Outcomes of Early Versus Delayed Colonoscopy in Lower Gastrointestinal Bleeding Using a Hospital Administrative...

Limited data exist on whether early colonoscopy for lower gastrointestinal bleeding (LGIB) alters 30-day mortality, performance of endoscopic intervention, or need for blood transfusion. Our primary objective was to determine whether early colonoscopy in LGIB is associated with decreased 30-day mortality using a large hospital administrative database.

Patients hospitalized between January 2008 and September 2015 were identified using a validated, machine learning algorithm for identifying patients with LGIB. "Early" colonoscopy occurred by day 2 of admission and "late" colonoscopy between days 3 and 5. A propensity score for early colonoscopy was constructed using plausible confounders. Univariable and multivariable logistic regression were used to determine factors associated with 30-day mortality, endoscopic intervention, and transfusion need. The propensity score was included as a confounding factor for mortality analysis in the multivariable model.

In total, 1204 patients underwent colonoscopy for LGIB. Of these, 295 patients (25%) underwent early colonoscopy, and these patients had a lower Charlson Comorbidity Index (P=0.001) and shorter length of stay (3 vs. 5 d, P=0.0001). Early colonoscopy was not associated with decreased 30-day mortality [odds ratio (OR), 0.73; confidence interval (CI), 0.27-1.69], but was associated with increased endoscopic intervention (OR, 2.62; CI, 1.37-4.95) and decreased need for transfusion (OR, 0.65; CI, 0.49-0.87). On multivariable analysis adjusting for timing of colonoscopy, age, and propensity score for early colonoscopy, early colonoscopy was not associated with a decrease in 30-day mortality (OR, 1.37; CI, 0.50-3.79).

Early colonoscopy does not affect 30-day mortality but may allow for earlier endoscopic intervention and decreased transfusion need.