SYMPTOM-SPECIFIC USE OF UPPER GASTROINTESTINAL ENDOSCOPY IN HUMAN IMMUNODEFICIENCY VIRUS-INFECTED PATIENTS YIELDS HIGH DIVIDENDS

The yield of upper gastrointestinal endoscopy (esophagogastroduodenosc opy; EGD) in human immunodeficiency virus (HIV)-infected patients base d on presenting symptoms has not been well studied. We studied consecu tive patients with documented HIV infection undergoing EGD at a large inner-city hospital between August 1, 1990 and December 31, 1993; all had presenting symptoms and indications for EGD prospectively recorded at the time of EGD. All endoscopic abnormalities were routinely subje cted to biopsy, and extensive histopathological evaluation was perform ed. EGD was considered helpful when the findings stimulated specific t herapeutic intervention other than antifungal or antacid medications. The specific indications for EGD in 156 patients were as follows: esop hageal symptoms, 102 patients (65%); abdominal pain, 18 (12%); upper g astrointestinal bleeding, 25 (16%); refractory nausea and vomiting, 11 (7%). Overall, pathologic findings were identified in 116 patients (7 4%): in refractory esophageal symptoms, 82%; upper gastrointestinal bl eeding, 92%; abdominal pain, 39%; nausea and vomiting, 27%. EGD with b iopsy identified a specifically treatable opportunistic disorder other than Candida in 80 patients (51%), including idiopathic esophageal ul cer (22%) or viral esophagitis and/or duodenitis (29%). EGD was not he lpful in 22.3% of cases, those involving Candida (12.3%) and peptic ul cer disease (PUD)-related causes (10%). The mean CD4 count of patients with opportunistic pathologic findings (24/mm(3), n = 79) was signifi cantly lower than that of patients with PUD/gastroesophageal reflux di sease (GERD) (167/mm(3), n = 9) or negative EGDs (165/mm(3), n = 35). Overall, the results of EGD influenced patient management in 78% of ca ses. We conclude that selective symptom-specific use of EGD, particula rly in patients with esophageal symptoms refractory to antifungal ther apy or gastrointestinal bleeding, usually identifies specifically trea table abnormalities, whereas EGD is less useful for the evaluation of abdominal pain or nausea and vomiting.