Abstract
Objective. We sought to determine the etiologies, manifestations, and risk factors for persistent (⩾7 days) diarrhea in human immunodeficiency virus type 1 (HIV-1)-infected persons in Peru.
Design. The present study is a case-control study of 147 HIV-1-infected case subjects with persistent diarrhea and 147 HIV-1-infected control subjects without diarrhea.
Methods. We obtained clinical, demographic, and exposure data, CD4 lymphocyte counts, and stool samples for detection of enteric parasitic and bacterial pathogens and rotavirus.
Results. One or more enteric pathogen was identified in 55% of case subjects and 21% of control subjects (odds ratio adjusted for CD4 lymphocyte count, 3.8; 95% confidence interval, 2.2–6.5). The median CD4 lymphocyte count was highest with pathogen-free diarrhea and lowest with Cryptosporidium infection. Cryptosporidium species (the most frequent pathogen), Giardia lamblia, Aeromonas species, Campylobacter species, and rotavirus were all significantly associated with diarrhea. Bacterial pathogens were significantly associated with G. lamblia and rotavirus infection. Of the bacterial pathogens (Aeromonas, Campylobacter, Salmonella, and Vibrio species and enterotoxigenic Escherichia coli), only 24% were susceptible to cotrimoxazole, whereas 90% were susceptible to ciprofloxacin. In no case did the sensitivity or positive predictive value of specific clinical and laboratory findings for curable enteric infections exceed 50%.
Conclusions. Several enteric pathogens were associated with diarrhea in HIV-1-infected case subjects in Peru, especially among those who were heterosexual. Clinical findings were poor predictors of detectable microbial etiology. The guidelines for initial management of chronic diarrhea with sulfamethoxazole-trimethoprim in HIV- 1-infected persons require revision, at least in settings where prophylaxis with this agent is common.
In developing countries, persistent (⩾7 days) diarrhea affects up to 95% of persons with AIDS, frequently causing malabsorption, significant weight loss (slim disease), higher rates of extraintestinal opportunistic infections, and increased mortality [1–3]. Studies in developing countries, including Peru, have confirmed very high rates of acute and chronic diarrhea in HIV-1-infected persons ([4–7] and unpublished data, the Peruvian Ministry of Health HIV/AIDS surveillance system), and chronic diarrhea with slim disease in an HIV-1-infected person is an AIDS-defining criterion, according to theWorld Health Organization (WHO) classification [8].
Shigellosis [9], Campylobacter [10] infection, and cryptosporidiosis [4, 11] occur relatively more frequently in HIV-1-infected persons than in persons without HIV-1 infection. Some agents produce diarrhea almost exclusively in HIV-1-infected persons (e.g., Mycobacterium avium complex, cytomegalovirus, and HIV-1 enteropathy) [12, 13]. Others cause more-severe, moreprolonged, or more often recurrent diarrhea in the presence of HIV-1 infection (e.g., Cryptosporidium species [4, 11, 14-]; Isospora species [18]; Salmonella species [19–22]; astrovirus, adenovirus, calicivirus [23, 24], and, perhaps, Microsporidium species [25, 26]; Cyclospora cayetanensis [27]; Shigella species; and Campylobacter species). Some agents apparently have unaltered courses but occur commonly in HIV-1-infected persons (e.g., Clostridium difficile) [28].
With the myriad of etiologies and sometimes-altered natural history of enteric infections in HIV-1-infected persons with diarrhea, it remains uncertain how well clinical manifestations or risk factors predict microbial etiology or are able to guide the empirical choice of therapy for an individual patient. For example, although the presence of right-upper-quadrant pain associated with biliary-tract involvement has also been reported in HIV-1-infected persons infected with Cryptosporidium species [29, 30], the remaining characteristics of persistent diarrhea (e.g., duration, frequency, and characteristics of stools) and gastrointestinal symptoms were similar in HIV-1-infected persons in Zaire with and without Cryptosporidium infection [4].
Few published studies have examined themicrobial etiologies of chronic diarrhea in HIV-1-infected persons in Latin America. Most commonly implicated have been Microsporidium species, found in 29 (33%) of 89 persons with chronic diarrhea in Chile [31], and Cryptosporidium species, found in 5 (11%) of 46 HIV-1-infected persons in Rio de Janeiro [32] and in 8 (20%) of 40 HIV-1-infected persons with persistent diarrhea in another Brazilian study [33]. Among 166 HIV-1-infected persons in Fortaleza, Brazil, only Cryptosporidium and Microsporidium species were associated with diarrhea [34]. The latter study was not restricted to persistent diarrhea.
In Peru, diarrhea is common in persons with AIDS [35], and Cryptosporidium and Isospora species have been found in 34% of persons at the time of diagnosis of AIDS [36]. In Lima, Peru, Salmonella typhi and Salmonella paratyphi infections were more frequent and more severe in HIV-1-infected persons than in control subjects without HIV-1 infection [37]. A 1996 study of 93 HIV-1-infected adults with acute or chronic diarrhea in Lima found C. difficile to be the most common pathogen (29% of cases), followed by Cryptosporidium parvum (23% of cases) [38].
The primary objectives of the present study were to define themicrobial etiologies of persistent diarrhea in HIV-1-infected men and women in Peru and to determine the correlation between the clinical manifestations of persistent diarrhea and microbial etiology. The secondary objectives were to define risk factors for specific microbial etiologies of persistent diarrhea, determine antimicrobial susceptibility of bacterial enteric pathogens, and assess patterns of physicians' management of diarrhea and patients' responses to therapy in relation to microbial etiology.
SUBJECTS, MATERIALS, AND METHODS
Study population and study design. The present study was conducted prospectively between June 1998 and January 2000 at 3 facilities (Santa Rosa II at Hospital Dos de Mayo, Arzobispo Loayza Hospital, and Instituto de Medicina Tropical Alexander Von Humboldt) that provide inpatient and outpatient services to HIV-1-infected persons in Lima. The facilities provide services mostly to persons of low socioeconomic status. Through February 2000, these 3 facilities combined had reported cumulatively 29% of the HIV-1 infections reported in Peru and 42% of the HIV-1 infections reported in Lima.
HIV-1-infected inpatients and outpatients seeking care at these facilities were informed about the study by their physician or by signs posted in the waiting areas. Those qualifying as case or control subjects and willing to participate were recruited after giving written, informed consent.
Case subjects were consecutive adult men and nonpregnant women with confirmed HIV-1 infection and persistent diarrhea. An episode of persistent diarrhea was defined as the occurrence of an average of at least 3 liquid bowel movements daily for at least 1 week before the visit to the clinic. Persons who had received antibiotic/antiparasitic treatment for diarrhea within the past 14 days were excluded; those receiving antibiotics as prophylaxis for opportunistic infections were not excluded. An unmatched control subject was recruited for each case subject from among men and nonpregnant women with confirmed HIV-1 infection who was seeking care at the same facility and who had not had diarrhea or started new antibiotic treatment within the 2 weeks before enrollment.
Case and control subjects underwent a face-to-face interview administered by use of a questionnaire concerning general demographic characteristics, risk factors for HIV-1 infection, previous CD4 lymphocyte counts, suspected duration of HIV-1 infection, medication and mineral supplements taken during the last 2 months, and history of weight loss, opportunistic infections, and other medical conditions. The questionnaire also concerned general living conditions, including sanitation in the home (running water and access to a toilet), and other factors that may increase the risk of infectious diarrhea. In addition, the presence of symptoms related to enteritis and/or colitis was recorded for the case subjects. Case subjects were asked to identify diagrams that best depicted the form of their stools [39] and to rate the frequency of each form recognized.
After completing the questionnaire, both case and control subjects provided a blood sample in an anticoagulated, trace element-free tube and were given 3 containers to be returned with 1 fresh stool sample/day for 3 days. Subjects' clinical records were subsequently reviewed to determine the treatment prescribed by the physician (treatment was not started until after the interview and first stool specimen were obtained), and subjects were asked to return for follow-up evaluation by the study team in 2 weeks.
The University ofWashington Human Subjects Review Committee and local institutional review boards at Cayetano Heredia University, Dos de Mayo Hospital, and PRISMA (a local nongovernmental organization) approved the study protocol. All study subjects gave written, informed consent.
Laboratory studies. The 3 fresh stool specimens were transported to the pathology laboratory at the Universidad Peruana Cayetano Heredia, for studies to determine the presence of inflammatory cells, blood, and parasites. Aliquots of the first specimen for each subject were transported frozen and in swabs in Cary-Blair transport media, for C. difficile toxin assays, rotavirus ELISAs, and cultures for pathogenic bacteria.
Inflammatory cells were quantitated in stool samples by light microscopy in methylene blue-stained smears, and scores were assigned according to the average leukocyte count in 50 fields at ×40 magnification (1–50, 51–100, and >100 leukocytes). The presence of occult blood was determined by use of the Hemoccult Sensa kit (Beckman Coulter), in accordance with the manufacturer's protocol.
To determine the presence of parasites, formalin-ether-concentrated stool samples were examined by light microscopy. Additionally, all samples were subjected to a modified acid-fast stain [40] and the Baermann technique for Strongyloides stercoralis [41]. A modified trichrome preparation was used to detect Microsporidium species, in accordance with a published protocol [42]. If microscopy results suggested the presence of C. cayetanensis, modified safranin staining [43] was performed to confirm the finding. For Cryptosporidium and Microsporidium species, scores were assigned according to the load of parasites in 50 fields at ×100 magnification (1–50, 51–100, and >100 parasites).
For detection of C. difficile toxin in frozen stool samples, the Premier C. difficile Toxin A EIA test (Meridian Diagnostics) was used. For detection of rotavirus, the Premier Rotaclone EIA kit was used (Meridian Diagnostics).
Selective and differential media were used for detection of enteric pathogens from the Cary-Blair transport media. All suspicious colonies were identified by standard bacteriology procedures and/or serological testing for enteric pathogens. For detection of Vibrio species, specimens were plated onto thiosulfate citrate bile sucrose medium and enriched with alkaline peptone water. For the isolation of Salmonella, Shigella, Aeromonas, Campylobacter, and Plesiomonas species, samples were inoculated onto Hektoen Enteric and xylose-lysine-desoxycholate agars, Butzler's medium in an anaerobic environment, and MacConkey medium. Selenite F enrichment broth was used to enhance recovery of Salmonella and Shigella species [44]. To detect enterotoxigenic Escherichia coli (ETEC), lactose-fermenting Escherichia coli isolates from the MacConkey medium were tested for production of heat-labile (LT) and heat-stable (ST) 1a and ST 1b toxins, by gene probe hybridization. Suspicious isolates were confirmed by use of P32-labeled probes [45].
Antibiotic susceptibility of bacterial pathogens was determined by the Kirby-Bauer disk diffusion technique [46]. Isolates were tested for susceptibility to sulfamethoxazole-trimethoprim (SMX-TMP), ciprofloxacin, norfloxacin, erythromycin, cephalothin, doxycycline, tetracycline, gentamicin, kanamycin, chloramphenicol, and nalidixic acid.
Blood samples were collected for determination of CD4 lymphocyte counts by standard flow cytometry, by use of a FACScan (Becton Dickinson) cytometer. Measurement of C-reactive protein (CRP) was preformed in serum that previously had been frozen, by use of a high-sensitivity latex-enhanced immunonephelometric method, with reagents provided by the manufacturer (Dade Behring Diagnostics), in a Clinical Laboratory Improvement Amendments-certified clinical laboratory using standard quality-control techniques. Copies of all stool and blood test results were given to the subjects and their physicians as they became available.
Analysis. To test for the significance of associations, the χ2 test, Fisher's exact test, and Student's t test were used in univariate analyses. Unconditional logistic regression was used in multivariate models to adjust for effects of covariates.
To examine correlations between clinical manifestations and the microbial etiology of persistent diarrhea, all clinical variables were ranked according to their positive predictive value (PPV) for each specific etiology. Sensitivities for the predictors with the best PPV were also calculated.
RESULTS
A total of 322 persons consented to participate and responded to the baseline questionnaire; 28 provided no stool sample and were excluded from the analysis. Thus, 147 case subjects and 147 control subjects were included in the final analyses. No demographic differences were found between subjects excluded and those included.
Subject characteristics. Median (range) ages were 32 (19–63) years for case subjects and 31 (19–64) years for control subjects. Men accounted for 66% of case subjects and 69% of control subjects. Case subjects were significantly less likely than control subjects to have completed high school (18% vs. 31%; P = .01), have running water in the home (75% vs. 85%; P = .04), and live in a house with a toilet or a shared toilet (80% vs. 93%; P = .002).
Most subjects (79%) had received prophylactic antibiotics during the 2 months before enrollment. The most commonly used antibiotic was SMX-TMP, which was received by 66% of the subjects as a prophylaxis for Pneumocystis jirovecii (formerly Pneumocystis carinii) pneumonia. Among subjects with a CD4 lymphocyte count <200 cells/mm3, SMX-TMP prophylaxis was used by 95 (79%) of 121 with diarrhea and 52 (68%) of 76 without diarrhea (P = .16); among subjects with a CD4 lymphocyte count ⩾200 cells/mm3, SMX-TMP prophylaxis was used by 16 (67%) of 24 with diarrhea and 30 (42%) of 71 without diarrhea (P = .07). This latter difference may reflect the fact that, in Peru, chronic diarrhea represents an indication for SMX-TMP prophylaxis against P. carinii infection in HIV- 1-infected persons who have not undergone determination of CD4 lymphocyte count.
Current tuberculosis treatment and prophylaxis were reported by 9%and 37% of case and control subjects, respectively. Recent use of antimycotics and of antiretroviral drugs was reported by 8% and 9% of case and control subjects, respectively. No differences in reported use of antituberculosis, antimycotic, and antiretroviral drugs were observed between case and control subjects. For 98 control subjects who could estimate the date of their last episode of diarrhea, the median time from the last episode of diarrhea to enrollment was 70 days (range, 15 days to 18 months), and 60% of these control subjects reported at least 1 episode of diarrhea during the 12 months before enrollment.
Microbiological results.Giardia lamblia and/or Cryptosporidium species in stool samples were strongly associated with diarrhea (adjusted odds ratio [OR], 5.7; 95% confidence interval [CI], 1.7–18.8, for G. lamblia, and adjusted OR, 2.7; 95% CI, 1.2–5.9, for Cryptosporidium species; table 1). Aeromonas species, Campylobacter species, and rotavirus were also significantly associated with diarrhea, as was the presence of ⩾1 bacterial enteric pathogen, irrespective of type. Fourteen case subjects (10% of all case subjects and 18% of subjects infected with a detectable pathogen) had multiple infections; detection of bacterial infections was significantly associated with detection of G. lamblia (P = .002) and rotavirus (P = .001). Commensal parasites found included Entamoeba coli (60 subjects), Chilomastix mesnili (30 subjects), Endolimax nana (22 subjects), and Iodamoeba butschlii (1 subjects). Of these, Entamoeba coli and E. nana were actually associated with the absence of diarrhea (P = .004 and P = .01, respectively); the others had no association with diarrhea.
Of 25 subjects reporting use of antiretrovirals during the 2 months before enrollment, 10 were enrolled as case subjects and 15 were enrolled as control subjects. Pathogens isolated from the 10 case subjects included Giardia species (2 subjects), rotavirus (1 subject), Campylobacter species (2 subjects), and Cryptosporidium species (4 subjects). No enteric pathogens were isolated from control subjects. Although most subjects (82%) turned in the 3 stool samples requested for parasite studies, most parasite infections were detected in the first stool sample obtained (e.g., 91% for giardiasis and 77% for cryptosporidiosis).
Antimicrobial susceptibility testing was performed for 42 of the 43 bacterial strains isolated from case and control subjects and demonstrated susceptibility to SMZ-TMP for only 10 (24%) of 42 bacterial pathogens isolated from HIV-1-infected subjects with and without diarrhea, including 3 (16%) of 19 Shigella species, 1 (14%) of 7 Aeromonas species, 0 of 4 Campylobacter species, 4 (50%) of 8 ETEC, 1 (33%) of 3 Salmonella species, and 1 (100%) of 1 Vibrio alginolyticus. However, among Shigella, Salmonella, and Aeromonas species and ETEC, all but 1 were susceptible to ciprofloxacin. All Campylobacter species were susceptible to doxycycline, with 3 (75%) of 4 also susceptible to erythromycin.
Clinical, laboratory, and risk predictors of enteric infection. Any enteric pathogen was identified among 8 (35%) of 23 case subjects who identified themselves as men who have sex with men (MSM) and 71 (57%) of 124 heterosexual case subjects (P = .08) and was identified among 3 (10%) of 29 control subjects who identified themselves as MSM and 28 (24%) of 118 heterosexual control subjects (P = .18). The differences between case and control subjects, for specific infections, were evident irrespective of sexual orientation.
To explore differences in clinical and laboratory manifestations caused by different enteric pathogens among case subjects, subjects were first grouped as having either parasitic, bacterial, viral (rotavirus), or pathogen-free diarrhea (table 2). No clear pattern of clinical manifestations emerged for any category of pathogen. For parasitic infections, upper-abdominal pain had the highest PPV (42%) but was present in only 39% of case subjects with parasite-associated diarrhea. Similarly, although 43% of case subjects with blood detected in a stool sample had enteric bacterial infections, blood was detected in stool samples from only 48% of subjects with bacterial infection. Seven (22%) of 32 case subjects with any bacterial infection, compared with 6 (9%) of those with no pathogen detected, reported having ducks at home (P = .16).
Mean (range) plasma zinc concentrations were 146 (25–194) mg/dL for case subjects and 146 (15–179) mg/dL for control subjects (P = .33). Mean CRP values were 3.6 mg/dL for case subjects and 1.1 mg/dL for control subjects (P < .0001). However, mean CRP values were similar for case subjects with an enteric infection (3.8 mg/dL) and case subjects without enteric infection (3.5 mg/dL) (P = .72).
Peripheral blood leukocyte counts and hemoglobin concentration did not predict enteric infection. Neutrophils were present somewhat more often in stool samples from case subjects with enteric bacterial infection (17 [53%] of 32) than in stool samples from case subjects without enteric bacterial infection (38 [34%] of 111) (P = .07). Among stool samples from case subjects with parasitic infections, fecal neutrophils were present in 32% with Cryptosporidium species only, 31% with Giardiasis species, and 50% with Microsporidium species (P, not significant). Blood was detected in stool samples from 13 (48%) of 27 subjects with bacterial enteric infection and 17 (18%) of 92 subjects without such infection (P = .004).
Similar analyses were then performed for the most common specific parasitic and bacterial agents detected among case subjects (Cryptosporidium species, G. lamblia, and Shigella species). Again, PPVs and sensitivity for the best predictors were low: report of pigeons at home, for Cryptosporidium infections (PPV, 40%; sensitivity, 14%); upper abdominal pain, for G. lamblia infections (PPV, 21%; sensitivity, 50%); and report of ducks at home, for Shigella infections (PPV, 22%; sensitivity, 33%).
CD4 lymphocyte counts ranged from 0 to 2021 cells/mm3 (mean, 115 cells/mm3) in case subjects and from 1 to 1537 cells/mm3 (mean, 243 cells/mm3) in control subjects (P <.0001). As shown in figure 1, median CD4 lymphocyte counts and 25%–75% (interquartile) ranges were lower for subjects with diarrhea associated with any pathogen (mean CD4 lymphocyte count, 122 cells/mm3) than for subjects with diarrhea not associated with a pathogen (mean CD4 lymphocyte count, 216 cells/mm3) (P < .0001), and CD4 lymphocyte counts were significantly lower for the 38 subjects with Cryptosporidium infection (mean CD4 lymphocyte count, 99 cells/mm3; 89% of this group had CD4 lymphocyte counts <150 cells/mm3) than for those without such infection (P = .002).
Adequacy of therapy prescribed. Data on antibiotics prescribed could be retrieved from the clinical records of 82 (56%) of the case subjects included in the analysis. Among these 82 case subjects, evidence of prescription of metronidazole was obtained from only 3 of 7 case subjects with Giardia infection, and evidence of prescription of an effective antibiotic (active in vitro against the pathogen isolated) was obtained from 8 of 19 case subjects with enteric bacterial infection. On the basis of further information from interviews of 104 case subjects returning for their follow-up visit, only 1 case subject with Giardia infection and 1 case subject with enteric bacterial infection actually received adequate treatment (even those prescribed an effective antimicrobial often failed to obtain or take the medication).
DISCUSSION
Our objectives were to determine the etiology, manifestations, and risk factors for persistent diarrhea in HIV-1-infected persons in Lima, as well as to define patterns of treatment and the antimicrobial susceptibilities of the enteric bacterial pathogens encountered. Enteric pathogens were identified in 21% of HIV-1-infected control subjects without diarrhea and 55% of HIV- 1-infected case subjects with diarrhea. Tests to detect a wide range of microbial pathogens were performed, but no histopathological studies were possible to document cytomegalovirus infection or other gastrointestinal diseases in this population. Specific pathogens associated with diarrhea included Cryptosporidium species, G. lamblia, several bacterial pathogens, and rotavirus. Most clinical manifestations correlated rather poorly with etiology; neutrophils and blood in stool samples were best correlated with isolation of bacterial pathogens, but both were absent in stool samples from 52% of subjects with bacterial enteric infections. Most bacterial pathogens were resistant to SMZ-TMP. Surprisingly, enteric pathogens were detected somewhat less frequently among MSM than among heterosexuals.
A 1996–1997 study of 93 HIV-1-infected patients with diarrhea at a hospital in Lima found that C. difficile toxin and Cryptosporidium species were the most common microbial etiologies identified [38]. Some parasites, such as Microsporidium species, Isospora belli, and C. cayetanensis, were also found with higher frequency in that study than in the present one. Unlike the previous study, in which rotavirus was found to be more common among diarrhea-free control subjects, our study found no rotavirus infection among control subjects. Differences between the 2 studies could be related to differences in the sensitivity or specificity of the tests performed, differences in inclusion criteria for case subjects (only our study was restricted to persistent diarrhea), or real differences in the characteristics of the HIV-1-infected populations seeking care at the study sites. Evidence for the latter is the widespread use of SMXTMP as prophylaxis for PCP, which has been subsidized by the Peruvian Ministry of Health since 1997.
WHO guidelines for the management of chronic diarrhea in HIV-1-infected adults [47] recommend initial therapy with SMX-TMP given empirically or when an enteric bacterial infection is suspected or confirmed. The high prevalence of resistance to SMX-TMP in the bacterial strains that we isolated, along with the high prevalence of G. lamblia infections and the relatively low prevalence of Cyclospora and Isospora infections, suggests that these algorithms are not optimal for Peru. In settings where SMX-TMP prophylaxis is widely used in HIV- 1-infected patients, SMX-TMP may not be the first option for the treatment of diarrhea in persons who have received such prophylaxis. Most bacterial pathogens that we found remained susceptible to ciprofloxacin, which is also widely available in Peru. The inclusion of metronidazole as a first-line drug for patients unable to return to the clinic frequently or as a treatment for those who do not respond to fluoroquinolones should also be considered, in view of the relatively high frequency of giardiasis. Further studies will be necessary to evaluate the effect of revised management algorithms.
In summary, our study has prospectively assessed, in a standardized format, the clinical and microbiological correlates of persistent diarrhea in HIV-1-infected persons. Although some clinical and epidemiological findings were detected somewhat more often for subjects with specific infections and although the presence of neutrophils or blood in stool samples was associated with enteric bacterial pathogens, no signs or symptoms were adjudged to be sufficiently predictive of particular enteric infections to determine who should and should not receive antimicrobials. Furthermore, several case subjects had multiple infections. Thus, a common management algorithm including ciprofloxacin and metronidazole may warrant further evaluation for initial syndromic management of diarrhea in HIV-1-infected persons in settings where laboratory tests are not routinely available and early follow-up for test results is impractical.
Acknowledgement
We wish to thank Roger Glass (Viral Gastroenteritis Unit, Centers for Disease Control and Prevention) for his support with tests for detection of rotavirus. We would also like to thank the US Navy Medical Research Center Detachment, Lima, Peru, for their assistance and support.
References
Figures and Tables
Box plots of CD4 lymphocyte count, by the pathogen identified. The circles inside the boxes represent the median, the boxes represent the 25%-75% (interquartile) range, and the vertical lines (whiskers) represent the 5%-95% percentile range of CD4 lymphocyte counts for each category. ETEC, enterotoxigenic Escherichia coli.
Box plots of CD4 lymphocyte count, by the pathogen identified. The circles inside the boxes represent the median, the boxes represent the 25%-75% (interquartile) range, and the vertical lines (whiskers) represent the 5%-95% percentile range of CD4 lymphocyte counts for each category. ETEC, enterotoxigenic Escherichia coli.
Enteric microbial pathogens identified in HIV-1-infected case subjects with persistent diarrhea and HIV-1-infected control subjects without diarrhea.
Enteric microbial pathogens identified in HIV-1-infected case subjects with persistent diarrhea and HIV-1-infected control subjects without diarrhea.
Demographic, behavioral, sanitation, clinical, and laboratory findings, by category of pathogen.
Demographic, behavioral, sanitation, clinical, and laboratory findings, by category of pathogen.
Financial support: Fogarty International Center (International AIDS Research and Training Program grant D43 TW00007-06); University of Washington (UW) Center for AIDS Research; UW School of Medicine Anderson Foundation; Centers for Disease Control and Prevention.
