independence Transfusion independence and persistent transfusion dependence (both not defined in the article)

Hisashige et al. [19], Japan

Oral UFT vs. surgery alone, CRC

CEA and CUA; Boag model{

Hsu et al. [14], Taiwan

Capecitabine

CUA; Markov

vs. 5-FU/LV, CRC

Lifetime; Chinese public health care system perspective

(not defined in the article) CCyR, which is associated with longterm survival in CML

structure; length of cycle not defined in the article

Three levels of time horizon#; National Health Insurance Japanese payer perspective Lifetime; national health care payer perspective

Main finding

Sunitinib dominatedy

Indirect comparison

One-way

from four phase II or III RCTs and expert panel Phase III RCT (called COMPARZ), and other published sources

deterministic and probabilistic sensitivity analysis Probabilistic sensitivity analyses

Pazopanib dominated sunitinib

Previously published trials

One-way, multivariate probabilistic and scenario sensitivity analyses One-way and multiway

Erlotinib had similar clinical efficacy and lower cost compared with alternatives Capecitabine dominated

Phase III RCT

sensitivity analyses

both SFN and BEV/ IFN

5-FU/LV

Previously published literature

One-way sensitivity analyses

Imatinib dominated IFN

Phase III RCT

One-way analysis

Capecitabine dominated 5-FU/LV

12-wk head-to-head RCT

One-way and probabilistic multivariate

Phase III RCT

sensitivity analyses None were conducted

Clinical trials

One-way sensitivity

Published literature Patient interviews Treatment guidelines Expert panel

analyses Multiway probabilistic sensitivity analysesjj

Dasatinib dominated its alternative

Capecitabine dominated 5-FU/LV Lenalidomide showed dominance

QALYs PFLYs LYs

RFS OS (not defined in the article)

Retrospectively based on a previous multicenter RCT

One-way, multivariate probabilistic and scenario sensitivity analyses

UFT dominated surgery alone

QALMs

RFS

RCT

One-way sensitivity

Capecitabine

OS (not defined in the article)

analysis

dominated 5-FU/LV

(continued on next page)

V A L U E I N H E A L T H R E G I O N A L I S S U E S 1 6 ( 2 0 1 8 ) 1
1 5

Amdahl et al. [25], Canada

Sensitivity analyses

Khan et al. [22], United Kingdom

Le Lay et al. [30], United Kingdom

Oral VNB vs. IV third-generation agents: VNB, paclitaxel, docetaxel, gemcitabine; NSCLC Erlotinib vs. gefitinib, NSCLC

Martikainen et al. [21], Finland

TMZ vs. PCV, GBM

Maroun et al. [34],

Oral UFT/folinic acid

Canada

Paz-Ares et al. [13], Spain

vs. parenteral FU/ folinic acid, mCRC

Sunitinib vs. BSC, GIST

CUA; no modeling was conducted

CMA; Markov structure; 1-wk cycle

CUA; Markov structure; 1-mo cycle

CUA; Markov structure; 1-mo cycle CMA; no modeling was conducted

CUA; Markov structure; 6-wk

Until disease progression or death; health care payer perspective

52 wk (1 y); UK National Health System perspective

Lifetime; health care payer perspective

Lifetime; Finnish health care payer perspective 1-yxx; hospital and government perspectives

6 y; Spanish National Health System

cycle Perrocheau et al. [28], France

Shiroiwa et al. [15], Japan

XELOX regimen vs. FOLFOX-6 regimen, CRC

Capecitabine vs. IV 5-FU/LV, CRC

CMA; no modeling was conducted

CUA; Markov structure; 3-mo cycle

LYs PFLYs QALYs

Cost for each treatment arm

QALYs PFLYs LYs

LMs PFLMs QALMs Total average cost per patient per treatment and per cycle QALYs PFLYs LYs

3.5 y (May 2003
December 2006); French health insurance perspective

Short-term (1 y) and long-term (up to 15 y); health care payer perspective

Disease management cost per patient Overall length of stay

QALYs

OS PFS Progression was based on RECIST criteria

RCT designed prospectively

Six health states: induction, remission with or without dose

Previously published trials (A literature search was

reduction, dropout, progression, death (not defined in the article) PFS Progression Death

conducted and the largest phase III RCT for each agent was selected.) Indirect comparison based on four published RCTs (one

Progression-free state** Progression stateyy Deathzz Not defined in the article

for erlotinib and three for gefitinib) Previously published trials (medical literature) Two large multicenter, phase III RCTs

PFS Progression Death (not defined in the article) Not defined in the article

RFS OS (not defined in the article)

RCT

One-way and probabilistic sensitivity analyses

Scenario sensitivity analyses

Erlotinib has about 80% chance of being costeffective in a subset of elderly poor performance patients with NSCLC unfit for chemotherapy who develop first-cycle (28 d) rash VNB appears as the most costminimizing therapeutic option

Scenario sensitivity analyses

Erlotinib is costeffective compared with gefitinib

Probabilistic sensitivity analyses

TMZ dominated PCV

None were

Oral UFT/folinic acid

conductedjjjj

Deterministic and probabilistic

cost of treatment was less than that of IV FU/folinic acid Sunitinib dominated BSC

sensitivity analyses Phase III RCT

Phase III RCT A Japanese phase II trial Other published

None were conducted

One-way and probabilistic sensitivity analyses

XELOX allowed a significant reduction in the disease management cost per patient, as well as the overall length of stay Capecitabine dominated 5-FU/LV

TAGEDNV A L U E I N H E A L T H R E G I O N A L I S S U E S 1 6 ( 2 0 1 8 ) 1
1 5

Lee et al. [32], Hong Kong

Erlotinib vs. supportive care (placebo), NSCLC

resources

(continued on next page)

5

6

Table 1 – continued Study, country

Ting et al. [31], United States

Zeng et al. [33], China

Comparators, neoplasm Erlotinib, afatinib, and cisplatinpemetrexed

CUA; Markov structure; 1-mo cycle

Time horizon; perspective Lifetime; health care payer perspective (although not

Primary outcome

Definition of effectiveness

Source of primary clinical data

Sensitivity analyses

Main finding

QALYs

DFS (not defined in the article)

Two phase III RCTs

Probabilistic sensitivity analyses

Erlotinib is the preferred treatment option

QALYs

PFS (not defined in the article)

Multicenter, doubleblind phase III RCT

One-way and probabilistic sensitivity analyses

Gefitinib maintenance therapy was not costeffective

mentioned in the article) CUA; semi-Markov structure; 3-wk length cycle

10 y; Chinese health care perspective

AP, accelerated phase; BEV/IFN, bevacizumab/interferon-a; BP, blast phase; BSC, best supportive care; CCyR, complete cytogenetic response; CEA, cost-effectiveness analysis; CMA, cost-minimization analysis; CML, chronic myeloid leukemia; CP, chronic phase; CRC, colorectal cancer; CUA, cost-utility analysis; EPO, recombinant erythropoietin; FOL FOX-6, oxaliplatin + LV + IV bolus 5-FU; 5-FU/LV, 5-fluorouracil/leucovorin; GBM, glioblastoma multiforme; GIST, gastrointestinal stroma tumors; IV, intravenous; LM, life-month; LY, life-year; mRCC, metastatic renal cell carcinoma; NHS, National Health Service; NSCLC, non
small cell lung cancer; OS, overall survival; PCV, procarbazine, lomustine + vincristine; PFLM, progressionfree life-month; PFLY, progression-free life-year; PFS, progression-free survival; DFS, disease-free survival; QALM, quality-adjusted life-month; QALY, quality-adjusted life-year; RCT, randomized clinical trial; RECIST, response evaluation criteria in solid tumors; RFS, relapse-free survival; SFN, sorafenib; TMZ, temozolomide; UFT, uracil-tegafur; VNB, vinorelbine; XELOX, oxaliplatin + capecitabine. * Time from randomization to the first documentation of objective disease. y A dominant treatment option is better in both clinical and economic aspects. z CP is the initial stable phase of CML, whereas AP and BP are both considered as advanced CML; resistance to treatment was also defined in detail in the article. x Time before the need of blood transfusion. jj Using Monte-Carlo simulation techniques. { Boag model combined with the competing risk model. # Observational period (5.6 y), 10-y follow-up and over lifetime. ** Time from the beginning of chemotherapy to the second relapse. yy Time from the second relapse to death. zz Death was modeled as an absorbing state. xx RCT recruitment period. jjjj Nevertheless, minimum and maximum costs for an outpatient oncology visit were included in their calculations.

V A L U E I N H E A L T H R E G I O N A L I S S U E S 1 6 ( 2 0 1 8 ) 1
1 5

(multiple comparisons), NSCLC Maintenance gefitinib vs. placebo, NSCLC

Analysis type; model structure

TAGEDNV A L U E I N H E A L T H R E G I O N A L I S S U E S 1 6 ( 2 0 1 8 ) 1
1 5

similar health states differently. For instance, progression-free survival was defined in two studies as the time from randomization to the first documentation of the disease of interest [2,12]. In another study [21], it was reported that progression-free survival is the time from the beginning of chemotherapy to the second relapse, progression state is the time from the second relapse to death, and death was modeled as an absorbing state. Other definitions included relapse events, which were defined consistently in only four of the studies as instances of relapse, new colon cancer, or death due to colon cancer or cancer treatment [14,16
18]. In a study by Ghatnekar et al. [29], different definitions were used in relevance to clinical effectiveness, with these being resistance to standard-dose treatment, rising white blood cell count after the initiation of treatment, failure to achieve complete hematological response after 3 to 6 months of therapy, a loss of complete hematological response at any time under therapy, failure to achieve major cytogenetic responses after 12 months of therapy, or a loss of major cytogenetic responses at any time during therapy. In the article published by Chen et al. [24], response to treatment was differently defined as the achievement of complete cytogenetic response in a period of 12 months, with this not clearly defined. Goss et al. [20] defined clinical effectiveness, which was the reduced transfusion independence, as a 50% or greater reduction in transfusion requirement. Regarding safety considerations, several studies focused on grade III or IV adverse drug reactions [12,13,19,20,23,25,31,32], and these were obtained mostly from RCTs. Interestingly, only one of the studies took all the side effects into consideration [22]. Some studies, however, reported only costs of treating adverse effects and/or medications used for them, without identifying the actual adverse events [15,16,18,34]. In the study by Chen et al. [24], the management cost of adverse effects was not included because of the nonavailability of data.

Sensitivity Analysis Sensitivity analyses were conducted in all the studies, except two [17,28]. Most authors performed deterministic and/or probabilistic one-way sensitivity analyses (n = 18), where relevant parameters in the latter varied by around §20%. Scenario analyses were conducted in four studies [19,23,30,32], and multivariate sensitivity analysis was an additional sensitivity analysis in five studies [18
20,23,29]. Several variables were reported as most influential in the incremental cost-effectiveness ratio (ICER) outcomes. In the study by Paz-Ares et al. [13], for example, the ICER was most influenced by the overall survival hazard ratio, whereas this was the utility value for a subgroup in the study by Zeng et al. [33]. Hsu et al. [14] concluded that a 20% variation in the cost of each drug administration visit leads to a 2.25-fold change in cost-savings, whereas Chen et al. [24] indicated that the ICER is sensitive to the discount rate and the acquisition cost of OACDs. In the analysis by Carlson et al. [23], the estimates of the treatment duration were among the most influential parameters. In the Canadian comparison by Amdahl et al. [25], the dominance of pazopanib was indicated via 79% of multivariate simulations to be within a predefined threshold of cost-effectiveness ratio. The conclusions of the included studies confirmed the robustness of the results against uncertainties.

Discussion This systematic review is the first to identify the characteristics, trends, and reporting quality of published research in economic evaluations about the use of OACDs, regardless of drug or cancer type. Other reviews do exist in the literature, but those are drugor cancer-specific and/or do not focus on critically appraising the

7

methodological characteristics of various economic studies. These previous reviews are limited, however, as compared with the present one, in the ability to comprehensively identify methodological gaps in the literature and the ability to make recommendations for future research on OACDs, regardless of medication or underlying disease. This is important to note because, unlike the previous reviews, and as already discussed earlier, this review is not to generate evidence or does not come with recommendations about how a particular medication should be used or how a specific underlying cancer should be handled. A recent review by Shen et al. [43] summarized the economic evaluations of targeted oral chemotherapies, focusing on budget impact analyses only. Two studies compared an OACD (i.e., topotecan) with best supportive care by conducting systematic reviews of previously published RCTs [44,45], whereas other review articles discussed the clinical effects of an OACD, namely, capecitabine, against the parenteral alternatives mostly [46
51]. Discussion of economic evaluation methods was included in two review articles [52,53], but these were for a single type of cancer (advanced colorectal cancer), which included only the “costeffectiveness” evaluations and/or included articles till 2009 only. This review included 21 economic evaluations of OACDs. Although the results of the studies provided answers to the questions for which they were conducted, a broad range of methodological trends and gaps were observed across studies. This related to the type of evaluation, time horizon, time adjustment, outcome measures, modeling, perspective, and transparency. Methodological gaps in the included studies reflected a general lack of adherence to current standards for conducting and reporting economic evaluations, such as those by the Panel on CostEffectiveness in Health and Medicine [54] or by the British Medical Journal’s guidelines for economic submissions [55]. This mostly resulted in different ranges of ICERs, limiting the robustness of the body of evidence and the guidance to decision makers in other settings. The use of health care resources and the work productivity are anticipated to be primarily affected in a cancer population. Nevertheless, only two studies indicated that they are targeting the broader social perspective, with these also not including the indirect cost of lost productivity, which is inappropriate [56]. All remaining studies were from the payer perspective. This is a common research perspective and typically includes direct medical expenses. Most of the studies included modeling analyses, which is appropriate because modeling is valuable for better understanding the different effects and costs of an intervention. Nevertheless, the validity of modeling and its results relies on the evidence and assumptions it is based on. Here, about the evidence, all reviewed studies were nonexperimental in design, whereby they relied on secondary data available at external resources. In fact, most of the modeling studies built on the same source of data, which was a single published RCT [13,14,16
19,28,33]. This is unfortunate because prospective RCTs would have generated more robust and convincing evidence for the local setting. Even if RCTs are not feasible because of the limitations in resources, the incorporation of local-based cohort data would have provided more relevant results. About modeling assumptions, none of the modeling studies clearly indicated assumptions made, which would reflect on their reproducibility. As also noted in Table 2, several studies have provided partial or insufficient details regarding model choice and the assumptions made. The focus in these studies was largely on the longer term effects. Here, the Markov model is ideal for pharmacoeconomics evaluations of a recurrent disease such as cancer, where it has an advantage over decision-analytic models in relation to incorporating longer time horizons. It extends the results of clinical trials and extrapolates intermediate end points into final

8

Table 2 – Reporting quality of economic evaluations using CHEERS checklist. Section/item

Aller et al. [12]

Amdahl et al. [25]

Carlson et al. [23]

Cassidy et al. [18]

Chen et al. [24]

Di Costanzo et al. [16]

Ghatnekar et al. [29]

Sunitinib

Pazopanib

Erlotinib

Capecitabine

Imatinib

Capecitabine

Dasatinib

1 2 3

A (Acheived) A A

A A A

PA A PA

A PA A

A A A

A A A

A A A

4

A

A

A

A

A

A

A

5 6 7 8 9

A A A A A

A A A A A

A A A PA A

A A A A A

A A A PA PA

A A A PA A

A A A A

PA PA A A

PA PA A A

A PA PA PA

PA PA A A

PA PA A A

Choice of health outcomes Measurement of effectiveness Estimating resources and costs Currency, price date, conversion Choice of model Assumptions Analytical model Results Study parameters Incremental costs and outcomes Characterizing uncertainty Characterizing heterogeneity Discussion Study findings, limitations, generalizability, and current knowledge Other Source of funding Conflicts of interest Result of the study Cost-effectiveness state

10 11 12 13

A A A A PA (Partially acheived) PA A A PA

14 15 16

A A PA

A A A

A A A

PA A A

NA PA NA

PA A A

A A PA

17 18

A A

A A

A A

A PA

PA PA

A A

A A

19 20

A NA (Not acheived)

A NA

A NA

A NA

PA NA

PA NA

A NA

21

PA

A

PA

PA

PA

PA

A

22 23

A A + Dom

A A + Dom

A A + Dom

NA NA + Dom

PA PA + Dom

PA A + Dom

PA A + Cost-saving

Section/item

No. Glen and Cassidy [17] Goss et al. [20]

OACD Title, abstract, and introduction Title Abstract Background and objectives

1 2 3

Hisashige et al. [19]

Hsu et al. [14]

Khan et al. [22]

Le Lay et al. [30]

Lee et al. [32]

Capecitabine

Lenalidomide

Uracil-tegafur

Capecitabine

Erlotinib

Vinorelbine

Erlotinib

PA PA A

A PA A

A PA A

A A A

A A A

PA PA PA

A PA PA (continued on next page)

V A L U E I N H E A L T H R E G I O N A L I S S U E S 1 6 ( 2 0 1 8 ) 1
1 5

OACD Title, abstract, and introduction Title Abstract Background and objectives Methods Target population and subgroups Setting and location Study perspective Comparators Time horizon Discount rate

No.

Section/item OACD Title, abstract, and introduction Title Abstract Background and objectives Methods Target population and subgroups Setting and location Study perspective Comparators Time horizon Discount rate Choice of health outcomes Measurement of effectiveness

4

A

A

A

A

A

PA

A

5 6 7 8 9 10 11 12 13

A A A PA NA PA PA PA PA

A A A A NA PA A A A

A A A A A PA A A A

A A A PA A PA A A A

PA NA A A PA PA PA A A

PA A A A NA PA PA A PA

A PA A A A PA PA PA PA

14 15 16

PA NA PA

A A A

NA NA A

PA PA PA

PA PA A

PA PA PA

A PA A

17 18

PA NA

A A

A A

A A

PA A

A A

A A

19 20

NA NA

A NA

A NA

A NA

A PA

A NA

A NA

21

PA

A

A

PA

PA

PA

PA

22 23

PA NA + Dom (Dominance)

NA NA + Dom

NA NA + Dom

PA NA + Dom

PA A + Dom

NA NA + Cost-saving

A A + Cost-effective

No.

Maroun et al. [34]

Martikainen et al. [21] Perrocheau et al. [28] Paz-Ares et al. 2008 Shiroiwa et al. [15] Ting et al. [31]

Uracil-tegafur

Temozolomide

Capecitabine

Sunitinib

Capecitabine

Erlotinib and afatinib Gefitinib

1 2 3

A A A

A A A

A A A

A A A

A A A

A PA PA

A A A

4

A

A

A

A

A

A

A

5 6 7 8 9 10 11

A A A NA NA PA PA

A A A PA PA A A

A A A NA NA PA NA

A A A PA A PA A

A A A A PA PA PA

A PA A A PA PA PA

A A A A A PA A

Zeng et al. [33]

9

(continued on next page)

TAGEDNV A L U E I N H E A L T H R E G I O N A L I S S U E S 1 6 ( 2 0 1 8 ) 1
1 5

Methods Target population and subgroups Setting and location Study perspective Comparators Time horizon Discount rate Choice of health outcomes Measurement of effectiveness Estimating resources and costs Currency, price date, conversion Choice of model Assumptions Analytical model Results Study parameters Incremental costs and outcomes Characterizing uncertainty Characterizing heterogeneity Discussion Study findings, limitations, generalizability, and current knowledge Other Source of funding Conflicts of interest Result of the study Cost-effectiveness state

10

Table 2 – continued No.

Estimating resources and costs Currency, price date, conversion Choice of model Assumptions Analytical model Results Study parameters Incremental costs and outcomes Characterizing uncertainty Characterizing heterogeneity Discussion Study findings, limitations, generalizability, and current knowledge Other Source of funding Conflicts of interest Result of the study

12 13

A A

A PA

A PA

A A

A A

PA PA

A A

14 15 16

NA NA PA

PA A A

NA NA NA

A A PA

PA A PA

PA PA PA

PA A A

17 18

PA NA

A A

A NA

A A

A A

A A

A A

19 20

NA NA

A NA

NA NA

A PA

A NA

A NA

A NA

21

PA

A

PA

PA

A

A

A

22 23

PA NA + (Support study drug) Cost-saving

PA PA +

PA NA =

NA NA +

A A +

A NA +

Dom

Cost-saving

Dom

Dom

Cost-effective

A A ¡ (Do not support study drug) Not cost-effective

Cost-effectiveness state

Aller et al. [12]

Amdahl et al. [25]

CHEERS, Consolidated Health Economic Evaluation Reporting Standards.

Carlson et al. [23]

Cassidy et al. [18]

Chen et al. [24]

Di Costanzo et al. [16]

Ghatnekar et al. [29] V A L U E I N H E A L T H R E G I O N A L I S S U E S 1 6 ( 2 0 1 8 ) 1
1 5

Section/item

11

TAGEDNV A L U E I N H E A L T H R E G I O N A L I S S U E S 1 6 ( 2 0 1 8 ) 1
1 5

Table 3 – A summary of study comparisons. Oral drug Capecitabine (or XELOX) Dasatinib Erlotinib

Gefitinib Imatinib Linalidomide Pazopanib Sunitinib Temozolomide Uracil-tegafur Vinorelbine

Comparator 5-FU (or FOLFOX-6) High-dose imatinib BSC Docetaxel and pemetrexed Gefitinib Afatinib or cisplatin-pemetrexed Placebo IFN PCV Sunitinib SFN and BEV/IFN, BSC EPO Surgery alone, 5-FU IV third-generation agents: VNB, paclitaxel, docetaxel, gecitabine

Number of comparisons 6 1 4

1 1 1 1 2 1 2 1

BSC, best supportive care; BEV/IFN, bevacizumab/interferon-a; EPO, recombinant erythropoietin; FOLFOX-6, oxaliplatin + LV + IV bolus 5-FU; 5FU/LV, 5-fluorouracil/leucovorin; IFN, interferon-a; PCV, procarbazine, lomustine + vincristine; SFN, sorafenib; VNB, vinorelbine; XELOX, oxaliplatin + capecitabine.

outcomes [57
59]. Although most of the modeling-based studies were indeed Markovian, about a third of the modeling studies favored a non-Markov modeling. Even in studies that performed Markov modeling, it seems that most studies included a shorter term horizon follow-up, where this was between 6 and 10 years of follow-up. It is believed that this is mostly due to an existing gap in the local clinical and QOL evidence available to populate the longer horizon multistate model. Only one of the studies used a combination of a decision tree and a Markov model to capture data for both the short-term and the longer term events of therapy. A conceptual framework element in modeling is the definition of the time horizon or the cycle length in Markov modeling. It is through this that health utilities and costs that are attached to any state are accumulated [60]. Here, extending life and enhancement in its quality are the fundamental aims of medical care. Hence, a time horizon that spans the patient’s entire life, representing life expectancy or QALY, has been well accepted in the clinical decision analyses. Nevertheless, in a third of the included studies (all Markov-based) that had a lifelong time horizon, the risks of events in modeling were inappropriately time-constant, where researchers did not adjust risks over time. This is particularly relevant in a disease such as cancer. Markov simulation frequently assumes that the average health status annual hazard is constant over the entire lifetime. This can lead to a systematic error because it is inconsistent with actually reported risks of cancer disease, whereby mortality and morbidity decline with time since diagnosis. The clarity of the costing methodology and cost components is another aspect to discuss. The unclear reporting of cost components and measuring methods can make it difficult for the different settings to benefit from and apply results. For example, although most articles (n = 19) provided definitions of the cost components considered, none indicated whether costs used were hospital charges or costs. This is important because a hospital charge is not an ideal estimation of cost because this is decided on to compensate for the cost of other services and facilities provided by the hospital setting. Using charges instead of costs leads to less robust conclusions. QALY is crucial in most chronic diseases and is of particular importance in patients with cancer. QALY is a widely preferred summary multidimensional value of outcomes in pharmacoeconomics. It incorporates trade-offs between QOL and quantity of

life in a common metric. Indeed, most of the studies in this review assessed QALYs or quality-adjusted life-months [61
64]. Nevertheless, QALY calculations are based on patient value or preference-based estimations, whereby none of the included studies involved QOL data collection from patients, except the study by Goss et al. [20]. Despite the importance of QALY in decision making, decision makers may find it more clinically relevant to consider outcomes that are presented regarding health states, such as death, recovery, and relapse. Out of the included studies, however, none considered such outcomes. This is despite the fact that the published RCTs, on which most of the included studies are based, included health status outcomes. The common presentation of health outcomes in pharmacoeconomics as QALYs only, instead of both QALYs and specific health states, is a shortcoming that requires attention. Despite all method variability, the overall conclusions of studies about the favorable cost effectiveness of OACDs were mostly consistent. The exception to this was reported in a study by Zeng et al. [33], in which the high price of the oral drug was the most influential parameter in the model. Nevertheless, according to the article’s authors, reducing the price or increasing the willingness-to-pay threshold would shift the odds toward the cost effectiveness of the drug. Drug-related adverse events have a significant influence on the direct cost and cost effectiveness and, hence, are anticipated to be of primary consideration when differentiating between oral and intravenous chemotherapy. Although most studies included the analysis of adverse event costs, none of the studies modeled discontinuations due to adverse events. The extent of discontinuation and its cost are not clear in the studies, which were also not included in sensitivity analyses conducted. To consider the side effects that are associated with discontinuations as equivalent to those that are not is inappropriate when guiding decision making. Most studies demonstrated the results of sensitivity analyses when the value of a broad range of variables was changed, where studies were most sensitive to clinical outcomes as compared with cost. Nevertheless, none of these studies included justifications for the changes made. Also, there is limited variability in the types of sensitivity analysis conducted. Most studies carried out one-way sensitivity analyses only, which can, in the absence of correlation, underestimate uncertainty, even if interpreted correctly [65].

12

V A L U E I N H E A L T H R E G I O N A L I S S U E S 1 6 ( 2 0 1 8 ) 1
1 5

Table 4 – Health state utility values. Study, country Aller et al. [12], Spain

Amdahl et al. [25], Canada

Carlson et al. [23], United States

Cassidy et al. [18], United Kingdom

Chen et al. [24], China Di Costanzo et al. [16], Italy

Ghatnekar et al. [29], Sweden

Glen and Cassidy [17], United Kingdom Goss et al. [20], United States

Hisashige et al. [19], Japan Hsu et al. [14], Taiwan

Khan et al. [22], United Kingdom

Lee et al. [32], Hong Kong

Utility value by health states

Primary source of method and utility estimation

IFN: 0.730 Sunitinib: 0.718, 0.758* SFN: 0.731 BEV/IFN: 0.730 During second-line therapy with any other treatment: 0.6398 After termination of second-line therapy: 0.560 Pazopanib, stable/prerelapse: 0.7089 (0.0193) Pazopanib, postprogression vs. preprogression survival:
0.1580 (0.0395) Sunitinib, stable/prerelapse: 0.6832 (0.0236) Sunitinib, postprogression vs. preprogression survival:
0.1323 (0.0331)

From the Sunitinib trial QOL data (EQ-5D instrument)y [35] Expert panels Published reports

Stable disease on oral therapy: 0.67 Stable disease on IV therapy: 0.65 Recently progressed disease: 0.47 Stable disease plus neutropenia: 0.56 Stable disease plus febrile neutropenia: 0.56 Stable disease plus diarrhea: 0.61 Stable disease plus nausea: 0.61 Stable disease plus stomatitis: 0.61 Stable disease plus rash: 0.62 Dead: 0.00 Chemotherapy: 0.8 Stable/prerelapse: 0.86 Postrelapse: 0.59 Death: 0 Details not defined in the article Chemotherapy: 0.8 Stable/prerelapse: 0.86 Postrelapse: 0.59 Death: 0 CP responder: 0.90 CP nonresponder: 0.72 AP: 0.53 BP: 0.29 Death (from either CML-related or non
CML-related causes): 0 Details not defined in the article Transfusion-dependent: 0.50 50% reduced requirement: 0.81 Transfusion-independent: 0.91 Stage III after surgery: 0.75 (0.55, 0.85) Stage III after metastasis: 0.20 (0.00, 0.40) Chemotherapyx: 0.8 Prerelapse: 0.86 Postrelapse: 0.59 Death: 0 Preprogression: 0.6482 vs. 0.6438 Postprogression: 0.5517 vs. 0.5760 for erlotinib vs. placebo, respectively. In the rash subgroup, preprogression: 0.6407 vs. 0.6193 PFS: 0.653 Progression: 0.473 Death: 0

Functional Assessment of Chronic Illness Therapy—Fatigue with Its Additional Concerns Module, the Functional Assessment of Cancer Therapy
Kidney Symptom Index 19, the Seville Quality of Life Questionnaire, and the Cancer Therapy Satisfaction Questionnaire, collected from patients during the RCT Published literature by Nafees et al.z [36]

Ramsey et al. [37]

EQ-5D collected by Reed et al. [38] Ramsey et al. [37]

A TTO technique using the EQ-5D instrument among 100 laypersons in the United Kingdom [39]

From medical literature Health utility interviews with a group of patients with MDS in the United States (N = 8) Published study by Ness et al. [40] From QOL in survivors of colorectal carcinoma; Ramsey et al. [37]

EQ-5D instrument forms filled during the RCT

Published literature by Nafees et al. [36]

(continued)

13

TAGEDNV A L U E I N H E A L T H R E G I O N A L I S S U E S 1 6 ( 2 0 1 8 ) 1
1 5

Table ? 4 – continued Study, country Martikainen et al. [21], Finland

Paz-Ares et al. [13], Spain

Shiroiwa et al. [15], Japan

Ting et al. [31], United States

Zeng et al. [33], China

Utility value by health states S1jj: 0.55 S2: 0.41 S3: 0.43 S4: 0.31 S5: 0.14 Sunitinib, no disease progression: 0.712, 0.769{ BSC, no disease progression: 0.781 Disease progression: 0.577 Chemotherapy: 0.8 Disease-free: 0.86 Recurrence: 0.59 Death: 0 Stable disease under oral therapy: 0.67 Stable disease under IV therapy: 0.65 Stable disease with neutropenia: 0.56 Stable disease with diarrhea: 0.61 Stable disease with stomatitis/mucositis: 0.61 Stable disease with rash: 0.62 Disease progression: 0.47 PFS on oral therapy: 0.67 PFS plus rash: 0.62 PFS plus diarrhea: 0.61 PS: 0.47 (range 0.19
0.56) for both arms

Primary source of method and utility estimation QOL estimates and utility scores were gathered using proxy respondents and a VAS method [41]

From the Sunitinib trial QOL data (EQ-5D instrument) [35] Ramsey et al. [37]

From previous cost-effectiveness studies on the same population [42]

Published literature by Nafees et al. [36]

AP, accelerated phase; BEV/IFN, bevacizumab/interferon-a; BP, blast phase; BSC, best supportive care; CML, chronic myeloid leukemia; CP, chronic phase; EQ-5D, EuroQol five-dimensional questionnaire; 5-FU/LV, 5-fluorouracil/leucovorin; IFN, interferon-a; IV, intravenous; MDS, myelodysplastic syndrome; NSCLC, non
small cell lung cancer; PCV, procarbazine, lomustine + vincristine; PFS, progression-free survival; PS, progressed survival; QOL, quality of life; RCT, randomized controlled trial; SFN, sorafenib; TTO, time trade-off; VAS, visual analogue scale. * Sunitinib utility value: during the treatment period, 0.718; during the off-treatment period: 0.758. y Calculated using the weighted average utility values. z Community-based study from the United Kingdom, which used the EQ-5D and standard gamble interview to assess QOL in advanced NSCLC. x During the 24-wk chemotherapy for both capecitabine and 5-FU/LV. jj Disease stages are as follows: S1: 45-y-old patient with glioma who is having the first surgery; S2: same patient after surgery and radiotherapy on the tumor area; S3: chemotherapy alternative temozolomide; S4: chemotherapy alternative PCV; and S5: progression. { During the 4 wk with the treatment of each cycle: 0.712;during the 2 wk without the treatment of each cycle (rest weeks): 0.769.

Also, literature research has reported that in pharmacoeconomics research, favorable results were more associated with industry funding as compared with other types of funding [66]. In this review, it has been found that most of the studies are industry-funded in sponsorship of the oral formulation of the chemotherapy. Poor reporting of essential research aspects has been found, primarily relating to the studies’ time horizon, discount rate, choice of health outcomes, choice of model and assumptions made, characterizing heterogeneity, limitations, generalizability of results, source of funding, and conflicts of interest. On the basis of this study’s findings, several areas for improving future evidence can be proposed. Presentation of study details should be enhanced in published reports. There are several quality assessment checklists that authors can use to enhance reporting of important aspects of economic studies, including several published instruments such as the CHEERS reporting checklist [27], which has been found in the literature to be a well-established commonly used tool [67
69]. There are design concerns about models’ construction and analyses. Researchers should enhance their compliance to good practices by using existing good practice guidelines, such as the health economic evaluation quality appraisal instrument, those proposed by the International Society of Pharmacoeconomics and Outcome Research [27,70,71], and other good practice modeling guidelines published, for example, by Philips et al. [72,73] and Sculpher et al. [74].

Future evidence can benefit from research that better considers the nonmedical costs of therapies. This is particularly important in relation to OACDs because intravenous therapies have a significant impact on patients’ lives, where patients spend a considerable amount of time traveling to, waiting for, and receiving cancer care [75]. In cancer, it is possible that lost income has a larger impact than out-of-pocket costs [10]. Long-term modeling should be incorporated more often for costs and outcomes. Future research in local settings should better document and audit social effects of chemotherapies, including the association between intermediate and final results of interest. More clinically intuitive cost-effectiveness outcomes need to increase in literature, whereby a justifiable cost is required to be assessed against specific health state outcomes, instead of QALY outcomes only. For the purpose of economic models that guide decisions based on local practices, published RCTs can provide limited insight into generalizable effectiveness, intervention and patient management processes, contextual factors, experiences, and implementation bias. Hence, future local economic models should incorporate locally based sources of data. If local RCTs are not feasible for this purpose, locally relevant decision making can be achieved via designs that can replace local RCTs or be combined with published RCT data or by designs that can include discrete choice experiments and cohort data. Recent methods such as the multicriteria decision modeling should be used to connect

14

V A L U E I N H E A L T H R E G I O N A L I S S U E S 1 6 ( 2 0 1 8 ) 1
1 5

different sources of evidence into more relevant conclusions for local decision making [76,77]. Finally, several limitations are associated with this review, which are mainly related to its inclusion and exclusion criteria. First, there is the English-restricted literature search, which may have missed relevant non-English research, such as that in French, Chinese, or German. Authors do not have the resources to translate all non-English research literature. Second, although a comprehensive literature search was conducted in this study, as per quality assessment tools [78], relevant studies could have also been explored in other research database sources including Global Health, Cost-Effectiveness Analysis Registry, Open Grey, and Information Technology Assessment International. Finally, additional search terms to those in the study or other combinations of them are always possible and can generate further studies.

Conclusions Given the unavoidable certainty of opportunity costs in policy choices, it is vital that “economic evidence” plays some part in the decision-making process. A synthesis of evidence from a number of relevant research studies should be an integral component of the policymaking process. Economic evaluations provide a valuable basis either for decision making or for knowledge generation. This review has presented that despite several economic analyses of OACDs having been published, essential aspects remain under-reported, for example, modeling assumptions, cost calculations, time horizons, outcomes, presentation, and sources of data. These limit the generalizability and transferability of results, and hence the use of results from economic evaluations produced in other settings should be interpreted with caution. It is important to note that although the identified models varied in their methodologies, the conclusions seemed broadly consistent. But consistent conclusions may not have a cumulative value to them because of the heterogeneity of methods. This research can be used as guidance for further analysis that avoids the methodological shortcomings of existing studies, especially designed to the setting and combined with local data. Source of financial support: No funding was received for the purpose of this study.

Supplemental Materials Supplemental material accompanying this article can be found in the online version as a hyperlink at http://dx.doi.org/10.1016/j. vhri.2018.05.003 or, if a hard copy of article, at www.valuein healthjournal.com/issues(select volume, issue, and article). TAGEDH1R E F E R E N C E S TAGEDN

[1] World Health Organization. Cancer, fact sheet. 2015. Available from: http://www.who.int/mediacentre/factsheets/fs297/en/. [Accessed August 13, 2017]. [2] Betcher J, Dow E, Khera N. Oral chemotherapy in patients with hematological malignancies—care process, pharmacoeconomic and policy implications. Curr Hematol Malig Rep 2016;11:288–94. [3] Kavookjian J, Wittayanukorn S. Interventions for adherence with oral chemotherapy in hematological malignancies: a systematic review. Res Social Adm Pharm 2015;11:303–14. [4] Schott S, Shneeweiss A, Reinhardt J, et al. Acceptance of oral chemotherapy in breast cancer patients—a survey study. BMC Cancer 2011;11:129.

[5] Liu G, Franssen E, Fitch MI, et al. Patient preferences for oral versus intravenous palliative chemotherapy. J Clin Oncol 1997;15:110–5. [6] Foulan V, Schoffski P, Wolter P. Patient adherence to oral anticancer drugs: an emerging issue in modern oncology. Acta Clin Belg 2011;66:85–96. [7] Tefferi A, Kantarjian H, Rajkimar SV, et al. In support of a patientdriven initiative and petition to lower the high price of cancer drugs. Mayo Clin Proc 2015;90:996–1000. [8] Benjamin L, Cotte FE, Philippe C, et al. Physicians’ preferences for prescribing oral and intravenous anticancer drugs: a discrete choice experiment. Eur J Cancer 2012;48:912–20. [9] Jonsson B, Wilking N. A global comparison regarding patient access to cancer drugs. Ann Oncol 2007;18(Suppl. 3):iii1–77. [10] Wayne T. Benefits outweigh costs in universal healthcare: business case for reimbursement of take-home cancer medicines in Ontario and Atlantic Canada. Am J Med Med Sci 2014;4:126–38. €lle V, et al. Impact of the healthcare [11] Benjamin L, Buthion V, Gwenae payment system on patient access to oral anticancer drugs: an illustration from the French and United States contexts. BMC Health Serv Res 2014;14:274. [12] Aller EC, Maroto P, Kreif N, et al. Cost-effectiveness evaluation of sunitinib as first-line targeted therapy for metastatic renal cell carcinoma in Spain. Clin Transl Oncol 2011;13:869–77. [13] Paz-Ares L, Del Muro XG, Grande E, et al. Cost-effectiveness analysis of sunitinib in patients with metastatic and/or unresectable gastrointestinal stroma tumours (GIST) after progression or intolerance with imatinib. Clin Transl Oncol 2008;10:831–9. [14] Hsu TC, Chen HH, Yang MC, et al. Pharmacoeconomic analysis of capecitabine versus 5-fluorouracil/leucovorin as adjuvant therapy for stage III colon cancer in Taiwan. Value Health 2011;14:647–51. [15] Shiroiwa T, Fukuda T, Shimozuma K, et al. Cost-effectiveness analysis of capecitabine compared with bolus 5-fluorouracil/l-leucovorin for the adjuvant treatment of colon cancer in Japan. Pharmacoeconomics 2009;27:597–608. [16] Di Costanzo F, Ravasio R, Sobrero A, et al. Capecitabine versus bolus fluorouracil plus leucovorin (folinic acid) as adjuvant chemotherapy for patients with Dukes’ C colon cancer: economic evaluation in an Italian NHS setting. Clin Drug Investig 2008;28:645–55. [17] Glen H, Cassidy J. Redefining adjuvant chemotherapy in patients with stage III colon cancer: X-ACT trial. Expert Rev Anticancer Ther 2008;8:547–51. [18] Cassidy J, Douillard JY, Twelves C, et al. Pharmacoeconomic analysis of adjuvant oral capecitabine vs intravenous 5-FU/LV in Dukes’ C colon cancer: the X-ACT trial. Br J Cancer 2006;94:1122–9. [19] Hisashige A, Yoshida S, Kodaira S. Cost-effectiveness of adjuvant chemotherapy with uracil-tegafur for curatively resected stage III rectal cancer. Br J Cancer 2008;99:1232–8. [20] Goss TF, Szende A, Schaefer C, et al. Cost effectiveness of lenalidomide in the treatment of transfusion-dependent myelodysplastic syndromes in the United States. Cancer Control 2006;13(Suppl.):17–25. [21] Martikainen JA, Kivioja A, Hallinen T, et al. Economic evaluation of temozolomide in the treatment of recurrent glioblastoma multiforme. Pharmacoeconomics 2005;23:803–15. [22] Khan I, Morris S, Hackshaw A, et al. Cost-effectiveness of first-line erlotinib in patients with advanced non-small-cell lung cancer unsuitable for chemotherapy. BMJ Open 2015;5:e006733. [23] Carlson JJ, Reyes C, Oestreicher N, et al. Comparative clinical and economic outcomes of treatments for refractory non-small cell lung cancer (NSCLC). Lung Cancer 2008;61:405–15. [24] Chen Z, Wang C, Xu X, et al. Cost-effectiveness study comparing imatinib with interferon-alpha for patients with newly diagnosed chronic-phase (CP) chronic myeloid leukemia (CML) from the Chinese public health-care system perspective (CPHSP). Value Health 2009;12 (Suppl. 3):S85–8. [25] Amdahl J, Diaz J, Park J, et al. Cost-effectiveness of pazopanib compared with sunitinib in metastatic renal cell carcinoma in Canada. Curr Oncol 2016;23:e340–54. [26] Moher D, Liberati A, Tetzlaff J, Altman DG. PRISMA Group. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. PLoS Med 2009;6:e1000097. [27] Husereau D, Drummond M, Petrou S, et al. Consolidated Health Economic Evaluation Reporting Standards (CHEERS)—explanation and elaboration: a report of the ISPOR Health Economic Evaluation Publication Guidelines Good Reporting Practices Task Force. Value Health 2013;16:231–50. [28] Perrocheau G, Bennouna J, Ducreux M, et al. Cost-minimisation analysis in first-line treatment of metastatic colorectal cancer in France: XELOX versus FOLFOX-6. Oncology 2010;79:174–80. [29] Ghatnekar O, Hjalte F, Taylor M. Cost-effectiveness of dasatinib versus high-dose imatinib in patients with chronic myeloid leukemia (CML), resistant to standard dose imatinib—a Swedish model application. Acta Oncol 2010;49:851–8.

TAGEDNV A L U E I N H E A L T H R E G I O N A L I S S U E S 1 6 ( 2 0 1 8 ) 1
1 5

[30] Le Lay K, Myon E, Hill S, et al. Comparative cost-minimisation of oral and intravenous chemotherapy for first-line treatment of nonsmall cell lung cancer in the UK NHS system. Eur J Health Econ 2007;8:145–51. [31] Ting J, Ho T, Xiang P, et al. Cost-effectiveness and value of information of erlotinib, afatinib, and cisplatin-pemetrexed for first-line treatment of advanced EGFR mutation-positive non-small-cell lung cancer in the United States. Value Health 2015;18:774–82. [32] Lee VW, Schwander B, Lee HV. Effectiveness and cost-effectiveness of erlotinib versus gefitinib in first-line treatment of epidermal growth factor receptor
activating mutation-positive non
small-cell lung cancer patients in Hong Kong. Hong Kong Med J 2014;20:178–86. [33] Zeng X, Li J, Peng L, et al. Economic outcomes of maintenance gefitinib for locally advanced/metastatic non-small-cell lung cancer with unknown EGFR mutations: a semi-Markov model analysis. PLoS One 2014;9:e88881. [34] Maroun JA, Asche C, Romeyer F, et al. A cost comparison of oral tegafur plus uracil/folinic acid and parenteral fluorouracil for colorectal cancer in Canada. Pharmacoeconomics 2003;21:1039–51. [35] Demetri GD, Van Oosterom AT, Garrett CR, et al. Efficacy and safety of sunitinib in patients with advanced gastrointestinal stromal tumour after failure of imatinib: a randomised controlled trial. Lancet 2006;368:1329–38. [36] Nafees B, Stafford M, Gavriel S, et al. Health state utilities for non small cell lung cancer. Health Qual Life Outcomes 2008;6:84. [37] Ramsey SD, Andersen MR, Etzioni R, et al. Quality of life in survivors of colorectal carcinoma. Cancer 2000;88:1294–303. [38] Reed SD, Anstrom KJ, Li Y, et al. Updated estimates of survival and cost effectiveness for imatinib versus interferon-alpha plus low-dose cytarabine for newly diagnosed chronic-phase chronic myeloid leukaemia. Pharmacoeconomics 2008;26:435–46. [39] Levy A, Szabo SM, Tabberer M, et al. Utility values for health states for chronic myelogenous leukaemia (CML): estimates from laypersons in Australia, the United Kingdom (UK) and Canada. Value Health 2007;10:A344. [40] Ness RM, Holmes AM, Klein R, et al. Utility valuations for outcome states of colorectal cancer. Am J Gastroenterol 1999;94:1650–7. [41] Torrance GW, Feeny D, Furlong W. Visual analog scales: Do they have a role in the measurement of preferences for health states? Med Decis Making 2001;21:329–34. [42] Carlson J, Garrison L, Ramsey S, et al. The potential clinical and economic outcomes of pharmacogenomic approaches to EGFR-tyrosine kinase inhibitor therapy in non-small-cell lung cancer. Value Health 2009;12:20–7. [43] Shen C, Chien CR, Geynisman DM, et al. A review of economic impact of targeted oral anticancer medications. Expert Rev Pharmacoecon Outcomes Res 2014;14:45–69. [44] Hartwell D, Jones J, Loveman E, et al. Topotecan for relapsed small cell lung cancer: a systematic review and economic evaluation. Cancer Treat Rev 2011;37:242–9. [45] Loveman E, Jones J, Hartwell D, et al. The clinical effectiveness and cost-effectiveness of topotecan for small cell lung cancer: a systematic review and economic evaluation. Health Technol Assess 2010;14:1–204. [46] Jones L, Hawkins N, Westwood M, et al. Systematic review of the clinical effectiveness and cost-effectiveness of capecitabine (Xeloda) for locally advanced and/or metastatic breast cancer. Health Technol Assess 2004;8:1–143. iii, xiii
xvi,. [47] Twelves CJ. Xeloda in adjuvant colon cancer therapy (X-ACT) trial: overview of efficacy, safety, and cost-effectiveness. Clin Colorectal Cancer 2006;6:278–87. [48] Ward S, Kaltenthaler E, Cowan J, et al. Clinical and cost-effectiveness of capecitabine and tegafur with uracil for the treatment of metastatic colorectal cancer: systematic review and economic evaluation. Health Technol Assess 2003;7:1–93. [49] Ward S, Kaltenthaler E, Cowan J, et al. The clinical and economic benefits of capecitabine and tegafur with uracil in metastatic colorectal cancer. Br J Cancer 2006;95:27–34. [50] Best JH, Garrison LP. Economic evaluation of capecitabine as adjuvant or metastatic therapy in colorectal cancer. Expert Rev Pharmacoecon Outcomes Res 2010;10:103–14. [51] Comella P, Casaretti R, Sandomenico C, et al. Capecitabine, alone and in combination, in the management of patients with colorectal cancer: a review of the evidence. Drugs 2008;68:949–61. [52] Krol M, Koopman M, Uyl-de Groot C, et al. A systematic review of economic analyses of pharmaceutical therapies for advanced colorectal cancer. Expert Opin Pharmacother 2007;8:1313–28.

15

[53] Leung HW, Chan AL, Leung MS, et al. Systematic review and quality assessment of cost-effectiveness analysis of pharmaceutical therapies for advanced colorectal cancer. Ann Pharmacother 2013;47:506–18. [54] Gold M. Panel on cost-effectiveness in health and medicine. Med Care 1996;34:DS197–9. [55] Drummond M, Jefferson T. Guidelines for authors and peer reviewers of economic submissions to the BMJ. The BMJ Economic Evaluation Working Party. BMJ 1996;313:275–83. [56] Mattke S, Balakrishnan A, Bergamo G, et al. A review of methods to measure health-related productivity loss. Am J Manag Care 2007;13:211–7. [57] Kay R. A Markov model for analysing cancer markers and disease states in survival studies. Biometrics 1986;42:855–65. [58] Briggs A, Sculpher M. An introduction to Markov modelling for economic evaluation. Pharmacoeconomics 1998;13:397–409. [59] Barton P, Bryan S, Robinson S. Modelling in the economic evaluation of health care: selecting the appropriate approach. J Health Serv Res Policy 2004;9:110–8. [60] Sonnenberg F, Beck J. Markov models in medical decision making: a practical guide. Med Decis Making 1993;13:322–38. [61] Mehrez A, Gafni A. Quality-adjusted life years, utility theory, and health years equivalents. Med Decis Making 1989;9:142–9. [62] Gyrd-Hansen D. Willingness to pay for QALY. Pharmacoeconomics 2005;23:423–32. [63] Bravo Vergel Y, Sculpher M. Quality-adjusted life years. Pract Neurol 2008;8:175–82. [64] Rasanen P, Roine E, Sintonen H, et al. Use of quality-adjusted life years for the estimation of effectiveness of health care: a systematic literature review. Int J Technol Assess Health Care 2006;22: 235–41. [65] Claxton K. Exploring uncertainty in cost-effectiveness analysis. Pharmacoeconomics 2008;26:781–98. [66] Al-Badriyeh D, Alameri M, Al-Okka R. Cost-effectiveness research in cancer therapy: a systematic review of literature trends, methods and the influence of funding. BMJ Open 2017;7:e012648. [67] Donker T, Blankers M, Hedman E, et al. Economic evaluations of internet interventions for mental health: a systematic review. Psychol Med 2015;45:3357–76. [68] Mihalopoulos C, Chatterton M. Economic evaluations of interventions designed to prevent mental disorders: a systematic review. Early Interv Psychiatry 2015;9:85–92. [69] Gallagher J, McCarthy S, Byrne S. Economic evaluations of clinical pharmacist interventions on hospital inpatients: a systematic review of recent literature. Int J Clin Pharm 2014;36:1101–14. [70] Langer A. A framework for assessing health economic evaluation (HEE) quality appraisal instruments. BMC Health Serv Res 2012;12:253. [71] Weinstein MC, O’Brien B, Hornberger J, et al. Principles of good practice for decision analytic modeling in health-care evaluation: report of the ISPOR Task Force on Good Research Practices—Modeling Studies. Value Health 2003;6:9–17. [72] Philips Z, Ginnelly L, Sculpher M, et al. Review of guidelines for good practice in decision-analytic modelling in health technology assessment. Health Technol Assess 2004;8:1–158. iii
iv, ix
xi,. [73] Philips Z, Bojke L, Sculpher M, et al. Good practice guidelines for decision-analytic modelling in health technology assessment: a review and consolidation of quality assessment. Pharmacoeconomics 2006;24:355–71. [74] Sculpher M, Fenwick E, Claxton K. Assessing quality in decision analytic cost-effectiveness models: a suggested framework and example of application. Pharmacoeconomics 2000;17:461–77. [75] Findlay M, Minckwitz G, Wardley A. Effective oral chemotherapy for breast cancer: pillars of strength. Ann Oncol 2008;19:212–22. [76] Al-Badriyeh D, Alabbadi I, Fahey M, et al. Multi-indication pharmacotherapeutic multicriteria decision analytic model for the comparative formulary inclusion of proton pump inhibitors in Qatar. Clin Ther 2016;38:1158–73. [77] Al-Badriyeh D, Fahey M, Alabbadi I, et al. Statin selection in Qatar based on multi-indication pharmacotherapeutic multi-criteria scoring model, and clinician preference. Clin Ther 2015;37:2798–810. [78] BJ Shea, Reeves BC, Wells G, et al. AMSTAR, a critical appraisal tool: assessing the methodological quality of systematic reviews. 2017. Available from: https://amstar.ca/docs/AMSTAR-2.pdf. [Accessed August 13, 2017].