Big data provide opportunity for research, drive learning at the healthcare system level

Reviewed by Kate Anderton, B.Sc. (Editor)Nov 7 2019

ISPOR, the professional society for health economics and outcomes research-;held its third plenary session at ISPOR Europe 2019, "Big Data Healthcare: Endless Opportunities for Research and Learning," this morning in Copenhagen, Denmark.

Big data present a tremendous opportunity for the measurement and reporting of quality in healthcare that can enhance insight and decision making. Healthcare data vary and include discrete coded data elements, images of diagnostics tests, and unstructured clinical notes.

This type of high-volume and high-variety information demands innovative forms of information processing. In this session, panelists discussed a number of examples where experts are effectively using big data and for research and to drive learning at the healthcare system level.

Panelists for this session included:

• Dorte Gyrd-Hansen, PhD, MSc, Danish Centre for Health Economics,University of Southern Denmark, Odense, Denmark
• Laura Hatfield, PhD, Harvard Medical School, Boston, MA, USA
• Jeremy Rassen, ScD, Aetion, Inc, New York, NY, USA
• Michal Rosen-Zvi, PhD, IBM Research, The Hebrew University, Jerusalem, Israel

Dr Rassen believes that the debate over randomized controlled trials (RCTs) versus real-world evidence (RWE) is a false dichotomy.

He argued that, rather than being an "either/or situation," RWE can augment and extend RCT evidence. Rassen noted that the issue is not one of study methodology, but about establishing causality, and that both RCTs and RWE can "allow you to climb the causal ladder." He noted that if researchers apply "principled process" that RWE can be equally effective to RCTs.

Rassen identified principled process as including (a) thoughtful appropriateness assessment, (b) verification checks for real-world data reliability, (c) target-trial paradigm, and (d) transparency guidelines (as noted in the ISPOR/ISPE joint publication).

He stressed that without application of principled process, RWE may not be considered regulatory grade. He also noted that cases remain where RCTs are better suited, such as placebo-controlled pivotal trials and where comparison is inappropriate for RWE.

Dr Hatfield spoke about how we can "know what works" in a sea of healthcare ideas. She pointed out that experimentation is everywhere, yet often RCTs are impractical for assessment.

Healthcare produces much data through billing claims and electronic health records and Hatfield noted that this invites "quasi-experimental" analyses. She also provided an overview of "difference in differences" (aka "diff-in-diff") methodology that attempts to mimic a controlled trial by using observational data.

Dr Rosen-Zvi spoke about how electronic medical records can be used to provide better healthcare. She discussed the transformative power of machine learning and stated that it has both technical and nontechnical challenges.

The nontechnical challenges identified by 180 experts included regulatory, privacy and data protection, user acceptance and public opinion, lack of standardization, legal, and availability of skilled artificial intelligence (AI) experts. Rosen-Zvi illustrated the promise of AI with an example where AI improved breast cancer detection.

There was agreement among the panelists that the opportunities for big data in healthcare are vast and the challenges are great. New methods are needed to harness the potential for these data to have real impact.

ISPOR is recognized globally as the leading professional society for health economics and outcomes research and its role in improving healthcare decisions.

ISPOR Europe 2019 expects to draw more than 5000 healthcare stakeholders with an interest in HEOR, including researchers and academicians, assessors and regulators, payers and policymakers, the life sciences industry, healthcare providers, and patient engagement organizations.