Le A Denno Research Paper

Abstract

Background

Shigella is a leading cause of childhood diarrhea mortality in sub-Saharan Africa. Current World Health Organization guidelines recommend antibiotics for children in non cholera-endemic areas only in the presence of dysentery, a proxy for suspected Shigella infection.

Methods

To assess the sensitivity and specificity of the syndromic diagnosis of Shigella-associated diarrhea, we enrolled children aged 6 months to 5 years presenting to 1 of 3 Western Kenya hospitals between November 2011 and July 2014 with acute diarrhea. Stool samples were tested using standard methods for bacterial culture and multiplex polymerase chain reaction for pathogenic Escherichia coli. Stepwise multivariable logit models identified factors to increase the sensitivity of syndromic diagnosis.

Results

Among 1360 enrolled children, median age was 21 months (interquartile range, 11–37), 3.4% were infected with human immunodeficiency virus, and 16.5% were stunted (height-for-age z-score less than −2). Shigella was identified in 63 children (4.6%), with the most common species being Shigella sonnei (53.8%) and Shigella flexneri (40.4%). Dysentery correctly classified 7 of 63 Shigella cases (sensitivity, 11.1%). Seventy-eight of 1297 children without Shigella had dysentery (specificity, 94.0%). The combination of fecal mucous, age over 23 months, and absence of excessive vomiting identified more children with Shigella-infection (sensitivity, 39.7%) but also indicated antibiotics in more children without microbiologically confirmed Shigella (specificity, 82.7%).

Conclusions

Reliance on dysentery as a proxy for Shigella results in the majority of Shigella-infected children not being identified for antibiotics. Field-ready rapid diagnostics or updated evidence-based algorithms are urgently needed to identify children with diarrhea most likely to benefit from antibiotic therapy.

Shigella is a leading cause of diarrheal disease in children worldwide [1]. Children with Shigella-associated diarrhea are at risk of linear growth faltering and the consequences of this Gram-negative bacteria can be fatal [2–5]. Risk of death from Shigella is associated with young age, malnutrition, human immunodeficiency virus (HIV)-infection, and complications including encephalopathy, hyponatremia, and seizures [6–9]. Historically, Shigella dysenteriae type 1 infections were thought to be responsible for most Shigella-attributed deaths [10]. However, S dysenteriae type 1 prevalence appears to be decreasing globally, and evidence suggests that risk of death and other potentially lethal sequelae may not be Shigella species-specific [7, 8, 11–14]. Although a key clinical feature of S dysenteriae type 1 is bloody stool (dysentery), other species of Shigella are less likely to be dysenteric [8, 15]. For reasons possibly related to the virtual disappearance of S dysenteriae type 1 or more careful management of children presenting with dysenteric diarrhea, dysentery no longer appears to be associated with poor Shigella outcomes [2, 7, 9]. Reducing Shigella-associated morbidity and mortality may require increased attention to Shigella-infected children without dysentery.

Antibiotics reduce time to diarrhea resolution and bacterial clearing in children with dysentery due to Shigella [16, 17]. Antibiotic treatment of nondysenteric Shigella in patients with travelers’ diarrhea and in Shigella-challenged volunteers also confers similar benefits [18, 19]. Given the risk of poor outcome associated with nondysenteric species of Shigella, antibiotic treatment of all childhood Shigella infections may reduce morbidity and mortality attributed to this infection.

In resource-limited settings, where the highest morbidity and mortality burden from Shigella occurs, most clinical facilities lack access to pathogen detection diagnostics. In the absence of such laboratory capacity, health workers rely on clinical judgment and syndrome-based guidelines for management of diarrhea [20]. In World Health Organization (WHO) guidelines, including the updated 2014 Integrated Management of Childhood Illness (IMCI) algorithm for diarrhea, antibiotics (a 3-day course of ciprofloxacin) are only recommended for children with suspected Shigella infection, as determined by presence or history of dysentery, as well as for children with suspected cholera [20, 21].

Within a large cohort of children presenting with acute diarrhea at 3 hospitals in Western Kenya, we sought to determine the sensitivity and specificity of the syndromic definition of dysentery for diagnosing Shigella-associated diarrhea against the gold standard of stool culture. We compared dysentery-alone with alternative classifications of Shigella infection based on sociodemographic factors, clinical history, clinical presentation, and stool examination with the goal of improving the sensitivity of syndromic algorithms for identifying Shigella.

METHODS

Population

We enrolled children aged 6 months to 5 years presenting with acute diarrhea to outpatient departments at 3 hospitals in the Nyanza province of Western Kenya (Kisii Provincial, Homa Bay District, and Migori District Hospital) as part of an ongoing diarrhea and fever surveillance study from November 28, 2011 to July 31, 2014. Acute diarrhea was defined as 3 or more loose or watery stools in the last 24 hours lasting less than 14 continuous days. Written informed consent was obtained from primary caregivers of enrolled children and all study procedures were approved by the University of Washington and Kenya Medical Research Institute (KEMRI) Ethics Committees.

Data Collection

Clinical history and sociodemographic information were collected from the accompanying caregivers. Integrated Management of Childhood Illness-specified clinical signs by physical examination were sought, including general danger signs (not able to drink or breastfeed, excessive vomiting, convulsions, lethargy, stiff neck), dehydration signs (sunken eyes, slow recoil skin pinch, restlessness or irritability, drinking eagerly or thirsty), as well as dysentery (reported history of blood in the stool since the diarrhea episode began). Height, weight, and mid-upper arm circumference were measured by the nursing staff and height-for-age z-scores (HAZ), weight-for-age z-scores (WAZ), and weight-for-height z-scores (WHZ) were calculated using WHO ANTHRO software [22]. Stunting and wasting were defined as HAZ less than −2 and WHZ less than −2, respectively.

All caregivers were provided with stool collection kits and instructions for collection. If a child could not produce stool within 1 hour, 2 rectal swabs were obtained. Blood was collected from children for malaria and HIV testing (per Kenyan National Guidelines). Human immunodeficiency virus status was determined using antibody testing (Abbott Determine rapid test kit and confirmed using Uni-Gold) or HIV DNA polymerase chain reaction (PCR) assays if the child was <18 months old. Malaria parasitemia was assessed by both rapid testing ([RDT] Paracheck Pf Orchid Biomedical Services, India) and microscopy. The HIV status of consenting accompanying biological mothers was ascertained by self-report and confirmed with antibody testing if unknown or HIV negative.

Stool Specimen Processing

Rectal swabs or a portion of stool samples were transferred into Cary-Blair transport medium for bacterial culture; remaining stool samples were placed in 10% formalin for parasite determination. Samples were stored and shipped on cold packs daily to the KEMRI/United States Army Medical Research Unit Microbiology Hub laboratory in Kericho, Kenya, within 24 hours of collection. Upon arrival at the laboratory, stool was examined for gross blood, mucous, and appearance (watery). Bacteria culture and microscopy for parasite detection were performed as described elsewhere [23]. Four lactose fermenting and 2 sorbitol nonfermenting Escherichia coli isolates from children enrolled before October 2013 were used for identification of pathogenic E coli with multiplex PCR [24]. Antibiotic susceptibility testing was performed using the MicroScan Walkaway40 Plus automated bacterial identification platform.

Statistical Analysis

Among enrolled children, we described sociodemographic, clinical history, clinical presentation, macroscopic stool investigation, and enteric pathogen detection with frequencies and percentages or medians and interquartile ranges (IQR). Because enteroinvasive E coli (EIEC) and Shigella share the same virulence gene, ipaH, we also calculated the frequency of overlap between the 2 types of infections. We determined the diagnostic accuracy (sensitivity, specificity, positive predictive value [PPV], and negative predictive value [NPV]) of dysentery for identifying children infected with culture-confirmed Shigella. Dysentery was defined as either history or presence of bloody stool. For each measure of diagnostic performance, 95% confidence intervals (CIs) were calculated assuming a binomial distribution. We described the frequencies and percentages of various factors in children with and without Shigella infections and calculated the discriminatory ability of each factor using area under the curve (AUC) and associated bootstrapped 95% CIs.

Multivariable logit regression models were fit to determine the minimal set of variables that improved upon the discriminatory ability of dysentery for syndrome-based identification of Shigella infections. The following variables were considered as plausible discriminatory indicators of Shigella infection and ordered by increasing the level of resource and training requirements: Model 1, dysentery; Model 2–Model 1 with laboratory-observed mucous or watery stool; Model 3–Model 2 with easily collected history and presentation information (IMCI general danger signs, IMCI-dehydration signs, age, sex, history of fever, recent antibiotic use, current breastfeeding, and number months of exclusive breastfeeding); Model 4–Model 3 plus measures that require slightly more time, training, and equipment (axillary temperature, HAZ, and WHZ); and Model 5–Model 4 factors in addition to HIV and malaria RDT results, which, although becoming more frequently available in resource-limited settings, may not be feasible for the low-level health facilities. For each model, factors were added stepwise (starting with smallest P value in univariate analyses) into a logistic regression model, and variables were maintained in the model as long as the variable contributed to the discrimination of Shigella infections beyond the previously added variables (as measured by a decreasing Akaike Information Criteria [AIC] value estimated at each model iteration) [25]. The AUCs of the 5 models were compared, pairwise, using nonparametric methods based on a generalized U-statistic. Children were reclassified using the model that was a meaningful improvement on the previous model, but it was not statistically different from the subsequent model. The diagnostic accuracy of the newly created model was assessed.

RESULTS

Study Population

Among 2102 screened children, 742 did not meet prespecified inclusion criteria or declined participation (Figure 1). As a result, 1360 children were included in this analysis. Included children were a median of 21 months old (IQR, 11–37 months), approximately half (47.1%) were female, and the majority (94.4%) enrolled from Kisii or Homa Bay hospitals. Less than half (40.2%) reported a monthly household income of less than 5000 Kenyan shillings (∼$60.00), and 40.0% lived in overcrowded households (≥2 people per room). The median WAZ, WHZ, and HAZ was −0.6 (IQR, −1.4 to 0.3), −0.4 (IQR, −1.6 to 0.6), and −0.5 (IQR, −1.6 to 0.6), respectively; 10.4% had an HIV-infected mother, and 3.4% were infected with HIV (Table 1).

Table 1.

Characteristics of Enrolled Children (N = 1360)

Characteristic n (%) 
Median (IQR) 
Sociodemographic 
 Median age (months) 21 (11–37) 
 Male 720 (52.9%) 
 Site 
  Kisii 613 (45.1%) 
  Homa Bay 671 (49.3%) 
  Migori 76 (5.6%) 
 Monthly household income <5000 Kenyan Shillings 545 (40.2%) 
 Crowdinga507 (40.0%) 
 Livestockb ownership 1004 (73.9%) 
 Unprotected water sourcec186 (13.7%) 
 Reports treating drinking waterd1069 (78.9%) 
Clinical Historye
 Bloody stoolf79 (6.5%) 
 History of fever within last 48 hours 597 (43.9%) 
 Antibiotic used in last 7 days 167 (12.3%) 
 Median no. of months exclusively breastfed 6 (4–6) 
 Currently breastfeeding (among ≤24 monthsg545 (74.3%) 
Clinical Presentation 
 Presenting with any IMCI danger signs 396 (29.4%) 
  Unable to drink or breastfeed 55 (4.1%) 
  Excessive vomiting 344 (25.5%) 
  Convulsions 6 (0.5%) 
  Lethargy/Unconscious 24 (1.8%) 
 Presenting with any dehydration signs 419 (30.8%) 
  Restless/Irritable 222 (16.3%) 
  Sunken eyes 255 (18.8%) 
  Drinks eagerly, thirsty 155 (11.4%) 
  Skin pinch goes back slowly 83 (6.1%) 
  Sunken fontanelle 91 (6.7%) 
 IMCI Dehydration Classifications 
  Severeh81 (6.0%) 
  Somei188 (13.8%) 
  None 1091 (80.2%) 
 Stuntedj (HAZ≤) 220 (16.5%) 
 Wastedk (WHZ≤2) 243 (18.0%) 
 Axillary temperature ≥37.5°C at presentation 451 (33.2%) 
 HIV-exposed uninfectedl128 (10.4%) 
 HIV-infected 46 (3.4%) 
 HIV-associated immunosuppressionm16 (37.2%) 
Stool Examinationn
 Blood observed in stool 15 (1.2%) 
 Mucous observed in stool 696 (55.8%) 
 Watery 863 (69.2%) 
Characteristic n (%) 
Median (IQR) 
Sociodemographic 
 Median age (months) 21 (11–37) 
 Male 720 (52.9%) 
 Site 
  Kisii 613 (45.1%) 
  Homa Bay 671 (49.3%) 
  Migori 76 (5.6%) 
 Monthly household income <5000 Kenyan Shillings 545 (40.2%) 
 Crowdinga507 (40.0%) 
 Livestockb ownership 1004 (73.9%) 
 Unprotected water sourcec186 (13.7%) 
 Reports treating drinking waterd1069 (78.9%) 
Clinical Historye
 Bloody stoolf79 (6.5%) 
 History of fever within last 48 hours 597 (43.9%) 
 Antibiotic used in last 7 days 167 (12.3%) 
 Median no. of months exclusively breastfed 6 (4–6) 
 Currently breastfeeding (among ≤24 monthsg545 (74.3%) 
Clinical Presentation 
 Presenting with any IMCI danger signs 396 (29.4%) 
  Unable to drink or breastfeed 55 (4.1%) 
  Excessive vomiting 344 (25.5%) 
  Convulsions 6 (0.5%) 
  Lethargy/Unconscious 24 (1.8%) 
 Presenting with any dehydration signs 419 (30.8%) 
  Restless/Irritable 222 (16.3%) 
  Sunken eyes 255 (18.8%) 
  Drinks eagerly, thirsty 155 (11.4%) 
  Skin pinch goes back slowly 83 (6.1%) 
  Sunken fontanelle 91 (6.7%) 
 IMCI Dehydration Classifications 
  Severeh81 (6.0%) 
  Somei188 (13.8%) 
  None 1091 (80.2%) 
 Stuntedj (HAZ≤) 220 (16.5%) 
 Wastedk (WHZ≤2) 243 (18.0%) 
 Axillary temperature ≥37.5°C at presentation 451 (33.2%) 
 HIV-exposed uninfectedl128 (10.4%) 
 HIV-infected 46 (3.4%) 
 HIV-associated immunosuppressionm16 (37.2%) 
Stool Examinationn
 Blood observed in stool 15 (1.2%) 
 Mucous observed in stool 696 (55.8%) 
 Watery 

Background and Objectives: To identify the effects of global health electives over a decade in a pediatric residency program. Methods: This was an anonymous email survey of the Boston Combined Residency alumni funded for global health electives from 2002 to 2011. A test for trend in binomial proportions and logistic regression were used to document associations between elective and participant characteristics and the effects of the electives. Qualitative data were also analyzed. Results: Of the 104 alumni with available email addresses, 69 (66%) responded, describing 94 electives. Elective products included 27 curricula developed, 11 conference presentations, and 7 academic publications. Thirty-two (46%) alumni continued global health work. Previous experience, previous travel to the site, number of global electives, and cumulative global elective time were associated with postresidency work in global health or with the underserved. Conclusions: Resident global electives resulted in significant scholarship and teaching and contributed to long-term career trajectories.

Categories: 1

0 Replies to “Le A Denno Research Paper”

Leave a comment

L'indirizzo email non verrà pubblicato. I campi obbligatori sono contrassegnati *