Pitfalls To Avoid When Leveraging Clinical Imaging Data for RWE Programs

By Janak Joshi, senior vice president, chief technology officer and head of strategy, Life Image.

Janak Joshi

In December 2018, the FDA announced its new framework for the real world evidence (RWE) program, which would require including imaging data alongside claims, electronic health records (EHRs) and other datasets in clinical research. In issuing this new framework, regulators underlined the continued importance of using contextualized, quality datasets to make drug development faster, safer, more efficient and less expensive.

Because of this move to include authentic patient data in the drug development process, imaging data has become an essential part of RWE as it can accelerate the development cycle and improve the confidence in the final clinical arguments in support of drugs going to market.

Imaging data plays such a leading role in clinical decision-making because it is the most advanced diagnostic evidence for several diseases, and it can clearly show disease progression and drug impact across a variety of therapeutic areas, among other reasons. While EHRs and medical claims are the predominate sources of data, because they were initially designed for billing and payment purposes they do not have the depth and breadth needed to accurately capture the nuances of a patient’s full clinical history – nor do they contain imaging information.

Clinical researchers looking to achieve a holistic view of each patient’s healthcare journey by incorporating medical imaging into their RWE programs should avoid these three things.

Biased data

Institutional bias stems from using data from a single health system, which tends to follow a uniform set of treatment protocols, leading to homogenous evidence data. A diverse dataset includes variation, for instance in geography, which can influence socioeconomic and environmental factors, level of education, healthcare access, payer mix and demographics.

The most effective RWE incorporates medical data, including imaging, from varied populations that include both research and non-research settings, AMCs and community hospitals, publicly and privately funded institutions, and a mix of highly insured and uninsured patients. The ultimate goal of RWE is to be representative of any and all patients across the globe.

A limited, siloed data pool

Small datasets do not accurately reflect the “real world,” therefore RWE requires very large databases with various datasets in order to ensure data integrity and credibly match patients to appropriate clinical trials. This poses a challenge since much of today’s data is siloed. To make RWE representative of outcomes and context, clinical researchers must break down siloes to achieve a large, interoperable pool of quality data from a breadth of sources, which they can normalize and match across sets for optimal results.

Take, for example, a new drug trial that needs to involve 500 individuals meeting specific real-world data standards. For each participant, researchers may require four years of prescription details, four years of imaging data, five years of blood test results, as well as genomics and other relevant data. However, consider that over the years many of these patients likely went to various pharmacies, switched health plans and/or providers, and had imaging and blood tests performed at various facilities or out-of-network sites. As a result, each patient’s information may be spread out over multiple EHR systems and may even be in non-digital, fax or CD formats.

Even with state-of-the-art technology, finding such a large number of patients that meet inclusion and exclusion criteria would potentially require clinical researchers to process data from millions of claims across multiple payer organizations, with millions of pharmacy benefit manager records, and countless patient profiles from different EHR systems, all while analyzing non-digitized technology. Having access to massive databases of myriad datasets allows researchers to more efficiently and effectively identify patients for clinical research.

Using data out of context

RWE’s power lies in its ability to represent how healthcare is delivered and experienced in reality, not just in a trial setting. RWE programs must take into consideration that a drug compound, medical device or even a computerized diagnosis affects people differently due to factors ranging from biology to psychology. To improve the statistical models used to determine drug safety and effectiveness, treatment performance must be measured within the context of the patient.

Data must be standardized, normalized and interoperable to be useful; lack of interoperability will at best slow the RWE process and at worst bias data. Technical standards for images from devices or storage can differ greatly from manufacturer to manufacturer, therefore the ability to pull data agnostically from disparate sources of imaging is crucial to support RWE’s needs for data breadth, diversity and quantity.

By definition, RWE needs to come from the context of actual medical decisions and activity – but today medical data chiefly stems from EHR systems which are geared toward billing instead of the context of medical decisions. The persistent challenge of RWE is creating and expanding tools that make the unstructured data populated by clinicians for clinicians available, accessible and interoperable. Since comprehensive patient indexing technology has not been historically applied to imaging, organizations looking to enhance their data sets with imaging RWE must normalize and combine imaging with diverse data sets in order to follow the full patient journey.

Write a Comment

Your email address will not be published. Required fields are marked *