Do Adenoma Detection Rates Decline Before Holidays and During Major News Events?

Introduction: Variations in colonoscopy adenoma detection rate (ADR) may be due to a number of factors; for example, lower ADR has been observed in afternoon sessions compared to morning sessions. However, provider distraction is a factor that has not been thoroughly investigated. We aimed to determine if increased external distraction from looming holidays or major news events would negatively effect ADR. Methods: Using our institution’s endoscopy record, we performed a retrospective study of colonoscopies performed between 7/31/16 and 7/31/21 on patients ≥ age 50. The primary comparison was the difference in ADR during time periods of increased distraction compared to baseline. Time periods of increased distraction were divided into 2 categories: holiday eves, i.e. the Fridays before a 3 day weekend, and news events, i.e. the 24 hours after a notable national news story, as defined by a composite of year-end top news headlines from media companies including CBS, US News, and Newsweek. We compared the ADR on holiday eves to Fridays not preceding a holiday. We compared the ADR during major news events to that of colonoscopies done on other weekdays. We examined the independent associations of holiday eves and major new events with ADR after adjusting for age, gender, race, ethnicity, FHx of colon cancer, indication for colonoscopy, adequacy of preparation and presence of a fellow using multivariate logistic regression. Results: We identified 9,999 colonoscopies performed during the study period (69% screening, 31% surveillance), including 169 that occurred on a holiday eve and 176 that occurred during a news event. The ADR during news events was 33%, which was not significantly different to other colonoscopies (37%, P = 0.3). The ADR on holiday eves was 36%, which was not significantly different compared to other Friday colonoscopies (35%, P = 0.8). Multivariate analysis showed age, gender, colonoscopy preparation quality and surveillance colonoscopies were each independent predictors of ADR, but that holiday eves (OR 1.04; 95% CI 0.74-1.46) and news events (OR 0.85; 95% CI 0.61-1.16) were not. The presence of a fellow was associated with adenoma detection in the model that evaluated news events (see Table 1). Conclusion: Time periods of increased distraction were not associated with a significant decrease in ADR. Further studies should determine the strategies used to maintain attention during holiday eves and news events and incorporate them to combat periods of known decreased ADR such as afternoons. Table 1. - Holiday Eve Analysis Compared Colonoscopies Performed on Fridays. News Event Analysis Compared Colonoscopies Performed on All Weekdays Multivariate Analysis of ADR on Holiday Eve Fridays vs Regular Fridays (n=2,070) Multivariate Analysis of ADR in the 24hrs After News Events vs Regular Days (n=9,999) Variable OR 95% CI P value Variable OR 95% CI P value Holiday Eve No 0.8 News Event No 0.3 Yes 1.04 0.74-1.46 Yes 0.85 0.61-1.16 Age Range 50-59 < 0.001 Age Range 50-59 < 0.001 60-69 1.27 1.03-1.58 60-69 1.25 1.13-1.38 70-79 1.56 1.20-2.03 70-79 1.69 1.51-1.90 80+ 2.86 1.63-5.12 80+ 2.44 1.90-3.14 Gender Female < 0.001 Gender Female < 0.001 Male 1.51 1.25-1.82 Male 1.54 1.41-1.67 Race White 0.9 Race White 0.5 African American/Black 0.92 0.65-1.29 African American/Black 1.03 0.88-1.21 Unknown or Other 0.99 0.80-1.23 Unknown or Other 1.06 0.96-1.16 Ethnicity Non Hispanic or Latino < 0.001 Ethnicity Non Hispanic or Latino < 0.001 Hispanic or Latino 1.62 1.24-2.11 Hispanic or Latino 1.26 1.12-1.42 Unknown 2.23 0.90-5.43 Unknown 1.88 1.14-3.07 Family Hx of Cancer No 0.7 Family Hx of Cancer No >0.9 Yes 0.92 0.62-1.35 Yes 1 0.86-1.18 Colonoscopy Indication Screening < 0.001 Colonoscopy Indication Screening < 0.001 Surveillance 1.63 1.32-2.01 Surveillance 1.5 1.37-1.65 Adequate Preparation No < 0.001 Adequate Preparation No < 0.001 Yes 2.85 1.67-5.17 Yes 1.64 1.31-2.07 Fellow Presence No 0.2 Fellow Presence No < 0.001 Yes 1.16 0.92-1.47 Yes 1.33 1.2-1.47 Logistic regression modeling used to obtain odds ratios (ORs) and confidence intervals (CI).