Variation in polyp detection rates at screening colonoscopy

• Published in: Gastrointestinal endoscopy 2009-06, Vol.69 (7), p.1288-1295
• Main Authors: Imperiale, Thomas F., MD, Glowinski, Elizabeth A., RN, Juliar, Beth E., MS, MA, Azzouz, Faouzi, MS, Ransohoff, David F., MD
• Format: Article
• Language: English
• Subjects: Adenoma - diagnosis; Biological and medical sciences; Clinical Competence; Colonic Polyps - diagnosis; Colonoscopy - standards; Colorectal Neoplasms - diagnosis; Cross-Sectional Studies; Digestive system. Abdomen; Early Detection of Cancer; Endoscopy; Female; Gastroenterology and Hepatology; Gastroenterology. Liver. Pancreas. Abdomen; Humans; Investigative techniques, diagnostic techniques (general aspects); Male; Mass Screening; Medical sciences; Middle Aged; Quality of Health Care; Retrospective Studies
• ISSN: 0016-5107; 1097-6779
• DOI: 10.1016/j.gie.2007.11.043

Background Variation in polyp detection among endoscopists has been used to justify the need for establishing quality standards for colonoscopy performance. Objective To measure variation in polyp detection rates (PDRs) among endoscopists who perform screening colonoscopy and to identify associated factors. Design Cross-sectional analysis of summary-level data. Setting Endoscopy practices in central Indiana. Subjects Twenty-five endoscopists and their patients. Main Outcome Measurements Mean procedure time (MPT); proportions of patients with any polyp, any adenoma, any polyp ≥1.0 cm, and multiple adenomas; and variation in PDRs and identification of outliers. Multiple linear regression analysis identified factors that accounted for the variation in PDRs. Results A total of 2664 screening colonoscopies (1108 women and 1556 men) were performed. The mean patient age was 59 years; the mean proportion of women was 42%; the MPT was 17.1 minutes. Adenoma detection rates ranged from 7% to 44% ( P < .001) and from 0% to 13% for large polyps, which was not statistically significant ( P = .07). For all polyp categories, only 1 to 3 high outlier endoscopists (ie, higher than mean PDRs) were identified. Models that included the number of procedures, mean age, percentage of women, and MPT accounted for 36% to 56% of the variation in PDRs. In all models, only MPT was significantly associated with PDRs. Limitations Whether each endoscopist's cohort was at comparable risk for colorectal neoplasia was uncertain. In comparison with individual-level data, analysis of summary-level data is limited. Conclusions PDRs vary widely among endoscopists, although only a few (high) outliers were identified. Variation in PDRs was associated only with MPT. Further research is needed to determine the clinical importance of and reasons for this variation.