Conducting Research on the Internet: Medical Record Data Integration with Patient-Reported Outcomes

Abstract

BackgroundThe growth in the number of patients seeking health information online has given rise to new direct-to-patient research methods, including direct patient recruitment and study conduct without use of physician sites. While such patient-centric designs offer time and cost efficiencies, the absence of physician-reported data is a key concern, with potential impact on both scientific rigor and operational feasibility. ObjectiveTo (1) gain insight into the viability of collecting patient-reported outcomes and medical record information in a sample of gout patients through a direct-to-patient approach (ie, without the involvement of physician sites), and (2) evaluate the validity of patient-reported diagnoses collected during a patient-reported outcomes plus medical record (PRO+MR) direct-to-patient study. MethodsWe invited a random sample of MediGuard.org members aged 18 to 80 years to participate via email based on a gout treatment or diagnosis in their online profiles. Interested members clicked on an email link to access study information, consent to participate electronically, and be screened for eligibility. The first 50 consenting participants completed an online survey and provided electronic and wet signatures on medical record release forms for us to obtain medical charts from their managing physicians. ResultsA total of 108 of 1250 MediGuard.org members (8.64%) accessed study information before we closed the study at 50 completed surveys. Of these 108 members who took the screener, 50 (46.3%) completed the study, 19 (17.6%) did not pass the screening, 5 (4.6%) explicitly declined to participate due to the medical record requirement, and 34 (31.5%) closed the browser without completing the survey screener. Ultimately, we obtained 38 of 50 charts (76%): 28 collected using electronic signature and 10 collected based on wet signature on a paper form. Of the 38 charts, 37 cited a gout diagnosis (35 charts) or use of a gout medication (2 charts). Only 1 chart lacked any mention of gout. ConclusionsPatients can be recruited directly for observational study designs that include patient-reported outcomes and medical record data with over 75% data completeness. Although the validity of self-reported diagnosis is often a concern in Internet-based studies, in this PRO+MR study pilot, nearly all (37 of 38) charts confirmed patient-reported data.