Look beyond self-reported history
See how Contxt works alongside your existing recruitment process.
Challenge: Self-reported health histories cost the organization twice
A large clinical research organization had a database of nearly 2,000 people interested in participating in research, but the health histories in it were largely self-reported. This meant eligibility was only as accurate as a respondent’s ability to remember past diagnoses, recall the medications they took, and report history that affected matching.
That created two problems.
Sending the wrong candidates to sites
The first, and more immediate, was the risk of sending ineligible candidates to study sites. A respondent could look like a match on paper and be referred, only to be ruled out during site screening. That wasted site time, discouraged the candidate, and added avoidable recruitment costs.
Missing the right candidates for the next study
The second was missed opportunity. Willing participants are among the hardest parts of a trial to find, so a database like this is valuable. Someone who doesn’t fit one study may fit another, especially for a hard-to-recruit condition. When a new study opened, the database should have been the organization’s first stop, but only if the information in it was complete.
Solution: Prescription and medical data fill in what self-reports leave out
The organization worked with Contxt, by Milliman IntelliScript, on a retrospective analysis of 1,996 respondent records, adding longitudinal prescription and medical data to self-reported histories. Data came back for about 95% of the records.
Catching gaps before referral
Contxt flagged records where a respondent’s prescription or medical history didn’t line up with what they had reported, such as someone who said they didn’t have a condition while their data suggested otherwise. Those flags tell coordinators which histories to confirm before a candidate is sent to a site, not after.
Finding candidates already on file
Contxt also identified patterns pointing to diagnoses, treatments, and conditions that didn’t appear in the recruitment records. That gave the organization a way to search its database for more than what respondents had reported, including conditions that are hard to recruit for.
Results: A clearer view of who to refer and who to call next
Broader coverage
Prescription or medical data returned for about 95% of the 1,996 records reviewed.
Gaps to catch before referral
Nearly 300 records (about 15%) where respondents reported not having a condition, but their prescription or medical history suggested otherwise.
Leads for future studies
More than 520 condition signals missing from the recruitment records.
Signals tied to hard-to-recruit conditions, including heart valve disorders and schizophrenia.
