Comparative cost-effectiveness of antiviral therapies in patients with chronic hepatitis B: A systematic review of economic evidence

Authors


Professor You-Ping Ling, Chinese Evidence-Based Medicine Center, West China Hospital, Sichuan University, No. 37 Guo Xue Xiang, Chengdu, 610041, China. Email: sunyzmy@hotmail.com

Abstract

Background and Aim:  Economic efficiency of the alternative antiviral therapies for chronic hepatitis B has not been systematically investigated and their quality remains unknown. The aim of the present study was to systematically overview economic evidence of antiviral therapies for chronic hepatitis B.

Methods:  We searched six databases and eight major journals supplemented with screening references of eligible studies. Full economic evaluations comparing alternative antiviral therapies in patients with chronic hepatitis B virus infection were included. Two investigators assessed the study quality and transferability, independently. Data were analyzed qualitatively with adjustment when appropriate.

Results:  Fourteen studies (six modeling vs eight trials and database analyses) were included. Quality was high in five studies, moderate in one US and five Chinese studies, and low in three Chinese studies. The major problems of quality are costing methods and analysis and the presentation of results. In Australia and Poland, lamivudine-preferred strategies dominated interferon (IFN)-α and its related strategy from the health-care sector perspective. In the US, adefovir salvage produced US$8446 per additional quality-adjusted life years (QALY) compared with IFN-α. In Spain, the cost of adefovir was US$34 840 for additional virological response. In Taiwan, the use of pegylated IFN-α (pegIFN-α) produced US$11 711.4 per additional QALY, compared with lamivudine. In China, the incremental cost-effectiveness ratios of combination therapy lamivudine ranged from US$2860 to US$22 160 per additional loss of hepatitis B e antigen (HBeAg), and IFN-α versus lamivudine ranged from US$2490 to US$8890 per additional loss of HBeAg.

Conclusion:  The cost-effectiveness frontiers of treatment alternatives vary and are influenced by the comparators and socioeconomic conditions of countries. Lamivudine-containing therapy is cost-effective when newer antiviral agents (e.g. adefovir/pegIFN-α) were not available. Economic methods should be further improved in studies, particularly in China.

Introduction

Chronic hepatitis B (CHB) is an important health problem around the world.1 An estimated 50 million new cases of hepatitis B virus (HBV) infection are diagnosed annually with 5–10% adults and up to 90% infants becoming chronically infected.2 Of the 350 million such cases worldwide, 75% occur in Asia and 50% are Chinese. Although vaccines are effective in reducing the occurrence of hepatitis B infection in Asia,3 CHB is persistently a challenging problem.4 Therapeutic strategies are intensively investigated to achieve sustained response and long-term benefits for CHB.5

The traditionally-used antiviral agents for CHB include standard interferon-α (IFN-α), lamivudine, and thymosin-α (Tα).6–8 IFN-α and lamivudine have similar effects on serological response.9,10 However, IFN-α has frequent and potentially severe adverse effects, and viral resistance often occurs in long-term use of lamivudine.11 The effect of Tα has not been well established.12 New antiviral agents, including pegylated IFN-α (pegIFN-α), adefovir, and entecavir are more effective with fewer adverse effects and less occurrence of drug resistance, but are more expensive. With limited health resources available, the issue of economic efficiency of these medicines has received more attention.13

Published economic evidence may be used to influence decision making;14–16 however, individual studies are restricted in applicability because of the limitations of study characteristics. Also, the quality varies across studies, which might result in spurious findings.17 Unsystematic use of evidence is likely to cause biased decisions. Overview methods can be used to comprehensively search and identify relevant evidence, critically appraise the quality and applicability, and appropriately analyze the data so as to produce unbiased results.18

This study systematically reviews the economic evidence of alternative antiviral therapies for CHB, so as to assess their cost-effectiveness and identify their strengths and weaknesses to assist in future research.

Methods

Search strategy

We searched Medline, the Science Citation Index (expanded), Current Content Connect, the NHS Economic Evaluation Database, the Health Technology Assessment Database, and the Chinese Biomedical Database from 1 January 1980 to 25 December 2006. The key words of cost, cost-effectiveness, cost-utility, cost-benefit, economic evaluation (or economic appraisal or economic analysis), hepatitis B, and hepatitis B virus were used, either with MESH terms or free texts, for identifying published literature. We searched seven key relevant journals from January 1995 to December 2006; the references of eligible studies were also screened.

Eligibility criteria

The economic studies eligible for inclusion needed to fulfill the following criteria:

  • • Economic evaluations of any design that considered both costs and health outcomes, such as modeling, trial-based, and retrospective database studies. The studies that only considered costs were ineligible
  • • Patients with chronic HBV infection of any virological and serological presentations. Those with HIV and HCV co-infection and surgical procedures were ineligible
  • • At least two antiviral therapies were compared with or without standard care/no treatment as a base case. The comparison of antiviral therapies only with no treatment/standard care was ineligible, because it could not address comparative cost-effectiveness of antiviral therapies
  • • The titles and abstracts were primarily screened for eligibility. When necessary, the full texts were also used. We did not consider abstract publications, because they could not provide sufficient information.

Health outcomes measures

The primary measures for health outcomes were life-years gained (LYG) and quality-adjusted life years (QALY). The surrogate measures were considered if virological and serological responses were considered. We chose multiple measures because different studies might choose different measures on the basis of their questions of interest.

Assessment of methodological quality and transferability

We used a modified criteria checklist for assessing methodological quality (Appendix I). The modified list was derived from two established checklists: a 19-item checklist19 and a 10-item checklist.20 We excluded some of the quality items of these two checklists, because they were not directly related to methodological quality. The modified checklists were validated after three consultations among a panel of four experts. We used five essential factors to assess the transferability of the study results. At present, there is no consensus on the criteria for assessing the transferability of economic studies.21,22 We therefore conducted a three-round consultation among a panel of experts who decided the transferability items and criteria from a list of 12 potential items (Appendix II). Of the items, clinical practice variations and resource use were the most important factors. Two investigators (QW & SX) selected studies and assessed quality and transferability, independently. The overall rating of quality and transferability were conducted based on the judgment of the investigators; any disagreement was solved by discussion.

Data extraction, adjustment, and analysis

A data extraction spreadsheet was developed and validated by a piloted extraction. Data extraction was based on, but not limited to, the reported sources. When necessary, the principal authors of the studies were contacted by e-mail or regular mail for the details of the studies. Based on all the available information, we defined the characteristics and adjusted the results, if necessary. As there were significant variations across the studies, we qualitatively analyzed the results.23 To increase the homogeneity among the trials, we adjusted the cost data and updated them to the year of 2006, with a consumer price index of 3%. Cnew = Cold*(1 + 0.03)[2006–X], where Cnew was the adjusted costs, Cold was the reported costs, and X meant the year of unit prices. The costs in all studies were transferred to US dollars, with rates of 8.2 (US$1 = RMB 8.2), 4.0 (US$1 = PLN 4.0), 1.82 (US$1 = A$1.82), 0.80 (US$1 = €0.80), and 32.5 (US$1 = NTD 32.5).

The results of the Chinese studies, where possible, were plotted in a 2-D coordinate. The strategy of lamivudine monotherapy was used as the reference case, with which other therapeutic options were compared. We found that four studies appropriately collected and valued the cost and effectiveness, but inappropriately calculated the incremental cost-effectiveness ratios (ICER). We therefore calculated the ICER based on the formula: ICER = (Ca–Cb)/(Ea–Eb) = ΔC/ΔE, where C is the cost, E is the effectiveness, and a and b are the interventions.

Results

A total of 435 studies were searched, and 43 studies were primarily included after screening titles and abstracts. Using full texts, 14 studies were eligible for inclusion.

Methodological quality and transferability of studies

Four studies conducted in the USA,24 Australia,25 Poland,26 Spain,27 and Taiwan28 were high in quality, with at least 84.6% of quality items meeting the criteria. Five studies from mainland China29–33 and one US study34 were generally moderate, with 53.8–76.9% of quality items meeting the criteria. Three Chinese studies35–37 were poor in quality, with 31–46% of quality items meeting the criteria. All the studies were clear with competing alternatives and patient populations. The effectiveness of the data of competing alternatives was obtained in 92.9% (13/14) of the studies through randomized trials and overview of randomized trials. Twenty nine per cent (4/14) of the studies addressed the life-time economic outcomes. The quality of costing varied among the studies; 76.9% (10/13) of the studies adequately identified the costs. Cost valuation was adequately performed in 64.3% (9/14) of the studies. An incremental analysis was conducted in all the studies, but was inappropriately calculated in four Chinese studies. Sensitivity analyses were adequate in six studies (42.9%), whereas the studies from mainland China did not conduct sensitivity analyses.

The factors affecting the transferability of the results in 14 studies are presented in Table 1. Five studies were conducted in Australia, Poland, and the USA based on the clinical guidelines and local expert opinions. The Australian, Spanish, Taiwanese, and Polish studies adapted the health-care sector perspective, while two US studies from the third-party payer. One US study24 considered both the patients with hepatitis B e antigen (HBeAg)-negative and -positive patients, and the Spanish study27 included HBeAg-negative patients, while the other three only considered HBeAg-positive patients. Unit prices of the studies were mostly based on the local wholesale price of drugs and schedules for health-care costs. Eight studies from mainland China29–33,35–37 adapted various perspectives, but five of them only considered drug costs. Patients were HBeAg positive and replicative of HBV–DNA. Unit prices varied from hospital-regulated systems to the national registry.

Table 1.  Study characteristics of 12 economic evaluations
Study IDPerspectiveDesignTime horizonInterventionsOutcome measuresPatient characteristicsCosting methodsClinical settingClinical practice
  • Costing methods indicate identification, measurement, and valuation of costs.

  • IFN, 10 MIU, tiw, 4 months; LAM 100 mg qd, 12 months.

  • ADV, adefovir; ALT, alanine aminotransferase; HBeAG, Hepatitis B e antigen; IFN, interferon; LAM, lamivudine; OC1, HBeAg seroconversion; OC2, progression to cirrhosis; OC3, loss of HBeAg; OC4, loss of HBeAg/HBV DNA and normalization of ALT; pegIFN-α, pegylated IFN-α; pts, patients; MIU, million international units; QALY, quality-adjusted life years; qd, once daily; Tα, thymosin-α; tiw, three times a week; Tx, treatment; ULN, upper limits of normal; YLG, years of life gained; tiw, three times a week.

Sullivan et al. (2006)28Public health systemMarkov modelLifetime(a) pegIFN-α-2a, 180 μg, 48 week; (b) LAM, 48 weekLYG, QALYAged 32, Asian, HBeAg positive, ALT >2 ULNDirect medical costs included unit costs from 2004 Fee Schedule for Medical Service and Reference List for DrugsTaiwan health CareBased on expert Panel
Buti et al. (2006)27Public Health SystemDecision tree4 years(a) LAM, 4 years, if resistance: use ADV; (b) ADV, 4 yearsVirological responseHBeAg-,adults patients, ethnicity unknownDirect medical costs included; Unit costs from Medicinal Product Catalog (2003), SOIKOS (2004), market prices; Macro-costing (survey) methods used.Outpatient care—Spanish health careBased on expert panel and European consensus
Kanwal et al. (2005)24Third-party payerDecision tree and Markov modelLifetime(a) ‘doing nothing’; (b) IFN: 10 MIU, tiw; 26 week for HBeAg+ pts; 52 week for HBeAg- pts; (c) LAM; (d) ADV; (e) ADV salvageQALYStarting Tx at 40; elevated enzyme level, and no evidence of cirrhosis, 55% HBeAg negative; no pretreatmentDirect medical costs included; Costs data derived from published literature and guides, with costs adjusted to 2004 US dollars by the consumer price index; 3% used to discount costs (2004)US Primary careAccording to published guidelines in the USA
Orlewska (2002)26Health care payerDecision tree Statistical modeling1 year To 100  years of age(a) LAM-preferred; (b) IFN-preferred option; (c) IFN alone; (d) no treatment. Standard protocol usedOC1, OC2 YLGM:F = 0.4:0.6, age between 30 and 50, HBeAg+ without precore- mutant, elevated ALT, no progression to cirrhosis, IFN-naiveDirect medical costs included; Micro-costing method used, but physical units not reported. Cost data from expert opinions, and guideline. No discounting used for long-term results.Polish health careBased on Polish hepatologists who manage 90% of patients in Poland, and guidelines
Crowley et al. (2002)25Health care sectorDecision tree Markov model1 year 30 years(a) LAM-preference; (b) IFN alone; (c) no treatment. Standard protocol.OC1, OC2 YLG, QALYM:F = 0.7:0.3; age of starting Tx: 30; HBeAg plus HBV–DNA positive; IFN-naiveDirect medical costs included; Micro-costing method used; Physical units not reported; Adjustment of costs 5% used to discount costs and outcomesAustralian outpatient and inpatient departmentsBased on consensus opinion of an expert panel of six major Australian hepatologists
Brooks et al. (2001)34Third-party payerDecision tree1 year(a) LAM, 1 year; (b) IFN: 10 MIU, tiw, 16 weekOC1, OC2No details presentedDrug wholesale price of 1999 were used.UnknownUnknown
He and Yu (2002)29Third-party payerRandomized trial1 year(a) LAM, 1 year (b) IFN, 5 MIU, tiw, 0.5 yearsOC3HBeAg+, ALT >2 UNL, M:F = 0.65:0.35. Chinese.Hospital-regulated drug price (2001) used.Chinese hospital careChinese practice pattern
Wei (2000)30Not specifiedRetrospective database analysis1 year(a) Ta: 10 mg, qd 1 year; (b) LAM, 1 year; (c) LAM, 1 year + IFN: 3MIU, tiw, 49 weeksOC3M:F = 3:1, HBeAg+, ALT >2UNL, Chinese pts, aged 16–40. No HCV/HIV co-infectionDirect medical and indirect costs included; Micro-costing methods used; physical units reported. Human capital approach for indirect costs.Chinese inpatient health careChinese practice pattern
Wu et al. (2002)36Third-party payerControlled trial1 year(a) LAM >6 months; (b) IFN: 3 MIU, tiw, >6 months; (c) conventionalOC3M:F = 0.6:0.4, HBeAg+, no antiviral pretreatment, ALT >120 IU/LDrug costs included, without reporting the sources of unit prices and year of priceChinese hospital careChinese practice pattern
Yan (2004)37Third-party payerControlled trial1 year(a) LAM, 12 months (b) IFN, 5 MIU, tiw, 6 monthsOC3M:F = 0.66:0.34, HBsAg and HBeAg positive, ALT >120 IU/L, Chinese ptsHospital-regulated drug prices (2001)_ includedChinese hospital careChinese practice pattern
Yang and Chen (2001)31PatientRandomized trial9 months(a) LAM, 9 months (b) IFN, 3 MIU, tiw, 6 months (c) IFN + LAMOC3M:F = 2.38:1, HBeAg and HBV DNA positive, ALT >2 UNLGross-costing methods used, unit prices of 2001. Direct medical and non-medical costs included.Chinese outpatient health careChinese practice pattern
Zhang et al. (2002)32Health care sectorRandomized trial6 months(a) LAM, 6 months (b) IFN, 3 MIU, tiw, 6 months (c) conventionalOC3HBeAg+, ALT >2 UNL, Chinese patients.Direct medical costs, without sources of unit prices. Micro-costing methods used.Chinese Inpatient and outpatient careChinese practice pattern
Zhou et al. (2004)33Third-party payerRandomized trial1 year(a) LAM, 1 year (b) IFN, 5 MIU, tiw, 6 monthsOC3HBeAg+, ALT >2UNL, M:= 0.7:0.3, Chinese pts, Aged 21–52Medication costs included; Gross-costing methods used, of unit prices in 2002.Chinese hospital careChinese practice pattern
Wang and Ye (2002)35Third-party payerRandomized trial1 year(a) LAM, 12 months (b) IFN, 5 MIU, tiw, 6 monthsOC4M:F = 0.6:0.4, HBeAg+, no antiviral pretreatment, ALT >2 UNLDrug cost of 2000 included.Chinese hospital careChinese practice pattern

Characteristics of studies

Comparators

The Taiwanese study28 compared pegIFN-α with lamivudine; the Spanish study27 compared adefovir with lamivudine. One US24 study investigated five therapeutic options, including IFN-α, lamivudine, adefovir monotherapy, and adefovir-salvage therapy. The Australian25 and Polish26 studies mainly compared IFN-α-preference therapy with lamivudine-preference therapy, in which both IFN and lamivudine were assumed to be available to patients. Seven studies compared standard IFN-α with lamivudine, with or without standard care; two studies compared the combination of IFN-α and lamivudine with monotherapy, one of which also included thymosin.

Patients

Twelve studies were conducted for patients with HBeAg-positive CHB. One US24 study included both the HBeAg-positive and -negative patients; the proportion was assumed on the basis of patient epidemiology in the USA. The Spanish27 studies only included HBeAg-negative patients. All the patients in the studies were adults.

Outcome measures and length of follow up

The US, Australian, Polish, and Taiwanese studies investigated the lifetime outcomes, therefore, used LYG or QALY as the measures. The other studies only spanned from 6 months to 4 years, and therefore chose virological or serological measures.

Study design and settings

Modeling was used in six studies. The Australian, Polish, Taiwanese, and one US study used the Markov techniques to estimate lifetime outcomes. The Spanish studies, and another US study, used decision-tree analyses to investigate 1- and 4-year results. Studies from mainland China were conducted on the basis of short-term trials or databases. Seven studies chose third-party payer as the perspective; five studies used the health-care sector perspective; one study was conducted based on the patient perspective; and the last one did not describe the perspective. All of the studies were cost-effectiveness studies, of which four lifetime modeling studies included QALY in addition to LYG. Five studies with a time horizon over 1 year adapted the discounting rate ranging from 3 to 5%.

Health outcomes of antiviral therapies

In Australia, the real-life use of lamivudine gained an additional 3.1 QALY compared with IFN-α, and 4.7 QALY with no treatment.25 In Poland, hepatic cirrhosis reduced 37.76 years in the population who started treatment at the age of 30, and 20 years in the population starting at 50. In Spain, the virological response was achieved in 40.4% in patients using lamivudine, and 78.0% in patients using adefovir after 4 years of treatment. In Taiwan, pegIFN-α produced an additional 0.41 QALY, compared with lamivudine.

The Chinese studies assessed the serological response of antiviral therapies. The difference in loss of HBeAg between treatment alternatives and lamivudine can be seen in Figure 1. Clearly, the superiority of IFN-α and combination therapy were generally consistent across the studies, although one study32 indicated an opposite result. An overview of randomized trials also confirmed that the superiority of IFN-α to lamivudine in the loss of HBeAg (odds ratio = 0.63, 95% confidence interval = 0.43–0.83).38 A meta-regression analysis identified no linear relation of the duration of treatment with response.

Figure 1.

Incremental costs (vertical) and effectiveness (horizontal) of treatment alternatives compared to lamivudine in the Chinese studies. Clinical outcome was loss of hepatitis B e antigen (HBeAg). All the alternative therapies were compared to lamivudine. Numbers in the parentheses are the incremental cost-effectiveness ratios (X103 US dollars for an additional loss of HBeAg). Incremental cost-effectiveness ratios were only calculated only for the studies located at the north-east quadrant. Squares located at the north-west quadrant were dominated by lamivudine, and thus, not reported. Numbers under the study names are the rate differences and cost differences. Interferon (IFN)-α versus lamivudine: Yang (a), Wu (a), Zhang (a), Yan, He, Zhou; lamivudine plus IFN-α versus lamivudine: Yang (b), Wei (b); Standard therapy versus lamivudine: Wu (a), Zhang (b); thymosin-α versus lamivudine; Wei (a).

Costs of antiviral therapies

Cost data were available in 13 studies and direct medical costs were estimated in six studies.25–28,32,33 Five studies used drug acquisition costs,34–37 one study included direct medical and non-medical costs,31 and one study included direct medical and indirect costs.30 In Australia, 30-year costs were similar in alternative strategies (US$24 094.6 for lamivudine-preference vs US$23 649.7 for IFN-α preference). In Taiwan, the lifetime costs of pegIFN-α strategy were US$10 951.8 and those of lamivudine were US$6154.3. In Poland, lamivudine-preferred strategy (US$1981.50) was lower in cost than IFN-α-preferred strategy (US$3113.25) and IFN-α alone (US$2661.00). In Spain, adefovir was produced at twice the cost of lamivudine (US$14 335 vs US$26 299) in a 4-year period.

Two Chinese studies estimated the medical costs. One study33 indicated that lamivudine was produced at a lower cost than IFN-α (US$757.32 vs US$2225.61). Another study,32 although indicating a similar trend of resource use, produced a double estimate of costs. In China, drug acquisition costs were much lower in lamivudine (US$468.3–843.8) than those in IFN-α (US$1529.8–5393.2). The great variation of costs was as a result of the difference in unit prices and medication intensity. In the USA, the drug acquisition cost of lamivudine was lower than IFN-α (US$1583.3 vs US$5589.1). If the indirect costs were included, the costs of lamivudine increased to US$2310 in China, compared to US$2070.3 in Tα alone.

Incremental cost-effectiveness ratio and uncertainty

Lifetime cost-effectiveness was estimated in four studies.24–26,28 In Australia, the lamivudine-preferred strategy dominated IFN-α alone and no treatment. Comparing IFN-α to no treatment, the ICER was US$1367.58 per additional QALY. Univariate and multivariate analyses indicated that the results were sensitive to the rate of HBeAg seroconversion, progression to cirrhosis, and the proportion of patients with eligibility for each drug, but the cost-effectiveness frontier remained unchanged.

In Poland, the IFN-α-preferred strategy and IFN-α alone were dominated by the lamivudine-preferred strategy and no treatment and therefore was ruled out from the analyses. Compared to no treatment, the lamivudine-preferred strategy cost US$526.25 for achieving an additional LYG in patients starting treatment at 30, and cost US$994.5 per additional LYG in those starting treatment at 50. According to the suggested threshold for cost-effectiveness (US$15 600), the lamivudine-preferred strategy was cost-effective. A univariate analysis examined the robustness of the results. Although the results were sensitive to the probability of progression to cirrhosis, the cost-effectiveness frontiers were not changed.

In the US study, lamivudine and adefovir monotherapy dominated. The ICER of IFN-α against no treatment was US$6337 per additional QALY, and the ICER of the adefovir salvage strategy against IFN-α was US$8446 per additional QALY. A univariate approach identified six variables that the model was sensitive to, in which the cost-effectiveness frontiers changed. Monte Carlo analyses were used to compare adefovir salvage and IFN-α and presented the results using the cost-effectiveness acceptability curve. If the underlying cost-effectiveness cut-off value was set as US$50 000 per QALY, adefovir salvage was cost-effective.28 However, with a very restricted budget, IFN-α tended to be cost-effective.

In Spain,27 adefovir was not a dominant strategy compared with lamivudine, although it produced a much higher virological response. The ICER of adefovir versus lamivudine was US$34 380 per additional response. However, the underlying rationale was not presented to indicate the cost-effectiveness of these two alternative therapies.

In Taiwan, the ICER of pegIFN-α against lamivudine is US$11 711.4 per additional QALY and US$14 367 per additional YLG. The variations of the key variables (e.g. probability of developing compensate cirrhosis from CHB or seroconversion, and the efficacy rate of pegIFN-α) led to a range of US$9656.0–14 931.1 per additional QALY. On the basis of the threshold of US$17 300, pegIFN-α was more cost-effective.

Seven studies from mainland China investigated the short-term economic benefits (Fig. 1). The ICER of two studies30,31 comparing combination therapy against lamivudine varied substantially (US$2.86 vs US$22.16 × 103 per loss of HBeAg). The main reason for the difference could be the variation of the interventions, such as duration and intensity. Six studies29,31–33,36,37 comparing IFN-α with lamivudine indicated inconsistent results across the studies (Fig. 1). Two studies33,36 suggested that IFN-α was dominated by lamivudine. The ICER of the other four studies varied (US$2.49–8.89 × 103 per additional loss of HBeAg). The reason for this was mainly because of the difference of clinical outcomes across the studies. One study30 suggested that lamivudine was more effective, but required more use of health resources than Tα (US$1.14 × 103 per loss of HBeAg). The study3,5 using the composite measure, suggested a higher effect of IFN-α than lamivudine, but also produced higher costs, resulting in an ICER of US$53.5 × 103 per additional gain of outcome. None of the studies adequately conducted sensitivity analyses.

Discussion

Standard IFN-α, lamivudine, pegIFN-α, adefovir, entecavir, and Tα are currently approved antiviral drugs for treating chronic HBV infection. Due to their high acquisition costs and long-term use, it is imperative to understand the economic efficiency of these therapies. However, the limitations of the applicability of the studies restricted their use interchangeably between countries. The quality of the economic studies is also a concern of using the studies for influencing policy making. We therefore conducted an overview of economic evidence on antiviral therapy for CHB to produce unbiased results of cost-effectiveness. We do not expect that this study addresses all aspects of problems. The key roles of this overview are to scientifically present the state-of-art of economic research for CHB management and identify the points for future research.

Study quality

Five studies conducted in the USA, Australia, Poland, Spain, and Taiwan were generally high in quality. The validity of these studies can be justified from the methods of planning, identifying, measuring, and valuing the benefits and costs. The remaining studies from mainland China were concerned with the quality, since most of them had poorer performance in identifying, measuring and valuing costs, analyzing, and presenting the benefits and costs. Therefore, the results from mainland China might be biased, whereas those studies from other countries/regions were at less risk of bias.

Costs and cost-effectiveness of alternative therapies

Although the cost components and time horizons were differently included, the costs in the lamivudine-containing arm were generally lower than that of IFN-α. This is mostly because the acquisition costs of lamivudine were significantly lower than IFN-α, even though the duration of IFN-α might be shorter than lamivudine. In Australia, the cost of the lamivudine-preferred strategy was slightly higher than that of the IFN-α strategy, because in the real-life setting, the eligibility for administrating lamivudine is higher than IFN-α, causing a higher rate in the use of medicines in the lamivudine arm than the IFN-α arm. However, nearly all the studies did not consider the impact of the long-term use of lamivudine and its possible consequence of adverse effects on cost.

The cost-effectiveness frontiers of alterative therapies vary across countries. It seems likely that the difference of comparators causes the variations. In the Polish and Australian studies, the cost-effectiveness was analyzed based on the availability of lamivudine and IFN-α, and the results consistently indicated that the lamivudine-preferred strategy was more effective. In contrast, in the US, Taiwanese, and Spanish studies, the frontiers of cost-effectiveness changed from lamivudine to a newer generation of antiviral agents. In the US and Spanish studies, the use of adefovir improves economic efficiency compared to IFN-α or lamivudine. In Taiwan, the cost-effectiveness frontier is moved from lamivudine to pegIFN-α. It is implicitly indicated that the use of a new generation of antiviral medicines (e.g. adefovir) could be more economically efficient; however, more studies are needed to prove this hypothesis.

In mainland China, the results are less informative for decision making, because the ICER estimates varied significantly across the studies, and these studies did not adequately address the uncertainty when using sensitivity analyses. By stratifying the studies for comparisons, the combination therapy produced very high ICER (up to US$221 600 per loss of HBeAg). The comparison of IFN-α with lamivudine also produced high ICER (more than US$2490 per loss of HBeAg). Also, there is no consensus of a cut-off point for deciding the cost-effectiveness. However, the consultation of the experts in the national authority suggested that a ceiling ratio of US$1800 per additional loss of HBeAg was used to indicate the cut-off value. Clearly, all the ICER of the competing treatments against lamivudine exceeded this ceiling value, in reverse, indicating the cost-effectiveness of lamivudine.

Strengths and weaknesses of the study

In this study, we assessed the cost-effectiveness of all the competing antiviral therapies in different health-care setting. In particular, we included data from China, which accounts for approximately 50% of chronic HBV infection worldwide. We critically appraised the quality of the studies, which made the process more transparent. We exclusively developed a criteria checklist to assess the transferability of studies, which also contributed in making the results more transparent. This review included four studies from Australia, Poland, and the US, and nine studies from mainland China and Taiwan. Of these, 12 were from the Asia–Pacific region, where the prevalence of chronic HBV infection is highest,39 reflecting a close relevance to the efforts of CHB management in this region.

On the downside is the inherent weakness of a few of the included studies. The quality of a few of the studies, in particular the Chinese studies, remains the obstacle in further improving the robustness of results. The costing methods are the main disadvantage of these economic studies. However, it is the fact that the economic evaluation was introduced to the field of CHB management only in recent years, and the problem with costing widely exists in health care other than hepatitis management. Also, we conducted several rounds of contacts with investigators to reduce the ‘unclear’ items in the quality checklist, which helped improve the quality assessment. The difference in the choice of perspectives was another problem, especially in the Chinese studies. However, the results were generally consistent across studies using different perspectives. This also suggested that the choice of perspectives might not affect the cost-effectiveness frontiers.

We should acknowledge that the studies from mainland China need further improvement. In these studies, none consider long-term results; however, CHB is a chronic and progressive disease that needs long-term care. The outcome measures in the Chinese studies included loss of HBeAg, whereas the outcomes interesting to policy makers are final economic outcomes (e.g. YLG and QALY). The lack of these data limited their appropriateness for decision making. Quality was another key difficulty. The major problem was how to identify measure and value costs, thus causing a concern of biased estimates. The insufficient reporting also caused the problem of understanding the details of the study. We tried to contact authors to obtain additional information and this indeed improved the quality of the studies themselves. Last, variations of the studies also caused problems with pooling of the results, but this is common and an inherent difficulty in economic studies, where heterogeneity exists significantly across studies.

Implications of this study

This study has systematically reviewed economic evidence in the world that investigates the economic efficiency of antiviral therapies for CHB. It is likely that lamivudine-containing therapy might be more cost-effective than IFN-α and Tα, irrespective of the perspective chosen. However, the use of newer antiviral therapies (e.g. adefovir and pegIFN-α) might change the cost-effectiveness frontier. It is implied that the advantage of improved efficacy and decreased occurrence of viral resistance in the newer antiviral agents could justify the incremental costs. However, more studies are needed across regions (e.g. Australia and China) to prove this. Considering the economic affordability in different countries, the cost-effectiveness frontiers may also be different. For example, in the USA, the economic effectiveness ratio may be very high, in which case the use of more expensive but effective medicines is cost-effective. However, in developing countries, cost-effectiveness ratios may be relatively low, and lamivudine-containing therapy remains cost-effective in this context.

In summary, the cost-effectiveness frontiers can be varying, depending on the comparators. The lamivudine-containing strategy was more cost-effective than IFN-α and Tα; however, the introduction of newer antiviral agents could further improve the economic efficiency. On the basis of current evidence, it is suggested that: (i) the methods of economic evaluation should be educated and improved, particularly in countries such as China; (ii) economic efficiency of antiviral therapies should be further investigated and updated, particularly that for newer generation antiviral agents; and (iii) the cost-effectiveness frontiers of treatment alternatives might vary depending on the socioeconomic conditions.

Appendices

Appendix I

Criteria list for assessment of methodological quality of economic evaluations

Quality itemsAppraisal guides
Is a well-defined question posed?A research question has to identify clearly the alternatives being compared and the population for which the comparison is made.
Is study population clearly described?The relevant clinical characteristics, entry and eligibility criteria, as well as drop-out during follow up should be stated explicitly.
Are competing alternatives clearly described? Does the study provide evidence that the program is effective?A detailed description should be given of the competing interventions. Systematic review of randomized trials and individual randomized trials are used to justify the effectiveness of the program.
Is the time horizon appropriate to include relevant costs and consequences?The time horizon should always be equal for costs and outcomes if these are combined in a ratio. The time should be long enough to include all relevant costs and outcomes relating to the intervention.
Are all important and relevant outcomes identified?Identification of all important and relevant outcomes should be given in relation to the perspective and the research question.
Are all important and relevant outcomes measured appropriately?The outcome measurement should result from the outcome identification.
Are all important and relevant costs/resource use identified?Identification of all important and relevant costs should be given in relation to the perspective and the research question.
Are all important and relevant resource use measured in appropriate units?The costs/resource use should be measured appropriately in physical units.
Are all important and relevant resource use valued appropriately?The sources of valuation should be clearly stated for each cost price of every volume parameter and their reference year.
Is an incremental analysis of the consequences and costs of alternatives performed?An incremental analysis should examine the additional costs from on intervention over another, compared to the additional outcomes.
Are all future costs and outcomes discounted appropriately?Discounting is done appropriately if all costs and outcomes are converted to one single year, based on a motivated discount rate.
Is an adequate sensitivity analysis performed?All main areas of uncertainty are considered.

Appendix II

Criteria list for assessment of transferability of economic evaluations

FactorsAppraisal guides
  1. ALT, alanine aminotransferase; HBeAg, hepatitis B e antigen; HBV, hepatitis B virus.

PerspectiveThe desired perspective is from the health-care sector. Choice of perspective is transferable, if (a) conducted from the healthcare sector perspective, or (b) availability of disaggregated data from the societal perspective.
Patient characteristicsAccording to the Chinese recommendations, patients with HBeAg-positive and replication of HBV–DNA, and elevated level of ALT should receive antiviral therapy. Inclusion of patients is not transferable, if (a) the following criteria did not fulfill (b) patients with decompensate liver diseases were included, or (c) co-infection of HCV and HIV was considered eligible.
Clinical practice variationThe preferred clinical practice pattern is inclusive of inpatient and outpatient health care under Chinese health system. The clinical practice is not transferable, if pharmaceutical care, which might be informed by the practice guideline, was different from the Chinese pattern.
Resources useThe resources use is transferable if the study clearly documented the resources use data in separate physical units and in relation to the clinical practice pattern, the resources use pattern is similar to that in China.
Unit pricesThe unit costs in different geographic areas are the most apparent factor that influences the transferability of economic studies. Nearly all the economic evaluations conducted outside China are unlikely to be transferable to the Chinese setting.
However, adequate adjustment may be appropriate to transfer the published results to the Chinese setting.

Ancillary