- Email: [email protected]
Diagnostic yield and cost-effectiveness of endoscopy in chronic human immunodeficiency virus-related diarrhea
Diagnostic yield and cost-effectiveness of endoscopy in chronic human immunodeficiency virus-related diarrhea Edmund J. Bini, MD, Jonathan Cohen, MD New York, New York
Background: Endoscopy is commonly performed in patients with chronic human immunodeficiency virus (HIV)-related diarrhea after negative stool studies. The aim of this study was to determine the diagnostic yield and cost-effectiveness of endoscopy in this setting. Methods: Consecutive HIV-infected patients with chronic unexplained diarrhea who were referred for diagnostic endoscopy were identified. Patient charts, pathology reports, and endoscopy records were reviewed. Results: A total of 479 endoscopic procedures were performed in 307 patients. A pathogen was identified in 147 patients (47.9%); cytomegalovirus was the most frequent organism found (21.5%). The average cost of identifying a pathogen by endoscopy was $3822.94. Colonoscopy had a greater diagnostic yield than flexible sigmoidoscopy (38.7% vs. 22.4%, p = 0.009) and was more cost-effective. The yield of upper endoscopy was 29.6%. In patients with a CD4 count of less than 100/mm3, endoscopy had a higher diagnostic yield (62.8% vs. 8.3%, p < 0.0001) and a lower cost of identifying a pathogen ($2943.92 vs. $21,583.51) than in those with higher CD4 counts. Conclusions: Endoscopy frequently identifies a pathogen in HIV-related chronic diarrhea. Colonoscopy is the most cost-effective procedure. Endoscopic evaluation has a significantly higher diagnostic yield and is considerably more cost-effective in patients with a CD4 count of less than 100/mm3 than in those with higher CD4 counts. (Gastrointest Endosc 1998;48:354-61.) Chronic diarrhea is an important clinical problem in patients infected with HIV. As a result of immunosuppression, these patients are susceptible to gastrointestinal infections with bacteria, viruses, fungi, and protozoa.1-12 When stool examination fails to identify a pathogen, the decision to pursue a more aggressive diagnostic evaluation is controversial. Some clinicians advocate an intensive diagnostic evaluation consisting of stool studies and endoscopic biopsies of the small intestine and colon to identify all potential pathogens.1,10,11 Others have recommended a less aggressive approach consisting of stool studies followed by symptomatic treatment Received September 29, 1997. For revision January 30, 1998. Accepted April 24, 1998. From the Division of Gastroenterology, New York University Medical Center, Bellevue Hospital, and New York Veterans Administration Medical Center, New York, New York. Presented in part at the Presidential Plenary Poster Session during Digestive Disease Week, May 1997, Washington, D.C. (Gastrointest Endosc 1997;45:AB45). Reprint requests: Jonathan Cohen, MD, 232 East 30th St., New York, NY 10016. 37/1/91307 354
GASTROINTESTINAL ENDOSCOPY
with antidiarrheal agents if no pathogens are identified.13,14 We investigated the results of an endoscopic evaluation in a large group of HIV-infected patients with chronic diarrhea and a negative stool examination to delineate the diagnostic yield and cost-effectiveness of endoscopy in this setting. PATIENTS AND METHODS Patients All patients with documented HIV infection and chronic diarrhea who were referred for diagnostic endoscopy between October 1, 1993, and October 31, 1996, were identified for analysis in this study. Patients were referred from the primary care outpatient clinics, the virology clinics, the emergency department, and the inpatient wards. Data were collected by reviewing our endoscopy database, pathology reports, and patient medical records. The CD4 lymphocyte count obtained closest to the time of endoscopy was recorded. Chronic diarrhea was defined as at least three loose watery stools per day for a minimum of 1 month. Those patients in whom endoscopy was performed for indications other than chronic diarrhea, such as odynophagia, dysphagia, gastrointestinal bleeding, abdominal pain, or acute diarrhea, were excluded from the analysis. As a matter of VOLUME 48, NO. 4, 1998
Diagnostic yield and cost-effectiveness of endoscopy in HIV-related diarrhea
policy in the gastroenterology unit, fellows could only schedule an endoscopic evaluation of patients with HIVassociated diarrhea if they had documentation of at least one negative stool sample for Clostridium difficile, and two negative stool samples for bacterial culture, ova and parasites (including cryptosporidia), and acid-fast stain with culture for mycobacteria. Accordingly, these criteria were used to define a negative stool examination in this study. Stool examination Examination of stool for Cryptosporidium was performed using light microscopy after specimens were stained with modified Ziehl-Neelsen stain. Stools for ova and parasites were evaluated after staining with trichrome stain. Routine evaluation of stool specimens for microsporidia, Isospora, and Cyclospora were not performed at our hospital. Bacterial culture of stool specimens were examined for Salmonella, Shigella, Yersinia, and Campylobacter; routine examination of stool for Escherichia coli was not performed. Stool was also examined using Ziehl-Neelsen stain and mycobacterial culture. Enzyme immunoassay of stool for Clostridium difficile toxin A was performed by the pathology department. Immunoassays for Giardia and Cryptosporidium and polymerase chain reaction for cytomegalovirus were not performed. Endoscopic techniques All endoscopic procedures were performed by gastroenterology fellows with an attending physician present. The extent of the endoscopic evaluation was based on the presenting symptoms of each patient and was determined by the gastroenterologist evaluating each patient before any procedure. Although the choice of which endoscopic procedure(s) to perform was left to the attending physician and fellow, mucosal biopsy specimens were obtained during every examination regardless of the presence of endoscopic abnormalities. Consent was obtained from all patients before endoscopy. Upper endoscopy was performed under conscious sedation and biopsy specimens were obtained in all patients from the duodenum or proximal jejunum for light-microscopic examination. Mucosal biopsy specimens for electron microscopy and small bowel aspirates were obtained at the discretion of the endoscopist. Aspirates were submitted for ova and parasite examination; bacterial cultures were not performed. Additional biopsy specimens were obtained for mycobacterial staining and culture. All mucosal abnormalities were noted and multiple biopsy specimens were obtained. Preparation for flexible sigmoidoscopy was performed using bisacodyl tablets and Fleet enemas (C. B. Fleet Co., Inc., Lynchburg, Va.). The sigmoidoscope was advanced to 60 cm whenever possible and multiple biopsy specimens were obtained from the rectosigmoid colon for lightmicroscopic examination in all patients. In addition, biopsy specimens were obtained from all mucosal abnormalities, and additional specimens were taken for mycobacterial staining and culture. Routine biopsies for electron microscopy, viral cultures, or fungal cultures were not performed. VOLUME 48, NO. 4, 1998
E Bini, J Cohen
Table 1. Characteristics of HIV-infected patients undergoing endoscopy (n = 307) Male Age (yr)* CD4 count/mm3* CD4 count < 200 cells/mm3 CD4 count < 100 cells/mm3 Risk factors for HIV infection Intravenous drug use Homosexual Heterosexual Unknown Number of stools per day* Duration of diarrhea (weeks)* Empiric antidiarrheal therapy Weight loss of at least 5 kg
243 (79.2%) 38.1 ± 0.4 (19-63) 98.3 ± 6.9 (0-500) 261 (85.0%) 223 (72.6%) 175 (57.0%) 57 (18.6%) 49 (16.0%) 26 (8.4%) 8.0 ± 0.2 (3-20) 19.6 ± 0.8 (4-60) 94 (30.6%) 168 (54.7%)
*Data expressed as mean ± SEM (range).
Colonoscopy was performed under conscious sedation after the colon was prepared using bisacodyl tablets, Fleet enemas, and either Fleet phosphosoda or polyethylene glycol–balanced electrolyte solution. The colonoscope was advanced to the cecum and on withdrawal, multiple biopsy specimens were obtained from the proximal colon and rectosigmoid colon in all patients. The specimens were placed in separate formalin jars and labeled according to tissue location. Additional specimens for mycobacterial staining and culture were obtained from the proximal colon and rectosigmoid colon and placed in sterile containers. In addition to routine biopsy specimens obtained from the proximal and rectosigmoid colon, multiple biopsy specimens were obtained from all endoscopic abnormalities, such as ulcerations or colitis. Colonoscopic biopsy specimens were not submitted for viral cultures, fungal cultures, or electron microscopy. Pathology All biopsy specimens were evaluated by pathologists experienced in HIV-related gastrointestinal infections. Biopsy specimens for light microscopy were routinely fixed in formalin, embedded in paraffin, and stained with hematoxylin and eosin, Grocott’s methenamine silver stain for fungal organisms, and acid-fast stain for mycobacteria. Immunohistochemical stains to confirm the presence of cytomegalovirus or herpes simplex virus were performed when hematoxylin and eosin stain revealed possible cytopathic changes without definitive evidence of inclusion bodies. We do not routinely obtain viral cultures in an effort to avoid detecting clinically insignificant positive results.12 In situ DNA hybridization was not available at our hospital center. Mycobacterial cultures were performed on biopsy specimens submitted in sterile containers. Biopsy specimens obtained for electron microscopy were placed in glutaraldehyde fixative and embedded in plastic. Electron microscopy was performed using a transmission electron microscope. Small bowel aspirates were evaluated for ova and parasites using light microscopy. GASTROINTESTINAL ENDOSCOPY
355
E Bini, J Cohen
Diagnostic yield and cost-effectiveness of endoscopy in HIV-related diarrhea
Cost of endoscopy The cost of endoscopy was estimated by adding the physician fee under Medicaid reimbursement, the facility fee for endoscopy, and the pathology fee for the biopsy specimens. Medicaid reimbursement values were used in the cost analysis because this payment method is most applicable to our patient population. The total estimated cost for each endoscopic procedure with biopsies, including the physician fee, the facility fee, and the pathology fee, is as follows: flexible sigmoidoscopy $749.42; upper endoscopy $946.81; and colonoscopy $1079.64. The cost for additional procedures, such as examination of small bowel aspirates for ova and parasites ($8.91) and electron microscopy for small bowel biopsies ($610.00) were included in our analysis for those patients in whom these examinations were performed. The cost of identifying a pathogen was calculated by combining the cost of endoscopy and histologic examination, as well as the cost of electron microscopy and small bowel aspirates when performed, and dividing the total cost by the number of patients who had a pathogen identified. Data analysis For each patient analyzed in this study, the specific diagnostic tests performed, the cost of this evaluation, and the pathogens identified by each test were recorded. The diagnostic yield and the cost of identifying a pathogen by upper endoscopy, flexible sigmoidoscopy, and colonoscopy were calculated, stratifying the results according to CD4 lymphocyte count. For colonoscopy data, the yield of right and left-sided biopsies were compared. Univariate analysis was performed to compare groups using the Student’s t test and the Mann-Whitney test for continuous variables and the Fisher’s exact test for categorical variables. Multivariate analysis was performed using forward stepwise logistic regression to determine which demographic, laboratory, and clinical factors were independently associated with having a pathogen identified by endoscopy. A two-tailed p value of < 0.05 was considered statistically significant. All data are expressed as means ± standard error.
RESULTS During the study period, 307 consecutive HIVinfected patients with chronic diarrhea and a negative stool evaluation were identified. Clinical features of these patients are shown in Table 1. In this patient population, intravenous drug use was the most common risk factor for HIV infection. The mean CD4 count, which was obtained 32.6 ± 0.9 days before endoscopy, was 98.3 ± 6.9 cells/mm3. Two hundred twenty-three patients (72.6%) had a CD4 count of less than 100 cells/mm3, and 273 (88.9%) met the 1993 CDC criteria for acquired immunodeficiency syndrome (AIDS).15 Endoscopy was performed on an outpatient basis in 235 patients (76.5%); the remaining 72 (23.5%) were hospitalized at the time of endoscopy. No complications directly related to endoscopy were observed. 356
GASTROINTESTINAL ENDOSCOPY
Before endoscopy, the mean number of stool specimens examined were the following: ova and parasite examination, including Cryptosporidia (3.3 ± 0.08); bacterial culture (2.2 ± 0.05); mycobacterial stain and culture (2.2 ± 0.05); and C. difficile (1.4 ± 0.03). Stool was examined for fecal leukocytes in 271 patients (1.6 ± 0.05) and was positive in 97 (35.8%). A total of 479 endoscopic procedures were performed in 307 patients, including 203 upper gastrointestinal endoscopies, 191 colonoscopies, and 85 flexible sigmoidoscopies. Colonoscopy was performed to the cecum in 187 of the 191 patients (97.9%), and flexible sigmoidoscopy was complete in 79 of the 85 patients (92.9%). There was no significant difference in the number of upper endoscopies, flexible sigmoidoscopies, or colonoscopies between inpatients and outpatients or between patients with a CD4 count of less than 100 cells/mm3 and those with higher CD4 counts. A pathogen was identified by endoscopy and biopsy in 147 patients (47.9%). The specific pathogens identified by each endoscopic procedure and for the entire cohort of patients are shown in Table 2. Cytomegalovirus was the most common pathogen identified in this patient population (21.5%). Six patients had more than one pathogen identified by endoscopy and biopsy. In addition, 34 patients (11.1%) were found to have Kaposi’s sarcoma and 8 patients (2.6%) had lymphoma. Villus atrophy in the absence of an identifiable pathogen was found on small bowel biopsy in 34 of the 203 patients examined (16.7%). The diagnostic yield for each endoscopic procedure stratified by CD4 lymphocyte count is shown in Table 3. Of all endoscopic procedures performed, colonoscopy had the highest diagnostic yield (38.7%). The diagnostic yield of colonoscopy was significantly higher than the yield of flexible sigmoidoscopy (38.7% vs. 22.4%, p = 0.009). In the 191 patients in whom colonoscopy was performed, mucosal abnormalities were noted in 102 (53.4%) and were significantly more common in individuals who had a pathogen identified than in those in whom no organism was detected (73.0% vs. 41.0%, p < 0.0001). These mucosal abnormalities could be categorized into five patterns as follows: colitis (45.1%), single or multiple ulcers (23.5%), colitis with ulcers (17.6%), a colonic mass (9.8%), and colitis with a colonic mass (3.9%). Of the 74 patients in whom a pathogen was identified by colonoscopy and biopsy, 21 (28.4%) had a diagnosis established only on biopsies obtained from the right colon. The pathogens identified in the 21 patients in whom a diagnosis was made exclusively from the right colon were as follows: cytomegalovirus (17 patients); Mycobacterium tuberculosis (2 patients); Mycobacterium avium complex VOLUME 48, NO. 4, 1998
Diagnostic yield and cost-effectiveness of endoscopy in HIV-related diarrhea
E Bini, J Cohen
Table 2. Pathogens identified by endoscopy in chronic HIV-related diarrhea
Pathogen Cytomegalovirus Microsporidia Mycobacterium avium complex Cryptosporidium Clostridium difficile Giardia Mycobacterium tuberculosis Isospora Hymenolepis nana Strongyloides stercoralis
Upper endoscopy (n = 203)
Flexible sigmoidoscopy (n = 85)
Colonoscopy (n = 191)
Overall (n = 191)
10 26 5 12 0 6 0 3 0 0
6 0 5 2 5 0 1 0 0 0
53 0 13 0 1 0 3 0 3 2
66 26 23 14 6 6 4 3 3 2
Three individuals had cytomegalovirus detected on more than one examination, and six patients had more than one pathogen identified.
Table 3. Diagnostic yield of endoscopic procedures in HIVrelated chronic diarrhea All patients Upper endoscopy (n = 203) Flexible sigmoidoscopy (n = 85) Colonoscopy (n = 191) Overall endoscopic workup (n = 307)
CD4 < 100 CD4 ≥ 100 cells/mm3 cells/mm3
29.6% 22.4%
37.9%* 27.9%*
8.6% 0%
38.7% 47.9%
50.0%* 62.8%*
4.3% 8.3%
*p < 0.05 compared with CD4 ≥ 100 cells/mm3.
(1 patient); and both cytomegalovirus and Mycobacterium tuberculosis (1 patient). Fourteen of these 21 patients (66.7%) and 11 of the 17 individuals (64.7%) with right-sided cytomegalovirus infection had intravenous drug use as their risk factor for HIV infection. Right-sided cytomegalovirus infection was noted in 11 of the 109 intravenous drug users who had a colonoscopy (10.1%). In contrast, only 3 of 39 (7.7%) homosexual patients who had a colonoscopy were found to have isolated right-sided cytomegalovirus infection. Thirty-four percent of the patients with cytomegalovirus had disease localized only in the right colon. The endoscopic appearance of cytomegalovirus colitis could be categorized into three patterns as follows: colitis (36.5%), single or multiple ulcers (28.8%), and colitis with ulcers (26.9%); only four patients (7.7%) with cytomegalovirus infection had normal-appearing mucosa. When isolated in the right colon, cytomegalovirus most often appeared as ulcers without colitis (52.9%). In contrast, cytomegalovirus in the left colon most often appeared as colitis (37.5%) or colitis with ulcers (37.5%). VOLUME 48, NO. 4, 1998
Table 4. Characteristics of HIV-infected patients with and without a pathogen identified by endoscopy Characteristic Age (yr) CD4 count/mm3* Fever Abdominal pain Weight loss (kg)* Hematocrit (%)* Albumin (gm/dL)* Fecal leukocytes Number of stools per day Duration of diarrhea (weeks) Empiric antidiarrheal therapy
Pathogen (n = 147)
No pathogen (n = 160)
p Value
38.1 ± 0.7 39.7 ± 2.9 44.2% 55.1% 9.6 ± 0.5 33.9 ± 0.3 3.1 ± 0.04 51.9% 8.2 ± 0.3
37.8 ± 0.5 152.1 ± 11.3 36.3% 49.4% 4.3 ± 0.3 35.4 ± 0.4 4.0 ± 0.04 19.9% 7.8 ± 0.3
0.75 < 0.0001 0.16 0.36 < 0.0001 0.0009 < 0.0001 < 0.0001 0.36
18.2 ± 1.1
20.9 ± 1.1
0.08
32.7%
28.8%
0.54
*Independent predictor of having a pathogen identified by endoscopy using multivariate analysis.
The diagnostic yield of endoscopy was significantly higher in patients with a CD4 lymphocyte count of less than 100 cells/mm3 than in those with higher CD4 counts (62.8% vs. 8.3%, p < 0.0001). Only 7 patients with a CD4 count of 100 cells/mm3 or greater had a pathogen identified by endoscopy. The pathogens identified in these 7 patients were Cryptosporidium, Strongyloides stercoralis, and Giardia in 2 patients each, and Isospora in 1 patient. In none of the 46 patients with a CD4 count of 200 cells/mm3 or greater was a pathogen identified. Clinical features of patients in whom a pathogen was identified were compared with those in whom no pathogen was found on endoscopy. The results of this comparison is shown in Table 4. Patients in whom a pathogen was identified were significantly GASTROINTESTINAL ENDOSCOPY
357
E Bini, J Cohen
Diagnostic yield and cost-effectiveness of endoscopy in HIV-related diarrhea
Table 5. Cost of identifying a pathogen by endoscopy Cost per Cost per Cost per pathogen pathogen CD4 pathogen CD4 < 100/mm3 ≥ 100/mm3 Upper endoscopy Flexible sigmoidoscopy Colonoscopy Overall endoscopic workup
$4867.68 $3352.67
$3717.44 $2682.13
$17,520.28 NA*
$2786.64 $3822.94
$2159.28 $2934.92
$25,371.54 $21,583.51
*No patient with a CD4 count of ≥ 100/mm3 had a pathogen identified by sigmoidoscopy.
more likely to have a low CD4 lymphocyte count, weight loss, fecal leukocytes, and lower hematocrit and serum albumin levels. Of the 7 individuals with a CD4 count of 100 cells/mm3 or greater in whom a pathogen was identified, weight loss (mean 13.5 kg, range 4.2 to 22.9 kg) was present in all but one patient. Inpatients were significantly more likely to have a pathogen identified by endoscopy than outpatients (58.3% vs. 44.7%, p = 0.04). Homosexuals were no more likely to have a pathogen identified than were those with other risk factors for HIV infection (50.9% vs. 47.2%, p = 0.66). Multivariate analysis identified serum albumin (p < 0.0001), inpatient status (p < 0.0001), CD4 lymphocyte count (p = 0.0001), hematocrit level (p = 0.02), and weight loss (p = 0.04) as independent predictors for having a pathogen identified by endoscopy. In the 203 patients undergoing upper endoscopy, aspiration of fluid from the duodenum or jejunum was obtained for ova and parasite analysis in 185 individuals (91.1%) and mucosal biopsies were evaluated by electron microscopy in 161 (79.3%). Examination of small bowel aspirates yielded a diagnosis in 3 of the 185 patients (1.6%); all 3 individuals were found to have Giardia. However, 2 of these 3 patients also had Giardia identified by lightmicroscopic examination of small bowel biopsies. Electron microscopy identified a pathogen in 44 patients (27.3%): microsporidia (25 patients); Cryptosporidium (11 patients); Giardia (5 patients); and Isospora (3 patients). Of the 25 patients with microsporidia, 21 (84%) were identified as Enterocytozoon bieneusi and 4 (16%) were identified as Encephalitozoon intestinalis. All of the patients with Cryptosporidium, Giardia, and Isospora had the diagnosis made by light microscopy in addition to electron microscopy. Among the 25 patients with microsporidia identified on electron microscopy, 17 (68%) had the pathogen identified by light microscopy as well. The use of electron microscopy 358
GASTROINTESTINAL ENDOSCOPY
in these 161 patients therefore resulted in the diagnosis of 8 patients with microsporidia (all Enterocytozoon bieneusi) that would have been missed if light microscopy had been used alone. Cost analysis The average cost of an endoscopic evaluation in this study was $1830.53 per patient. The cost of identifying a pathogen for each endoscopic procedure stratified by CD4 lymphocyte count is shown in Table 5. The average cost of identifying a pathogen by endoscopy in this patient population was $3822.94. Evaluation of small-bowel biopsy specimens by electron microscopy and small-bowel aspirates for ova and parasites resulted in an incremental cost of $12,276.25 and $1648.35, respectively, for each additional pathogen identified. DISCUSSION In HIV-infected patients with chronic unexplained diarrhea, data assessing the diagnostic yield of upper and lower endoscopy are limited. Only two previous studies have addressed this issue in patients with negative stool studies before endoscopy.16,17 Connolly et al.16 evaluated 33 HIV-infected patients with seemingly pathogen-negative diarrhea. In their study, patients underwent upper endoscopy and proctoscopy with biopsy, Schilling and D-xylose testing, microbiologic examination of stool, small-bowel barium examination, and barium enema. They identified a pathogen in 12 patients (36%) and found cryptosporidia to be the most common organism. In another study, Wilcox et al.17 prospectively evaluated 48 HIV-infected patients with chronic unexplained diarrhea. In their study, patients were evaluated by upper endoscopy and colonoscopy with biopsy. A potential cause of diarrhea was identified in 21 patients (44%). In contrast to Connolly16 and Wilcox17 and their associates found cytomegalovirus to be the most common pathogen. In our study of 307 consecutive HIVinfected patients with chronic unexplained diarrhea, cytomegalovirus was also the most common pathogen isolated. In addition, we found the diagnostic yield of endoscopy (47.9%) to be similar to that reported by Wilcox et al.17 In this study, upper endoscopy identified a pathogen in 60 of 203 patients (29.6%). Microsporidia was the most common pathogen found (26 patients). In a retrospective study of 57 HIV-infected patients with chronic diarrhea and negative stool studies, Bown et al.18 reported a similar diagnostic yield for upper endoscopy (26%). In their study, no patient had a pathogen identified by examination of duodenal aspirate for ova and paraVOLUME 48, NO. 4, 1998
Diagnostic yield and cost-effectiveness of endoscopy in HIV-related diarrhea
sites. We also found that small-bowel aspirates have a low diagnostic yield (1.6%). Bown et al.18 identified a pathogen on electron microscopy in 6 of 12 patients (50%). In contrast, we found the diagnostic yield of electron microscopy in 161 patients to be much lower (27.3%). In addition, the majority of patients in whom a pathogen was identified by electron microscopy also had the diagnosis established by light microscopy. All of the pathogens that were identified only by electron microscopy were found to be microsporidia (Enterocytozoon bieneusi), an organism for which no effective therapy is yet available.4,19-21 The presence of villous atrophy in our series (16.7%) has been well described.12 However, the etiology and clinical significance of this entity remains poorly defined. The high cost of identifying a pathogen by upper endoscopy is a result of the low diagnostic yield of this procedure and the use of electron microscopy in our patient population. If electron microscopy was deleted from our evaluation, the cost of identifying a pathogen by upper endoscopy would decrease from $4867.68 to $3727.90. Although elimination of electron microscopy results in substantial savings without apparent adverse clinical consequences, the cost of detecting a pathogen by upper endoscopy is still higher than that of flexible sigmoidoscopy and colonoscopy. Furthermore, some of the patients with pathogens detected by upper endoscopy may have had a diagnosis established by stool studies if immunofluorescent assays for Giardia and Cryptosporidium, enzyme immunoassay for Giardia antigen, or fluorescent staining of stool specimens for microsporidia were done before an endoscopic evaluation. The cost-benefit of performing these specialized tests routinely, with the potential for reduced endoscopic procedures, remains to be determined. In HIV-infected patients with chronic unexplained diarrhea undergoing lower endoscopy, controversy exists concerning the best diagnostic approach. Some clinicians recommend flexible sigmoidoscopy,12,17,19,22-24 whereas others suggest performing colonoscopy as the initial endoscopic procedure.10,25,26 In this study, the diagnostic yield of colonoscopy was significantly higher than the yield of flexible sigmoidoscopy (38.7% vs. 22.4%). In addition, 28% of the pathogens identified by colonoscopy and biopsy were found only in the right colon. Of note, 95% of the pathogens that were identified only in the right colon were organisms for which effective therapy is currently available.1,27-32 Right-sided cytomegalovirus infection of the colon has not been well described as a cause of chronic diarrhea. However, treatment of the patients with right-sided VOLUME 48, NO. 4, 1998
E Bini, J Cohen
cytomegalovirus colitis resulted in complete resolution of diarrhea in 66.7%. This therapeutic response strongly suggests that cytomegalovirus was the cause of chronic diarrhea in these patients. Cytomegalovirus accounted for 81% of pathogens located exclusively in the right colon. Conversely, in 34% of the patients with cytomegalovirus a diagnosis was established only on biopsy specimens taken from the cecum and ascending colon. These data are consistent with the findings of Dieterich and Rahmin26 who reported that 39% of cytomegalovirus colitis was found only in the cecum. This contrasts with the results of Wilcox et al.,17 who found only 1 of 11 patients (9.1%) with cytomegalovirus colitis whose diagnosis could not be made with left-sided biopsies alone.17 This discrepancy may be explained by differences in the patient population evaluated, given the higher proportion of patients in this study who were intravenous drug users. The finding of right-sided cytomegalovirus infection in 3 of 39 homosexual patients (7.7%) undergoing colonoscopy is similar to that of Wilcox et al.,17 who evaluated primarily homosexual patients and found right-sided cytomegalovirus infection in 1 of 48 individuals examined (2.1%). Although the cost of each colonoscopy was higher than the cost of flexible sigmoidoscopy, the greater diagnostic yield of colonoscopy resulted in a considerably lower cost of identifying a pathogen ($2786.64 vs. $3352.67). Given the 22.4% yield of flexible sigmoidoscopy and assuming from our data that 11% of patients with negative left-sided biopsies on sigmoidoscopy would have a pathogen identified on right-sided biopsies during subsequent colonoscopy, the cost per diagnosis of such a two-step strategy would be $4822.85. Performing colonoscopy only after negative flexible sigmoidoscopy would be 47% more costly than performing colonoscopy as the initial test in all patients. HIV-infected patients who are severely immunocompromised (CD4 count of less than 100 cells/mm3) are at increased risk for opportunistic infections, including cytomegalovirus, Mycobacterium avium complex, microsporidia, and cryptosporidia.1,12,25 The diagnostic yield of endoscopic investigation of chronic HIV-related diarrhea has been shown to correlate with the degree of immunosuppression.12,17 In this study, patients with a CD4 lymphocyte count of less than 100 cells/mm3 were significantly more likely to have a pathogen identified by endoscopy than those with higher CD4 counts. Because the extent of the endoscopic evaluation was left to the discretion of the endoscopist, there was potential for selection bias. However, there were no significant differences in the number and type of endoscopic GASTROINTESTINAL ENDOSCOPY
359
E Bini, J Cohen
Diagnostic yield and cost-effectiveness of endoscopy in HIV-related diarrhea
procedures performed between sicker patients and those who were less ill (inpatients vs. outpatients and CD4 count < 100 cells/mm3 vs. CD4 count ≥ 100 cells/mm3). Only 7 of 84 patients (8.3%) with a CD4 count of 100 cells/mm3 or greater had a pathogen identified by endoscopy. Weight loss was present in all but one of these 7 patients. In their prospective study of 48 patients, Wilcox et al.17 identified weight loss and a shorter duration of diarrhea as independent predictors for establishing a diagnosis by endoscopy and biopsy using multivariate analysis. Other investigators have also found weight loss to be an important clinical parameter.16 Neither the frequency of bowel movements nor the duration of diarrhea before endoscopy were significant predictors of identifying a pathogen in this study. Several diagnostic strategies have been recommended for the evaluation of chronic diarrhea in HIV-infected patients. Most investigators agree that stool studies should be the first diagnostic test because they are simple to perform, inexpensive, and noninvasive and have a high diagnostic yield.12,33,34 Many authors suggest at least three stool samples before proceeding with an endoscopic evaluation. It is possible that requiring three instead of two negative stool specimens would have obviated the need for endoscopy in some of our patients. However, there have been no studies demonstrating the marginal diagnostic yield of the third stool specimen in this patient population. When stool studies fail to identify a pathogen, endoscopic evaluation is frequently recommended.1,12 To date, the cost-effectiveness of upper endoscopy, flexible sigmoidoscopy, and colonoscopy in this setting has not been well defined. A medical decision analysis published in 1990 evaluated several diagnostic strategies for the evaluation of AIDS-related diarrhea in 1000 hypothetical patients. After taking cost into account, they recommended a single stool culture as the most cost-effective approach, with endoscopy reserved only for patients who fail symptomatic treatment.14 However, the outcome assumptions used were based on limited successful therapies available at the time and on information about diagnostic yield and pathogens identified drawn from studies involving small groups of patients. Determining the most cost-effective approach requires better data about the diagnostic yield of various endoscopic procedures, the frequency of finding pathogens for which effective treatment is available, and the comparative direct and indirect costs of empiric symptomatic versus pathogen-based therapy. The average cost of detecting a pathogen in this study ($3822.94) is relatively low compared 360
GASTROINTESTINAL ENDOSCOPY
with the cost per diagnosis of other entities, such as an adenomatous polyp detected by screening flexible sigmoidoscopy ($8766.00).35 Justifying this cost depends on further outcome data using current available therapies and including both cost and quality of life assessments. The critical question of the effectiveness of pathogen specific therapy in our study population is the subject of ongoing investigation. This study provides a clearer understanding of the diagnostic yield of various endoscopic procedures in a large group of patients with chronic HIVrelated diarrhea and the cost incurred in making these diagnoses. The data demonstrate a much higher diagnostic yield and more cost-effective evaluation in individuals with low CD4 lymphocyte counts, suggesting that this subset of patients is most likely to benefit from an endoscopic evaluation rather than a symptomatic treatment approach. The benefit of small-bowel aspirates and particularly electron microscopy appears to be minimal. If clinicians elect to use endoscopy to pursue specific treatable pathogens, colonoscopy with biopsies from the right and left colon is the most cost-effective examination with the highest diagnostic yield. In patients without weight loss and with a CD4 lymphocyte count of 100 cells/mm3 or greater, endoscopy is unlikely to identify an etiology. These patients may be managed best by empiric treatment with antidiarrheal medications, reserving endoscopy for those in whom diarrhea persists despite antidiarrheal therapy. REFERENCES 1. Smith PD, Quinn TC, Strober W, Janoff EN, Masur H. NIH Conference: gastrointestinal infections in AIDS. Ann Intern Med 1992;116:63-77. 2. Antony MA, Brandt LJ, Klein RS, Bernstein LH. Infectious diarrhea in patients with AIDS. Dig Dis Sci 1988;33:1141-6. 3. Gazzard BG. Diarrhea in human immunodeficiency virus antibody-positive patients. Semin Liver Dis 1992;12:154-66. 4. Goodgame RW. Understanding intestinal spore-forming protozoa: cryptosporidia, microsporidia, isospora, and cyclospora. Ann Intern Med 1996;124:429-41. 5. Rene E, Marche C, Regnier B, Saimot AG, Vilde JL, Perrone C, et al. Intestinal infections in patients with acquired immunodeficiency syndrome: a prospective study in 132 patients. Dig Dis Sci 1989;34:773-80. 6. Janoff EN, Orenstein JM, Manischewitz JF, Smith PD. Adenovirus colitis in the acquired immunodeficiency syndrome. Gastroenterology 1991;100:976-9. 7. Grohmann GS, Glass RI, Pereira HG, Monroe SS, Hightower AW, Weber R, et al. Enteric viruses and diarrhea in HIVinfected patients. N Engl J Med 1993;329:14-20. 8. Dancygier H. AIDS and gastrointestinal endoscopy. Endoscopy 1992;24:169-75. 9. Smith PD. Infectious diarrhea in patients with AIDS. Gastroenterol Clin North Am 1993;22:535-48. 10. Simon D, Brandt LJ. Diarrhea in patients with the acquired VOLUME 48, NO. 4, 1998
Diagnostic yield and cost-effectiveness of endoscopy in HIV-related diarrhea
11.
12.
13. 14.
15.
16.
17.
18.
19.
20.
21.
22. 23.
immunodeficiency syndrome. Gastroenterology 1993;105: 1238-42. Smith PD, Lane HC, Gill VJ, Manischewitz JF, Quinnan GV, Fauci AS, et al. Intestinal infections in patients with the acquired immune deficiency syndrome: etiology and response to therapy. Ann Intern Med 1988;108:328-33. Wilcox CM, Rabeneck L, Friedman S. Malnutrition and cachexia, chronic diarrhea, and hepatobiliary disease in patients with human immunodeficiency virus infection. Gastroenterology 1996;111:1724-52. Brady CE. AIDS-related diarrhea: What’s a gastroenterologist to do? Am J Gastroenterol 1991;86:1685-6. Johanson JF, Sonnenberg A. Efficient management of diarrhea in the acquired immune deficiency syndrome (AIDS): a medical decision analysis. Ann Intern Med 1990;112:942-8. Center for Disease Control and Prevention. 1993 revised classification system for HIV infection and expanded surveillance case definition for AIDS among adolescents and adults. MMWR 1992;41:1-19. Connolly GM, Forbes A, Gazzard BG. Investigation of seemingly pathogen-negative diarrhoea in patients infected with HIV-1. Gut 1990;31:886-9. Wilcox CM, Schwartz DA, Cotsonis G, Thompson SE. Chronic unexplained diarrhea in human immunodeficiency virus infection: determination of the best diagnostic approach. Gastroenterology 1996;110:30-7. Bown JW, Savides TJ, Mathews C, Isenberg J, Behling C, Lyche KD. Diagnostic yield of duodenal biopsy and aspirate in AIDS-associated diarrhea. Am J Gastroenterol 1996;91:228992. Blanshard C, Francis N, Gazzard BG. Investigation of chronic diarrhoea in acquired immunodeficiency syndrome. A prospective study of 155 patients. Gut 1996;39:824-32. Rabeneck L. Diagnostic workup strategies for patients with HIV-related chronic diarrhea: What is the end result? J Clin Gastroenterol 1993;16:245-50. Johanson JF. To scope or not to scope: the role of endoscopy in the evaluation of AIDS-related diarrhea. Am J Gastroenterol 1996;91:2261-2. Sharpstone D, Gazzard B. Gastrointestinal manifestations of HIV infection. Lancet 1996;348:379-83. Connolly GM, Forbes A, Gleeson JA, Gazzard BG. The value
24.
25. 26.
27. 28.
29.
30. 31.
32.
33.
34.
35.
E Bini, J Cohen
of barium enema and colonoscopy in patients infected with HIV. AIDS 1990;4:687-9. Connolly GM, Shanson D, Hawkins DA, Harcourt Webster JN, Gazzard BG. Non-cryptosporidial diarrhea in human immunodeficiency virus (HIV) infected patients. Gut 1989;30:195-200. Mayer HB, Wanke CA. Diagnostic strategies in HIV-infected patients with diarrhea. AIDS 1994;8:1639-48. Dieterich DT, Rahmin M. Cytomegalovirus colitis in AIDS: presentation in 44 patients and a review of the literature. J Acquir Immune Defic Syndr 1991;4(Suppl 1):S29-35. Goodgame RW. Gastrointestinal cytomegalovirus disease. Ann Intern Med 1993;119:924-35. Dieterich DT, Kotler DP, Busch DF, Crumpacker C, Du Mond C, Dearmand B, et al. Ganciclovir treatment of cytomegalovirus colitis in AIDS: a randomized, double-blind, placebo-controlled multicenter study. J Infec Dis 1993;167:278-82. Barnes PF, Bloch AB, Davidson PT, Snider DE. Tuberculosis in patients with human immunodeficiency virus infection. N Engl J Med 1991;324:1644-50. Iseman MD. Treatment of multidrug-resistant tuberculosis. N Engl J Med 1993;329:784-91. Simon D, Weiss LM, Brandt LJ. Treatment options for AIDSrelated esophageal and diarrheal disorders. Am J Gastroenterol 1992;87:274-81. Nelson MR, Connolly GM, Hawkins DA, Gazzard BG. Foscarnet in the treatment of cytomegalovirus infection of the esophagus and colon in patients with the acquired immune deficiency syndrome. Am J Gastroenterol 1991;86:876-81. Crotty B, Smallwood RA. Investigating diarrhea in patients with acquired immunodeficiency syndrome. Gastroenterology 1996;110:296-8. Blackman E, Binder S, Gaultier C, Benveniste R, Cecilio M. Cryptosporidiosis in HIV-infected patients: diagnostic sensitivity of stool examination, based on number of specimens submitted. Am J Gastroenterol 1997;92:451-3. Lieberman D. Cost-effectiveness of colon cancer screening. Am J Gastroenterol 1991;86:1789-94.
Availability of Journal back issues As a service to our subscribers, copies of back issues of Gastrointestinal Endoscopy for the preceding 5 years are maintained and are available for purchase from Mosby until inventory is depleted at a cost of $14.00 per issue. The following quantity discounts are available: 25% off on quantities of 12 to 23, and one third off on quantities of 24 or more. Please write to Mosby, Inc., Subscription Services, 11830 Westline Industrial Dr., St. Louis, MO 63146-3318, or call 800-453-4351 or 314-453-4351 for information on availability of particular issues. If unavailable from the publisher, photocopies of complete issues may be purchased from UMI, 300 N. Zeeb Rd., Ann Arbor, MI 48106; 313-761-4700.
VOLUME 48, NO. 4, 1998
GASTROINTESTINAL ENDOSCOPY
361
Background: Endoscopy is commonly performed in patients with chronic human immunodeficiency virus (HIV)-related diarrhea after negative stool studies. The aim of this study was to determine the diagnostic yield and cost-effectiveness of endoscopy in this setting. Methods: Consecutive HIV-infected patients with chronic unexplained diarrhea who were referred for diagnostic endoscopy were identified. Patient charts, pathology reports, and endoscopy records were reviewed. Results: A total of 479 endoscopic procedures were performed in 307 patients. A pathogen was identified in 147 patients (47.9%); cytomegalovirus was the most frequent organism found (21.5%). The average cost of identifying a pathogen by endoscopy was $3822.94. Colonoscopy had a greater diagnostic yield than flexible sigmoidoscopy (38.7% vs. 22.4%, p = 0.009) and was more cost-effective. The yield of upper endoscopy was 29.6%. In patients with a CD4 count of less than 100/mm3, endoscopy had a higher diagnostic yield (62.8% vs. 8.3%, p < 0.0001) and a lower cost of identifying a pathogen ($2943.92 vs. $21,583.51) than in those with higher CD4 counts. Conclusions: Endoscopy frequently identifies a pathogen in HIV-related chronic diarrhea. Colonoscopy is the most cost-effective procedure. Endoscopic evaluation has a significantly higher diagnostic yield and is considerably more cost-effective in patients with a CD4 count of less than 100/mm3 than in those with higher CD4 counts. (Gastrointest Endosc 1998;48:354-61.) Chronic diarrhea is an important clinical problem in patients infected with HIV. As a result of immunosuppression, these patients are susceptible to gastrointestinal infections with bacteria, viruses, fungi, and protozoa.1-12 When stool examination fails to identify a pathogen, the decision to pursue a more aggressive diagnostic evaluation is controversial. Some clinicians advocate an intensive diagnostic evaluation consisting of stool studies and endoscopic biopsies of the small intestine and colon to identify all potential pathogens.1,10,11 Others have recommended a less aggressive approach consisting of stool studies followed by symptomatic treatment Received September 29, 1997. For revision January 30, 1998. Accepted April 24, 1998. From the Division of Gastroenterology, New York University Medical Center, Bellevue Hospital, and New York Veterans Administration Medical Center, New York, New York. Presented in part at the Presidential Plenary Poster Session during Digestive Disease Week, May 1997, Washington, D.C. (Gastrointest Endosc 1997;45:AB45). Reprint requests: Jonathan Cohen, MD, 232 East 30th St., New York, NY 10016. 37/1/91307 354
GASTROINTESTINAL ENDOSCOPY
with antidiarrheal agents if no pathogens are identified.13,14 We investigated the results of an endoscopic evaluation in a large group of HIV-infected patients with chronic diarrhea and a negative stool examination to delineate the diagnostic yield and cost-effectiveness of endoscopy in this setting. PATIENTS AND METHODS Patients All patients with documented HIV infection and chronic diarrhea who were referred for diagnostic endoscopy between October 1, 1993, and October 31, 1996, were identified for analysis in this study. Patients were referred from the primary care outpatient clinics, the virology clinics, the emergency department, and the inpatient wards. Data were collected by reviewing our endoscopy database, pathology reports, and patient medical records. The CD4 lymphocyte count obtained closest to the time of endoscopy was recorded. Chronic diarrhea was defined as at least three loose watery stools per day for a minimum of 1 month. Those patients in whom endoscopy was performed for indications other than chronic diarrhea, such as odynophagia, dysphagia, gastrointestinal bleeding, abdominal pain, or acute diarrhea, were excluded from the analysis. As a matter of VOLUME 48, NO. 4, 1998
Diagnostic yield and cost-effectiveness of endoscopy in HIV-related diarrhea
policy in the gastroenterology unit, fellows could only schedule an endoscopic evaluation of patients with HIVassociated diarrhea if they had documentation of at least one negative stool sample for Clostridium difficile, and two negative stool samples for bacterial culture, ova and parasites (including cryptosporidia), and acid-fast stain with culture for mycobacteria. Accordingly, these criteria were used to define a negative stool examination in this study. Stool examination Examination of stool for Cryptosporidium was performed using light microscopy after specimens were stained with modified Ziehl-Neelsen stain. Stools for ova and parasites were evaluated after staining with trichrome stain. Routine evaluation of stool specimens for microsporidia, Isospora, and Cyclospora were not performed at our hospital. Bacterial culture of stool specimens were examined for Salmonella, Shigella, Yersinia, and Campylobacter; routine examination of stool for Escherichia coli was not performed. Stool was also examined using Ziehl-Neelsen stain and mycobacterial culture. Enzyme immunoassay of stool for Clostridium difficile toxin A was performed by the pathology department. Immunoassays for Giardia and Cryptosporidium and polymerase chain reaction for cytomegalovirus were not performed. Endoscopic techniques All endoscopic procedures were performed by gastroenterology fellows with an attending physician present. The extent of the endoscopic evaluation was based on the presenting symptoms of each patient and was determined by the gastroenterologist evaluating each patient before any procedure. Although the choice of which endoscopic procedure(s) to perform was left to the attending physician and fellow, mucosal biopsy specimens were obtained during every examination regardless of the presence of endoscopic abnormalities. Consent was obtained from all patients before endoscopy. Upper endoscopy was performed under conscious sedation and biopsy specimens were obtained in all patients from the duodenum or proximal jejunum for light-microscopic examination. Mucosal biopsy specimens for electron microscopy and small bowel aspirates were obtained at the discretion of the endoscopist. Aspirates were submitted for ova and parasite examination; bacterial cultures were not performed. Additional biopsy specimens were obtained for mycobacterial staining and culture. All mucosal abnormalities were noted and multiple biopsy specimens were obtained. Preparation for flexible sigmoidoscopy was performed using bisacodyl tablets and Fleet enemas (C. B. Fleet Co., Inc., Lynchburg, Va.). The sigmoidoscope was advanced to 60 cm whenever possible and multiple biopsy specimens were obtained from the rectosigmoid colon for lightmicroscopic examination in all patients. In addition, biopsy specimens were obtained from all mucosal abnormalities, and additional specimens were taken for mycobacterial staining and culture. Routine biopsies for electron microscopy, viral cultures, or fungal cultures were not performed. VOLUME 48, NO. 4, 1998
E Bini, J Cohen
Table 1. Characteristics of HIV-infected patients undergoing endoscopy (n = 307) Male Age (yr)* CD4 count/mm3* CD4 count < 200 cells/mm3 CD4 count < 100 cells/mm3 Risk factors for HIV infection Intravenous drug use Homosexual Heterosexual Unknown Number of stools per day* Duration of diarrhea (weeks)* Empiric antidiarrheal therapy Weight loss of at least 5 kg
243 (79.2%) 38.1 ± 0.4 (19-63) 98.3 ± 6.9 (0-500) 261 (85.0%) 223 (72.6%) 175 (57.0%) 57 (18.6%) 49 (16.0%) 26 (8.4%) 8.0 ± 0.2 (3-20) 19.6 ± 0.8 (4-60) 94 (30.6%) 168 (54.7%)
*Data expressed as mean ± SEM (range).
Colonoscopy was performed under conscious sedation after the colon was prepared using bisacodyl tablets, Fleet enemas, and either Fleet phosphosoda or polyethylene glycol–balanced electrolyte solution. The colonoscope was advanced to the cecum and on withdrawal, multiple biopsy specimens were obtained from the proximal colon and rectosigmoid colon in all patients. The specimens were placed in separate formalin jars and labeled according to tissue location. Additional specimens for mycobacterial staining and culture were obtained from the proximal colon and rectosigmoid colon and placed in sterile containers. In addition to routine biopsy specimens obtained from the proximal and rectosigmoid colon, multiple biopsy specimens were obtained from all endoscopic abnormalities, such as ulcerations or colitis. Colonoscopic biopsy specimens were not submitted for viral cultures, fungal cultures, or electron microscopy. Pathology All biopsy specimens were evaluated by pathologists experienced in HIV-related gastrointestinal infections. Biopsy specimens for light microscopy were routinely fixed in formalin, embedded in paraffin, and stained with hematoxylin and eosin, Grocott’s methenamine silver stain for fungal organisms, and acid-fast stain for mycobacteria. Immunohistochemical stains to confirm the presence of cytomegalovirus or herpes simplex virus were performed when hematoxylin and eosin stain revealed possible cytopathic changes without definitive evidence of inclusion bodies. We do not routinely obtain viral cultures in an effort to avoid detecting clinically insignificant positive results.12 In situ DNA hybridization was not available at our hospital center. Mycobacterial cultures were performed on biopsy specimens submitted in sterile containers. Biopsy specimens obtained for electron microscopy were placed in glutaraldehyde fixative and embedded in plastic. Electron microscopy was performed using a transmission electron microscope. Small bowel aspirates were evaluated for ova and parasites using light microscopy. GASTROINTESTINAL ENDOSCOPY
355
E Bini, J Cohen
Diagnostic yield and cost-effectiveness of endoscopy in HIV-related diarrhea
Cost of endoscopy The cost of endoscopy was estimated by adding the physician fee under Medicaid reimbursement, the facility fee for endoscopy, and the pathology fee for the biopsy specimens. Medicaid reimbursement values were used in the cost analysis because this payment method is most applicable to our patient population. The total estimated cost for each endoscopic procedure with biopsies, including the physician fee, the facility fee, and the pathology fee, is as follows: flexible sigmoidoscopy $749.42; upper endoscopy $946.81; and colonoscopy $1079.64. The cost for additional procedures, such as examination of small bowel aspirates for ova and parasites ($8.91) and electron microscopy for small bowel biopsies ($610.00) were included in our analysis for those patients in whom these examinations were performed. The cost of identifying a pathogen was calculated by combining the cost of endoscopy and histologic examination, as well as the cost of electron microscopy and small bowel aspirates when performed, and dividing the total cost by the number of patients who had a pathogen identified. Data analysis For each patient analyzed in this study, the specific diagnostic tests performed, the cost of this evaluation, and the pathogens identified by each test were recorded. The diagnostic yield and the cost of identifying a pathogen by upper endoscopy, flexible sigmoidoscopy, and colonoscopy were calculated, stratifying the results according to CD4 lymphocyte count. For colonoscopy data, the yield of right and left-sided biopsies were compared. Univariate analysis was performed to compare groups using the Student’s t test and the Mann-Whitney test for continuous variables and the Fisher’s exact test for categorical variables. Multivariate analysis was performed using forward stepwise logistic regression to determine which demographic, laboratory, and clinical factors were independently associated with having a pathogen identified by endoscopy. A two-tailed p value of < 0.05 was considered statistically significant. All data are expressed as means ± standard error.
RESULTS During the study period, 307 consecutive HIVinfected patients with chronic diarrhea and a negative stool evaluation were identified. Clinical features of these patients are shown in Table 1. In this patient population, intravenous drug use was the most common risk factor for HIV infection. The mean CD4 count, which was obtained 32.6 ± 0.9 days before endoscopy, was 98.3 ± 6.9 cells/mm3. Two hundred twenty-three patients (72.6%) had a CD4 count of less than 100 cells/mm3, and 273 (88.9%) met the 1993 CDC criteria for acquired immunodeficiency syndrome (AIDS).15 Endoscopy was performed on an outpatient basis in 235 patients (76.5%); the remaining 72 (23.5%) were hospitalized at the time of endoscopy. No complications directly related to endoscopy were observed. 356
GASTROINTESTINAL ENDOSCOPY
Before endoscopy, the mean number of stool specimens examined were the following: ova and parasite examination, including Cryptosporidia (3.3 ± 0.08); bacterial culture (2.2 ± 0.05); mycobacterial stain and culture (2.2 ± 0.05); and C. difficile (1.4 ± 0.03). Stool was examined for fecal leukocytes in 271 patients (1.6 ± 0.05) and was positive in 97 (35.8%). A total of 479 endoscopic procedures were performed in 307 patients, including 203 upper gastrointestinal endoscopies, 191 colonoscopies, and 85 flexible sigmoidoscopies. Colonoscopy was performed to the cecum in 187 of the 191 patients (97.9%), and flexible sigmoidoscopy was complete in 79 of the 85 patients (92.9%). There was no significant difference in the number of upper endoscopies, flexible sigmoidoscopies, or colonoscopies between inpatients and outpatients or between patients with a CD4 count of less than 100 cells/mm3 and those with higher CD4 counts. A pathogen was identified by endoscopy and biopsy in 147 patients (47.9%). The specific pathogens identified by each endoscopic procedure and for the entire cohort of patients are shown in Table 2. Cytomegalovirus was the most common pathogen identified in this patient population (21.5%). Six patients had more than one pathogen identified by endoscopy and biopsy. In addition, 34 patients (11.1%) were found to have Kaposi’s sarcoma and 8 patients (2.6%) had lymphoma. Villus atrophy in the absence of an identifiable pathogen was found on small bowel biopsy in 34 of the 203 patients examined (16.7%). The diagnostic yield for each endoscopic procedure stratified by CD4 lymphocyte count is shown in Table 3. Of all endoscopic procedures performed, colonoscopy had the highest diagnostic yield (38.7%). The diagnostic yield of colonoscopy was significantly higher than the yield of flexible sigmoidoscopy (38.7% vs. 22.4%, p = 0.009). In the 191 patients in whom colonoscopy was performed, mucosal abnormalities were noted in 102 (53.4%) and were significantly more common in individuals who had a pathogen identified than in those in whom no organism was detected (73.0% vs. 41.0%, p < 0.0001). These mucosal abnormalities could be categorized into five patterns as follows: colitis (45.1%), single or multiple ulcers (23.5%), colitis with ulcers (17.6%), a colonic mass (9.8%), and colitis with a colonic mass (3.9%). Of the 74 patients in whom a pathogen was identified by colonoscopy and biopsy, 21 (28.4%) had a diagnosis established only on biopsies obtained from the right colon. The pathogens identified in the 21 patients in whom a diagnosis was made exclusively from the right colon were as follows: cytomegalovirus (17 patients); Mycobacterium tuberculosis (2 patients); Mycobacterium avium complex VOLUME 48, NO. 4, 1998
Diagnostic yield and cost-effectiveness of endoscopy in HIV-related diarrhea
E Bini, J Cohen
Table 2. Pathogens identified by endoscopy in chronic HIV-related diarrhea
Pathogen Cytomegalovirus Microsporidia Mycobacterium avium complex Cryptosporidium Clostridium difficile Giardia Mycobacterium tuberculosis Isospora Hymenolepis nana Strongyloides stercoralis
Upper endoscopy (n = 203)
Flexible sigmoidoscopy (n = 85)
Colonoscopy (n = 191)
Overall (n = 191)
10 26 5 12 0 6 0 3 0 0
6 0 5 2 5 0 1 0 0 0
53 0 13 0 1 0 3 0 3 2
66 26 23 14 6 6 4 3 3 2
Three individuals had cytomegalovirus detected on more than one examination, and six patients had more than one pathogen identified.
Table 3. Diagnostic yield of endoscopic procedures in HIVrelated chronic diarrhea All patients Upper endoscopy (n = 203) Flexible sigmoidoscopy (n = 85) Colonoscopy (n = 191) Overall endoscopic workup (n = 307)
CD4 < 100 CD4 ≥ 100 cells/mm3 cells/mm3
29.6% 22.4%
37.9%* 27.9%*
8.6% 0%
38.7% 47.9%
50.0%* 62.8%*
4.3% 8.3%
*p < 0.05 compared with CD4 ≥ 100 cells/mm3.
(1 patient); and both cytomegalovirus and Mycobacterium tuberculosis (1 patient). Fourteen of these 21 patients (66.7%) and 11 of the 17 individuals (64.7%) with right-sided cytomegalovirus infection had intravenous drug use as their risk factor for HIV infection. Right-sided cytomegalovirus infection was noted in 11 of the 109 intravenous drug users who had a colonoscopy (10.1%). In contrast, only 3 of 39 (7.7%) homosexual patients who had a colonoscopy were found to have isolated right-sided cytomegalovirus infection. Thirty-four percent of the patients with cytomegalovirus had disease localized only in the right colon. The endoscopic appearance of cytomegalovirus colitis could be categorized into three patterns as follows: colitis (36.5%), single or multiple ulcers (28.8%), and colitis with ulcers (26.9%); only four patients (7.7%) with cytomegalovirus infection had normal-appearing mucosa. When isolated in the right colon, cytomegalovirus most often appeared as ulcers without colitis (52.9%). In contrast, cytomegalovirus in the left colon most often appeared as colitis (37.5%) or colitis with ulcers (37.5%). VOLUME 48, NO. 4, 1998
Table 4. Characteristics of HIV-infected patients with and without a pathogen identified by endoscopy Characteristic Age (yr) CD4 count/mm3* Fever Abdominal pain Weight loss (kg)* Hematocrit (%)* Albumin (gm/dL)* Fecal leukocytes Number of stools per day Duration of diarrhea (weeks) Empiric antidiarrheal therapy
Pathogen (n = 147)
No pathogen (n = 160)
p Value
38.1 ± 0.7 39.7 ± 2.9 44.2% 55.1% 9.6 ± 0.5 33.9 ± 0.3 3.1 ± 0.04 51.9% 8.2 ± 0.3
37.8 ± 0.5 152.1 ± 11.3 36.3% 49.4% 4.3 ± 0.3 35.4 ± 0.4 4.0 ± 0.04 19.9% 7.8 ± 0.3
0.75 < 0.0001 0.16 0.36 < 0.0001 0.0009 < 0.0001 < 0.0001 0.36
18.2 ± 1.1
20.9 ± 1.1
0.08
32.7%
28.8%
0.54
*Independent predictor of having a pathogen identified by endoscopy using multivariate analysis.
The diagnostic yield of endoscopy was significantly higher in patients with a CD4 lymphocyte count of less than 100 cells/mm3 than in those with higher CD4 counts (62.8% vs. 8.3%, p < 0.0001). Only 7 patients with a CD4 count of 100 cells/mm3 or greater had a pathogen identified by endoscopy. The pathogens identified in these 7 patients were Cryptosporidium, Strongyloides stercoralis, and Giardia in 2 patients each, and Isospora in 1 patient. In none of the 46 patients with a CD4 count of 200 cells/mm3 or greater was a pathogen identified. Clinical features of patients in whom a pathogen was identified were compared with those in whom no pathogen was found on endoscopy. The results of this comparison is shown in Table 4. Patients in whom a pathogen was identified were significantly GASTROINTESTINAL ENDOSCOPY
357
E Bini, J Cohen
Diagnostic yield and cost-effectiveness of endoscopy in HIV-related diarrhea
Table 5. Cost of identifying a pathogen by endoscopy Cost per Cost per Cost per pathogen pathogen CD4 pathogen CD4 < 100/mm3 ≥ 100/mm3 Upper endoscopy Flexible sigmoidoscopy Colonoscopy Overall endoscopic workup
$4867.68 $3352.67
$3717.44 $2682.13
$17,520.28 NA*
$2786.64 $3822.94
$2159.28 $2934.92
$25,371.54 $21,583.51
*No patient with a CD4 count of ≥ 100/mm3 had a pathogen identified by sigmoidoscopy.
more likely to have a low CD4 lymphocyte count, weight loss, fecal leukocytes, and lower hematocrit and serum albumin levels. Of the 7 individuals with a CD4 count of 100 cells/mm3 or greater in whom a pathogen was identified, weight loss (mean 13.5 kg, range 4.2 to 22.9 kg) was present in all but one patient. Inpatients were significantly more likely to have a pathogen identified by endoscopy than outpatients (58.3% vs. 44.7%, p = 0.04). Homosexuals were no more likely to have a pathogen identified than were those with other risk factors for HIV infection (50.9% vs. 47.2%, p = 0.66). Multivariate analysis identified serum albumin (p < 0.0001), inpatient status (p < 0.0001), CD4 lymphocyte count (p = 0.0001), hematocrit level (p = 0.02), and weight loss (p = 0.04) as independent predictors for having a pathogen identified by endoscopy. In the 203 patients undergoing upper endoscopy, aspiration of fluid from the duodenum or jejunum was obtained for ova and parasite analysis in 185 individuals (91.1%) and mucosal biopsies were evaluated by electron microscopy in 161 (79.3%). Examination of small bowel aspirates yielded a diagnosis in 3 of the 185 patients (1.6%); all 3 individuals were found to have Giardia. However, 2 of these 3 patients also had Giardia identified by lightmicroscopic examination of small bowel biopsies. Electron microscopy identified a pathogen in 44 patients (27.3%): microsporidia (25 patients); Cryptosporidium (11 patients); Giardia (5 patients); and Isospora (3 patients). Of the 25 patients with microsporidia, 21 (84%) were identified as Enterocytozoon bieneusi and 4 (16%) were identified as Encephalitozoon intestinalis. All of the patients with Cryptosporidium, Giardia, and Isospora had the diagnosis made by light microscopy in addition to electron microscopy. Among the 25 patients with microsporidia identified on electron microscopy, 17 (68%) had the pathogen identified by light microscopy as well. The use of electron microscopy 358
GASTROINTESTINAL ENDOSCOPY
in these 161 patients therefore resulted in the diagnosis of 8 patients with microsporidia (all Enterocytozoon bieneusi) that would have been missed if light microscopy had been used alone. Cost analysis The average cost of an endoscopic evaluation in this study was $1830.53 per patient. The cost of identifying a pathogen for each endoscopic procedure stratified by CD4 lymphocyte count is shown in Table 5. The average cost of identifying a pathogen by endoscopy in this patient population was $3822.94. Evaluation of small-bowel biopsy specimens by electron microscopy and small-bowel aspirates for ova and parasites resulted in an incremental cost of $12,276.25 and $1648.35, respectively, for each additional pathogen identified. DISCUSSION In HIV-infected patients with chronic unexplained diarrhea, data assessing the diagnostic yield of upper and lower endoscopy are limited. Only two previous studies have addressed this issue in patients with negative stool studies before endoscopy.16,17 Connolly et al.16 evaluated 33 HIV-infected patients with seemingly pathogen-negative diarrhea. In their study, patients underwent upper endoscopy and proctoscopy with biopsy, Schilling and D-xylose testing, microbiologic examination of stool, small-bowel barium examination, and barium enema. They identified a pathogen in 12 patients (36%) and found cryptosporidia to be the most common organism. In another study, Wilcox et al.17 prospectively evaluated 48 HIV-infected patients with chronic unexplained diarrhea. In their study, patients were evaluated by upper endoscopy and colonoscopy with biopsy. A potential cause of diarrhea was identified in 21 patients (44%). In contrast to Connolly16 and Wilcox17 and their associates found cytomegalovirus to be the most common pathogen. In our study of 307 consecutive HIVinfected patients with chronic unexplained diarrhea, cytomegalovirus was also the most common pathogen isolated. In addition, we found the diagnostic yield of endoscopy (47.9%) to be similar to that reported by Wilcox et al.17 In this study, upper endoscopy identified a pathogen in 60 of 203 patients (29.6%). Microsporidia was the most common pathogen found (26 patients). In a retrospective study of 57 HIV-infected patients with chronic diarrhea and negative stool studies, Bown et al.18 reported a similar diagnostic yield for upper endoscopy (26%). In their study, no patient had a pathogen identified by examination of duodenal aspirate for ova and paraVOLUME 48, NO. 4, 1998
Diagnostic yield and cost-effectiveness of endoscopy in HIV-related diarrhea
sites. We also found that small-bowel aspirates have a low diagnostic yield (1.6%). Bown et al.18 identified a pathogen on electron microscopy in 6 of 12 patients (50%). In contrast, we found the diagnostic yield of electron microscopy in 161 patients to be much lower (27.3%). In addition, the majority of patients in whom a pathogen was identified by electron microscopy also had the diagnosis established by light microscopy. All of the pathogens that were identified only by electron microscopy were found to be microsporidia (Enterocytozoon bieneusi), an organism for which no effective therapy is yet available.4,19-21 The presence of villous atrophy in our series (16.7%) has been well described.12 However, the etiology and clinical significance of this entity remains poorly defined. The high cost of identifying a pathogen by upper endoscopy is a result of the low diagnostic yield of this procedure and the use of electron microscopy in our patient population. If electron microscopy was deleted from our evaluation, the cost of identifying a pathogen by upper endoscopy would decrease from $4867.68 to $3727.90. Although elimination of electron microscopy results in substantial savings without apparent adverse clinical consequences, the cost of detecting a pathogen by upper endoscopy is still higher than that of flexible sigmoidoscopy and colonoscopy. Furthermore, some of the patients with pathogens detected by upper endoscopy may have had a diagnosis established by stool studies if immunofluorescent assays for Giardia and Cryptosporidium, enzyme immunoassay for Giardia antigen, or fluorescent staining of stool specimens for microsporidia were done before an endoscopic evaluation. The cost-benefit of performing these specialized tests routinely, with the potential for reduced endoscopic procedures, remains to be determined. In HIV-infected patients with chronic unexplained diarrhea undergoing lower endoscopy, controversy exists concerning the best diagnostic approach. Some clinicians recommend flexible sigmoidoscopy,12,17,19,22-24 whereas others suggest performing colonoscopy as the initial endoscopic procedure.10,25,26 In this study, the diagnostic yield of colonoscopy was significantly higher than the yield of flexible sigmoidoscopy (38.7% vs. 22.4%). In addition, 28% of the pathogens identified by colonoscopy and biopsy were found only in the right colon. Of note, 95% of the pathogens that were identified only in the right colon were organisms for which effective therapy is currently available.1,27-32 Right-sided cytomegalovirus infection of the colon has not been well described as a cause of chronic diarrhea. However, treatment of the patients with right-sided VOLUME 48, NO. 4, 1998
E Bini, J Cohen
cytomegalovirus colitis resulted in complete resolution of diarrhea in 66.7%. This therapeutic response strongly suggests that cytomegalovirus was the cause of chronic diarrhea in these patients. Cytomegalovirus accounted for 81% of pathogens located exclusively in the right colon. Conversely, in 34% of the patients with cytomegalovirus a diagnosis was established only on biopsy specimens taken from the cecum and ascending colon. These data are consistent with the findings of Dieterich and Rahmin26 who reported that 39% of cytomegalovirus colitis was found only in the cecum. This contrasts with the results of Wilcox et al.,17 who found only 1 of 11 patients (9.1%) with cytomegalovirus colitis whose diagnosis could not be made with left-sided biopsies alone.17 This discrepancy may be explained by differences in the patient population evaluated, given the higher proportion of patients in this study who were intravenous drug users. The finding of right-sided cytomegalovirus infection in 3 of 39 homosexual patients (7.7%) undergoing colonoscopy is similar to that of Wilcox et al.,17 who evaluated primarily homosexual patients and found right-sided cytomegalovirus infection in 1 of 48 individuals examined (2.1%). Although the cost of each colonoscopy was higher than the cost of flexible sigmoidoscopy, the greater diagnostic yield of colonoscopy resulted in a considerably lower cost of identifying a pathogen ($2786.64 vs. $3352.67). Given the 22.4% yield of flexible sigmoidoscopy and assuming from our data that 11% of patients with negative left-sided biopsies on sigmoidoscopy would have a pathogen identified on right-sided biopsies during subsequent colonoscopy, the cost per diagnosis of such a two-step strategy would be $4822.85. Performing colonoscopy only after negative flexible sigmoidoscopy would be 47% more costly than performing colonoscopy as the initial test in all patients. HIV-infected patients who are severely immunocompromised (CD4 count of less than 100 cells/mm3) are at increased risk for opportunistic infections, including cytomegalovirus, Mycobacterium avium complex, microsporidia, and cryptosporidia.1,12,25 The diagnostic yield of endoscopic investigation of chronic HIV-related diarrhea has been shown to correlate with the degree of immunosuppression.12,17 In this study, patients with a CD4 lymphocyte count of less than 100 cells/mm3 were significantly more likely to have a pathogen identified by endoscopy than those with higher CD4 counts. Because the extent of the endoscopic evaluation was left to the discretion of the endoscopist, there was potential for selection bias. However, there were no significant differences in the number and type of endoscopic GASTROINTESTINAL ENDOSCOPY
359
E Bini, J Cohen
Diagnostic yield and cost-effectiveness of endoscopy in HIV-related diarrhea
procedures performed between sicker patients and those who were less ill (inpatients vs. outpatients and CD4 count < 100 cells/mm3 vs. CD4 count ≥ 100 cells/mm3). Only 7 of 84 patients (8.3%) with a CD4 count of 100 cells/mm3 or greater had a pathogen identified by endoscopy. Weight loss was present in all but one of these 7 patients. In their prospective study of 48 patients, Wilcox et al.17 identified weight loss and a shorter duration of diarrhea as independent predictors for establishing a diagnosis by endoscopy and biopsy using multivariate analysis. Other investigators have also found weight loss to be an important clinical parameter.16 Neither the frequency of bowel movements nor the duration of diarrhea before endoscopy were significant predictors of identifying a pathogen in this study. Several diagnostic strategies have been recommended for the evaluation of chronic diarrhea in HIV-infected patients. Most investigators agree that stool studies should be the first diagnostic test because they are simple to perform, inexpensive, and noninvasive and have a high diagnostic yield.12,33,34 Many authors suggest at least three stool samples before proceeding with an endoscopic evaluation. It is possible that requiring three instead of two negative stool specimens would have obviated the need for endoscopy in some of our patients. However, there have been no studies demonstrating the marginal diagnostic yield of the third stool specimen in this patient population. When stool studies fail to identify a pathogen, endoscopic evaluation is frequently recommended.1,12 To date, the cost-effectiveness of upper endoscopy, flexible sigmoidoscopy, and colonoscopy in this setting has not been well defined. A medical decision analysis published in 1990 evaluated several diagnostic strategies for the evaluation of AIDS-related diarrhea in 1000 hypothetical patients. After taking cost into account, they recommended a single stool culture as the most cost-effective approach, with endoscopy reserved only for patients who fail symptomatic treatment.14 However, the outcome assumptions used were based on limited successful therapies available at the time and on information about diagnostic yield and pathogens identified drawn from studies involving small groups of patients. Determining the most cost-effective approach requires better data about the diagnostic yield of various endoscopic procedures, the frequency of finding pathogens for which effective treatment is available, and the comparative direct and indirect costs of empiric symptomatic versus pathogen-based therapy. The average cost of detecting a pathogen in this study ($3822.94) is relatively low compared 360
GASTROINTESTINAL ENDOSCOPY
with the cost per diagnosis of other entities, such as an adenomatous polyp detected by screening flexible sigmoidoscopy ($8766.00).35 Justifying this cost depends on further outcome data using current available therapies and including both cost and quality of life assessments. The critical question of the effectiveness of pathogen specific therapy in our study population is the subject of ongoing investigation. This study provides a clearer understanding of the diagnostic yield of various endoscopic procedures in a large group of patients with chronic HIVrelated diarrhea and the cost incurred in making these diagnoses. The data demonstrate a much higher diagnostic yield and more cost-effective evaluation in individuals with low CD4 lymphocyte counts, suggesting that this subset of patients is most likely to benefit from an endoscopic evaluation rather than a symptomatic treatment approach. The benefit of small-bowel aspirates and particularly electron microscopy appears to be minimal. If clinicians elect to use endoscopy to pursue specific treatable pathogens, colonoscopy with biopsies from the right and left colon is the most cost-effective examination with the highest diagnostic yield. In patients without weight loss and with a CD4 lymphocyte count of 100 cells/mm3 or greater, endoscopy is unlikely to identify an etiology. These patients may be managed best by empiric treatment with antidiarrheal medications, reserving endoscopy for those in whom diarrhea persists despite antidiarrheal therapy. REFERENCES 1. Smith PD, Quinn TC, Strober W, Janoff EN, Masur H. NIH Conference: gastrointestinal infections in AIDS. Ann Intern Med 1992;116:63-77. 2. Antony MA, Brandt LJ, Klein RS, Bernstein LH. Infectious diarrhea in patients with AIDS. Dig Dis Sci 1988;33:1141-6. 3. Gazzard BG. Diarrhea in human immunodeficiency virus antibody-positive patients. Semin Liver Dis 1992;12:154-66. 4. Goodgame RW. Understanding intestinal spore-forming protozoa: cryptosporidia, microsporidia, isospora, and cyclospora. Ann Intern Med 1996;124:429-41. 5. Rene E, Marche C, Regnier B, Saimot AG, Vilde JL, Perrone C, et al. Intestinal infections in patients with acquired immunodeficiency syndrome: a prospective study in 132 patients. Dig Dis Sci 1989;34:773-80. 6. Janoff EN, Orenstein JM, Manischewitz JF, Smith PD. Adenovirus colitis in the acquired immunodeficiency syndrome. Gastroenterology 1991;100:976-9. 7. Grohmann GS, Glass RI, Pereira HG, Monroe SS, Hightower AW, Weber R, et al. Enteric viruses and diarrhea in HIVinfected patients. N Engl J Med 1993;329:14-20. 8. Dancygier H. AIDS and gastrointestinal endoscopy. Endoscopy 1992;24:169-75. 9. Smith PD. Infectious diarrhea in patients with AIDS. Gastroenterol Clin North Am 1993;22:535-48. 10. Simon D, Brandt LJ. Diarrhea in patients with the acquired VOLUME 48, NO. 4, 1998
Diagnostic yield and cost-effectiveness of endoscopy in HIV-related diarrhea
11.
12.
13. 14.
15.
16.
17.
18.
19.
20.
21.
22. 23.
immunodeficiency syndrome. Gastroenterology 1993;105: 1238-42. Smith PD, Lane HC, Gill VJ, Manischewitz JF, Quinnan GV, Fauci AS, et al. Intestinal infections in patients with the acquired immune deficiency syndrome: etiology and response to therapy. Ann Intern Med 1988;108:328-33. Wilcox CM, Rabeneck L, Friedman S. Malnutrition and cachexia, chronic diarrhea, and hepatobiliary disease in patients with human immunodeficiency virus infection. Gastroenterology 1996;111:1724-52. Brady CE. AIDS-related diarrhea: What’s a gastroenterologist to do? Am J Gastroenterol 1991;86:1685-6. Johanson JF, Sonnenberg A. Efficient management of diarrhea in the acquired immune deficiency syndrome (AIDS): a medical decision analysis. Ann Intern Med 1990;112:942-8. Center for Disease Control and Prevention. 1993 revised classification system for HIV infection and expanded surveillance case definition for AIDS among adolescents and adults. MMWR 1992;41:1-19. Connolly GM, Forbes A, Gazzard BG. Investigation of seemingly pathogen-negative diarrhoea in patients infected with HIV-1. Gut 1990;31:886-9. Wilcox CM, Schwartz DA, Cotsonis G, Thompson SE. Chronic unexplained diarrhea in human immunodeficiency virus infection: determination of the best diagnostic approach. Gastroenterology 1996;110:30-7. Bown JW, Savides TJ, Mathews C, Isenberg J, Behling C, Lyche KD. Diagnostic yield of duodenal biopsy and aspirate in AIDS-associated diarrhea. Am J Gastroenterol 1996;91:228992. Blanshard C, Francis N, Gazzard BG. Investigation of chronic diarrhoea in acquired immunodeficiency syndrome. A prospective study of 155 patients. Gut 1996;39:824-32. Rabeneck L. Diagnostic workup strategies for patients with HIV-related chronic diarrhea: What is the end result? J Clin Gastroenterol 1993;16:245-50. Johanson JF. To scope or not to scope: the role of endoscopy in the evaluation of AIDS-related diarrhea. Am J Gastroenterol 1996;91:2261-2. Sharpstone D, Gazzard B. Gastrointestinal manifestations of HIV infection. Lancet 1996;348:379-83. Connolly GM, Forbes A, Gleeson JA, Gazzard BG. The value
24.
25. 26.
27. 28.
29.
30. 31.
32.
33.
34.
35.
E Bini, J Cohen
of barium enema and colonoscopy in patients infected with HIV. AIDS 1990;4:687-9. Connolly GM, Shanson D, Hawkins DA, Harcourt Webster JN, Gazzard BG. Non-cryptosporidial diarrhea in human immunodeficiency virus (HIV) infected patients. Gut 1989;30:195-200. Mayer HB, Wanke CA. Diagnostic strategies in HIV-infected patients with diarrhea. AIDS 1994;8:1639-48. Dieterich DT, Rahmin M. Cytomegalovirus colitis in AIDS: presentation in 44 patients and a review of the literature. J Acquir Immune Defic Syndr 1991;4(Suppl 1):S29-35. Goodgame RW. Gastrointestinal cytomegalovirus disease. Ann Intern Med 1993;119:924-35. Dieterich DT, Kotler DP, Busch DF, Crumpacker C, Du Mond C, Dearmand B, et al. Ganciclovir treatment of cytomegalovirus colitis in AIDS: a randomized, double-blind, placebo-controlled multicenter study. J Infec Dis 1993;167:278-82. Barnes PF, Bloch AB, Davidson PT, Snider DE. Tuberculosis in patients with human immunodeficiency virus infection. N Engl J Med 1991;324:1644-50. Iseman MD. Treatment of multidrug-resistant tuberculosis. N Engl J Med 1993;329:784-91. Simon D, Weiss LM, Brandt LJ. Treatment options for AIDSrelated esophageal and diarrheal disorders. Am J Gastroenterol 1992;87:274-81. Nelson MR, Connolly GM, Hawkins DA, Gazzard BG. Foscarnet in the treatment of cytomegalovirus infection of the esophagus and colon in patients with the acquired immune deficiency syndrome. Am J Gastroenterol 1991;86:876-81. Crotty B, Smallwood RA. Investigating diarrhea in patients with acquired immunodeficiency syndrome. Gastroenterology 1996;110:296-8. Blackman E, Binder S, Gaultier C, Benveniste R, Cecilio M. Cryptosporidiosis in HIV-infected patients: diagnostic sensitivity of stool examination, based on number of specimens submitted. Am J Gastroenterol 1997;92:451-3. Lieberman D. Cost-effectiveness of colon cancer screening. Am J Gastroenterol 1991;86:1789-94.
Availability of Journal back issues As a service to our subscribers, copies of back issues of Gastrointestinal Endoscopy for the preceding 5 years are maintained and are available for purchase from Mosby until inventory is depleted at a cost of $14.00 per issue. The following quantity discounts are available: 25% off on quantities of 12 to 23, and one third off on quantities of 24 or more. Please write to Mosby, Inc., Subscription Services, 11830 Westline Industrial Dr., St. Louis, MO 63146-3318, or call 800-453-4351 or 314-453-4351 for information on availability of particular issues. If unavailable from the publisher, photocopies of complete issues may be purchased from UMI, 300 N. Zeeb Rd., Ann Arbor, MI 48106; 313-761-4700.
VOLUME 48, NO. 4, 1998
GASTROINTESTINAL ENDOSCOPY
361