Network meta-analyses in comparative effectiveness research: an interview with Ulf Staginnus

In this feature, we discuss Ipsen’s (Paris, France) use of systematic reviews and network meta-analyses in comparative effectiveness research with its Senior Vice President of Global Market Access and Pricing, Ulf Staginnus.

Go to the profile of The Evidence Base
Feb 12, 2020
1
0

At the ISPOR Europe 2019 (2–6 November; Copenhagen, Denmark) meeting, Ipsen (Paris, France) representatives presented two key studies with the aim of increasing stakeholders’ understanding of treatment pathways, demonstrating value to payers and positively impact patient outcomes. One study involved an assessment of the budget impact of long-acting somatostatin analogues (LA-SSAs) in the treatment of acromegaly and gastroenteropancreatic neuroendocrine tumors, considering attributes related to drug delivery of LA-SSAs in the UK.

The second study was a network meta-analysis analyzing the effect of cabozantinib as a first-line treatment, versus standard-of-care comparators, on the progression-free and overall survivals of individuals with advanced renal cell carcinoma.

In this feature, Ipsen’s Senior Vice President of Global Market Access and Pricing, Ulf Staginnus, discusses Ipsen’s use of systematic reviews and network meta-analyses in comparative effectiveness research, and the future of modern data-mining approaches in the field.


What are some of the benefits of network meta-analyses for conducting comparative effectiveness research?

Network meta-analyses are an extension of traditional meta-analyses and use new statistical methods to incorporate evidence from both direct and indirect comparisons of treatments and trials. They generate new information for key stakeholders in the healthcare industry, including physicians, pharmacists, healthcare insurers and treatment providers. This new evidence can give physicians additional information to consider, improving treatment options available for patients, including treatment access and reducing costs for payers and society.

Network meta-analyses…generate new information for key stakeholders in the healthcare industry, including physicians, pharmacists, healthcare insurers and treatment providers.”

In the area of treatment provision for advanced renal cell carcinoma, as an example, Ipsen used a network meta-analysis to respond to the challenge that healthcare practitioners have faced in the last year with the introduction of immunotherapies and with no way to objectively, directly compare their effectiveness. While double-blinded, randomized trials are considered the gold standard for comparative effectiveness research, it would take several years before head-to-head comparisons could be completed. In some cases, it’s not always practical or possible to test every medicine against the other medicines that treat the same condition to answer pressing clinical questions.

In these cases, network meta-analyses may offer some helpful insights; for example, they suggest that cabozantinib significantly increases progression-free survival in intermediate- and poor-risk patients, compared with current standards-of-care. We can therefore conclude that cabozantinib may be an efficient, first-line treatment option for advanced renal cell carcinoma.


What are some of the challenges associated with methods used to economically estimate budget impacts of therapies, and what limitations may these impose?

The budget impact of a therapy is only one piece of the value of a drug and, therefore, results should be interpreted considering the clinical, safety and cost–effectiveness aspects. One limitation to keep in mind when developing a budget-impact model, at the time of launch, is that part of it relies on assumptions – such as evolution of market share after introduction of the new drug – and on data coming from randomized controlled trials, for example, data on dosing or treatment duration, which may not be representative of future real-world use in clinical practice. Therefore, there is great value in updating budget-impact models with real-world data once available. Generation of real-world evidence through ‘traditional’ methods can be quite time and resource consuming.

Modern data-mining techniques are greatly beneficial as they make real-world evidence generation much more efficient.”

Modern data-mining techniques are greatly beneficial as they make real-world evidence generation much more efficient. We’ve seen the generation of real-world evidence increase significantly these past few years, allowing treatment-decision algorithms to be refined beyond randomized controlled trial data, and ultimately further improve patients’ outcomes. However, data-mining approaches do have potential limitations as well. Accessibility of data can be one, particularly when it comes to global analyses, as the inputs for these systems often come from a range of settings and a variety of systems are used.

Once accessed, problems surrounding lack of data standardization can arise if data are missing or inconsistent with the format required. Both limitations make analysis and accurate prediction more difficult, which can result in more expensive projects, both in terms of time and money.


What novel healthcare insights may modern quantitative data reviews offer?

Data mining is the process of learning data from information technology to identify hidden structures that allow knowledge discovery and predictions on the evolution of a phenomenon.

Data-mining techniques can extrapolate new patterns and knowledge from collected datasets, which can be applied to the development of decision-making models. In the healthcare space, this can assist clinical decision making and help improve procedures such as prognosis, diagnosis and treatment planning. Ipsen is already working on integrating this new technique into its projects across several products and indications.

Quantitative data are essential for evaluating and guiding improvements in healthcare. These data can be taken from different levels of the healthcare system, from patients to service to organization level. Budget-impact model data, together with machine learning, are great tools to generate additional real-world evidence and address questions that randomized controlled trials cannot tackle.

Budget-impact model data, together with machine learning, are great tools to generate additional real-world evidence and address questions that randomized controlled trials cannot tackle.”

At the patient level, for example, when several treatment options (A, B, C) are equally recommended, data mining can help by categorizing patients’ profiles into those who are likely to respond to A, B or C based on multiple characteristics – genomics, biomarker etc. This allows for a more personalized approach to treatment


How do you hope these fields will evolve in the future?

In the future, data-mining techniques may be commonly used in healthcare and other industries. This is thanks to the use of automated data-capture systems and more affordable computing power. Ipsen is committed in using both recognized health economics and outcomes research methods and data-mining techniques to bring new, effective treatments to patients in shorter time frames while considering both cost and access. We seek to improve value in healthcare beyond cost-containment and short-term interventions and we are engaged in accelerating the use of modern data mining that are disruptive, innovative and rapid.

To increase adoption of such techniques, we hope that the quality of electronic health records will continue to improve in order to provide more exhaustive information, ideally in a standardized format. We hope that health authorities – regulatory and health technology assessors – will consider those evidence as valid to support their decision making; at this stage it is still unclear how relevant it is for those stakeholders.

Go to the profile of The Evidence Base

The Evidence Base

Community, Future Medicine

The Evidence Base is a community site covering the latest news, opinion and insight into the collection and application of real-world data to real-world problems.
85 Contributions
5 Followers
0 Following

No comments yet.