FAVOR Factors Associated with Variability Of Radiologists 2000

  • Slides: 16
Download presentation
FAVOR Factors Associated with Variability Of Radiologists (2000 to present) Supported by grants from

FAVOR Factors Associated with Variability Of Radiologists (2000 to present) Supported by grants from the National Cancer Institute (R 01 CA-107623; K 05 CA-104699) and the Agency for Healthcare Research and Quality

Development of FAVOR My interest in mammography My frustration over data from academic centers

Development of FAVOR My interest in mammography My frustration over data from academic centers and experts, not community setting Invitation from Steve Taplin to attend the 1998 BCSC meeting

FAVOR I (2000 to 2005) Specific Aims 1. Radiologists’ characteristics (new survey data collected)

FAVOR I (2000 to 2005) Specific Aims 1. Radiologists’ characteristics (new survey data collected) 2. Facility level characteristics (new survey data collected) 3. Statistical modeling Multi-site project: BCSC sites (GHC, CO, NH), the Univ of WA and Duke

Computer-Aided Detection (CAD) Marks on Screening Mammogram

Computer-Aided Detection (CAD) Marks on Screening Mammogram

Impact of CAD on Performance Recall rate 10% to 13% (p<0. 001) Biopsies completed

Impact of CAD on Performance Recall rate 10% to 13% (p<0. 001) Biopsies completed per 15 to 18 (p<0. 001) 1, 000 screens increased Sensitivity (% of women with cancer that have positive tests) 80% to 84% J Fenton et al. , NEJM 2007

CAD Use and Overall Accuracy AUCs for facilities without vs. with CAD are 0.

CAD Use and Overall Accuracy AUCs for facilities without vs. with CAD are 0. 919 vs. 0. 871 (p=0. 005). Adjusted for patient, radiologist, and facility characteristics, registry, and facility-level random effects.

Impact of CAD on Performance % of cancers that are DCIS increased with CAD

Impact of CAD on Performance % of cancers that are DCIS increased with CAD 25% to 34% (p<0. 04) J Fenton et al. , NEJM 2007

FAVOR I Medical Malpractice • Malpractice experience (N=127 MD’s) – 52% reported prior claim

FAVOR I Medical Malpractice • Malpractice experience (N=127 MD’s) – 52% reported prior claim – 15% reported prior mammography related claim • Malpractice effect on clinical practice – 76% expressed concern about the impact on their practice – 59% concerned that malpractice increased their recommendations for biopsy Elmore et al. Radiology 2005.

Radiologists’ Estimates of the 5 -year risk of a breast cancer diagnosis. “A 41

Radiologists’ Estimates of the 5 -year risk of a breast cancer diagnosis. “A 41 -year old white woman. . . ” Calculated risk Egger et al. Med Dec Mak 2005.

FAVOR II (2006 to present) Specific Aims 1. Statistical methods 2. Radiologists’ characteristics (new

FAVOR II (2006 to present) Specific Aims 1. Statistical methods 2. Radiologists’ characteristics (new survey collection) 3. Continuing Medical Education (CME) 4 BCSC sites initially Expanded to all 7 BCSC sites to gather survey data on interpretive volume and practice locations for the AIM study (Funding from the American Cancer Society/NCI)

FAVOR Involvement of Junior Investigators • Medical Students – He. Yon Sohng • Residents

FAVOR Involvement of Junior Investigators • Medical Students – He. Yon Sohng • Residents – John Dick – Laurel Desnick • Ph. D students – – Gavino Puggioni Prashni Paliwal Carolyn Prouty Dawn Woodard • MPH students – Connie Nakano • Junior Faculty Members – – – – Amy Baernstein John Choe Joshua Fenton Tom Gallagher Sara Jackson Hillary Liss Shin Ping Tu

FAVOR Supplement Disclosure of Medical Errors • Vignette added to FAVOR II survey on

FAVOR Supplement Disclosure of Medical Errors • Vignette added to FAVOR II survey on the topic of medical errors in mammography (T. Gallagher et al. , Radiology 2009) • Supplement funding to support qualitative research project.

Beyond FAVOR New R 01 on Breast Pathology • “BPATH” – A study of

Beyond FAVOR New R 01 on Breast Pathology • “BPATH” – A study of variability in interpretation of breast pathology, the impact of digital images and double reading • 5 -year multicenter study (R 01 CA- 140560) • Using pathology cases from BCSC sites in NH, VT • Random stratified sampling, with emphasis on atypia and DCIS • Development of test sets and future CME

Beyond FAVOR Other Areas to Investigate Comparative effectiveness of new technology in the real

Beyond FAVOR Other Areas to Investigate Comparative effectiveness of new technology in the real community setting Medicare linkage data elements Diagnostic evaluation of breast abnormalities

FAVOR Summary • Synergistic impact of collaborations with BCSC • Collaboration with BCSC has

FAVOR Summary • Synergistic impact of collaborations with BCSC • Collaboration with BCSC has allowed multiple cost effective R 01’s with unique linkage to performance data • FAVOR provided augmented data on individual radiologists and facilities and infrastructure to expand support the AIM study • Supportive of junior investigators • Development of long term collaborations