epidemiology of shigella-associated diarrhea in rural ... - CiteSeerX

8 downloads 0 Views 73KB Size Report
warm months increased the risk of developing Shigella-associated diarrhea, while breastfeeding ... gions.1 Shigellosis is common among children less than five.
Am. J. Trop. Med. Hyg., 71(3), 2004, pp. 367–372 Copyright © 2004 by The American Society of Tropical Medicine and Hygiene

EPIDEMIOLOGY OF SHIGELLA-ASSOCIATED DIARRHEA IN RURAL EGYPTIAN CHILDREN REMON R. ABU-ELYAZEED, THOMAS F. WIERZBA, ROBERT W. FRENCK, SHANNON D. PUTNAM, MALLA R. RAO, STEPHEN J. SAVARINO, KARIM A. KAMAL, LEONARD F. PERUSKI JR., IBRAHIM A. ABD-EL MESSIH, SAHAR A. EL-ALKAMY, ABDOLLAH B. NAFICY, AND JOHN D. CLEMENS U.S. Naval Medical Research Unit No. 3, Cairo, Egypt; U.S. Naval Research Unit No. 2, Jakarta, Indonesia; National Institute of Child Health and Human Development, National Institutes of Health, Bethesda, Maryland; Abu Homos Field Research and Training Center, Egyptian Ministry of Health and Population, Beheira, Egypt; International Vaccine Institute, Seoul, Korea

Abstract. During the period from February 1995 to February 1998, the epidemiology of Shigella diarrhea was studied among children less than three years of age residing in Egypt’s Nile Delta. Children were visited twice a week and a stool sample was collected from any of them with diarrhea. The incidence of Shigella-associated diarrhea was 0.2 episodes/ child-year, with S. flexneri being the most common serogroup isolated (55% of Shigella episodes). Younger age and the warm months increased the risk of developing Shigella-associated diarrhea, while breastfeeding was protective. Children with Shigella were ill for a mean of four days and passed a mean of six stools per day. Common symptoms included fever (35%), vomiting (19%), and dehydration (16%). Dysentery, however, was unusual, occurring in only 11% of the cases. In conclusion, Shigella-associated diarrhea remains relatively common in Egyptian children and supports the need for additional control measures including vaccine development. km southeast of Alexandria. Children less than 24 months of age and all newborns, without congenital abnormalities or a history of hospitalization, were identified and recruited for study enrollment. Children remained in the cohort until they reached 36 months of age or February 1998, whichever came first. Each enrolled child was visited at home twice a week by a trained field worker, and a standardized questionnaire was administered to the caregiver. In addition, a determination was made regarding any diarrheal symptoms since the previous visit. Those children reported to have loose or liquid stools had a rectal swab and fresh stool specimen collected. The rectal swab was immediately placed in Cary-Blair (CB) transport media and the stool specimen placed in a plastic capped container and both were refrigerated and subsequently transport to the Naval Medical Research Unit No. 3. For the purpose of case-control comparison, cross-sectional surveys of all enrolled children were conducted at two-month intervals over the three-year period of the study. In these surveys, rectal swabs and stool specimens were collected from all the children regardless of their diarrheal symptoms. A diarrheal day was defined as the occurrence of ⱖ3 unformed stools (or ⱖ1, if bloody) in a 24-hour period. For breastfed infants with non-bloody stools, the mother must also have indicated that there had been an increase in frequency or a decrease in consistency of stools in relation to the normal pattern of defecation. An episode of diarrhea began on the first diarrheal day after three or more consecutive non-diarrheal days and ended on the last diarrheal day, followed by three consecutive non-diarrheal days. Shigella diarrhea denoted an episode in which Shigella was isolated from any specimen collected during that episode of diarrhea, regardless of whether other pathogens were also isolated from the stool. Shigella infection was considered to be asymptomatic whenever diarrhea did not occur within seven days before or after the isolation of Shigella from feces. The presence of blood in stools was based on the report of the child’s caregiver. A child was classified as breast-fed if breast milk constituted any portion of the child’s diet. A mother was considered to be educated if she received any formal schooling. Using our previously described methods,23 socioeconomic status of the

INTRODUCTION Shigella infections remain a global public health concern, causing diarrhea in both the developing and developed regions.1 Shigellosis is common among children less than five years of age in developing countries and in persons who travel from industrialized to less developed countries.2,3 In the developing world, it is estimated that 113 million episodes of shigellosis occur annually, resulting in more than 400,000 deaths.4 Industrialized countries also report outbreaks of Shigella infections among high-risk populations such as children attending day care,5–7 persons with human immunodeficiency virus/acquired immunodeficiency syndrome,8–11 and inmates of custodial institutions.12 Prevention of Shigella infections has proven difficult, mainly due to the low inoculum needed to produce disease13,14 and inadequate empirical therapy options secondary to antimicrobial resistance.15–18 The World Health Organization has called for the development of candidate vaccines against Shigella.19 In considering the efficacy of a Shigella vaccine, it is necessary to have reliable estimates of the burden of disease in targeted endemic areas, including age-specific incidence data, risk factors, and the distribution of the various serogroups and subtypes of Shigella-causing infection. In Egypt, limited data exists regarding Shigella-associated diarrhea, with the last population-based study20 of Shigella conducted more than 15 years ago. Additionally, Zaki and others20 did not fully characterize the Shigella isolates with respect to serogroup. Therefore, this study was performed and the results of a three-year longitudinal study of diarrheal diseases in children living in the Nile River Delta in Egypt are reported. METHODS Study population and surveillance. Details about the study population, surveillance methodologies, specimen collection, and transport and laboratory processing have been previously reported by Naficy and others.21,22 Briefly, a prospective cohort study was conducted in two villages in Abu Homos, Egypt, a rural agricultural district located in the Nile Delta, 40

367

368

ABU-ELYAZEED AND OTHERS

family was expressed as a score of 0–28, based on the type of housing (apartment versus detached home) and the number of amenities belonging to the family (washing machine, car, television, etc.). Poor households were classified as those with scores in the lower third of the scale. Assessment of crowding was based on the number of residents per sleeping room. Laboratory evaluation. Standard microbiology laboratory techniques were used to isolate and identify Shigella, Salmonella, Campylobacter, Escherichia coli, and Vibrio sp. from the rectal swab in CB media, while the stool sample in buffered-glycerol-saline was cultured only on Salmonella-Shigella plates.24–27 Shigella serogroups and serotypes were determined using standard commercially available antisera (Difco Laboratories, Detroit, MI),28 and S. flexneri were ascertained using a panel of monoclonal antibody reagents that were kindly provided by Nils Carlin (Swedish Biological Laboratories, Vaccin, Stockholm, Sweden). Rotavirus testing was performed on the frozen stool specimens as previously described.21 Statistical analyses. Incidence rates were calculated by dividing the number of diarrhea episodes or the number of Shigella-associated episodes by child-years at risk of followup. To identify variables predictive of Shigella-associated diarrhea, the strength of association between the dependent and independent variables was computed using relative rates. Crude relative rates were obtained by adding a single independent variable to a Poisson regression model. To adjust for confounding among variables, a Poisson regression model was fit by simultaneously adding all independent variables to a single multivariate model. Since repeated measures on the same child are likely correlated, generalized estimating equations (GEEs) were used in the multivariate model. Both crude and adjusted relative risks (aRRs) were obtained from these models by exponentiating the model coefficients. Ninety-five percent confidence intervals (CIs) were calculated using model coefficients and empirical standard errors and exponentiating those results. The virulence of Shigella by species was assessed by comparing the frequency distribution of each sign or symptom (i.e., vomiting, bloody stools, fever, etc.) between S. flexneri, S. sonnei, and S. dysenteriae using a chi-square test. Too few episodes of S. boydii or episodes with more than one Shigella sp. were detected to include these groups in the analysis. To evaluate pathogenicity, a case-control analysis was performed to examine the association between the occurrence of diarrheal symptoms and excretion of Shigella. To minimize the effect of prior infections with Shigella and potential immunity to subsequent infections with the organism, analysis was restricted to the first Shigella infection, as either a case or control experienced by the subject since beginning enrollment in the study. Cases were defined as study subjects with Shigella diarrhea detected during the twice a week home visits. Controls were defined as children without diarrhea who had cultures performed on fecal samples obtained during the every-other-month routine surveys. Since the controls had a single fecal specimen collected at each visit, only the microbiologic result of the first fecal specimen from each episode was included in the analysis to avoid detection bias. To identify signs or symptoms related to Shigella-associated diarrhea, a second case-control analysis was conducted. Cases were children with Shigella-associated diarrhea without copathogens and controls were children with diarrhea not asso-

ciated with Shigella. The strength of the association between Shigella infection and independent variables was measured using crude and adjusted odds ratios (aORs). Crude ratios were obtained by fitting a single independent variable into the logistic regression model. To control for confounding between variables, multivariate models were fit by entering all independent variables to the regression model. As with the previous multivariate models, GEEs were also used to adjust for the correlation between repeated measures on the same individual. Parameter estimates from both the crude and multivariate models were used to calculate odds ratios and 95% CIs and obtain P values. All statistical tests were two-tailed. P values