Health outcomes data: When, where, and how to collect it
July 24, 2019
Let’s throw a pie in the sky and imagine that all healthcare providers have ready access to data on a medication’s real-world performance, so they know whether to prescribe it to the patient across their desk; that payers can use health outcomes data to decide whether to list or delist a drug, or how to price it; and that manufacturers of high-performing medications can leverage data to reach patients more efficiently and expand their global presence.
Why is this so hard to imagine, let alone implement?
Real-world data, real-world challenges
Capturing data is one thing, using and sharing it quite another. Within a specific therapeutic area—leukemia or multiple sclerosis, for example—an academic institution may launch a database, a manufacturer may support a registry, and a hospital or clinic may gather internal data. But can a physician at the hospital access the registry? Can the academic database mine the clinic’s data? Without effective cross-talk between the databanks, they stagnate in their respective siloes and their power remains limited.
Adding a further layer of complexity, the technologies used to capture data are constantly evolving, leading to glitches along the way. Take artificial intelligence (AI), for example. In theory, AI systems can crunch a vast number of data points from previous patients and use the information to predict how a new patient will respond to treatment X, Y or Z. The problem: AI algorithms are only as good as the inputs from which they learn. If the inputs contain biases, the algorithms will carry these biases forward. As an example, an AI algorithm that learns from previous images of moles in fair-skinned individuals may accurately identify a cancerous mole in a patient with fair skin, but miss the mark in a darker-skinned patient.1
Bias can lurk in the most unexpected places. In one instance, a Toronto-based start-up discovered the AI technology they were using to identify patients with neurologic disorder only worked for native speakers of Canadian English.2 When applied to people with different accents, the technology was liable to misinterpret speech patterns and identify a disease where none existed.2
Creating the infrastructure to enable such data communication takes time, money and resources, not to mention agreement from several parties. Concerns about privacy and data integrity can easily derail the process. And with development budgets under constant scrutiny, the vision of a “data highway” may seem a distant mirage.
Moving in the right direction
For all these obstacles, interest in generating and using real-world data has ballooned in recent years. This is a welcome development, as real-world data offers two important advantages over clinical-trial data: there is a lot more of it, and it reflects the experience of real-world patients in all their complexity, rather than the narrowly defined cohorts in clinical trials.3
Not surprisingly, clinical trial data and real-world data do not always align—as exemplified in an analysis of lung cancer patients on PD-1 inhibitors, which found that real-world patients had shorter survival times than their clinical-trial counterparts.3 Such discrepancies drive home the point that real-world evidence can—and should—supplement clinical-trial results. In support of this principle, the FDA based its 2016 decision to approve a new indication for an aortic valve replacement on real-world data from a product registry.4
Real-world data can also help contain costs by shining a lens on treatments that work—and those that don’t. According an American Society of Clinical Oncology (ASCO) study, fewer than one out of five recently approved cancer drugs significantly improve survival outcomes, making it more urgent than ever to link payment to value.5 With this objective in mind, some Canadian preferred provider networks (PPNs) are mining health outcomes data to substantiate pay-for-performance agreements between payers and manufacturers.
The wider the data net, the greater its strength. The next big step in data capture will be to link databases throughout the country—and beyond—to create a data highway that prescribers, payers and manufacturers can use to improve the patient experience. A number of countries have taken significant steps in this direction--notably the United Kingdom, where pharmaceutical companies can purchase access to an EMR family-physician database.6
Canada is not so far behind. Throughout the country, initiatives in health-outcomes data are already bringing the dream of “health data connectivity” to our corner of the world.
Real-world data drivers
Patient registry:7 A collection of standardized information about a group of patients who share a condition or experience.
Specialty medication data: Data that reveals how a specialty medication performs in different patient groups.
Patient support program (PSP): These programs provide educational and logistical support to patients taking specialty medications. PSPs can also be harnessed to capture health outcomes data. In recognition of the data’s strategic importance, more and more PSPs are weaving data-collection capabilities into their upfront design.
References:
Cuttler M. Transforming healthcare: how artificial intelligence is reshaping the medical landscape. CBC.ca. April 26, 2018. https://www.cbc.ca/news/health/artificial-intelligence-health-care-1.5110892
Gershgorn D. If AI is going to be the world’s doctor, it needs better textbooks. Quartz. September 6, 2018. https://qz.com/1367177/if-ai-is-going-to-be-the-worlds-doctor-it-needs-better-textbooks/
A study showing real world use of “real world data.” Datavant 2018. https://datavant.com/2019/01/14/a-study-showing-real-world-use-of-real-world-data/
Panner M. Real-world health data push brings challenges and opportunities. Forbes Community Voice. December 12, 2018. https://www.forbes.com/sites/forbestechcouncil/2018/12/12/real-world-health-data-push-brings-challenges-and-opportunities/#4005060f3647
Value-based contracting for oncology drugs. A NEHI white paper. https://www.nehi.net/writable/publication_files/file/nehi_vbconcology_final.pdf
Cavlan O et al. Real-world evidence: From activity to impact in healthcare decision making. McKinsey & Co. https://www.mckinsey.com/industries/pharmaceuticals-and-medical-products/our-insights/real-world-evidence-from-activity-to-impact-in-healthcare-decision-making
Workman TA. Engaging Patients in Information Sharing and Data Collection: The Role of Patient-Powered Registries and Research Networks. Agency for Healthcare Research and Quality, 2013. https://www.ncbi.nlm.nih.gov/books/NBK164514/