Purpose: Colonoscopy can decrease colorectal cancer (CRC) mortality; however, performing colonoscopy more frequently than recommended may increase cost and risk. The primary aim of this study was to determine rates and correlates of physician nonadherence to guidelines for colonoscopy screening and polyp surveillance intervals. The effectiveness of cancer screening and prevention programs will be diminished if capacity is reduced by overutilization of colonoscopy due to physician nonadherence to published guidelines. Results from this study can inform efforts to improve these types of programs, enhance provider education, and better align use of colonoscopy for CRC screening and surveillance with clinical guidelines.
Methods: The VA electronic medical records of 2,443 patients aged 50 to 64 years, with a fiscal year 2008 VA facility-performed colonoscopy, and no colonoscopy in prior 10 years, were reviewed and data were abstracted. Additional physician data were merged from the American Medical Association master file. Patients with incomplete colonoscopies or inadequate bowel preparations were excluded from analysis (n = 988). Nonadherence was defined as any recommendation that did not match the guideline follow-up interval. Data were collected from 25 VA hospitals that were randomly selected from 85 qualifying facilities stratified by academic affiliation (academic/nonacademic), geographic region, resource level (complexity score = high vs medium/low), and subgroups that performed ≥ 500 colonoscopies in fiscal year 2008. Patients were grouped into clinical scenarios based on index colonoscopy results including no polyp/normal tissue (n = 893), hyperplastic polyp only (n = 203), low risk adenoma (1 to 2 adenomas < 1 cm; n = 231), and higher risk adenoma (3-10 adenomas or any adenoma ≥ 1 cm or high grade dysplasia; n=128). The proportion of nonadherent recommendations was calculated for each of these 4 groups. We calculated adjusted odds ratios (OR) and 95% confidence intervals (CI) for the association between nonadherence, risk group, and physician characteristics, using SUDAAN software (v11.0.0) to account for our complex sample design.
Results: The overall nonadherence rate was 36% and ranged from 3% to 80% across medical facilities. Nonadherence varied by colonoscopy results: 28% among normal colonoscopies, 52% for hyperplastic polyps, 45% for low risk adenomas, and 49% for higher risk adenomas. All nonadherent recommendations dictated a shorter interval. In adjusted analyses, nonadherence was significantly higher for colonoscopies with hyperplastic (OR = 3.1, 95% CI [1.7, 5.6], or adenomatous polyps [low risk adenoma OR = 2.1], 95% CI [1.1, 4.2]; high risk adenoma OR = 2.8, 95% CI [1.2, 6.7]) compared with normal colonoscopies. Nonadherence was also significantly associated with geographic region, Charlson comorbidity score, and colonoscopy type. There were higher adjusted odds of nonadherence for colonoscopies performed in the Northeast than those of the Midwest (OR = 4.6, 95% CI [1.8, 11.8]); for increasing comorbidity (OR = 1.2 for a 1-unit Charlson score increase, 95% CI [1.1, 1.]), and for colonoscopies in which the indication was surveillance rather than screening (OR = 2.5, 95% CI [1.1, 5.6]). No other statistically significant associations were noted.
Conclusions: In a managed care setting with salaried physicians, more than one-third of recommendations suggested a need for repeat colonoscopy sooner than guidelines. The strongest predictors of nonadherence were colonoscopy findings and geographic region. Further research should investigate if targeting nonadherence to colonoscopy guidelines can reduce colonoscopy overuse and associated health care costs.
