Study objectives: --To determine the prevalence of infectious agents in patients with human immunodeficiency virus infection and odynophagia or dysphagia; the utility of endoscopic, histologic, cytologic, and virologic testing for the diagnosis of esophagitis; and the yield of blind brushings of the esophagus in this setting.
Design: --Prospective clinical case study.
Setting: --Urban county hospital.
Patients: --One hundred ten consecutive patients with esophageal symptoms and documented human immunodeficiency virus infection.
Intervention: --Blind brushing of the esophagus via orogastric tube followed by endoscopy with esophageal brushing for fungal stain, Papanicolau smear, and viral cultures and esophageal biopsies for histologic examination and viral culture.
Main results: --Seventy-two (65%) of the 110 patients had a total of 100 esophageal infections. Thirty-three (30%) had Candida alone, 22 (20%) had Candida and cytomegalovirus, two (1.8%) had Candida with cytomegalovirus and herpes simplex virus, seven (6%) had cytomegalovirus alone, six (5%) had herpes simplex virus alone, and two (1.8%) had both viruses. Fifty of 55 patients with plaques alone had Candida, and two (4%) had only viral infection. Of 19 patients with erosions or ulcers, 11 (58%) had a viral infection, two (11%) had Candida alone, and six (30%) had no etiologic agent identified. The sensitivity of endoscopic brushings (95%) was better than that of histologic examination (70%) in the diagnosis of Candida esophagitis. Likewise, viral cultures of brushings or biopsy specimens were more sensitive (67%) than histologic examination (35%) for viral esophagitis. Blind brushing of the esophagus had a sensitivity and specificity for infectious esophagitis of 84% and 75%, respectively. Oral thrush had a sensitivity of 53% and a positive predictive value of 77% for Candida esophagitis.