A 5-Year Statewide Analysis of Unplanned Hospital Visits for EGD, Colonoscopy, Combined EGD/Colonoscopy, and ERCP

Conventional complication rates for gastrointestinal endoscopic procedures may underestimate the broader risk represented by postprocedure unplanned hospital visits (UHVs). We aimed to characterize UHVs for procedures in Maryland and the District of Columbia from 2014 to 2018.

Data for all esophagogastroduodenoscopies (EGDs), colonoscopies, combined EGDs/colonoscopies, and endoscopic retrograde cholangiopancreatographies (ERCPs) performed between 2014 and 2018 was provided by the Maryland Health Information Exchange (Chesapeake Regional Information System for our Patients'). Patient demographics, timing of UHV within 14 days postprocedure, distance traveled, facility site ("home" vs "away" institution), and International Classification of Diseases codes for the UHV were analyzed. Only UHVs potentially attributable to the endoscopic procedures were included.

Among 304,786 endoscopic procedures and 3904 unplanned visits, the 14-day UHV rates were 1.7%, 0.6%, 1.3%, and 5.2% for EGD, colonoscopy, combined EGD/colonoscopy, and ERCP procedures respectively. From 2014 to 2018, the UHV rate on an annual basis remained stable for all procedure types except for ERCPs which increased. Patients who experienced UHVs were statistically different in sex, race, age, and distance traveled. UHVs most often occurred on postprocedure day 1; emergency department visits occurred most commonly. UHVs for all procedures, except ERCPs, were more likely to occur at a "home" institution. Overall, patients were more likely to be admitted postprocedure at an "away" institution.

Postendoscopic procedure UHV rates were generally low. However, UHV rates for EGDs and colonoscopies were significantly higher than conventional complication rates. As 30%-60% of all unplanned visits occurred at an "away" institution, endoscopists should consider a broad approach to detecting postprocedure complications and not rely on a single institution for data capture.