Tuesday, 19 March, 2024
HomeResearch IssuesResearchers caution about endemic sponsorship bias in cost-effective analyses

Researchers caution about endemic sponsorship bias in cost-effective analyses

Sponsorship bias in cost effective analyses (CEAs), a requirement of most manufacturers when applying for insurance coverage, is significant, systemic and present across a range of diseases and study designs, notes MedicalBrief.

This is the conclusion of researchers who have called for independent bodies to conduct CEAs to provide payers with more ability to negotiate lower prices. This impartiality is especially important for countries that rely on published CEAs to inform policy making for insurance coverage because of limited capacity for independent economic analysis, they conclude in a study published in the British Medical Journal.

Global health systems are dealing with scarce resources to cover healthcare services due to the rising demand and marketing of expensive drugs.

Cost effectiveness evidence has been widely accepted and used to inform price negotiation and healthcare insurance coverage policy making. In fact, manufacturers of new medicines are required to submit CEAs when applying for coverage approval by public or private payers in many countries. As a result, the literature for economic evaluations has expanded rapidly over the past decades.

Previous studies have found that about 20% of published CEAs were funded by the drug industry.

Seeking insurance coverage approval is strategically and financially important for industry. A new drug or device covered by insurance plans can generate much higher profit than those without such coverage. Industry’s economic ties to insurance coverage approval could lead to sponsorship bias in CEAs.

Evidence has consistently shown that industry-funded economic evaluations were more likely to report favourable cost effectiveness results to the sponsor.

Most of the published studies, however, were limited to specific diseases or treatments – for example, cancer, antidepressants, herpes zoster vaccine and statins.

The most recent analysis of bias in cost effectiveness studies based on systematic literature review was published more than 15 years ago.

Given the increasingly important role of CEAs in coverage policy making, up-to-date analysis on sponsorship bias is needed. Feng Xie and Ting Zhou of McMaster University, Canada, therefore conducted a systematic and comprehensive assessment on the sponsorship bias in CEAs by quantifying the association between industry sponsorship and incremental cost effectiveness ratio (ICER).

Study details

Industry sponsorship bias in cost effectiveness analysis: registry based analysis

Feng Xie, Ting Zhou.

Published in The BMJ on 22 June 2022

Abstract

Objective
To assess the association between industry sponsorship (drug, medical device, and biotechnology companies) and cost effectiveness results in cost effectiveness analysis (CEA).

Data source
The Tufts Cost-Effectiveness Analysis Registry was used to identify all CEAs published in Medline between 1976 and 2021.

Eligibility criteria for selecting studies
CEAs that reported incremental cost effectiveness ratio (ICER) using quality adjusted life year and provided sufficient information about the magnitude or location of the ICER.

Methods
Descriptive analyses were used to describe and compare the characteristics of CEAs with and without industry sponsorship. Logistic regression was used to identify the association between industry sponsorship and the cost effective conclusion using selected threshold values ($50 000 (£40 511; €47 405), $100 000, and $150 000). Robust linear regression was used to assess the association between industry sponsorship and the magnitude of ICER. All regression analyses were adjusted for disease and study design characteristics.

Results
8192 CEAs were eligible and included in the analysis, with 2437 (29.7%) sponsored by industry. Industry sponsored CEAs were more likely to publish ICERs below $50 000 (adjusted odds ratio 2.06, 95% confidence interval 1.82 to 2.33), $100 000 (2.95, 2.52 to 3.44), and $150 000 (3.34, 2.80 to 3.99) than non-industry sponsored studies. Among 5877 CEAs that reported positive incremental costs and quality adjusted life years, ICERs from industry sponsored studies were 33% lower (95% confidence interval −40 to −26) than those from non-industry sponsored studies.

Conclusions
Sponsorship bias in CEAs is significant, systemic, and present across a range of diseases and study designs. Use of CEAs conducted by independent bodies could provide payers with more ability to negotiate lower prices. This impartiality is especially important for countries that rely on published CEAs to inform policy making for insurance coverage because of limited capacity for independent economic analysis.

In an accompanying editorial, “Promoting confidence in cost-effectiveness analyses”, Adam Raymakers, senior health economist and Prof Aaron Kesselheim write:

Solutions to tackle bias are more important than ever

Xie and colleagues report a comprehensive investigation into the presence of bias in cost-effectiveness analyses conducted with sponsorship from the pharmaceutical or medical device industry. The authors used the Tufts Cost-Effectiveness Analysis Registry to analyse 8192 cost-effectiveness analyses, of which 29.7% were sponsored by industry.

Industry-sponsored studies were twice as likely to report the intervention being studied as cost-effective at a willingness-to-pay threshold of $50, 000 per quality adjusted life years (QALY) gained (odds ratio 2.06; 95% confidence interval 1.82-2.33). These findings are important because the studies listed in this comprehensive registry might directly influence organisations that use cost-effectiveness analysis in decision making.

Cost-effectiveness analysis can be a useful tool to aid system-level decision making about the adoption of new health technologies. However, this usefulness is directly related to the quality of the analysis and the confidence decision-makers might have in its accuracy. Assumptions made during the economic modelling process can influence the findings of any analysis.

As such, the convention in cost-effectiveness analysis is to be conservative about the benefits ascribed to the new intervention being considered, largely because substantial uncertainty is associated with new interventions relative to those in use as the standard of care.

Industry sponsored cost-effectiveness analyses might more frequently fall beneath a willingness-to-pay threshold of $50, 000 per QALY than those done by independent analysts for several reasons. As stated previously, analysts must make assumptions to make their model functional, so choices are made that might bias the results in a particular direction.

These assumptions include: over-estimation of the treatment effect (or durability of that effect), underestimation of the cost of the new treatment (through neglect of real world considerations, such as drug wastage due to vial size or the hospital resources required to deliver an intervention); omission of adverse events or underestimation of the adverse event rate; suboptimal model structure; problems with the comparator arm (including misspecification of the comparator); or time horizon.

For example, vials of infused cancer therapies that are oversized or undersized relative to their recommended dose can have substantial impacts on the cost of delivering these therapies and should be considered in cost-effectiveness analysis.

Given the uncertainty and risk of bias inherent in cost-effectiveness analyses, greater transparency in the modelling process is needed. For example, better reporting of model structure, model assumptions, and data to make models more replicable, as detailed in the Consolidated Health Economic Evaluation Reporting Standards (CHEERS) checklist.

Another approach would be to make economic models open source and available to researchers who wish to use the models in novel analyses, or to verify analyses conducted by others.

For any cost-effectiveness analysis to aid decision-making in health care, confidence in the findings is essential. In particular, greater transparency around model construction and assumptions would help to ensure industry sponsored analyses are scrutinised closely and can be replicated by independent analysts if necessary. A registry of cost-effectiveness analyses in which analytical plans and protocols could be registered in advance might facilitate greater scrutiny.

Additionally, increasing the funding and capacity for independent analysts to conduct this work might decrease reliance on industry-sponsored studies and their inherent risk of bias. A funding veil – a central repository of resources from public and private sources, managed by the government, that distributes funds to the best research proposals via an open mechanism— is one solution that separates funders of cost effectiveness analyses from the analysts conducting the work.

As costs of drugs and technologies increase along with the uncertainty associated with decisions about coverage being based on earlier and more immature evidence (such as phase 2 trials), the need for cost-effectiveness analyses to conform to best practices is becoming increasingly important.

The study by Xie and colleagues highlights that decision makers should exercise caution when using any published cost-effectiveness analysis in coverage decisions. Additionally, they should be acutely aware that analyses conducted by industry are likelier than independent analyses to be favourable to the drug or technology under investigation.

Potential solutions to mitigate these issues include improved reporting of methods and results, increased transparency, cost effectiveness models, or increased capacity and funding for independent institutions to conduct these analyses. Any of these approaches would reduce decision makers’ reliance on industry sponsored analyses.

 

BMJ article – Industry sponsorship bias in cost effectiveness analysis: registry based analysis (Open access)

 

BMJ accompanying editorial – Promoting confidence in cost-effectiveness analyses (Open access)

 

See more from MedicalBrief archives:

 

Top breast cancer researcher failed to disclose pharma payments

 

Study finds ‘spin’ in 26% of research – rising in non-randomised trials

 

‘Worrying’ rise in alcohol industry-funded research into alcohol impacts

 

Most first-wave COVID-19 clinical trials have ‘major design shortcomings’

 

First-in-human clinical trial confirms new HIV vaccine approach

 

 

 

MedicalBrief — our free weekly e-newsletter

We'd appreciate as much information as possible, however only an email address is required.