Journal of Infection (2002) 45: 99±106

doi:10.1053/jinf.2002.1002, available online at http://www.idealibrary.com on

Diarrhoea, CD4 Counts and Enteric Infections in a

Community-Based Cohort of HIV-Infected Adults in Uganda
A.-K. Brink1, C. MaheÂ1, C. Watera1, E. Lugada1, C. Gilks2,
J. Whitworth1 and N. French*2
1
MRC Programme on AIDS, Uganda and 2Liverpool School of Tropical Medicine, Liverpool, UK

Objectives: To examine relationships between diarrhoea, CD4 cell counts and stool pathogens in a community-based
cohort of HIV-infected adults in Uganda.
Patients and methods: Stool specimens, obtained between October 1995 and December 1997, were linked to patients'
symptoms and laboratory results. The relationship between CD4 counts and symptoms was tested using the Wilcoxon
rank-sum test and those between organisms and diarrhoea using first a univariate Mantel±Haenszel analysis and then
a logistic regression model adjusted for CD4 count and multiple organisms.
Results: 1213 HIV-infected individuals (70% women, median CD4 cell count at enrollment 215 cells/ml) were followed
for 1224 person years of observation (pyo). 484 stool samples were examined, 357 from patients with diarrhoea. The
rate of diarrhoea was 661 episodes per 1000 pyo. CD4 counts were significantly lower in individuals with diarrhoea
than those without (P < 0.001, Wilcoxon rank-sum test). Forty-nine percent of diarrhoeal stools and 39% of stools
from asymptomatic patients contained enteric pathogens. The most frequent isolates were helminths (29.5% of all
stools), followed by bacteria (19.2%) and then protozoa (8.9%). Rates of isolation of diarrhoea-associated pathogens
were 29% from diarrhoeal stools and 17% from asymptomatic stools (P ˆ 0.01, x2 test). The association between
diarrhoea and infection with bacteria or protozoa was weak and there was no association with helminths.
Cryptosporidium parvum infection alone was associated with low CD4 counts.
Conclusions: Diarrhoea was common and most strongly associated with low CD4 counts. Bacteria were frequently
found, even in stools from asymptomatic individuals. Over two-thirds of diarrhoeal episodes were undiagnosed,
suggesting that unidentified agents or primary HIV enteropathy are important causes of diarrhoea in this population.
# 2002 The British Infection Society

Introduction pathogens in industrialised settings. These studies both

demonstrated a strong negative association between
Diarrhoea is a significant cause of morbidity in people
diarrhoea and CD4 counts. Rates of infection with pro-
infected with the human immunodeficiency virus (HIV)
tozoan parasites were higher than with bacteria, and
both in industrialised nations and particularly in Africa
many stool samples were not found to contain any
[1,2]. Numerous cross-sectional studies in Africa, mainly
enteric pathogens. Although exposure to enteric patho-
of selected patient groups, have provided descriptive data
gens is far higher in developing countries, no prospective
on the causes of diarrhoea and isolation rates of parti-
studies have been reported. We now report data on the
cular enteric pathogens, concentrating on chronic diar-
rate of diarrhoea and the associations between diar-
rhoea [3±11]. Much less information is available on the
rhoea, CD4 counts and enteric pathogens from a pro-
epidemiology of diarrhoea and its associated risk factors.
spective community-based cohort study conducted in a
Recently, the results of two prospective community-
semi-urban population in Entebbe, Uganda.
based cohort studies conducted in the USA [12] and in
Switzerland [13] have been published, giving more
information on rates of diarrhoea and isolation of enteric Patients and Methods

Patients
* Please address all correspondence to: N. French, Liverpool School of
Tropical Medicine, Pembroke Place, Liverpool L3 5QA, UK. Tel.: ‡44- Stool specimens were obtained between October 1995
151-708 9393; E-mail address: nfrenc@aol.com (N. French). and December 1997 from HIV 1-infected adults (aged

0163±4453/02/$35.00 # 2002 The British Infection Society

100 A.-K. Brink et al.
15 years and above) participating in a community-based
pneumococcal vaccine efficacy trial [14], centred in
Entebbe on the shores of Lake Victoria, Uganda, a semi-
urban setting. In this population, HIV infection is almost
exclusively transmitted by heterosexual contact. Patients
attended one of two clinics (The AIDS Support Organi-
sation of Uganda (TASO), Entebbe, and the Ministry of
Health clinic, Uganda Virus Research Institute (UVRI))
for routine follow-up visits at six monthly intervals,
when clinical staging using the WHO staging system
[15] and CD4 cell counts were performed, and also at
any intervening time if they were symptomatic (interim
visits). The study was approved by the Uganda Virus
Research Institute Science and Ethics Committee, the
Uganda National Council for Science and Technology
AIDS Subcommittee and by the ethical review board of
the Liverpool School of Tropical Medicine.
Definition of diarrhoea
Symptoms reported by patients were used to classify
diarrhoea into acute and chronic episodes. Diarrhoea
was defined as the passage of three or more unformed
stools in 24 h. An episode of diarrhoea was classified as
acute if it lasted for less than one month and the patient
had not had any diarrhoea in the preceding month. An
episode was defined as chronic when diarrhoea lasted
continuously for a month or more or was intermittent
and recurrent over a period of at least two months, with
diarrhoeal symptoms for at least half this time. Sub-
sequent diarrhoea was classified as a new episode if there
was a diarrhoea-free interval of at least one month. If the
duration of symptoms did not fit any of these definitions
the case was excluded from the analysis.
Sample collection
Stool samples were requested from patients who pre-
sented with diarrhoea at the time of consultation
whenever possible or as soon as possible thereafter.
Information on the duration of symptoms of diarrhoea
was recorded on the laboratory request form completed
by the clinician seeing the patient. These laboratory
forms were the primary data source. In those cases
where the information was incomplete it was retrieved
retrospectively from the case notes without prior knowl-
edge of the stool sample result or the CD4 count of the
patient. Stool samples were also requested from ran-
domly selected patients on routine follow-up visits who
had had no diarrhoea in the previous month in order to
look at the prevalence of asymptomatic infection.
Microbiology
Stool samples were examined by microscopy of direct
smears and after formol±ether concentration of an
aliquot. Cultures were set up on the same day and
incubated at 37 C on xylose lactose desoxycholate (XLD)
agar, MacConkey and Campylobacter-selective agar and
in Selenite broth, which was subcultured after 24 h.
Salmonella spp., Shigella spp., and Campylobacter spp.
were identified according to standard protocols. Cultures
for enterohaemorrhagic Escherichia coli were performed
on Sorbitol MacConkey agar if stool was bloody on visual
inspection. Parasites were detected by microscopy only,
culture for Strongyloides stercoralis was not routinely
performed. Cryptosporidium parvum was examined for
after staining with the modified Ziehl±Neelsen method.
Repeat samples positive for C. parvum were excluded
from the rate of isolation computation because in this
population of severely immunocompromised individuals
they probably represented persistent rather than new
infections. Protozoa not believed to be human pathogens
were not reported. The methods used in this laboratory
did not allow the detection of Clostridium difficile, Micro-
sporidia spp. or viruses and samples were not cultured for
the presence of mycobacteria. Satisfactory performance
by the laboratory was attained in the UK National
external quality assurance scheme (NEQAS).
Haematology
CD4 counts were measured routinely at six monthly
follow-up visits using a FACS count# system (Becton
Dickinson, San JoseÂ, California, USA). For this study,
each time a participant provided a stool sample their
most recent CD4 count, nearest in time either side of the
stool sample within a time limit of six months, was used
for the analysis. Stool samples from patients from whom
no CD4 count within six months of taking the sample
was available, were excluded from the analysis.
HIV serology
HIV-1 status was confirmed in all participants by posi-
tivity on dual EIAs (Recombigen1 HIV-1/2, Trinity
Biotech plc., Bray, Co Wicklow, Ireland and Wellcozyme*
HIV Recombinant, Murex Biotech Ltd., Dartford, UK).
Statistics
Analyses were performed using STATA statistical soft-
ware version 6.0 (Stata Corporation, College Station,
Texas, USA). The relationship between the CD4 count of
 Diarrhoea and Enteric InfectionÐUganda
the patient who provided the stool sample and diarrhoeal
symptoms was assessed using the Wilcoxon rank-sum
test. A univariate Mantel±Haenszel analysis was per-
formed to look at the association between organisms
isolated and diarrhoea. An adjusted analysis, using a
logistic regression model, was then performed in order to
assess whether or not organisms were linked to symp-
toms of diarrhoea, incorporating patients' CD4 counts
(in two groups: <200 and 200 cells/ml) and the pre-
sence of other organisms in the same stool sample. In
this logistic model, a robust variance-covariance matrix
was used in order to take into account the correlation
between stool samples pertaining to the same patient.
Results
The study period was from October 1995 to December
1997. During this time, 1213 HIV-infected participants
were enrolled in the pneumococcal vaccine cohort, con-
tributing 1224 person years of observation (pyo). The
characteristics of the cohort at enrollment are shown
in the first column of Table I and show a high proportion
of advanced HIV disease.
In order to establish a crude estimate of the rate of
diarrhoea in the cohort we looked at clinic attendance
records for 870 participants attending the TASO clinic
between June 1996 and December 1997 (contributing
799 pyo). During this period, 346 cohort members made
695 clinic visits with 528 distinct episodes of diarrhoea,
giving a crude rate of 661 diarrhoeal episodes (both
acute and chronic) per 1000 pyo.
Samples and symptoms
During the study period, 549 stool samples were col-
lected, 65 samples (12%) were excluded from the ana-
lysis. The breakdown of stools by symptoms and reasons
for exclusion are shown in Fig. 1. A total of 484 stool
samples suitable for analysis, taken from 358 partici-
pants, remained.
Table I. Enrollment characteristics of the whole cohort and of the subset of participants who gave stool samples.
Patient characteristics Whole cohort
n ˆ 1213
Median age (IQ range) 30 (26±36)
Number of women (%) 847 (70)
Median CD4 count (IQ range) 215 (70±454)
% in WHO stages III ‡ IV 56
a
P < 0.001 (Wilcoxon rank-sum test) for both comparison with the whole cohort and with the asymptomatic participants.
b
P < 0.001 (2 test) for comparison with the whole cohort, P ˆ 0.003 for comparison with the asymptomatic participants.
101
The 357 samples from the 251 patients with diar-
rhoea represented 48% of episodes of reported diarrhoea.
Those patients with diarrhoea (acute or chronic) who
gave stool samples were not different in terms of age, sex,
CD4 count or WHO stage to those others in the cohort
who had diarrhoea but from whom no stool sample was
obtained (data not shown). Of the patients who gave
stool samples during an episode of acute diarrhoea only
6% developed chronic diarrhoea within the next two
months, suggesting that a minority of acute diarrhoeal
episodes marked the onset of a chronic illness.
The characteristics of the participants who provided
stool samples are shown in Table I. The CD4 counts at
enrollment were significantly lower for individuals who
subsequently experienced diarrhoea and gave stool
samples, when compared to the whole cohort or the
asymptomatic participants who provided stool samples.
Similarly, a significantly higher proportion of patients
with diarrhoea had been in WHO clinical stages III and
IV at enrollment.
Relationships between CD4 Counts,
Symptoms and Enteric Pathogens
CD4 counts and diarrhoea
As at enrollment, CD4 counts among the study partici-
pants providing stool samples were significantly lower in
those with diarrhoea than in those without diarrhoea
(P < 0.001, Wilcoxon rank-sum test). Patients with
chronic diarrhoea had lower CD4 counts than those with
acute diarrhoea (P < 0.001, Wilcoxon rank-sum test),
see Table II.
Laboratory isolation of organisms
One or more potentially pathogenic organisms
were isolated from 46% (224/484) of the stool
samples: 138 samples contained only one organism,
Patients with diarrhoea Asymptomatic participants
who gave stool samples who gave stool samples
n ˆ 251 n ˆ 107
30 (25±36) 30 (26±37)
159 (63) 75 (70)
118 (28±294)a 292 (69±541)
70b 53
102 A.-K. Brink et al.

Figure 1. Profile of stool samples and symptoms. Numbers in bold font show the total number of stools from n ˆ number of individuals.

63 samples two, 18 three and five samples four organ-

Table II. CD4 counts, measured within six months of stool collection,
isms, yielding a total of 338 organisms identified. Hel- according to symptoms.
minths occurred most frequently, 190 isolates were
found in 143 of the 484 stools; followed by bacteria, 104 Symptoms Number Median CD4 IQ range % stools from
in 93 samples. Protozoa were least common, 44 were of stools count patients with
(cells/ml) CD4 count
found in 43 stools. One or more organisms were found in <200 cells/ml
49% of stools from patients with diarrhoea (rates were
similar between acute and chronic diarrhoea) and in Asymptomatic 127 267 (61±537) 46
39% of stools from patients without diarrhoea. Rates All diarrhoea 357 71 (20±205) 75
Acute 165 115 (38±275) 66
of isolation are presented in Table III. Isolation of Chronic 192 50 (10±144) 82
diarrhoea-associated pathogens (Campylobacter spp.,
 Diarrhoea and Enteric InfectionÐUganda 103
Table III. Rate of isolation of organisms and their association with diarrhoea.

Organism Rate of isolation % of stools containing % of stools containing

in stool samples %, organism amongst organism amongst
n ˆ 484a patients with diarrhoea patients without diarrhoea

Bacteria 19.2 21.0 14.2

Campylobacter spp. 3.5 3.1 4.7
Non-typhi Salmonella spp. 8.1 9.2 4.7
Shigella spp.b 9.5 10.9 5.5
Protozoa 8.9 10.1 5.5
Cryptosporidium parvum 5.0 5.9 2.4
Giardia lamblia 1.9 2.2 0.8
Isospora belli 0.8 1.1 0
Entamoeba histolyticac 1.4 1.1 2.4
Helminths 29.5 29.1 30.7
Schistosoma mansoni 6.8 6.7 7.1
Strongyloides stercoralis 11.4 12.0 9.5
Ascaris lumbricoides 2.9 3.4 1.6
Hookworm 16.1 14.9 19.7
Trichuris trichiura 1.4 1.7 0.8
Othersd 1.0 Ð Ð

a
Stool samples may contain more than one organism.
b
Including Shigella flexneri and sonnei, no S. dysenteriae was isolated.
c
Cysts only.

Table IV. Unadjusted and adjusted odds ratios for the association between diarrhoea, organisms and CD4 counts.

Organism Unadjusted 95% confidence P value Adjusteda 95% confidence P value

odds ratio interval odds ratio interval

Bacteria 1.6 (0.9±2.8) 0.09 1.8 (1.0±3.3) 0.04

Cryptosporidium parvum 2.6 (0.8±8.2) 0.12 1.8 (0.5±6.3) 0.35
Other protozoa 4.4 (0.6±34) 0.12 4.5 (0.5±40.1) 0.18
Helminths 0.9 (0.6±1.4) 0.74 0.9 (0.6±1.5) 0.71
CD4 count <200 cells/ml 3.5 (2.3±5.3) <0.001 3.4 (2.2±5.3) < 0.001

a
Logistic regression adjusted for the presence of multiple organisms in some stools and CD4 count in two groups: <200 and 200 cells/ml.

non-typhi Salmonella spp., Shigella spp., C. parvum, Giardia Symptoms and enteric pathogens
lamblia and Isospora belli) was 26% from all stools; 29%
The association of bacteria and protozoa with diarrhoea
from diarrhoeal stools and 17% from asymptomatic
was weak in this population. No association with diar-
stools (P ˆ 0.01, 2 test).
rhoea was shown for helminths, see Table IV.

CD4 counts and enteric pathogens Discussion

Infection with C. parvum was associated with very low
CD4 counts: the median CD4 count of patients who gave
stools containing C. parvum was 16 cells/ml (IQ range
5±49) compared with 91 cells/ml (IQ range 21±276) for
patients giving stools containing any other organism.
C. parvum was treated separately from other protozoa in
subsequent analyses of associations between diarrhoea
and infecting organisms.
Diarrhoea is well recognised as an important component
of HIV-related morbidity. Despite this, reports from
cohort-based studies of diarrhoeal disease in sub-
Saharan Africa have been lacking. This study, in a
community-based cohort of HIV-infected adults living in
a semi-urban setting in Uganda, has produced a number
of important findings. It has confirmed that diarrhoea is
common and that there is a strong association with low
104 A.-K. Brink et al.
CD4 counts, particularly for chronic diarrhoea. Rates of
identification of organisms from stool samples were
broadly similar to those found in other African studies,
however, there were notable differences: I. belli was
infrequently found, whereas diarrhoea-causing bacteria
were responsible for one-fifth of all episodes; helminth
infections were not associated with diarrhoea.
The cohort consisted of individuals who had come
forward voluntarily for HIV-testing, often as a result of
HIV-related symptoms. That the majority of people were
already in a relatively advanced state of HIV infection
was confirmed by clinical staging of infection and by
measuring CD4 counts at enrollment. In Uganda, the
prevalence of HIV infection is slightly higher among
women than men [16] but our cohort had a somewhat
greater excess of women. Thus, early HIV infections and
males were under-represented in this study.
Although few cohort studies are available for
direct comparison, our rate of 661 diarrhoeal episodes
per 1000 pyo is substantially greater than the rate of
142 diarrhoeal episodes per 1000 pyo in the
Swiss cohort [13] in which patient attendance (every
six months and at interim visits with diarrhoea), case
definition and median CD4 count of the cohort at base-
line (230, IQ range 80±410 cells/ml) were similar to
those in our study. This finding supports the view that
HIV-associated diarrhoeal disease occurs more com-
monly in Africa than in industrialised nations. Our rate
represents a minimum estimate of the burden of diar-
rhoea, particularly of acute diarrhoea, as episodes were
only counted if a patient came to the clinic. Diarrhoea
which was self-limiting or which responded to over-the-
counter medication, would not have come to our
attention.
Chronic diarrhoea in particular, but also acute diar-
rhoea, was strongly associated with lower CD4 counts.
This pattern of morbidity was expected based on
knowledge of the natural history of HIV in the developed
and developing world. Nevertheless, a fifth of chronic
diarrhoeal episodes and one-third of acute episodes
were experienced by patients with CD4 counts of
200 cells/ml, emphasising that, in this population,
diarrhoea is a frequent problem even in less advanced
stages of HIV-associated immunosuppression. Appro-
priate provision should be made for the management of
diarrhoea when planning HIV-care interventions.
An estimated 48% of all reported diarrhoeal episodes
in the cohort were investigated by examination of a stool
sample. Reasons for not having investigated more
episodes include the following: some clients would have
been unable or unwilling to provide a sample at the time
of the original consultation; only those whose symptoms
persisted or who lived very near to the clinics would have
brought a sample later during the same episode of diar-
rhoea; and there may have been variations in the practice
of requesting stool samples between different clinicians.
However, there were no differences in age, sex, CD4 count
or clinical stage between the patients with diarrhoea
who did and did not give stool samples, which suggests
there was no clear selection bias leading to dispropor-
tionate representation of particular groups of individuals.
An identifiable cause of diarrhoea was established for
the minority of episodesÐ71% of diarrhoeal stools were
not found to contain any diarrhoea-associated pathogen.
A number of factors are likely to have contributed to an
underestimation of the true prevalence of pathogenic
organisms in stool samples: There is widespread com-
munity use of drugs, including co-trimoxazole, metroni-
dazole and mebendazole, for the treatment of diarrhoea
and some patients would have received such therapy
prior to presentation. Although requested at the time of
presentation, some samples were not produced until after
the patient had been started on specific anti-microbial
therapy by the clinic physicians. Stool samples collected
by patients at home before coming to the clinic would not
have been suitable for identifying Entamoeba histolytica
trophozoites. Rates of identification of E. histolytica were
generally very low, as has been previously reported from
Zambia [2], no trophozoites were found and our own
internal and external quality control measures suggested
possible under-identification of this organism. Diagnostic
facilities for microsporidia were not available in our
laboratory, although results from other studies in Africa
have shown that microsporidia are an important cause of
diarrhoea in HIV-infected people [2,9,17]. Several other
organisms with the potential to cause diarrhoea could
not be identified in our laboratory (including C. difficile,
Mycobacterium avium intracellulare and viruses). Cultures
were not routinely performed for S. stercoralis but a sub-
sequent study conducted in our laboratory showed that
culture increased the rate of detection of S. stercoralis by
about 50% (W. Bailey, personal comm.)
Amongst helminth parasites Schistosoma mansoni,
S. stercoralis and hookworm were isolated with relatively
high frequency. No association with diarrhoea was
found. The study area is close to Lake Victoria, a fresh-
water body well known for harbouring schistosomes.
Although there was no significant association with
diarrhoea, some patients found to have schistosomes in
their stool had complained of lower abdominal pain and
passage of blood.
The numbers of potential pathogens identified in the
stool samples from asymptomatic participants were high.
This was not particularly surprising for helminths but
 Diarrhoea and Enteric InfectionÐUganda
recovery of bacterial pathogens was greater than
expected. Convalescent excretion and long-term carriage
of Shigella spp., non-typhi Salmonella spp. and Campylo-
bacter spp. is well recognised. However, no information
exists on rates and frequencies of carriage in adult
populations in sub-Saharan Africa. Consequently, we are
unable to determine whether these apparently high rates
are due to altered carriage characteristics due to HIV-
infection, diarrhoea preceding the one month asympto-
matic case-definition period or truly asymptomatic
infection. Irrespective of the cause, the high rates of
carriage also suggest that humans may be a major res-
ervoir of infection in this population. This is particularly
relevant to non-typhi Salmonella infections, where recent
studies have failed to establish an environmental or
zoonotic source in Africa (unlike the situation in Europe
where animal-derived food products are strongly impli-
cated). Further studies of the epidemiology of Salmonella
and Shigella infections in this setting are required.
The role of antibiotics in the management of diar-
rhoea is currently unclear, particularly in those indivi-
duals whose diarrhoea is chronic and not self-limiting.
Current recommendations of giving co-trimoxazole in
chronic diarrhoea based on the frequency of isospora
[18] are probably inappropriate in this population.
Resistance to co-trimoxazole is high in Enterobacteriacae
in East Africa [19], a finding which may be a reflection
of widespread community use of co-trimoxazole and
Fansidar1 for malaria. Moreover, I. belli was an
uncommon finding in this cohort in contrast to reported
rates of 13% and 12.5% in case-series of Ugandan
patients with enteropathic AIDS [6,17] and rates ran-
ging from 2% to 28% in other parts of Africa [2,3,5,8,9].
This may be because of differences in the study popula-
tion or prevalence of the organism in the environment.
Further work to establish the role of the microbiology
laboratory and to test antimicrobial and symptomatic
strategies of care is required.
In this study, low CD4 counts showed the strongest
association with diarrhoea, and therefore the level
of HIV-associated immunosuppression appears to be
an important determinant of symptoms. Increased
susceptibility to opportunistic pathogens at low CD4
T-cell counts was clearly demonstrated only for C. parvum
infection. Pathogen-negative diarrhoea is likely to
include cases caused by organisms not identified in this
study, by the HIV virus itself and perhaps to a lesser
extent by tumours (small bowel lymphoma, Kaposi's
sarcoma) and drugs (western and local). It is likely that
control of HIV-replication would have a marked impact
on diarrhoea in this population. In the absence,
however, of accessible antiretroviral therapy for this
105
population in the near future, increasing access to potent
symptomatic treatment should be a priority together
with further studies to identify environmental and social
risk factors for diarrhoea.
Acknowledgements
We thank the medical, nursing and support staff at the TASO Entebbe
and UVRI clinics for their role in recruitment, follow-up and medical
care of the patients and administration of the study and the
Medical Research Council and UVRI laboratory staff for analysis of
clinical samples. We thank the patients themselves for agreeing to take
part in the study. The work was supported by the UK Medical Research
Council (G9323636). Dr French is supported by the Wellcome Trust.
References
1 Colebunders RL, Latif AS. Natural history and clinical presentation
of HIV-1 infection in adults. AIDS 1991; 5(Suppl 1): S103±S112.
2 Kelly P, Baboo KS, Wolff M, Ngwenya B, Luo N, Farthing MJ. The
prevalence and aetiology of persistent diarrhoea in adults in urban
Zambia. Acta Trop 1996; 61: 183±190.
3 Conlon CP, Pinching AJ, Perera CU, Moody A, Luo NP, Lucas SB.
HIV-related enteropathy in Zambia: a clinical, microbiological, and
histological study. Am J Trop Med Hyg 1990; 42: 83±88.
4 Mwachari C, Batchelor BIF, Paul J, Waiyaki PG, Gilks CF. Chronic
diarrhoea among HIV-infected adult patients in Nairobi, Kenya.
J Infect 1998; 37: 48±53.
5 Khumalo-Ngwenya B, Luo NP, Chintu C et al. Gut parasites in HIV-
seropositive Zambian adults with diarrhoea. East Afr Med J 1994; 71:
379±383.
6 Sewankambo N, Mugerwa RD, Goodgame R et al. Enteropathic AIDS
in Uganda. An endoscopic, histological and microbiological study.
AIDS 1987; 1: 9±13.
7 Mengesha B. Cryptosporidiosis among medical patients with the
acquired immunodeficiency syndrome in Tikur Anbessa Teaching
Hospital, Ethiopia. East Afr Med J 1994; 71: 376±378.
8 Gumbo T, Sarbah S, Gangaidzo IT et al. Intestinal parasites in
patients with diarrhea and human immunodeficiency virus infection
in Zimbabwe. AIDS 1999; 13: 819±821.
9 Tarimo DS, Killewo JZJ, Minjas JN, Msamanga GI. Prevalence of
intestinal parasites in adult patients with enteropathic AIDS in
north-eastern Tanzania. East Afr Med J 1996; 73: 397±399.
10 Hunter G, Bagshawe AF, Baboo KS, Luke R, Prociv P. Intestinal
parasites in Zambian patients with AIDS. Trans R Soc Trop Med Hyg
1992; 86: 543±545.
11 Batchelor BIF, Kimari JN, Brindle RJ. Microbiology of HIV associated
bacteraemia and diarrhoea in adults from Nairobi, Kenya. Epidemiol
Infect 1996; 117: 139±144.
12 Navin TR, Weber R, Vugia DJ et al. Declining CD4‡ T-lymphocyte
counts are associated with increased risk of enteric parasitosis and
chronic diarrhoea: Results of a 3-year longitudinal study. J AIDS
Hum Retrovirol 1999; 20: 154±159.
13 Weber R, Ledergerber B, Zbinden R et al. Enteric infections and
diarrhea in human immunodeficiency virus-infected persons. Arch
Intern Med 1999; 159: 1473±1480.
14 French N, Nakiyingi J, Carpenter LM et al. 23-valent pneumococcal
polysaccharide vaccine in HIV-1-infected Ugandan adults: double-
blind, randomised and placebo controlled trial. The Lancet 2000;
355: 2106±2111.
15 World Health Organisation. Acquired immunodeficiency syndrome
(AIDS): interim proposal for a WHO staging system for HIV infection
and disease. Wkly Epidem Rec 1990; 65: 221±224.
16 Kamali A, Carpenter LM, Whitworth JAG, Pool R, Ruberantwari A,
Ojwiya A. Seven-year trends in HIV-1 infection rates, and changes
106 A.-K. Brink et al.
in sexual behaviour, among adults in rural Uganda. AIDS 2000; 18 Lynen L. In: Biot M (ed.) Clinical AIDS Care Guidelines for Resource Poor
14: 427±434. Settings. MeÂdecins Sans FrontieÁres, Belgium, Luxembourg, 2001:
17 Ravera M, Reggiori A, Riccioni G, Ciantia F, Vison R. Prevalence pp. 8.6±8.16.
of Microsporidium, Cryptosporidium parvum and Isospora belli among 19 Shapiro RL, Kumar L, Phillips-Howard P et al. Antimicrobial-
AIDS patients with chronic diarrhoea admitted in Hoima Hospital, resistant bacterial diarrhea in rural western Kenya. JID 2001; 183:
Uganda. IX International Conference on AIDS and STD in Africa. 1701±1704.
Kampala, December 1995 (abstract TuB567).