P340: Integrating Patient-Reported Outcomes and Electronic Health Records Data for Innovative Real-World Evidence Generation
Poster Presenter
Mei Sheng Duh
Managing Principal
Analysis Group, Inc. United States
Objectives
To evaluate the feasibility of routine prospectively collected patient-reported outcome measures (PROMs) linked to electronic health records (EHR) from a large academic medical center to generate enhanced real-world evidence, using an ovarian cancer case study.
Method
EHR data from the Mass General Brigham (MGB) Research Patient Data Registry were linked with MGB PROMs data. Adult patients with ovarian cancer initiating a novel therapy were followed from therapy initiation (index date). Availability and completeness of PROMs were assessed pre- and post-index.
Results
Among 264 patients with ovarian cancer starting a specific novel therapy identified in the EHR, 60% had at least one linked PROM of interest (i.e., EORTC QLQ-C30, EORTC QLQ-OV28, PROMIS Emotional Support, PROMIS Instrumental Support, and PROMIS-10). PROMs were available for 48% of patients pre-index and 44% post-index; 33% had measures in both periods. During follow-up, patients with =1 PROM had an annualized mean of 9.5 measures (SD 14.2), with an average interval of approximately 3 months between sequential assessments.
PROM capture via routine clinical care varied by instrument but the patient response rates and survey completion rates are markedly higher than traditional one-off patient PRO survey studies. The EORTC instruments were most frequently completed. Among patients with =1 EORTC, approximately 60% had measures in both pre- and post-index periods, with an average of ~7 measures pre-index and ~5 annually during follow-up. The percentages of patients with at least one pre-index AND one post-index PRO measurements (ie, =2 survey responses per patient) were high, in the range of 27%-61%, for the following instruments: EORTC-QLQ-C30 (61%), EORTC-OV-28 (57%), PROMIS- emotional support (45%), PROMIS- instrumental support (45%), and PROMIS- Global 10 (27%).
PROM administration context influenced completeness. Capture was higher when PROMs were deployed as part of a structured surgical series compared with appointment-anchored collection. These findings highlight the importance of understanding clinical workflow and implementation context when interpreting PROM-derived real-world data.
Conclusion
Regulators and health technology assessment authorities increasingly recognize the importance of integrating patient perspectives into assessments of novel therapies. Traditionally, real-world patient-centered evidence are generated through prospective surveys. This traditional method comes with several important limitations: (1) low consent rates (in the range of 5%-25%), raising questions about responder bias and generalizability, (2) the need for prospective longitudinal follow-up, resulting in a rather lengthy study duration (typically 3-5 years), and (3) high study cost to support the trial-like infrastructure for prospective survey data collection. The uniqueness of the MGB system through their quality of care improvement program by integrating patient perspectives into clinical care has created an innovative data ecosystem that offers an opportunity to systematically combine patient-centered insights and clinical data to strengthen real-world evidence.
In this real-world case study, over half of ovarian cancer patients initiating a novel therapy had at least one linked PROM, and a substantial subset had repeated, longitudinal measures.
Leveraging this uniquely integrated PROM–EHR dataset enables more nuanced evaluation of patient experience alongside clinical outcomes, strengthening the evidence base for benefit–risk and reimbursement decision-making. Routine prospectively collected PROMs represent a cost-efficient alternative to traditional prospective survey data collection, which is often resource- and time-intensive. By linking PROMs with EHR data, this approach enables scalable generation of patient-centered real-world evidence, supports characterization of holistic disease burden, and informs clinical, regulatory, and reimbursement decision-making.
