Trends in Viral Hepatitis Cost-Effectiveness Studies | Page 2

From the more than 850 published papers related to hepatitis B and C infection, 77 met our entry criteria (Figure 1). Out of the 77 original cost-effectiveness analyses, 31 were hepatitis B related and 46 were hepatitis C related. Of the 46 hepatitis C treatment cost-effectiveness studies, 5 studies (10.9%) assessed different durations of interferon therapy,^20-24 8 studies (17.4%) compared interferon and ribavirin with interferon, ^25-32 5 studies (10.9%) compared pegylated interferon and ribavirin with no treatment,^33-37 and 15 studies (32.6%) compared pegylated interferon and ribavirin with interferon and ribavirin.^38-52 Of 46 hepatitis C–related studies, 5 (10.9%) compared the use of pegylated interferon alfa 2a plus ribavirin with pegylated interferon alfa 2b plus ribavirin.^53-57 In 5 of 46 studies (10.9%), pegylated interferon and ribavirin were compared with pegylated interferon or interferon monotherapy. ^51,52,58-60 In 18 of 46 studies (39.1%), there was an additional no-treatment arm.^{23,24,27,29,30,33-37,50,51,53,57-63} One study (2.2%) compared 2 different durations of treatment with pegylated interferon alfa 2a plus ribavirin,⁶⁴ while one other study (2.2%) compared peginterferon with interferon and interferon plus ribavirin treatments.⁶⁵

Of the 31 hepatitis B cost-effectiveness studies, 4 studies (12.9%) compared interferon with no treatment,^66-69 5 studies (16.1%) compared interferon with nucleos(t)ide analogue,^70-74 and 7 studies (15.2%) compared pegylated interferon with nucleos(t)ide analogue.75-81 In 3 of 31 studies (9.7%) there was an additional comparison with interferon.^76,79,80 Of the 31 studies, 13 (41.9%) compared nucleotide with nucleos(t) ide analogues.^82-95 One study (3.2%) specifically compared the use of a nucleoside analogue with no treatment.⁹⁶

The characteristics of the studies are shown in Table 1. Most studies (46 of 77 [59.74%]) utilized a cost-effectiveness ratio below $20,000/QALY in reaching a main conclusion for the study (P = .0002) (Figure 2). A total of 55 of 77 (71.4%) studies utilized ratios below the $50,000/QALY benchmark (P <.0001). Only 5 of 77 (6.49%) studies utilized ratios above $50,000/QALY. Of the 77 studies, 71 (92.2%) reported a favorable cost-effectiveness ratio in their conclusions (P <.0001).

Most of the studies (53 of 77 [68.8%]) fell within the intermediate quality score range of 12 to 14 (Figure 3). With regard to funding source, 39 of 77 (50.6%) papers reported funding by private industry and 11 of 77 (14.3%) papers reported academic funding. Of the 77 studies, 51 (66.2%) did not have a statement of disclosure. The locations in which most of the studies were carried out were the United States and Europe: 29 of 77 (37.7%) and 28 of 77 (36.3%), respectively. Most of the studies (60 of 77 [77.9%]) were published in journals with an impact factor of less than 4, and 46 of 77 (59.7%) papers were published between the years 2005 and 2011 (Figure 4).

We used the Kruskal-Wallis test to determine whether there was any relationship between cost-effectiveness ratios and study characteristics. We did not find any study characteristic that was associated with the reported costeffectiveness ratio based on our cutoff P value of .05. However, a few P values close to significance deserved our attention, including funding source (P = .0665) and country of origin (P = .0775). Academic-funded studies tended to have higher reported cost-effectiveness ratios than privately funded studies and studies funded in other ways, while other countries tended to report lower cost-effectiveness ratios than those reported in Asia, the United States, and Europe (Table 2).

Of the 45 corresponding authors for the 77 publications, 26 responded to our e-mail. Among the corresponding authors who responded, 85% stated that all their written costeffectiveness studies were published and 81% stated that they had submitted cost-effectiveness studies regardless of their results (favorable or unfavorable).

DISCUSSION

An attractive tool to justify the introduction of a new strategy into clinical practice is to demonstrate that it is cost effective. Indeed, with increasing healthcare costs, new treatments must be compared against a standard barometer. Cost-effectiveness studies can play an important role in clinical practice.⁹⁷

Our review demonstrates that a majority of published pharmacoeconomic analyses of antiviral therapy for both hepatitis B and C reported that the therapy was cost-effective. Most of the funding for these studies was from private industry, and the authors did not include financial disclosures. This raises the suspicion of pharmaceutical industry bias or influence on study publication, whether intentional or inadvertent.98 Moreover, fewer than 15% of the published studies had high- quality scores, and a journal impact factor greater than 4 was found in fewer than one- fourth of studies.

Publication bias occurs when undisclosed factors play a role in the likelihood of results getting published or the rate at which they are published.⁹⁹ It has been widely reported that there appears to be a bias in scientific journals toward publishing research that show positive versus negative results.^11,100,101 It has also been shown that studies with positive results are published before those with negative results.¹⁰¹ We suspect that the source of publication bias in favor of positive results may be the authors. Indeed, Dickersin et al demonstrated that negative results are a major reason for lack of manuscript writing and submission.⁵ Regional publication bias may also exist.¹⁰² Specifically, researchers may disproportionately pursue cost-effectiveness studies that they believe will be favorable and/or disproportionately submit for publication costeffectiveness studies that have favorable results. Indeed, most of the corresponding authors stated that few articles were unpublished or unsubmitted. One interpretation of this corresponding survey may be that researchers only attempt cost-effectiveness studies when there is a high likelihood of a positive result.