Probability to develop acute, chronic, or any diarrhea, depending on the CD4 lymphocyte count stratum at baseline.
Distribution of CD4 lymphocyte counts in patients with acute (AD) or chronic (CD) diarrhea of unknown cause and patients with diarrhea associated with different intestinal pathogens.
Diagnostic yield and CD4 lymphocyte counts in individual patients in whom enteric pathogens or no pathogens were identified by stool examination only, by upper endoscopy, or by ileocolonoscopy. The median CD4 cell count is depicted by a horizontal line.
Weber R, Ledergerber B, Zbinden R, Altwegg M, Pfyffer GE, Spycher MA, Briner J, Kaiser L, Opravil M, Meyenberger C, Flepp M, for the Swiss HIV Cohort Study. Enteric Infections and Diarrhea in Human Immunodeficiency Virus–Infected PersonsProspective Community-Based Cohort Study. Arch Intern Med. 1999;159(13):1473-1480. doi:10.1001/archinte.159.13.1473
Persons infected with human immunodeficiency virus (HIV) are at increased risk for diarrhea and enteric infections. We studied (1) the epidemiology of enteric pathogens associated with diarrhea, (2) the diagnostic yield of stool examination and endoscopic evaluation, (3) risks to develop diarrhea, and (4) the impact of diarrhea on patients' survival.
A total of 1933 participants in the Swiss HIV Cohort Study were prospectively followed up for a median of 25.5 months. A total of 560 diarrheal episodes were evaluated by standardized stool examination. Endoscopic evaluation was performed in 25% of patients with chronic diarrhea.
The incidence of diarrhea was 14.2 per 100 person-years (95% confidence interval, 13.0-15.4). Among patients with CD4 cell counts below 0.05×109/L, the probability to develop diarrhea within 1, 2, and 3 years was 48.5%, 74.3%, and 95.6%, respectively. The risk to develop diarrhea was increased among patients with severe immunodeficiency, homosexual men, and patients taking antiretroviral therapy. Pneumocystis carinii chemoprophylaxis did not reduce the risk of diarrhea. Diarrhea was an independent negative predictor of survival. Enteric pathogens were detected in 16.5% of 212 acute diarrheal episodes and in 46% of 348 chronic diarrheal episodes. The sensitivity of histological and stool examination was similar except for the diagnosis of intestinal cytomegalovirus infection and leishmaniasis, which required invasive evaluation.
Intestinal infections were diagnosed in less than 50% of chronic diarrheal episodes. The prevalence of enteric pathogens tended to decrease during the observation period, possibly because of improved antiretroviral therapy. Endoscopic evaluation did not improve the diagnostic yield compared with stool examination except for the diagnosis of cytomegalovirus enteritis and leishmaniasis.
DIARRHEA IS frequently stated to cause substantial morbidity and mortality in persons infected with human immunodeficiency virus (HIV), occurring at some time during the illness in up to 60% of patients in industrialized nations and in up to 90% of patients in developing countries.1- 3 However, few systematic data on the epidemiology of diarrhea and intestinal infections are available.4 Most reports are based on cross-sectional data of selected or small patient groups, and the reported prevalence rates of different intestinal pathogens vary widely.4 Recently, a changing epidemiological and clinical pattern of opportunistic complications was observed in industrialized countries that was explained by the influence of chemoprophylaxis and antiretroviral therapy, but little is known as to whether these interventions also had an impact on enteric infections. Furthermore, practical questions regarding the diagnostic approach of HIV-associated diarrhea, in particular the intensity of the diagnostic workup, remain unanswered or controversial.4- 9
Primary objectives of this study were to systematically investigate the spectrum of enteric pathogens associated with diarrhea in unselected patients including all transmission groups as well as women, and to compare the results of a standardized stool examination with the findings of endoscopic evaluation. Secondary objectives were to assess the incidence of acute and chronic diarrhea, to identify risks associated with diarrhea, and to determine the impact of diarrhea on patients' survival. Therefore, we studied all consecutive patients who were examined because of diarrhea at our HIV outpatient clinic during a 45-month period, and we prospectively followed up these patients within the Swiss HIV Cohort Study.
We prospectively collected information on all patients who were examined because of diarrhea at the HIV outpatient clinic of the University Hospital, Zurich, Switzerland, between July 1, 1992, and March 31, 1996, and who were also followed up within the Swiss HIV Cohort Study. Follow-up included information up to June 30, 1997. Data on diarrhea and enteric pathogens were obtained at the regular visits within the cohort study and also between study visits if patients attended our clinic for evaluation of diarrhea.
Between July 1, 1992, and June 30, 1994, participants of the cohort study without diarrhea were asked to provide 1 stool sample.
The Swiss HIV Cohort Study, initiated in 1988, is a prospective, community-based observational study of HIV-infected adults with ongoing enrollment.10,11 Enrollment is independent of sex, mode of HIV acquisition, HIV disease stage, or treatment. Informed consent is obtained from all participants. Patients are followed up at 6-monthly intervals. Standardized information is collected, including a history of HIV-associated diseases, the course of treatments, and laboratory data.
Diarrhea was defined as 2 or more fluid or 3 or more soft stools per day. A diarrheal episode was evaluated if it lasted for at least 3 days and was considered acute if it lasted less than 4 weeks and chronic if it lasted for 4 weeks or longer.
Stool samples of patients without diarrhea were examined for microsporidial spores and coccidial oocysts. All patients with diarrhea underwent a standardized stool examination including bacterial cultures, ova, and parasites (including microsporidia and cryptosporidia). In patients with chronic diarrhea, examinations additionally included mycobacteria, Clostridium difficile, and Strongyloides larvae. If no pathogens were detected or if diarrhea persisted after initial evaluation, at least an additional 2 stool specimens collected on different days were examined. Prepared sets of stool vials and instructions for stool collection were provided by a study nurse.
Patients with chronic diarrhea underwent esophagogastroduodenoscopy or ileocolonoscopy if the responsible physician at our institution decided that invasive evaluation was indicated and if the patient agreed. Ileal biopsies were performed in all patients who underwent lower endoscopy. Three biopsy specimens were obtained for histological examination, and additional tissue specimens were obtained for electron microscopy and bacterial and mycobacterial cultures. Intestinal fluid was aspirated and examined for the presence of protozoa.
A standard ova and parasite examination of the stool was performed.12Entamoeba histolytica was not differentiated from nonpathogenic Entamoeba dispar. Microsporidial spores were detected by means of the chromotrope staining technique.13,14 Microsporidial species were identified by electron microscopic examination or molecular analysis.15 Coccidial oocysts were detected in sediments obtained by the formalin–ethyl acetate concentration technique and stained with Kinyoun acid-fast stain.16 Bacterial cultures included Salmonella, Shigella, Campylobacter, Yersinia, Aeromonas, Plesiomonas, and C difficile. A C difficile toxin assay was also used. Verotoxin-producing and enterotoxigenic Escherichia coli were diagnosed by molecular techniques.17 Detection of mycobacteria included examination of fluorochrome-stained smears and cultures with the use of solid (Löwenstein-Jensen, Middlebrook 7H10/sel7H11 agar; Becton Dickinson and Company, Franklin Lakes, NJ) and liquid (Bactec 12 B; Becton Dickinson) media. Strongyloides larvae were detected by microscopic stool examination or culture.
Sediments of duodenal or colonic aspirates were stained with Giemsa, Gram, and chromotrope and examined by light microscopy. In addition to routine histological examination, Brown-Brenn–stained sections were examined at 630- and 1000-fold magnification to visualize microsporidia.13 Cytomegalovirus was detected by means of in situ hybridization. Glutaraldehyde-fixed tissues were examined by electron microscopy.18
To determine a possible change of the prevalence of enteric pathogens, patients were classified into 2 groups according to the date of the evaluation of diarrhea (July 1, 1992, to June 30, 1994; and July 1, 1994, to March 31, 1996). A patient could be attributed to both observation periods if he or she was followed up in both periods. An episode of chronic diarrhea that began in the first period and lasted until the second period was counted only once. Categorical variables between groups were compared by χ2 or Fisher exact tests; continuous variables were compared by Wilcoxon rank-sum tests.
The incidence of diarrhea was calculated as the number of episodes per 100 patient-years.
The risk to develop diarrhea was calculated for patients who had no diarrhea when they provided a first stool sample between July 1, 1992, and June 30, 1994. We chose this stool examination as time zero. Kaplan-Meier estimates of the cumulative probability of developing diarrhea were assessed from time zero to the date of either the first episode of acute or chronic diarrhea or the last visit before the end of the observation period for those without diarrhea. Three baseline CD4 strata were compared with the log-rank test. Cox proportional hazards regression models were used for multivariate analysis of the risk to develop diarrhea. Treatments (chemoprophylaxis, antiretroviral therapy) were modeled as time-dependent covariates. Models were adjusted for CD4 lymphocyte counts (as a continuous variable). SAS software, version 6.11 (SAS Institute Inc, Cary, NC) was used.
Predictors of survival were assessed with Cox proportional hazards regression models adjusted for CD4 lymphocyte count. Time was measured from the date of the first visit within the observation period to the date of death or the most recent follow-up. Treatments, opportunistic complications, and diarrheal episodes were modeled as time-dependent covariates.
Among the 1933 patients of the Swiss HIV Cohort Study who attended our outpatient clinic during the 45-month study period, we evaluated a total of 560 diarrheal episodes: 212 episodes of acute diarrhea and 348 of chronic diarrhea. Median follow-up per participant was 25.5 months (range, 0-57 months), accumulating to 3953 person-years. The proportion of women was 24.5%. Other demographic and clinical variables are summarized in Table 1.
The incidence of diarrhea was 14.2 per 100 person-years overall (95% confidence interval [CI], 13.0-15.4) (Table 1). During the 2 observation periods (July 1, 1992, to June 30, 1994, vs July 1, 1994, to March 31, 1996), the incidence of acute diarrhea was 6.53 (95% CI, 5.5-7.7) and 3.91 (95% CI, 3.1-5.0) (P<.001), and that of chronic diarrhea, 8.54 (95% CI, 7.4-9.9) and 9.13 (95% CI, 7.8-10.7) (P=.55), respectively (Table 2).
For patients without diarrhea who had CD4 lymphocyte counts below 0.05×109/L, the probability to develop diarrhea within 1, 2, and 3 years was 48.5% (95% CI, 38.5%-58.3%), 74.3% (61.1%-86.9%), and 95.6% (82.2%-100%), respectively, which was significantly higher than for patients classified into the higher CD4 strata (P<.001) (Figure 1).
In the multivariate analysis adjusted for CD4 cell count, the risk to develop diarrhea was significantly higher for homosexual men, patients with a history of opportunistic complications, and patients taking antiretroviral therapy (Table 3). In addition, female sex was an independent negative predictor of chronic diarrhea. Chemoprophylaxis with sulfamethoxazole-trimethoprim or dapsone-pyrimethamine had no impact on the risk to develop diarrhea.
Diarrhea was found to be an independent negative predictor of survival (Table 4). Other negative risks were low CD4 cell counts, opportunistic complications, older age, and intravenous drug use as the mode of HIV acquisition (Table 4). The risk of death was reduced by 45% (relative risk, 0.55; 95% CI, 0.45-0.68) by antiretroviral therapy.
The presence of cryptosporidial oocysts or microsporidial spores in stool specimens was strongly associated with diarrhea. Among 949 patients without diarrhea at study entry, cryptosporidia were detected in 7 (0.7%) and microsporidia in 4 (0.4%).
Enteric pathogens were detected in 18.2% of 143 and 13.0% of 69 acute diarrheal episodes during the 2 observation periods (Table 2).
Among the patients with chronic diarrhea, at least 1 intestinal pathogen was identified in 50.3% of 187 diarrheal episodes and 41.0% of 161 episodes evaluated during observation periods 1 and 2, respectively (Table 2). The detection rate of enteric opportunistic protozoa, nontuberculous mycobacteria, and cytomegalovirus associated with chronic diarrhea was lower during the second observation period than the first period, whereas the rate of enteric bacterial infections remained unchanged. The most common pathogens were cryptosporidia (observation period 1, 15.5%; period 2, 11.8%), nontuberculous mycobacteria (14.4%; 8.1%), microsporidia (10.7%; 5.3%), and bacteria (9.6%; 11.8%). Cytomegalovirus colitis was diagnosed in 5 (15%) of 34 and 3 (15%) of 20 patients who underwent colonoscopy during observation periods 1 and 2, respectively. In addition, cytomegalovirus duodenitis was found in 2 (5%) of 38 patients who underwent duodenoscopy during observation period 1, and autopsy disclosed cytomegalovirus enteritis in 1 patient who died of bowel perforation.
All enteric pathogens, except for E histolytica, were strongly associated with severe immunodeficiency (Figure 2). A total of 66.7%, 77.4%, and 90.2% of all pathogens were found in patients with CD4 lymphocyte counts below 0.05, 0.10, and 0.20×109/L, respectively.
Intestinal coinfections were diagnosed in 2.4% of 212 acute and 9.8% of 348 chronic diarrheal episodes. Among the 160 patients with chronic diarrhea in whom at least 1 intestinal pathogen was identified, 21.3% had dual or multiple intestinal coinfections. Intestinal coinfection was found among 48% of patients with intestinal nontuberculous mycobacteriosis, 38% with cryptosporidiosis, 29% with microsporidiosis, 27% with cytomegalovirus infection, and 26% with enteric bacterial infections.
The examination of tissue specimens obtained by gastroduodenoscopy disclosed pathogens that had not been diagnosed by previous stool examinations in 4 (8%) of 51 patients with chronic diarrhea, including cytomegalovirus duodenitis in 2 and intestinal leishmaniasis in 1 patient. Among the patients with chronic diarrhea who underwent ileocolonoscopy, at least 1 additional pathogen was detected in 10 (20%) of 50. Most frequently cytomegalovirus colitis was found (Table 5).
The diagnostic yield of endoscopic evaluations as well as of stool examinations correlated with the severity of immunodeficiency (Figure 3). All patients in whom pathogens were found by upper or lower endoscopy had CD4 cell counts below 0.12×109/L (median, 0.01×109/L).
To overcome possible limitations of cross-sectional studies, we prospectively investigated the epidemiological characteristics of HIV-associated enteric pathogens and diarrhea in a large cohort of unselected patients and sought to assess an efficient diagnostic approach to diarrhea by comparing the diagnostic yield of stool examination and endoscopic evaluation.
The participants in the Swiss HIV Cohort Study are likely to be representative of HIV-infected patients in general practice and include persons from each of the 3 predominant HIV transmission groups as well as women.10,11 Although examination of intestinal biopsy specimens was considered the criterion standard for diagnosis of various intestinal pathogens,13,19 endoscopic examination was not an inclusion criterion for this study because we believed that a mandatory invasive workup might have deterred some patients from consenting to evaluation for diarrhea. The panel of stool examinations, in contrast, was strictly controlled.
We found an incidence of 14.2 diarrheal episodes per 100 person-years in our cohort. Compared with the 2 other cohort studies reporting corresponding data, our figure is higher than the incidence of 2.6 among a cohort of homosexual and bisexual men followed up in 3 US cities20 and the rate of 4.3% per person-semester observed among homosexual and bisexual men in the US Multicenter AIDS Cohort Study.21 The differences may partially be explained by different proportions of severely immunodeficient patients in the 3 cohorts.
The results of our study may reflect only a minimal estimate of the burden of diarrheal disease because diarrheal episodes were counted only if they were present and reported at the time of a visit at our institution. Evidently, the number of diarrheal episodes was the cause of substantial morbidity. Moreover, not only was diarrhea a clinical surrogate marker of the immune function, but (as shown in a CD4 lymphocyte count–adjusted multivariate analysis) we found diarrhea to be an independent predictor of an increased mortality, probably because of the impact of enteric infections for which no treatment option is available (eg, cryptosporidiosis, Enterocytozoon bieneusi infection).
Diarrhea, the presence of enteric pathogens, and the proportion of microbiologically positive findings of stool examination or endoscopic evaluation were strongly associated with severe immunodeficiency, particularly with CD4 lymphocyte counts below 0.05×109/L.19,22 In contrast, the risk to develop diarrhea or the rate of enteric infections was low in patients with CD4 cell counts above 0.20×109/L.
The prevalence of some enteric pathogens was different from what we expected. First, bacterial infections were infrequent in patients with chronic or acute diarrhea and were almost exclusively diagnosed in the same CD4 strata as opportunistic pathogens, although enteric bacteria are usually not considered pathogens occurring primarily in immunocompromised hosts. Second, we had anticipated a higher rate of microsporidial infections, particularly because we thought that these newly described pathogens would explain a substantial proportion of previously unexplained diarrheal episodes, and cross-sectional studies had suggested prevalence rates of up to 50%.18 Third, in more than 20% of patients harboring enteric pathogens, dual or multiple intestinal coinfections were found. Fourth, the prevalence of cytomegalovirus enteritis was low but may be underestimated because the proportion of patients with endoscopic evaluation was only 25%. However, we believe that we did not significantly underdiagnose this disease because endoscopic evaluation was primarily performed in patients with low CD4 counts who were at highest risk for cytomegalovirus enteritis. Furthermore, since fever is present in at least 80% of patients with cytomegalovirus colitis,19 colonoscopy was performed in all patients with unexplained fever and persistent diarrhea.
The overall diagnostic yield in our cohort may appear remarkably low because in less than 50% of cases of diarrhea lasting more than 4 weeks, and in less than 20% of acute diarrheal episodes, a causative agent could be identified. Of note, an invasive workup did not result in a substantially improved diagnostic yield among our patient group with chronic diarrhea. Stool examination was at least as sensitive as histological and electron microscopic examination of intestinal tissue for all pathogens, except for cytomegalovirus and Leishmania.23 No other data of long-term cohort studies on the epidemiological characteristics of enteric pathogens are available, yet recently published results of cross-sectional studies reported a similar or even lower diagnostic yield of endoscopic evaluation.6,24 In contrast, detection rates of enteric pathogens of up to 70% to 83% have also been described.25- 27 Most of these results were obtained in small studies of possibly selected patients. Some investigators have reported using techniques to detect enteric viruses other than cytomegalovirus to increase their diagnostic yield.25,28 We did not include such examinations because no diagnostic techniques for routine use were available (except for rotavirus), and no options to treat such viral infections were known.
Our study was not designed to investigate pathogen-negative diarrhea. There is evidence that the current diagnostic methods may fail to detect some clinically relevant organisms. Novel diagnostic techniques have been proposed, including improved methods to culture Campylobacter species,29 cell culture assays, molecular methods, or electron microscopy to detect enteroaggregative E coli or other enteropathogenic bacteria,30,31 and nucleic acid–based methods to detect intestinal protists. In routine practice, however, such techniques are not available.32 A segment of our patients certainly had diarrhea of noninfectious origin, particularly drug-associated diarrhea, as indicated by the risk analysis summarized in Table 3 indicating that antiretroviral therapy was an independent predictor of diarrhea.
Whereas the number of episodes of chronic diarrhea appeared to remain stable over time, the prevalence of most intestinal pathogens as well as the proportion of positive findings of the diagnostic workup tended to decrease during the study period. Since the diagnostic strategies remained unchanged over time, the decrease of enteric pathogens may be the result of an improved antiretroviral therapy. Preliminary results indeed suggested that an improvement of the immune functions by using effective antiretroviral treatment can lead to a clinical improvement of microsporidiosis or cryptosporidiosis, in parallel to the clearance of these protists.33 Although our study was performed before widespread use of protease inhibitors, our cohort study could demonstrate a substantial survival benefit of patients receiving antiretroviral treatment.11 Double antiretroviral combination therapy became the standard of care at our institution during the study period, and protease inhibitors became available at the end of 1995.
In conclusion, diarrhea and intestinal pathogens were particularly associated with a large burden of disease in patients with CD4 cell counts below 0.05×109/L. Stool examinations proved as sensitive as endoscopic evaluation for all pathogens except cytomegalovirus and Leishmania. Chronic diarrhea remained unexplained in a considerable number of patients, and a comprehensive and invasive diagnostic workup proved efficient only in patients with CD4 lymphocyte counts below 0.10×109/L. Consequently, endoscopic evaluation appears rarely indicated in patients with higher CD4 cell counts. The prevalence of intestinal pathogens tended to decrease during the study period, probably related to improved antiretroviral therapy.
Accepted for publication December 8, 1998.
This work was supported in part by the Swiss National Program for AIDS Research, Bern, Switzerland (grant 91-7066), and the Swiss HIV Cohort Study. The Swiss HIV Cohort Study is funded by the Federal Office of Public Health, Bern.
We are indebted to Walter Bossart, PhD (Institute of Medical Virology, Peter Deplazes, VMD, and Alexander Mathis, PhD (Institute of Parasitology), Thomas Bächi, PhD, and Ruth Keller (Institute for Electron Microscopy), and Guido Funke, MD (Department of Medical Microbiology), University of Zurich, Zurich, Switzerland; Rudolf Ammann, MD, Ruedi Lüthy, MD, Josef Jost, MD, Christian Ruef, MD, and Bärbel Sauer for excellent consultative and technical assistance and patient care; and Karola Hodapp, study nurse (University Hospital, Zurich), for data collection.
Reprints: Rainer Weber, MD, Division of Infectious Diseases and Hospital Epidemiology, University Hospital, CH-8091 Zurich, Switzerland (e-mail: email@example.com).
Manuel Battegay, MD (Cochair of the Scientific Board); Enos Bernasconi, MD; Philippe Bürgisser, MD; Matthias Egger, MD; Peter Erb, PhD (Chairman of the Laboratories Group); Walter Fierz, MD; Markus Flepp, MD (Chairman of the Clinics Group); Patrick Francioli, MD (President); Hans-Jakob Furrer, MD; Peter Grob, MD; Bernard Hirschel, MD (Chairman of the Scientific Board); Laurent Kaiser, MD; Bruno Ledergerber, PhD; Raffaele Malinverni, MD; Lukas Matter, MD; Milos Opravil, MD; Fred Paccaud, MD; Giuseppe Pantaleo, MD; Luc Perrin, MD; Werner Pichler, MD; Jean-Claude Piffaretti, PhD; Martin Rickenbach, MD (Head of Data Center); Philippe Sudre, PhD; Jurg Schüpbach, PhD; Amalio Telenti, MD; Pietro Vernazza, MD; and Rainer Weber, MD.