Impact of childhood nutritional status on pathogen prevalence and severity of acute diarrhea

Children with acute and chronic malnutrition are at increased risk of morbidity and mortality following a diarrheal episode. To compare diarrheal disease severity and pathogen prevalence among children with and without acute and chronic malnutrition, we conducted a cross-sectional study of human immunodeficiency virus-uninfected Kenyan children aged 6–59 months, who presented with acute diarrhea. Children underwent clinical and anthropometric assessments and provided stool for bacterial and protozoal pathogen detection. Clinical and microbiological features were compared using log binomial regression among children with and without wasting (mid-upper arm circumference ≤ 125 mm) or stunting (height-for-age z score ≤ _2). Among 1,363 children, 7.0% were wasted and 16.9% were stunted. After adjustment for potential confounders, children with wasting were more likely than nonwasted children to present with at least one Integrated Management of Childhood Illness danger sign (adjusted prevalence ratio [aPR]: 1.3, 95% confidence interval [CI]: 1.0 to 1.5, P = 0.05), severe dehydration (aPR: 2.4, 95% CI: 1.5 to 3.8, P < 0.01), and enteroaggregative Escherichia coli recovered from their stool (aPR: 1.8, 1.1–2.8, P = 0.02). There were no differences in the prevalence of other pathogens by wasting status after confounder adjustment. Stunting was not associated with clinical severity or the presence of specific pathogens. Wasted children with diarrhea presented with more severe disease than children without malnutrition which may be explained by a delay in care-seeking or diminished immune response to infection. Combating social determinants and host risk factors associated with severe disease, rather than specific pathogens, may reduce the disparities in poor diarrhea-associated outcomes experienced by malnourished children. This analysis used data collected from children under 5 years of age presenting with acute diarrhea to the outpatient and inpatient units at Kisii Provincial Hospital, Migori District Hospital, and Homa Bay District Hospital between November 2011 and June 2014. A description of this study has been published previously.13,14 Briefly, the parent study recruited children aged 6 months to 15 years presenting to the health facility with acute diarrhea (≥ 3 loose stools in the previous 24 hours and lasting less than 14 days). Eligible participants were excluded if they were not accompanied by a legal guardian or biological parent, if study staff were unable to collect a stool sample or rectal swab or if the primary caregiver refused human immunodeficiency virus (HIV) testing on behalf of the child. Written informed consent was obtained from the child’s caregiver. For this analysis, we included HIV-uninfected children 6 months to 5 years of age recruited at Homa Bay or Kisii hospitals. Migori Hospital patients were excluded as the low numbers recruited would be challenging to appropriately include in statistical models. The study was approved by the Kenya Medical Research Institute and University of Washington ethical review boards. At enrollment the primary caregiver completed a standardized clinical and sociodemographic questionnaire. All children were examined by clinical staff who recorded the presence of World Health Organization (WHO) Integrated Management of Childhood Illness (IMCI) danger signs (unable to drink or feed, vomiting everything, convulsions, lethargic, or unconscious) and hydration status (severe dehydration defined as at least two of the following: lethargic or unconscious, sunken eyes, not able to drink, reduced skin turgor). Weight, mid-upper arm circumference (MUAC), and height (or length) were measured and HAZs, weight-for-age z-scores (WAZ), and weight-for-height z-scores (WHZ) were calculated using WHO ANTHRO software.15 Z-scores of less than −7 or greater than 7 were deemed implausible and set to missing. All children were tested for HIV using antibody testing (Abbott Determine™ rapid test kit [Abbott Park, IL] and confirmed using Uni-Gold™ [Trinity Biotech, Bray, Ireland]) or HIV polymerase chain reaction (PCR) if under 18 months of age. Maternal HIV status was determined by antibody rapid testing or self-report. Malaria status was determined by a combination of rapid test (Paracheck Pf® Orchid Biomedical Services, Verna, Goa, India) and microscopy. Stool samples were collected in stool collection containers by caregivers or study staff. When a child was not able to produce stool, three rectal swabs were obtained. Fecal samples were collected prior to antibiotic administration, if indicated. Samples were received in the nearby U.S. Army Medical Research Directorate—Kenya (USAMD-K) Microbiology Hub in Kericho within 24 hours of collection after being placed in transport media (Cary–Blair for bacterial culture and 10% formalin for protozoal testing) and maintained at 2–8°C. Bacterial pathogens (E. coli, Shigella spp., Campylobacter spp., Salmonella spp.) were identified using traditional culture and serotyping methods confirmed using the MicroScan WalkAway 40 Plus (Beckman Coulter, Brea, CA) automated platform. Escherichia coli isolates were further tested for virulence factors using multiplex PCR and classified by pathotype: ETEC (heat labile or heat stable enterotoxin), EPEC (bundle forming pilus and, after March 2013, intimin), enteroaggregrative E.coli (EAEC) (aatA and, after March 2013, aaiC), enteroinvasive E.coli (invasion plasmid antigen H), or enterohemmorhagic E.coli (Shiga toxin 1,2 and variant). Protozoal infections were identified from whole stool samples using microscopy after stool concentration using Mini Parasep® Solvent Free concentration kit (DiaSys, Berkshire, England). No viral testing was performed. Analyses were conducted to examine whether children with acute malnutrition (wasting, MUAC < 12.5 cm) or chronic malnutrition (stunted HAZ < −2) had a higher prevalence of detectable enteric pathogens at presentation or presented with more severe disease (defined as the presence of at least one IMCI danger sign and/or severe dehydration). WHZ was not used to define acute malnutrition because it is influenced by fluid status and children with diarrhea often have some degree of dehydration. All prevalences were compared using log-binomial regression and associated χ2 tests.16 Pathogens and danger signs that were significantly associated with nutritional status in univariate regression at an alpha of 0.05 were further assessed in multivariable models. Potential confounders were evaluated for inclusion in each model in a stepwise manner and maintained in the model if the prevalence ratio changed by more than 10%. The following potential confounding variables were considered: age, gender, hospital site, currently receiving any breast milk (yes/no), maternal HIV status, family income (dichotomized at above or below 5,000 Kenyan Shillings), caregiver education (primary education or less, some secondary education, or secondary or greater educational level), persons per room, and access to improved sanitation (as defined by WHO/United Nations Children's Fund Joint Monitoring Program).17 Finally, the linearity of significant univariate associations with acute malnutrition were examined graphically by further categorizing MUACs into: < 11.5 (SAM), ≥ 11.5 to < 12.5 (moderate acute malnutrition), ≥ 12.5 to < 13.5, ≥ 13.5 to < 14.5 and ≥ 14.5. No similar descriptive analysis was done for stunting as the initial results were not significant. All analyzes were conducted using Stata 13.1 (Statacorp; College Station, TX).