Knowing patient survival rates is critical to informing research and care delivery, particularly in the oncology space. But what if survival data is inaccurate or incomplete?
Well, that could jeopardize the development of new treatments and the ability for providers to address adverse outcomes. So argues COTA, a tech company providing oncology real-world data abstraction, curation and analytics to providers and life sciences companies. It pulls data from EHRs of partner providers and commercially available data from obituaries and the Social Security Administration to curate a longitudinal and de-identified dataset.
“We’re here to provide clarity to all touched by cancer,” COTA’s CMO C. K. Wang, M.D., told Fierce Healthcare.
In a recent study, the findings of which were presented at the annual American Society of Hematology meeting earlier this month, COTA looked at how its data compares to that of the National Death Index, a gold standard for mortality data. Data on 21,567 patients was included. Exact date concordance between COTA’s data and the National Death Index was observed in 88% of patients, a rate the peer-reviewed study determined to be high.
EHRs have revolutionized the way providers capture data, but they are not always accurate or complete. One quality improvement study of nearly 11,700 seriously ill patients at one academic health system recently found that 19% of those deceased were marked alive in the EHR. This inaccuracy hinders efficient health management, billing, advanced illness interventions and measurement, per the study. It also impedes the health system’s ability to learn from adverse outcomes.
Skewed data could then generate inaccurate or biased large language models or algorithms, according to COTA’s Chief Medical Officer C. K. Wang, M.D. “Datasets need to be as complete as possible and as representative as possible,” Wang told Fierce Healthcare.
While the data sources COTA pulls from are publicly available and anyone could access the same information, Wang noted, many are not even aware they have a data completeness issue. Others may not have the technical capabilities or resources to compile the complex data.
“It is quite an effort for providers or an institution to try to track down that patient. I don’t think anyone is incentivized to do that honestly,” Wang said. “Everything comes back to ROI and how much do I really have to put into this and whether it’s worth it.”
A lack of diversity in clinical trials is a well-documented problem. Implicit bias in oncology in particular also persists, affecting patient care, especially for vulnerable populations. Over the years, COTA has explored how a lack of representative sexual orientation and gender identity data impacts understanding of how they are impacted by cancer and various treatments, Wang said. The same can be said of mortality data in oncology.
While there isn’t much of a concrete benchmark for mortality data in medicine yet, COTA hopes that its study’s findings help establish a best practice methodology to be leveraged by stakeholders and help fill gaps in mortality data.