Shigellosis in an Urban Slum in Kenya: Risk Factors and Antimicrobial Resistance

Abstract

Diarrhea is an important global health challenge, and countries in sub-Saharan Africa and Asia bear the biggest morbidity and mortality burden among all ages [1] due to inefficient water, sanitation and hygiene (WaSH) practices, and limited access to healthcare. Shigella spp are the second most important cause of diarrheal disease and death globally [2]. In 2016, Shigella spp accounted for approximately 13.2% (212 438/1 609 379) of diarrheal deaths across all ages, including 63 713 deaths (41 191–93 611) among children <5 years old [2]. Shigella spp were also associated with diarrhea among adults, with increasing proportions in elderly persons [2]. In developing countries, the incidence of Shigella diarrhea is estimated to be between 0.4 and 43 cases per 100 child-years among children <2 years old [3–6].

Treatment center surveillance conducted from 2007 to 2010 in Nairobi's largest urban slum, Kibera informal settlement, reported a Shigella spp diarrhea prevalence of 24% among all age groups [7]. However, 2 studies found Shigella spp prevalences of 18.7% in rural Kenya [8] and 4% in Nairobi among children <5 years old [9]. Effective antibiotics can shorten symptom duration, reduce severe complications and death, and limit disease spread. However, antimicrobial resistance is a growing concern for management of shigellosis, with 2 studies [7, 8] reporting high resistance (>85%) of Shigella to trimethoprim-sulfamethoxazole. In 2024, the World Health Organization listed Shigella as one of the pathogens that needs to be prioritized for the development of new interventions [10]. Although there is currently no vaccine against Shigella spp, several candidates are at different stages of clinical evaluation [11, 12].

In many settings in lower- and middle-income countries, there is a lack of comprehensive epidemiologic data and understanding of the local disease burden and transmission factors. These are imperative for the effective development of preventive strategies, including vaccines against Shigella. This data would be crucial especially for urban informal settlements, which are expanding rapidly in sub-Saharan Africa, where the morbidity and mortality burden is highest. In the current study, we aimed to determine the isolation rate, antimicrobial resistance rates, risk factors (socioeconomic, behavioral, and WaSH), and clinical characteristics associated with Shigella diarrhea among persons of all ages in Mukuru informal settlement and neighboring villages, by leveraging data collected through a large Salmonella epidemiologic study conducted between 2020 and 2022 [13].

METHODS

Study Site

Mukuru informal settlement (Figure 1) is located 20 km east of Nairobi city. It has a population of approximately 700 000 persons and is subdivided into 9 villages: Mukuru Kwa Reuben, Mukuru Kwa Njenga, Sinai, Paradise, Jamaica, Kingstone, Mariguini, Fuata Nyayo, and Kayaba. The parent Salmonella epidemiologic study from which this Shigella substudy was derived was conducted in 2 villages, Mukuru Kwa Njenga and Mukuru Kwa Reuben, and the surrounding settlements, which have a population of approximately 150 000. The 2 villages have previously been identified as hot spots for diarrheal diseases [14, 15], and they are further subdivided into 9 and 8 zones, respectively. In the study area, families typically have 4–8 members and reside in corrugated iron huts that measure approximately10 × 10 ft². This densely populated informal settlement lacks proper sanitation facilities and solid waste management. Due to inadequate water infrastructure, water sources for the residents are public watering points supplied by the county, and vendors hawk the water to residents.

Figure 1.

Mukuru informal settlement (Mukuru Kwa Njenga and Mukuru Kwa Reuben). The 2 villages are divided into zones. The lines indicate boundaries between the different zones in Mukuru Kwa Njenga and Mukuru Kwa Reuben (sources: Esri, Maxar, Earthstar Geographics, and the GIS User Community).

Open in new tabDownload slide

Recruitment of Study Participants and Sample Collection

We analyzed data collected between November 2020 and December 2022, as part of a larger study investigating the burden of Salmonella disease among participants presenting to treatment settings with fever, diarrhea or both. Participants were recruited from 5 healthcare outpatient facilities (Medical Missionaries of Mary, City council clinic in Embakasi, Mama Lucy hospital, Mukuru Kwa Reuben hospital, and Our Lady of Nazareth hospital) serving the population in Mukuru slums and the immediate surrounding villages. The normal outpatient clinic operation is between 8 Am and 4 Pm, Monday through Friday, and participants were recruited between 8 Am and 1 Pm, Monday through Thursday. Each participant provided a written informed consent before recruitment to the study. Stool or rectal swab samples were then collected and stored in transport medium (Cary-Blair) in cooler boxes before being transported to the Kenya Medical Research Institute (KEMRI) laboratory 17 km away for laboratory analysis, within 4 hours after collection. Stool samples were prioritized. When that was not possible, a rectal swab sample was collected instead.

Inclusion Criteria and Definitions

The parent Salmonella study enrolled individuals of all ages presenting with diarrhea and/or fever who sought medical care at the 5 healthcare facilities serving the population residing in Mukuru informal settlement and the surrounding villages. The same dataset was used for this analysis.

Diarrheal cases were defined as ≥3 episodes of loose stool within the 24 hours before presentation at the health facilities. Fever was defined as a history of fever for 24 hours or an axillary temperature >38°C at presentation.

Assessment of Clinical, Sociodemographic, WaSH, and Environmental Factors

After recruitment on the day of presentation at the healthcare facility, the participant's clinical presentation was assessed by a clinical officer and recorded in a clinical report form. For those who fulfilled the inclusion criterion, a sample (stool/rectal swab sample) was collected. Data on sociodemographic, WaSH and environmental factors, such as contaminated water sources, waste management procedures, drinking water storage, for the household of the patient was collected through an interview by a trained research assistant using the Epicollect5 [16] mobile-based data collection tool.

Laboratory Analysis

Detection and Identification of Bacteria

Stool and rectal swab samples were initially enriched in selenite fecal broth at 37°C overnight. The broth cultures were then subcultured on MacConkey and xylose lysine deoxycholate (Oxoid) agar and incubated at 37°C overnight. Suspected Shigella bacteria (non–lactose fermenting colonies) were subsequently subcultured on Mueller-Hinton agar (Oxoid) and identified through biochemical characterization using triple sugar iron slants and then Analytical Profile Index (API) 20E strips and further speciated using Shigella-specific serologic antiserum.

Antimicrobial Susceptibility Testing

Antimicrobial susceptibility was assessed using the Kirby-Bauer disk diffusion method on Mueller-Hinton agar for all commonly used antimicrobials. The antibiotics used include ampicillin (10 µg), amoxicillin-clavulanic acid (30 µg), co-trimoxazole (25 µg), chloramphenicol (30 µg), gentamicin (10 µg), kanamycin (30 µg), tetracycline (30 µg), ceftriaxone (30 µg), ceftazidime (30 µg), cefpodoxime (10 µg), cefotaxime (30 µg), nalidixic acid (30 µg), ciprofloxacin (5 µg), and azithromycin (15 µg). Results were then interpreted based on the Clinical and Laboratory Standards Institute guidelines (M100) [8]. Escherichia coli (American Type Culture Collection [ATCC] 25922) and Staphylococcus aureus (ATCC 25923) were used to assure the testing performance of the potency of antibiotic discs and the quality of the medium. Isolates expressing resistance to ≥3 classes of antibiotics were designated as multidrug resistant.

Ethical Considerations

Ethical approval to undertake the study was acquired from the KEMRI Scientific and Ethics Review Unit (protocol SERU no. 3582).

Statistical Analysis

Descriptive statistics for each variable are represented as counts and percentages for categorical variables and as measures of central tendency for continuous variables—mean (SD) for normal and median (interquartile range) for skewed distributions. The analysis of the association between Shigella diarrhea and demographic characteristics, household WaSH factors, household behavioral factors, and household socioeconomic factors was performed using diarrheal patients who shed Shigella as case patients and 2 strategies for selection of controls: (1) Shigella negative without diarrhea (with only fever) and (2) Shigella negative with diarrhea. Shigella-negative cases without diarrhea were used to evaluate background predictive factors, such as sociodemographic, WaSH, and household behavioral factors, whereas the analysis of participants' clinical characteristics associated with Shigella positivity was performed using Shigella-negative patients with diarrhea as the control group to look for factors that discriminated between Shigella and non-Shigella diarrhea.

To identify factors associated with Shigella positivity (before adjustment for confounding), crude estimation of the statistical significance of the association between Shigella positivity with each independent variable was done using Pearson χ² or Fisher exact tests (informed by the mean expected counts per cell) for categorical variables; continuous variables were compared using Student t or Mann-Whitney U tests when data were not distributed normally. Crude odds ratios (ORs) with corresponding 95% confidence intervals (CIs), calculated with test-based methods, were estimated to measure the strength of association with Shigella-positive diarrhea for categorical variables.

To identify factors independently associated with Shigella positivity and to control for possible confounding variables, we considered all factors identified at bivariate analysis with a P value <.1 for the association, as well as the age and sex of the participant. A binary logistic regression model was built by stepwise selection of factors performed using both backward and forward conditional methods, with inclusion criteria set at P < .1. Key plausible factors, age and sex, were forced into the model. Two multivariable logistic models were constructed: (1) a model for demographic characteristics, household WaSH factors, household behavioral factors, and household socioeconomic factors, whose interpretation is based on the parsimonious models obtained from the first control strategy; and (2) a separate parsimonious model for subject clinical characteristics, which used the second control strategy. Adjusted ORs (aORs) with 95% CIs were determined by exponentiation of the regression coefficients and use of their standard errors. The threshold for significance was set at P < .05 (2 tailed) for all inferential statistics. Statistical analysis was performed using SAS software.

RESULTS

Frequency of Shigella Diarrhea and Antimicrobial Resistance Patterns

These analyses included 4689 participants, with equal proportions of female and male participants (Figure 2). Children <5 years old constituted 49.3% of the participants, and individuals <16 years accounted for 66.7%. Of the fecal samples from all 4689 participants, 67 (1.4%) were positive for Shigella spp (Figure 2). Among the 67 Shigella spp isolates, the majority (40 [59.7%]) were Shigella flexneri (Table 1). Antimicrobial resistance testing revealed that 34 (85%) of the S flexneri isolates were resistant to trimethoprim-sulfamethoxazole, with 5 (12.5%) resistant to ampicillin, 3 (7.5%) resistant to azithromycin, and 1 (2.5%) resistant to ciprofloxacin. All S flexneri isolates were susceptible to amoxicillin-clavulanate, ceftazidime, ceftriaxone, and cefpodoxime. Most Shigella sonnei isolates were resistant to trimethoprim-sulfamethoxazole 23 (85.2%) and tetracycline 18 (66.7%). In addition, 6 S sonnei isolates (22.2%) were resistant to ampicillin, with 4 (14.8%) showing resistance to each of the cephalosporins (cefotaxime, ceftriaxone, and cefpodoxime), while 1 isolate (3.7%) was resistant to ciprofloxacin. Furthermore, more than half (22 [55%]) of the S flexneri isolates showed intermediate resistance to azithromycin. Similarly, 14 (51.9%) of the S sonnei strains exhibited intermediate resistance to azithromycin, and 3 isolates (11.1%) showed intermediate resistance to ciprofloxacin (Table 1). Of note, multidrug resistance was seen more often in S sonnei (13 [48.1%]) than in S flexneri.

Figure 2.

Shigella diarrhea data analysis (CONSORT flow diagram).

Open in new tabDownload slide

Association Between Sociodemographic and Economic Factors and Shigella Diarrhea

Bivariate analysis of demographic characteristic indicated that although the majority of patients with Shigella diarrhea were female, this association was of borderline significance (crude OR, 1.66 [95% CI, .97–2.87]). (Table 2). Similarly, a borderline significant association was found between Shigella diarrhea and the educational level of the household head (crude OR, 3.73, [95% CI, .74–18.94]; P = .09).

After adjustment for sociodemographic, socioeconomic, and environmental factors, Shigella positivity was significantly associated with older age (aOR, 1.06 [95% CI, 1.03–1.09; P < .001), and residence in makeshift housing made of corrugated iron, timber, or mud walls (2.53 [1.31–4.87]; P = .006).

Associations Between WaSH and Environmental Characteristics; and Shigella Diarrhea

Bivariate analysis of WaSH characteristics revealed that Shigella diarrhea within the informal settlement was significantly associated with nontreatment of drinking water (crude OR, 2.09 [95% CI, 1.20–3.64]; P = .008). Moreover, consuming street food more than once a week (crude OR, 2.99 [95% CI, 1.14–7.85]; P = .02) and uncooked vegetables (3.50 [1.75–6.98]; P < .001) were significantly associated with Shigella diarrhea. The presence of contamination, such as an open sewer system or visibly polluted surface water, within 20 m of a domestic water source, was also significantly associated with Shigella diarrhea (crude OR, 2.00 [95% CI, 1.05–3.85]; P = .03). Finally, while the majority of patients kept animals in the compound, this association was borderline significant (crude OR, 1.72 [95% CI, .99–2.94]; P = .055) (Table 2).

After controlling for sociodemographic, WaSH, and environmental variables, Shigella positivity had a significant association with the consumption of uncooked vegetables (aOR, 3.39 [95% CI, 1.57–7.33]; P = .002). In addition, there was a marginal association with residing in a household that consumed street food at least once per week (aOR, 2.71 [95% CI, .98–7.48]; P = .055).

Clinical Characteristics of Shigella Diarrhea

Bivariate analysis showed that Shigella positivity was associated with fever in the past 3 days (crude OR, 2.28 [95% CI, 1.22–4.27]; P = .008), bloody diarrhea (4.29 [2.44–7.52]; P < .001), severe/moderate dehydration (3.30 [1.01–10.82]; P = .049), coated tongue (8.28 [1.02–67.15]; P = .02), and current high fever (temperature ≥38.0°C) (2.64 [1.42–4.92]; P = .002), (Table 3). Multivariable analysis, adjusting for age and sex of the participants, showed that Shigella positivity was significantly associated only with current high fever (aOR, 2.02 [95% CI, 1.04–3.94]; P = .04).

DISCUSSION

In the current study, we observed that Shigella isolation rates among all individuals presenting with diarrhea was 1.4%, with a rate of 1.5% among children <5 years old. S flexneri was the most commonly isolated species in the Mukuru informal settlement. The antimicrobial-resistant profile showed no resistance to co-amoxiclav, which is commonly used for treatment of shigellosis in Kenya for both S flexneri and S sonnei. However, both species exhibited high resistance (85%) to trimethoprim-sulfamethoxazole, commonly used for uncomplicated Shigella infections due to its effectiveness and affordability. Ampicillin, azithromycin, and ciprofloxacin remained largely effective, though concerning trends of intermediate resistance were observed against azithromycin and ciprofloxacin. Third-generation cephalosporins such as ceftazidime, ceftriaxone, cefpodoxime, and cefotaxime were effective against S flexneri. Notably, S sonnei showed higher multidrug resistance than S flexneri. These findings underscore the critical need for prudent antimicrobial stewardship to combat evolving antibiotic resistance in Shigella species and to optimize patient treatment outcomes. Clinically, Shigella diarrhea was associated with bloody diarrhea, severe/moderate dehydration, and coated tongue.

One limitation of this study was the restricted recruitment schedule, which may have led to the exclusion of potential participants, as the recruitment period did not cover all clinic hours, possibly causing some eligible individuals to be missed. However, we believe that these potential participants did not differ in any respect from those we recruited into the study. In addition, despite conducting 2-year surveillance involving a large population and extensive culture testing, all our analyses ultimately rely on a modest number of Shigella cases. Furthermore, we found an unexpectedly lower prevalence of Shigella diarrhea than previously reported among all age groups (24% in Kibera in 2013) [7] and among children <5 years old (3.2% in Nairobi and 8.4% and 20.1% in rural Kenya) [8, 17, 18]. The prevalence observed in this study was also lower than that observed in other East African countries (Ethiopia, Somalia and South Sudan), which were reported to have a Shigella spp prevalence between 4% and 20.6% [19].

Other studies in sub-Saharan Africa have also reported higher prevalence. The Vaccine Impact on Diarrhea in Africa (VIDA) study reported that the average Shigella prevalence in The Gambia, Kenya, and Mali was 7.4%, but the prevalence in Mali was 0.7% [8]. While the prevalence observed in this study could mean that the disease burden of Shigella diarrhea is low in our study site, one cannot draw this conclusion without population-based incidence data, since a low prevalence rate can still be seen with a substantial incidence rate if the comparative burden of other diarrheal pathogens is high. Future research to measure disease incidence is thus needed, and using more sensitive detections methods would be crucial. Finally, the use of Shigella-negative individuals with fever as a control group poses limitations when drawing inferences about sociodemographic, socioeconomic, and WaSH risk factors in the community. While this group may not be an ideal comparator, it does offer findings that can be corroborated in future studies.

Our analysis underscores the predominance of S flexneri in Shigella diarrhea, with >59% of detected serogroups, consistent with prior studies in Kenya [7, 8, 17, 20, 21] and the Global Enteric Multicenter Study (GEMS) [22]. The VIDA study also reported that the dominant serogroup in the 3 countries was S flexneri [8]; unlike the VIDA study, however, we did not detect Shigella boydii or Shigella dysentriae. These findings are indeed critical in informing vaccine development. We observed high antimicrobial resistance and intermediate resistance rates, with >80% resistance to co-trimoxazole and > 50% intermediate resistance to azithromycin (Table 1), mirroring patterns seen in Kenya, The Gambia, Mali [7, 8], Asia [23], and South America [24]. Of concern is ciprofloxacin-resistant Shigella spp, as ciprofloxacin is an important second-line antimicrobial used in shigellosis management [10], complicating management amid rising azithromycin nonsusceptibility. In addition, the rising azithromycin nonsusceptibility is worrying, given that it is one of the primary antibiotics used in shigellosis management. Although low (about 15%), the resistance to third-generation cephalosporins (ceftriaxone, cefpodoxime and cefotaxime) (Table 1) is notable and needs to be monitored. The increasing antimicrobial resistance is likely to contribute to increase in cost of treatment and therapy failure leading to higher mortality rates attributable to Shigella infections. Consequently, addressing antimicrobial resistance is crucial, given that alternatives are costly or availability is limited, especially in low-resource settings.

While Shigella vaccines are primarily considered for children, our results underscore the importance of targeting interventions across all age groups. Furthermore, our study reveals that distinguishing Shigella from non-Shigella diarrhea is challenging, except for the presence of high fever and dysentery, highlighting the necessity for meticulous clinical assessment and laboratory confirmation for accurate diagnosis and treatment. We identified that improving household factors related to food preparation and handling could potentially reduce the incidence and burden of Shigella infections in the community. In addition, given that the study setting was an informal slum, interventions such as improving the WaSH infrastructure and practices are likely to contribute to reducing the burden of Shigella infections.

Acknowledgments

The authors thank the director general of KEMRI for his support in publishing this data, and they thank Jerome Kim, MD and Florian Marks, PhD from the International Vaccine Institute (IVI) for their support of the proposal. They acknowledge the support from their collaborating institutions, icddr,b, IVI, and the Armauer Hansen Research Institute. The IVI is supported by the governments of Korea, Sweden, India, Finland, Denmark, the Philippines, and Thailand. The icddr,b is grateful to the governments of Bangladesh and Canada for providing unrestricted support.

Disclaimer. The funders had no role in study design, data collection and analysis, the decision to publish, or preparation of the manuscript.

Data sharing. Deidentified individual participant data and a data dictionary can be made available for passive surveillance, serosurvey, census, and census update data on request to the principal investigator ([email protected]) with a research proposal and signed data usage agreement.

Financial support. This work was supported by the National Institute of Allergy and Infectious Diseases, National Institutes of Health (grants R01 AI099525 and R01 AI116917) and the Bill & Melinda Gates Foundation (grant INV-062435).

References

Author notes