Use of National-scope Administrative Healthcare Claims Data for Targeting Physician Learners and Measuring Performance Change: Experiences from the RAPID® Primary Care CME Initiative on Rheumatoid Arthritis

Stephen Bender, Daniel S. Duch, Derek T. Dietze



Objective assessment of performance changes resulting from participation in national-scope Continuing Medical Education (CME) initiatives is a challenge for CME professionals and organizations. Also, accurately identifying potential clinician learners on a national basis who most need to improve their performance is an important related challenge. This article briefly describes the 6-year history and ongoing evolution of outcomes measurement for a CME initiative titled RAPID®--Rheumatoid Arthritis: Primary care Initiative for improved Diagnosis and outcomes, focusing on the use of national-scope aggregated administrative healthcare claims data to address these challenges.



Examination of a national scope administrative healthcare (medical) claims database representing the activity of 870,000 US clinicians was used to identify 61,382 clinicians who underperformed in diagnosing RA (defined as a high concentration of higher-risk patients, fewer than 15 RA diagnoses in 2008, and the fewest number of patients co-managed with a specialist) yet they had the perfect patient profiles under their care to make an impact in this disease state. These clinicians were managing a community of 17,748,309 patients who were living with a higher risk for RA and had not yet received a diagnosis of RA. We targeted this group of clinicians for participation in RAPID® activities by specifically recommending the activities to them (e.g., by email/direct mail). For the analysis, the group of underperforming clinicians who later participated in RAPID® was designated as “targeted” learners. “Non-targeted” learners participated in the activity because of professional interest, but without prior outreach. The proportion of RA diagnoses per patient population made by RAPID® initiative clinicians in the full year prior to participation in the RAPID® initiative (2008) was compared with the proportion of RA diagnoses made post-participation in the initiative (during 2011). Therefore, the post-activity RA diagnoses counted in the analysis were determined to be ≥ 2 years post-participation in the RAPID® program. The same performance comparisons between 2008 and 2011 were made among highly matched control groups of non-participants, with approximately 40 non-participant controls for every participating clinician. Within this overall analysis population, changes in RA diagnoses were also compared between internists (IM) and general/family practitioners (GP/FP).



A paired analysis of 784 RAPID®-participating clinicians of whom 256 diagnosed at least 1 RA patient in 2008. The number of clinicians making at least 1 diagnosis of RA increased after RAPID® participation. Overall, there was a 47% increase in the number of RAPID® providers who had at least one patient diagnosed with RA: 256/784 (2008), 376/784 (2011). For all RAPID® participating clinicians, there was an 11% statistically significant increase (P<0.001) in the proportion of RA diagnoses per total patient population in 2011 (1.25%) compared with 2008 (1.13%). In contrast, there was no change in the proportion (0.43%) of RA diagnoses per patient population in the control group. As a group, we found that IMs had better baseline performance than GP/FPs in diagnosing RA. This held true when comparing targeted, non-targeted, and control group IMs with the corresponding subgroups of GP/FPs. Non-targeted IMs who took the RAPID® course had the greatest proportion of RA diagnoses among their patients in the year before they took the course. Targeted GP/FPs had the lowest proportion of pre-RAPID® RA diagnoses. Targeted IMs (n=169) had a significant increase in RA diagnoses (16%; P=0.05) post-RAPID® participation, whereas the non-targeted group (n=109) had a significant decrease (-43%) in RA diagnoses post-RAPID® participation. The targeted GP/FPs, the group with the lowest proportion of RA diagnoses pre-participation, also had the largest increase in the proportion of RA diagnoses post-participation (27%; P<0.001). Of the 61,382 targeted prior underperforming clinicians, 1,691 of them completed at least one RAPID® CME activity and those 1,691 clinicians currently managed 265,834 of the 17,748,309 patients living at higher risk for RA. Overall, these performance improvements resulted in 1,837 newly diagnosed RA patients, and have the long-term potential to impact more than 265,834 higher-risk patients currently under the care of the participants of the RAPID® CME initiative.



This was the first study of which we are aware that used national-scope medical claims data to identify and target poorer-performing clinicians (i.e., low proportion of patients diagnosed with RA), then assessed whether participation in a national CME initiative improved performance (i.e., increased the proportion of patients with an RA diagnosis) by using the claims data to compare performance before and after the initiative. Participation in the RAPID® initiative was associated with positive changes in clinician performance, as measured by increased RA diagnoses. Results also indicated that identifying and inviting poorer performing clinicians through analysis of the medical claims database to specific CME activities improves the post-participation performance of the targeted group more than for non-targeted participants. 

Full Text:



  • There are currently no refbacks.