Role of health plan administrative claims data in participant recruitment for pragmatic clinical trials: An Aspirin Dosing: A Patient-centric Trial Assessing Benefits and Long-term Effectiveness (ADAPTABLE) example.

AIM: The purpose of this study is to evaluate HealthCore/Anthem Research Network recruitment strategies, compare response and enrollment rates for different recruitment strategies, and describe demographic and clinical characteristics of responders and enrollees. METHODS: HealthCore/Anthem Research Network, a part of the Health Plan Research Network of the Patient-Centered Clinical Data Research Network, used administrative claims data to identify eligible health plan members for potential participation in the Aspirin Dosing: A Patient-centric Trial Assessing Benefits and Long-term Effectiveness study. We approached health plan members, identified with a validated Patient-Centered Clinical Data Research Network common data model computable phenotype, and their clinical providers during November 2017 to August 2018. Providers were offered the option to exclude their patients' participation in Aspirin Dosing: A Patient-centric Trial Assessing Benefits and Long-term Effectiveness prior to our direct patient (member) outreach. Member identification was in two phases: Phase 1: 1 January 2006 to 1 April 2017, and Phase 2: 1 January 2006 to 2 February 2018. Phase 1 consisted of two batches of mail and one phone call per patient. In Phase 2, which included two similar batches of patients, outreach was via either mail or brochure and one phone call. RESULTS: Phase 1 and Phase 2 included 133,373 and 51,777 members, respectively. We engaged 28,593 providers in Phase 1, and 5077 in Phase 2. In Phase 1, 264,158 mixed email/mail messages were delivered to 133,373 members, followed by 90,481 phone calls from November 2017 to February 2018. In Phase 2, after simple randomization to letter or brochure, 51,777 members were sent email/mail or mailed brochure in three waves from May 2018 to July 2018. In this 9-week period, 51,623 communications were sent to 25,914 members in the email/mail group, and 50,160 brochures to 25,863 in the brochure group. Following email/mail or mailed brochure outreach, 16,624 and 16,580 calls were made to the groups, respectively. Overall, 1549 health plan members visited the study portal by 1 September 2018; 355 electronically signed the Informed Consent Form and enrolled. Mailed brochures drove more portal visits in Phase 2, but a lower percentage of responders enrolled. Recruitment was better in Phase 2-2.3 enrollees per 1000 outreach members versus 1.8 in Phase 1. CONCLUSION: This study showed the ability of a health plan within Patient-Centered Clinical Data Research Network to identify potential study participants with administrative claims, and use different outreach methods to facilitate recruitment and enrollment for pragmatic clinical trials.

Comparison of visual assessment of coronary stenosis with independent quantitative coronary angiography: Findings from the Prospective Multicenter Imaging Study for Evaluation of Chest Pain (PROMISE) trial.

BACKGROUND: The outcomes in patients by visual assessment and quantitative coronary angiography (QCA) for obstructive coronary artery disease (CAD) are not known. Our objectives were to compare visual and QCA estimates of obstructive CAD and to assess their relationship to outcomes in stable patients with symptoms of CAD. METHODS: The PROMISE trial randomized 10,003 patients with CAD symptoms to anatomical or functional testing. Site reports of invasive angiography detailing visual stenosis and independent, blinded QCA were performed for obstructive CAD (≥50% stenosis). Disagreement between methods was determined and compared with outcomes (death, myocardial infarction, unstable angina hospitalization, or major procedural complications). RESULTS: Of 929 patients (9.3% of PROMISE cohort) with angiograms assessed by sites and QCA, 593 (64%) had obstructive CAD per site reports, whereas 428 (46%) had stenosis ≥50% per QCA. Results differed in 177 patients (disagreement rate 19.1%, κ=0.63), of whom 171 had CAD per sites but not per QCA. One-year unadjusted Kaplan-Meier event rates were highest (5.1%) when QCA and visual assessment agreed for CAD, lowest (0.9%) when the 2 agreed for no obstructive CAD, and intermediate (3.1%) for patients who had CAD per visual assessment but not per QCA. CONCLUSIONS: Visual estimation of angiograms results in more frequent diagnosis of obstructive CAD as compared with QCA. Concordance of results for presence or absence of obstructive CAD was associated with high and low event rates, respectively. Disagreement was associated with intermediate event rates, suggesting that cardiologists integrated clinical information into routine visual assessment of angiograms.