A study of the safety of current gastrointestinal endoscopy (EGD)

Previous attempts at assessing the safety of upper gastrointestinal endoscopy have been hampered by incomplete data collection. We aimed to assess the 30-day mortality associated with esophagogastroduodenoscopy (EGD) and assess the important risk factors.

A retrospective cohort study was conducted of patients who underwent endoscopy at Ninewells Hospital in Dundee between 1 June 2000 and 31 May 2003. A total of 11 501 EGDs were performed in 8926 patients. These patients were record-linked to the death registry and the database of hospital admissions in order to calculate the all-cause 30-day mortality. An expert panel judged whether EGD had caused or contributed to the deaths. Logistic regression analysis was performed on outcomes of all-cause and EGD-contributed mortality.

The median age of the patients was 62 years (interquartile range 48 - 74 years), 54 % were women, and 94 % of procedures were diagnostic. A total of 395 patients died within 30 days (all-cause 30-day mortality rate 4.4 %). One patient death was caused directly by the EGD (procedure-caused mortality rate 1 in 9000). EGD was judged to have contributed to patient deaths at a rate of 1 in 182, based on majority agreement of experts: some factors associated with these deaths were percutaneous endoscopic gastrostomy insertion (odds ratio [OR] 18.39, 95 % confidence interval [CI] 5.71 - 59.22), melena or hematemesis indications (OR 9.01, 95 % CI 3.53 - 22.99), and esophageal varices (OR 6.28, 95 % CI 1.54 - 25.60).

A causal death rate of 1 in 9000 suggests that EGD is very safe. However, certain patient groups have an increased mortality, and the risks and benefits of EGD should be carefully evaluated in each patient.