Download Etiology of Bloody Diarrhea in Bolivian Children: Implications for

1527
Etiology of Bloody Diarrhea in Bolivian Children: Implications for Empiric
Therapy
John M. Townes,* Robert Quick, Oscar Y. Gonzales,
Miriam Linares, Esther Damiani, Cheryl A. Bopp,
Suzanne P. Wahlquist, Lori C. Hutwagner,
Erica Hanover, Eric D. Mintz, and Robert V. Tauxe, for
the Bolivian Dysentery Study Group†
Foodborne and Diarrheal Diseases Branch and Biostatistics and
Information Management Branch, Division of Bacterial and Mycotic
Diseases, and Biology and Diagnostics Branch, Parasitic Disease
Division, National Center for Infectious Diseases, Centers for Disease
Control and Prevention, Atlanta, Georgia; Proyecto de Salud Infantil y
Comunitaria, and Instituto Nacional de Laboratorios en Salud, La Paz,
Bolivia
In Bolivia, few data are available to guide empiric therapy for bloody diarrhea. A study was
conducted between December 1994 and April 1995 to identify organisms causing bloody diarrhea
in Bolivian children. Rectal swabs from children õ5 years old with bloody diarrhea were examined
for Salmonella, Shigella, and Campylobacter organisms; fecal specimens were examined for Entamoeba histolytica. A bacterial pathogen was identified in specimens from 55 patients (41%). Shigella
organisms were found in 39 specimens (29%); 37 isolates (95%) were resistant to ampicillin, 35
(90%) to trimethoprim-sulfamethoxazole, and 24 (62%) to chloramphenicol, but all were susceptible
to nalidixic acid. Only 1 of 133 stool specimens contained E. histolytica trophozoites. Multidrugresistant Shigella species are a frequent cause of bloody diarrhea in Bolivian children; E. histolytica
is uncommon. Clinical predictors described in this study may help identify patients most likely to
have Shigella infection. Laboratory surveillance is essential to monitor antimicrobial resistance and
guide empiric treatment.
In Bolivia, ú10,000 deaths caused by diarrheal disease occur
each year among children õ5 years old [1]. While diarrheal
disease control programs have emphasized the importance of
oral rehydration therapy for watery diarrhea and antimicrobial
therapy for bloody diarrhea, they have not made laboratory
surveillance a priority. Limited financial and laboratory resources generally preclude the use of routine stool cultures
and antimicrobial drug susceptibility testing to guide therapy.
Hence, therapeutic decisions are often made without supporting
data and may be ineffective or even harmful.
Many Latin American physicians treat bloody diarrhea as if it
were amebiasis. A study of antimicrobial therapy of childhood
diarrhea conducted in La Paz, Bolivia, in 1990 reviewed treatment received by a small series of patients presenting with
Received 16 August 1996; revised 6 December 1996.
Presented in part: 35th Interscience Conference on Antimicrobial Agents
and Chemotherapy, San Francisco, 20 September 1995.
Informed consent was obtained from parents or guardians of all patients.
Experimentation Guidelines of the US Department of Human Resources and
of the participating laboratories and hospitals were followed.
Financial support: Child and Community Health Project, La Paz, Bolivia;
Bolivian office of UNICEF. Entamoeba histolytica antigen detection kits for
this study were donated by Alexon,Inc. Sunnyvale, California.
Reprints or correspondence: Dr. John M. Townes, Centers for Disease Control and Prevention, Foodborne and Diarrheal Diseases Branch, Mailstop A38, Atlanta, GA 30333.
* Present affiliation: Oregon Department of Human Resources, Health Division, Center for Disease Prevention and Epidemiology, Portland, Oregon.
†
Study group members are listed after text.
The Journal of Infectious Diseases 1997;175:1527–30
q 1997 by The University of Chicago. All rights reserved.
0022–1899/97/7506–0036$01.00
/ 9d2a$$ju22
04-25-97 11:39:14
jinfa
bloody diarrhea; a high proportion were treated with metronidazole despite lack of evidence for amebiasis (CDC, unpublished
data). Trimethoprim-sulfamethoxazole and ampicillin are available for the treatment of bloody diarrhea at Bolivian hospitals
and health centers at no cost to patients. National programs
encourage physicians to treat all children with bloody diarrhea
with one of these agents and to consider the diagnosis of amebiasis for patients who do not respond to this therapy.
To assess the appropriateness of this strategy, we investigated cases of bloody diarrhea in children õ5 years old presenting to hospitals in three Bolivian cities between December
1994 and April 1995. Our goals were to identify the causes of
bloody diarrhea, to determine the antimicrobial susceptibility
patterns for Shigella isolates, and to identify clinical and laboratory correlates of Shigella infection.
Methods
Case enrollment. Four hospitals were included in the study:
Hospital del Niño Ovidio Aliaga Uria in La Paz, Hospital Germán
Urquidi in Cochabamba, and Hospital Boliviano-Japonés and Hospital de Niños in Santa Cruz. From December 1994 through April
1995, all children õ5 years old with bloody diarrhea who presented to these hospitals during daytime hours from Sunday
through Thursday were eligible for the study. Children were excluded from the study if another household member had already
been enrolled in the previous 7 days. Bloody diarrhea was defined
as §3 loose stools in a 24-h period, with any of these stools
containing visible blood as reported by a parent or adult guardian.
After an initial history, physical examination, and treatment plan
were completed by the attending physician in the clinic or emergency room, trained interviewers administered a standard question-
UC: J Infect
1528
Concise Communications
naire to parents of ill children. Questionnaires addressed demographic information, description of symptoms, and type of medical
care and medications previously given to the ill child.
Collection of specimens. Two rectal swabs were collected
from each patient by a member of the study team and placed
immediately in refrigerated Cary-Blair medium for daily transport
to the Enterobacteriology Laboratory at the National Institute of
Health Laboratories (INLASA) in La Paz. Freshly excreted whole
stool specimens were also collected and divided into three portions.
One portion was placed immediately in polyvinyl alcohol (PVA)
and transported to the Parasitic Diseases Laboratory at the Centers
for Disease Control and Prevention (CDC) in Atlanta. Specimens
not placed in PVA within 20 min of excretion were discarded. A
second portion of the fresh whole stool specimen was provided to
the local hospital laboratories for immediate direct examination,
and the remainder was sent to INLASA.
Examination of specimens. All rectal swab specimens were
cultured at INLASA, within 1 day after collection, for Shigella,
Salmonella, and Campylobacter organisms by use of standard techniques. Colonies consistent with Shigella and Salmonella were
identified and serotyped by standard techniques [2]. Those identified as Shigella were sent to the CDC Diarrheal Diseases Laboratory for confirmation and serotyping.
All Shigella isolates were tested at INLASA by use of the disk
diffusion method [3] for antimicrobial susceptibility to the following agents: ampicillin, trimethoprim-sulfamethoxazole, chloramphenicol, tetracycline, ciprofloxacin, nalidixic acid, sulfisoxazole,
streptomycin, gentamicin, ceftriaxone, cephalothin, and amoxicillin–clavulanic acid.
Permanent slides of fecal specimens fixed in PVA were stained
with trichrome (TC) stain and examined at CDC for protozoan parasites and fecal leukocytes by use of light microscopy. Fresh stool
specimens were examined at local hospitals by light microscopy with
saline and Lugol solution. Whole stool specimens were examined at
INLASA for Entamoeba histolytica–specific antigen by use of an
ELISA (ProSpecT Microplate Assay; Alexon, Sunnyvale, CA).
Statistical methods. Univariate and multivariate analyses of
clinical and laboratory data were conducted by use of generalized
estimating equations to control for clustering of patients among
study sites [4]. The two hospitals in Santa Cruz were considered
together as a single study site. These analyses were used as a
starting point for developing a set of clinical predictors for Shigella
infection. Combinations of the variables having a significant and
independent association with Shigella infection were then tested
against culture results to determine their sensitivity, specificity,
and positive predictive value for Shigella infection.
Results
A total of 133 patients with bloody diarrhea were enrolled
in the study, 55 (41%) in Cochabamba, 45 (34%) in La Paz,
and 33 (25%) in Santa Cruz. The median age of patients was
18 months (range, 3 – 58); 65 (49%) were girls.
Stool specimens were collected from all 133 children enrolled in the study. A bacterial pathogen was identified in specimens from 55 (41%); 2 specimens yielded more than one pathogen (Salmonella and Campylobacter species in 1 and Shigella
/ 9d2a$$ju22
04-25-97 11:39:14
jinfa
JID 1997;175 (June)
and Campylobacter species in another). The proportion of specimens yielding positive bacterial cultures did not differ significantly between sites. Shigella organisms were the most frequently identified pathogen. Cultures of fecal specimens from
39 (29%) children yielded Shigella organisms. Of these, 28
(72%) were Shigella flexneri (16 type 2a, 7 type 1b, 1 type 1a,
1 type 3a, 1 type 3b, 2 isolates not available for serotyping),
10 (26%) were Shigella sonnei, and 1 (3%) was Shigella dysenteriae type 2. No S. dysenteriae type 1 or Shigella boydii
organisms were identified. Campylobacter organisms were the
second most common pathogen, identified in 12 stool specimens (9%); Salmonella species were cultured from 6 specimens
(4.5%).
Local hospitals reported finding E. histolytica trophozoites
in 10 (7.5%) of 133 fresh stool specimens. Corresponding
PVA-fixed, TC-stained specimens examined at CDC showed
no E. histolytica trophozoites. Slides from 9 of these 10 specimens contained ú50 leukocytes/high-power field, and a bacterial pathogen was identified in cultures from 5 of these specimens (Shigella species, 3; Campylobacter species, 2).
Trophozoites of E. histolytica were identified in only 1 (0.75%)
of 133 TC-stained specimens examined at CDC. The ELISA
for E. histolytica antigen was positive for only 2 specimens.
Neither had E. histolytica trophozoites evident on microscopic
examination.
Only two Shigella isolates (1 S. sonnei and 1 S. dysenteriae
type 2) were susceptible to all of the antibiotics tested. Of 39
isolates, 37 (95%) were resistant to ampicillin and 35 (90%) were
resistant to trimethoprim-sulfamethoxazole; 34 isolates (87%)
were resistant to both of these drugs. Only 2 Shigella isolates
(5%) were resistant to cephalothin, and none was resistant to
ceftriaxone, nalidixic acid, or ciprofloxacin. The proportion of
resistant isolates did not vary significantly among study sites.
Antimicrobial resistance patterns differed by serotype. Resistance to chloramphenicol was limited to isolates of S. flexneri;
24 (86%) were resistant to this agent. Resistance to amoxicillin – clavulanic acid was also more common among S. flexneri
isolates; 19 (68%) of 28 S. flexneri isolates were resistant to
this agent, compared with 1 (10%) S. sonnei isolate.
Medications received before enrollment in the study could
be determined for 126 children; 51 children received at least
one antimicrobial agent. Prior treatment with ampicillin or trimethoprim-sulfamethoxazole did not decrease the proportion
of stool cultures that were positive for Shigella organisms;
Shigella organisms were isolated from 11 (32%) of 34 stool
specimens obtained from children who had been treated with
either trimethoprim-sulfamethoxazole or ampicillin and from
22 (29%) of 75 stool specimens obtained from children who
had not received any antimicrobial agent. All Shigella isolates
from patients treated with ampicillin or trimethoprim-sulfamethoxazole were resistant to these agents.
On univariate analysis of clinical and laboratory data, several
variables were significantly associated with Shigella infection
(table 1). On multivariate analysis, there remained a significant
UC: J Infect
JID 1997;175 (June)
Concise Communications
1529
Table 1. Clinical features associated with Shigella infection among 133 children with history of bloody diarrhea reported by parent or
guardian, Bolivia, December 1994 – April 1995 (univariate analysis).
Feature
Cries during defecation
Temperature (continuous variable)
Temperature ú38.47C
Stool frequency (continuous variable)
Stool frequency ú5/24 h
Blood observed in stool by examiner
ú50 leukocytes/high-power field
Listlessness
Seizures
Shigella organisms
in stools (n Å 38)*
No Shigella
organisms in stools
(n Å 94)
Odds ratio
P
95% confidence
interval
30 (79%)
—
16 (42%)
—
32 (84%)
33 (87%)
35 (92%)
38 (100%)
7 (18%)
43 (46%)
—
20 (21%)
—
53 (56%)
70 (74%)
50 (53%)
77 (82%)
6 (6.4%)
4.98
1.71
2.69
1.21
3.43
2.26
10.27
Undefined
3.31
õ.001
õ.001
õ.01
õ.001
õ.001
õ.05
õ.001
õ.05†
õ.001
2.12 – 11.74
1.33 – 2.21
1.36 – 5.31
1.1 – 1.33
1.80 – 6.52
1.02 – 5.00
2.16 – 48.77
Not applicable
1.29 – 6.14
* One patient with concurrent Campylobacter and Shigella infection was excluded from analysis.
†
Mantel-Haenszel x2.
independent association between Shigella infection and the following variables: crying during defecation, temperature
ú38.47C, stool frequency ú5/24 h, and ú50 leukocytes/highpower field seen on microscopic examination of a TC-stained
fecal smear examined at CDC. Combinations of these variables
yielded varying values for sensitivity, specificity, and positive
predictive value (table 2). In the absence of stool microscopy,
the presence of any three of four clinical variables (crying
during defecation, temperature ú38.47C, blood observed in
stool by clinician, and stool frequency ú5/24 h) offered the
best discrimination.
Discussion
Shigella infection was the most frequently identified cause
of bloody diarrhea in this study. The absence of S. dysenteriae
type 1 was not surprising. This strain, known for its tendency
to cause severe illness and large epidemics, has rarely been
encountered in Latin America in recent years [5].
The alarmingly high level of resistance to ampicillin and trimethoprim-sulfamethoxazole seen in this study suggests that children with bloody diarrhea who present to urban hospitals in Bolivia are unlikely to respond to treatment with these agents. No
resistance to nalidixic acid among Shigella isolates was observed
in this study, and its use has been shown in controlled clinical
trials to be effective in the treatment of shigellosis [6, 7]. For these
reasons, nalidixic acid, rather than ampicillin or trimethoprimsulfamethoxazole, should be promoted as the first-line agent for
the treatment of bloody diarrhea in this setting. To avert the rapid
development of resistance to nalidixic acid, physicians should be
encouraged to reserve antimicrobial therapy for the most severe
cases [8]. Rational guidelines for empiric therapy depend on
knowledge of evolving local antimicrobial resistance patterns.
Routine antimicrobial resistance monitoring is therefore an essential part of any diarrheal disease control program that advocates
empiric antimicrobial therapy.
Infection with E. histolytica was detected in stools from only
1 child in this study. This finding is consistent with previous
Table 2. Clinical and laboratory predictors of Shigella infection among 133 children with history of
bloody diarrhea as reported by parent or guardian, La Paz, Cochabamba, and Santa Cruz, Bolivia,
December 1994 – April 1995.
Combination of
features
Combination
Combination
Combination
Combination
1
2
3
4
Stools with
Shigella organisms
Stools without
Shigella organisms
Sensitivity
(%)
Specificity
(%)
Positive predictive
value (%)
32/38
34/38
29/38
27/38
43/94
34/94
27/94
15/94
84
89
76
71
54
64
71
84
43
50
52
64
NOTE. One patient with concurrent Campylobacter and Shigella infection was excluded from analysis. Combination 1 Å at least 2 of following: child cries during defecation, temperature ú38.47C, blood observed in stool by
examiner; combination 2 Å combination 1 plus ú50 leukocytes/high-power field seen on trichrome-stained slide of
fecal specimen preserved in polyvinyl alcohol; combination 3 Å at least 3 of following: cries during defecation,
rectal temperature ú38.47C, blood observed in stool by examiner, ú5 stools/24 h; combination 4 Å combination 3
plus ú50 leukocytes/high-power field seen on trichrome-stained slide of fecal specimen preserved in polyvinyl
alcohol.
/ 9d2a$$ju22
04-25-97 11:39:14
jinfa
UC: J Infect
1530
Concise Communications
studies that have demonstrated that amebiasis rarely causes
bloody diarrhea in young children [9, 10] and is greatly overdiagnosed [11 – 13]. Empiric antiamebic treatment is rarely indicated for children who do not respond to antimicrobial therapy
for shigellosis. These children are much more likely to have
drug-resistant shigellosis than amebiasis.
Because of limited laboratory resources, we confined our
bacteriologic investigation to Shigella, Salmonella, and Campylobacter organisms. If we had examined stools for additional
organisms, such as enterohemorrhagic, enteropathogenic, and
enteroinvasive Escherichia coli, or cultured multiple specimens
from each patient, the proportion of patients found to be infected with a bacterial pathogen might have been higher. The
etiology of bloody diarrhea caused by pathogens other than
Shigella, Salmonella, and Campylobacter species is a subject
for additional investigation.
The use of clinical predictors may make it possible to target
appropriate empiric antimicrobial therapy toward children most
likely to have Shigella infection, thereby substantially reducing
inappropriate use of antimicrobial agents. The four different
combinations of clinical and laboratory variables (table 2) are
potentially useful predictors of Shigella infection in Bolivian
children. Combinations 2 and 4 require microscopic examination of stools for fecal leukocytes. Although adding stool microscopy improves specificity and positive predictive value,
laboratory personnel must be able to correctly identify fecal
leukocytes and avoid mistaking them for pathogenic amebae.
For these two combinations to be useful, quality control of
laboratories is essential. Combinations 1 and 3 are sensitive
and have the advantage of not requiring any laboratory examinations. For these reasons, one of these two combinations may
prove to be the most useful criteria for treatment algorithms
proposed for rural areas, where access to laboratories may be
more limited. A study designed specifically to evaluate prospectively the validity and general applicability of these criteria
should be conducted before they are adopted as part of treatment algorithms.
/ 9d2a$$ju22
04-25-97 11:39:14
jinfa
JID 1997;175 (June)
Study Group Members
Other members of the Bolivian Dysentery Study Group are as
follows. Hospital Boliviano-Japonés, Santa Cruz: Héctor Soliz,
Orlando Jordán, Flavia Loaysa, David F. Ortiz, Nancy Terceros,
Jerges Nazareno; Hospital Germán Urquidi, Cochabamba: Rosalı́a
Sejas, Marianela Llanos, Pastor Velasquez, Ana Maria Moya; Hospital del Niño Ovidio Aliaga Uria, La Paz: Eduardo Aranda, Luis
Tamayo, Victor Lavayen, Raquel Bravo, Fernando Salazar, Antonieta Castillo, Daysi León; Ministerio de Desarollo Humano, Secretaria Nacional de Salud: Jaqueline Reyes Maldonado.
References
1. Gutierrez M, Ochoa LH, Raggers H. Encuesta nacional de demografia y
salud. La Paz, Bolivia: Instituto Nacional de Estadistica de Bolivia,
1994.
2. Ewing W. Identification of Enterobacteriaceae. 4th ed. New York: Elsevier
Science, 1986.
3. Woods G, Washington J. Antimicrobial susceptibility tests: dilution and
disc diffusion methods. In: Murray P, ed. Manual of clinical microbiology. 6th ed. Washington, DC: American Society for Microbiology,
1995:1327 – 41.
4. Zeger S, Liang K. Longitudinal data analysis for discrete and continuous
outcomes. Biometrics 1986; 42:121 – 30.
5. Centers for Disease Control and Prevention. Shigella dysenteriae type 1 —
Guatemala. MMWR Morb Mortal Wkly Rep 1991; 40:417 – 32.
6. Salam M, Bennish M. Therapy for shigellosis I. Randomized, double blind
trial of nalidixic acid in childhood shigellosis. J Pediatr 1988; 113:901 –
7.
7. Salam M, Bennish M. Antimicrobial therapy for shigellosis. Rev Infect
Dis 1991; 13(suppl):S332 – 41.
8. Tuttle J, Tauxe R. Antimicrobial resistant Shigella: the growing need for
prevention strategies. Infect Dis Clin Pract 1993; 2:55 – 9.
9. Huilan S, Zhen L, Mathan M, et al. Etiology of acute diarrhea among
children in developing countries: a multicenter study in 5 countries.
Bull World Health Organ 1991; 69:549 – 55.
10. Ronsmans C, Bennish M, Wierzba T. Diagnosis and management of dysentery by community health workers. Lancet 1988; 2:552 – 5.
11. Cruz S, Salazar L, Urroz L, Anderson M. Certeza diagnóstica de amebiasis
intestinal en Managua. Rev Cubana Med Trop 1991; 43:80 – 4.
12. Gonzáles-Ruiz A. Revisión del estado actual del diagnóstico diferencial
de las amibas en México. Salud Publica Mex 1990; 32:589 – 96.
13. Walsh J. Problems in recognition and diagnosis of amebiasis: estimation
of the global magnitude of morbidity and mortality. Rev Infect Dis
1986; 8:228 – 38.
UC: J Infect