Background & Aims: The aim of this study was to use a large national endosc
opic database to determine why routine endoscopy is performed in diverse pr
actice settings. Methods: A computerized endoscopic report generator was de
veloped and disseminated to gastrointestinal (GI) specialists in diverse pr
actice settings. after reports were generated, a data file was transmitted
electronically to a central databank, where data were merged from multiple
sites for analysis. Results: From April 1, 1997, to October 28, 1998, 276 p
hysicians in 31 practice sites in 21 states provided 18,444 esophagogastrod
uodenoscopy (EGD) reports, 20,748 colonoscopy reports, and 9767 flexible si
gmoidoscopy reports to the central databank, EGD was most commonly performe
d to evaluate dyspepsia and/or abdominal pain (23.7%), dysphagia (20%), sym
ptoms of gastroesophageal reflux without dysphagia (17%), and suspected upp
er GI bleeding (16.3%). Colonoscopy was most often performed for surveillan
ce of prior neoplasia(24%) and evaluation of hematochezia (19%) or positive
fecal occult blood test (15%), Flexible sigmoidoscopy was most commonly pe
rformed for routine screening (40%) and evaluation of hematochezia (22%). T
here were significant differences between academic and nonacademic sites, C
onclusions: The endoscopic database can be an important resource for future
research in endoscopy by documenting current practice patterns and changes
in practice over time.