The sensitivity of noninvasive stool microscopy and endoscopic biopsies from the upper and lower gastrointestinal tract in the diagnosis of Cryptosporidium parvum in patients with AIDS is not known. We evaluated 30 severely immunocompromised patients with AIDS and diarrhea caused by C. parvum. C. parvum was diagnosed by either stool microscopy, endoscopic biopsy, or both. Patients submitted a mean (+/-SEM) of 3.3 +/- 0.3 stool samples, each microscopically evaluated for ova and parasites. Upper and lower endoscopy were performed in all patients and endoscopic biopsies were taken throughout the gastrointestinal tract. Diarrhea had been present for a mean of 13.5 +/- 2.3 months and mean daily stool weight was 1224 +/- 127 g. Overall, individual stool samples were insensitive, as only 53% demonstrated C. parvum. When multiple stool samples were considered for each patient, 73% of subjects demonstrated C. parvum in at least one stool sample. The sensitivity of endoscopy with mucosal biopsy varied by anatomical location: stomach (11%), duodenum (53%), terminal ileum (91%), and colon (60%). The terminal ileum was significantly more likely than the duodenum to demonstrate C. parvum (P = 0.03). Thus, duodenal biopsies are much less sensitive than those from the terminal ileum in the diagnosis of C. parvum. In AIDS patients with diarrhea undergoing colonoscopy, intubation of the terminal ileum should be performed when feasible. Although individual stool samples are insensitive in detecting C. parvum, the diagnostic yield is improved by the collection of multiple samples.
