key: cord-008672-luoxomif authors: Mwachari, C.; Batchelor, B.I.F.; Paul, J.; Waiyaki, P.G.; Gilks, C.F. title: Chronic diarrhoea among HIV-infected adult patients in Nairobi, Kenya date: 2004-10-29 journal: J Infect DOI: 10.1016/s0163-4453(98)90561-8 sha: doc_id: 8672 cord_uid: luoxomif OBJECTIVES: Chronic diarrhoea and wasting are well recognized features of AIDS in Africa. However, because of resource constraints few ocmprehensive aetiological studies have conducted in sub-Saharan Africa which have included a broad range of microbiological investigations. We undertook a prospective cross-sectional study of adult patients admitted to a government hospital in Nairobi, Kenya, to determine possible bacterial, mycobacterial, parasitic and viral causes of diarrhoca; to consider which may be treatable; and to relate microbiological findings to clinical outcome. METHODS: Stool specimens from 75 consecutive HIV-seropositive patients with chronic diarrhoca admitted to a Nairobi hospital were subjected to microbiological investigation and results were compared with clinical findings and outcome. Stool samples were cultured for bacteria and mycobacteria and underwent light and electron microscopy; lawns of Escherichica coli were probed for pathogenic types and aliquots were tested for the presence of Clostridium difficile cytotoxin. Blood cultures for mycobacteria and other bacterial pathogens were performed as clinically indicated. RESULTS: Thirty-nine (52%) patients yielded putative pathogens, the most common being Cryptosporidium sp. (17%), Salmonella typhimurium (13%), and Mycobacterium tuberculosis (13%). Of 41 patients investigated for pathogenic Escherichia coli, enteroaggregative E. coli and diffusely adherent E. coli were each found in four patients. Thirty-one (41%) patients died. Detection of cryptosporidium cysts was the single most significant predictor of death (X(2) = 5.2, P<0.05). Many patients did not improve (21; 285) or self-discharged whilst still sick (5; 7%) but five (7%) were diagnosed ante mortem with tuberculosis and treated and a further 13 (17%) showed improvement by time of discharge. CONCLUSIONS: HIV-infected patients with chronic diarrhoea in Nairobi have a poor outcome overall, and even with extensive investigation a putative pathogen was identified in only just over half the patients. The most important step is to exclude tuberculosis: and the most useful investigation appears to be Ziehl-Neelsen staining. Other potentially treatable Gram-negative bacterial pathogens, S. typhimurium, Shigella sp. and adherent E. coli were, however, common but require culture facilities which are not widely accessible for definitive identification. Further studies focussing on simple ways to identify sub-groups of patients with treatable infections are warranted. Chronic diarrhoea and wasting, slim disease is a well recognized feature of African AIDS. 1'2 In Nairobi it is a common cause of hospital admission in HIV-infected individuals, ranking fourth in cause of admission (after tuberculosis, acute pneumonia and enteric fever-like illness) and third as cause of death. 3 Community studies also suggest that it is also a common cause of death, particularly when associated with significant body wasting. In Europe the aetiology of HIV-associated chronic diarrhoea has been comprehensively investigated 4 because * Please address all correspondence to: C. F. Gilks, Liverpool School of Tropical Medicine, Pembroke Place, Liverpool L3 5QA, U.K. Accepted for publication 20 April 1998. of the routine use of intensive diagnostic investigations, access to well equipped microbiology laboratories and frequent post-mortem studies. Such services are not often routinely available in most government hospitals in sub-Saharan Africa; and despite the frequency of HIV-related chronic diarrhoea, the aetiology is far less well defined. Most investigations into the cause of chronic diarrhoea have limited themselves to small patient numbers, 5' 6 to an intensive search for an individual pathogen 7-9 or have been narrowly based due to limited culture facilities, either ante-mortem or post-mortem. 6' ~o In particular, it is not clear whether any treatable causes of chronic diarrhoea are frequently missed because of the lack of appropriate diagnostic facilities. We undertook a 4-month study in Nairobi, Kenya, to determine possible bacterial, viral and parasitological causes of chronic diarrhoea and wasting using a broad range of investigations and related findings to patient outcome. From April to July 1992, consecutive patients reporting chronic diarrhoea (loose or watery stool for at least 4 weeks) were enrolled into a prospective clinical and microbiological study. All were adults (>16 years) admitted directly to the acute medical wards of the Kenyatta National Hospital (KNH), the main government hospital serving Nairobi. Patients referred from other hospitals were excluded. Recruitment occurred within 5 days of admission. All patients were examined, a brief history was taken and details were recorded on a standard clinical entry and follow-up form by a single observer (CM). Patients received HIV counselling and informed consent was obtained before recruitment and testing. Samples of stool and blood for serology were collected from all study patients. Blood culture for mycobacteria and other bacterial pathogens was performed on patients according to clinical assessment. Because of resource constraints there was limited access to the general diagnostic microbiology service when the study was conducted, and few stool or blood samples were routinely cultured in the service laboratory. All culture work was therefore carried out in the Kenya Medical Research Institute (KEMRI) microbiology research laboratory. Relevant results were given by the study team to the clinicians caring for the patients in KNH as soon as they were available. Patients were otherwise investigated and treated as was appropriate and in line with local policy, and followed daily or until death. The majority of patients (75%) were given cotrimoxazole with or without metronidazole on admission. CD4 counts are not routinely performed as a service investigation in KNH, and few families or patients were prepared to pay for them. The study had insufficient funds to carry out CD4 counts on all patients recruited. At the time of the study, sigmoidoscopy was not routinely carried out because of the limited capacity of the service, and no biopsies were taken. Post-mortems were occasionally requested, but usually only a few relatives consented. It was concluded that it would not be possible in this study to obtain biopsy or autopsy material in a consistent fashion, so this was not attempted. The study was approved by the Kenya Hospitals National Ethical Committee and the KEMRI research committee. Unless otherwise stated, the investigations described were applied to all patients. All specimens were processed for culture on the day of collection using Oxoid (Unipath, Basingstoke, U.K.) media and incubated in air at 37 °C for 18 h, unless specifically stated. Identification of isolates was by standard bacteriological techniques and confirmation of identification, along with phage typing and serotyping, where appropriate, were carried out by the Public Health Laboratory Service (PHLS), U.K. Culture of stool samples for Salmonella spp. and Shigella spp. was carried out by direct inoculation onto xylose lysine desoxycholate agar (XLD), brilliant green agar (BG) and selenite P broth. After incubation the broth was subcultured onto both XLD and BG agars. Campylobacter blood-free medium, with cefoperazone selective supplement, was directly inoculated and incubated microaerophilically at 37 °C for 48 h before examination. Cefsulodin-irgasan-novobiocin (CIN) medium was used to culture Yersinia spp. after direct inoculation and after cold enrichment in phosphate buffered saline at 4 °C for 14 days. CIN plates were incubated at 30 °C for 24h before examination. Direct inoculation onto thiosulphate citrate bile salt sucrose (TCBS) medium and enrichment in alkaline peptone water were used to culture Vibrio spp. Aeromonas spp. were cultured on Ryan's selective medium with 5 mg/1 ampicillin added. Blood culture was carried out, for 18 patients, in brain-heart infusion broth incubated in COa for up to 7 days. Patients were investigated for mycobacterial infection not as a putative cause of diarrhoea per se, but because a role for tuberculosis has been implicated in slim disease. 9 Mycobacterial culture from stool was in selective Kirchner medium (lOml), following decontamination with 5% sodium hydroxide for 15 min. Blood samples (2 ml) from 17 patients for mycobacterial culture were inoculated directly into Kirchner medium (lOml) at collection. All samples were incubated at 37 °C for up to 6 weeks. Phenol-auramine stained smears were examined directly by fluorescent microscopy for Mycobacterium spp. and Cl~ptosporidium sp. Modified Ziehl-Neelsen (ZN) staining was used to confirm equivocal results. Specimens were examined by light microscopy for ova, cysts and parasites directly and following formalin-ether concentration as wet preparations. The Uvitex 2B fluorescent method was applied to examine for microsporidia in faecal smears from 36 patients. In addition, aliquots of faeces were preserved in 50% (vol/vol) formalin solution and transported to Oxford PHL, U.K. Grids were prepared and examined for enteric viruses under a Phillips 301 electron microscope following negative staining with 1% methylamine tungstate. of clinical features and outcome of hospitalization are shown in Table I . One observer (CM) recruited all patients and 'marked weight loss' reflected the subjective impression of whether the patient was clinically wasted or not. Most patients were unable to state their pre-morbid weight. Thirty-seven (49%) patients reported fever and 18 (24%) were febrile on recruitment or during hospitalization. From stool samples from a random selection of 41 patients, lawns of presumptive E. coli were harvested and shipped to the Laboratory of Enteric Pathogens, Central Public Health Laboratories, U.K. on Columbia agar slopes for detection of pathogenic E. coli using DNA probes directed at the following groups: enteropathogenic E. coli Aliquots of faeces were stored at -70 °C and then transferred to Oxford PHL, U.K. for detection of C. difficile cytotoxin using MRC 5 human fibroblast cell lines. HIV antibody testing was performed using Wellcozyme HIV-1 (Wellcome Diagnostics, Dartford, U.K.) and confirmed using Enzynergost HIV 1 + 2, (Behringwerk AG, Marburg, Germany). One hundred and sixteen adults fulfilled the entry criteria for the study. All patients gave their consent and were entered into the study. Of these, 21 (18%) were found to be HIV-seronegative, three (3%) had equivocal HIV serology and 17 (15%) either had inadequate clinical information recorded or inadequate samples collected to enable any useful analysis to be performed; all were excluded. Results are presented for the remaining 75 patients. The wide range of potential pathogens detected can be seen in Table II . Clostridium difficile toxin, Vibrio spp. and Yersinia spp. were not detected. Light microscopy failed to detect ova, cysts or parasites normally associated with diarrhoea, although the following were detected: Chilomastix mesnili (one), Ascaris Iumbricoides (two), Entamoeba coli (two) and hookwork (two). No cases of Isospora belli were seen. No patient yielded ETEC, EPEC, EIEC or VTEC. One patient had both EAggEC and DAEC. The only Salmonella sp. isolated was S. typhimurium, of which the commonest phage type (PT) was PT 56 (40%). Four of 18 standard blood cultures (22%) were positive. One of 17 mycobacterial blood cultures was positive, yielding Mycobacterium avium-intracellulare. Fifteen patients had multiple pathogens isolated: most had just two, the commonest combinations being S. typhimurium with Cryptosporidium sp. (n=5). Potential pathogens (apart from HIV) were not found during the investigation of 36 (48%) patients. There was no signifcant difference in mortality rate between overall groups of patients yielding pathogens (17/39, 44%) and not yielding pathogens (14/36, 39%) (Table III) . However, mortality was significantly associated with detection of Cryptosporidium cysts. The mortality rate in patients with cryptosporidium cysts was 9/13 (69%) compared with 22/62 (36%) for cryptosporidium-negative patients (X2=5.2; P