Poultry Ownership in Urban Kenya is Associated with Increased Fecal Contamination in Household Soil

Diarrheal disease was the third leading cause of under-5 child mortality worldwide in 2017.¹ Exposure to enteric pathogens in animal feces has been associated with diarrheal disease in humans, and roughly 60% of known infectious diseases can be transmitted zoonotically.^2,3 Previous studies in low-income, rural households in Kenya, Bangladesh, and India detected Escherichia coli, a fecal indicator bacteria (FIB) of which some strains are diarrheagenic, originating from humans and domestic animals in household environments.^4–7 In Dhaka, Bangladesh, FIB and a ruminant-associated bacterial target were identified in hand rinses and floor samples from households both owning and not owning ruminants, suggesting that fecal contamination from animals is widespread in low-income, urban settings.⁸ Urban settlements are characterized by overcrowding and less land area per household; animal rearing practices differ from rural areas and humans may have high contact with animals because there is less space to separate animals from living quarters or for animals to graze.⁹ As urban migration increases, there is a need to better understand animal husbandry practices and transmission pathways of zoonotic enteropathogens in densely populated areas. We used household survey data and environmental (soil, water) samples in urban Kenya to assess (i) poultry husbandry practices and (ii) whether household poultry ownership was associated with E. coli contamination in environmental samples.

Between June–August 2019, 120 eligible households including at least one under-5 child were enrolled from two subcounties in Nairobi, Kenya: Kibera (n = 64) and Dagoretti South (n = 56) (Figure S1). Systematic sampling efforts were made to ensure approximately half of the households were from poultry-owning compounds. For every compound enrolled with poultry on a street, another compound without poultry was enrolled. Only one household was enrolled per compound. Household surveys and sample collection were conducted by trained field officers and assistants. Surveys with the primary caregiver of an under-5 child recorded household demographics; water, sanitation, and hygiene access and practices; and animal husbandry practices. A household soil and stored water sample was then collected from each household. Free chlorine levels were measured in a subset of stored water samples (47.5%; n = 57) to characterize chlorine levels in villages (see Supplementary Information for additional sample collection details). Samples were transported to Kenya Medical Research Institute (KEMRI) on ice and were processed within 6 hours of collection. E. coli were cultured on Tryptone Bile X-glucuronide (TBX) agar by incubation for 18 hours at 44.5 °C. Specifically, 100 mL of undiluted water samples were membrane filtered and incubated on TBX. Bacteria were eluted from approximately 5.0 g of each soil sample using 50 mL of distilled water and vortexing vigorously for 2 minutes. A 10^-2 dilution was prepared for each eluted sample; 10 mL of the eluant was membrane filtered and incubated on TBX. Following incubation, we enumerated E. coli from TBX plates with up to 500 colony-forming units (CFU). Written informed consent was obtained from all participants, and the study was approved by the KEMRI Scientific and Ethics Review Unit (12/3823) and Tufts Health Sciences Institutional Review Board (13205). The research permit was granted by the Kenyan government through the National Commission for Science, Technology and Innovation (19/2219/29298).

We conducted multivariable linear regressions to determine whether household poultry ownership was associated with (1) E. coli prevalence in household stored water and (2) log₁₀ E. coli concentration in household soil (CFU per dry gram). In a secondary analysis, we used the number of poultry owned by the household as our independent variable. For both analyses, we pre-screened the following covariates, which had <10% missingness and >5% prevalence, for association with each outcome: subcounty; number of households in the compound; number of household residents; asset ownership (television, bike, gas cooker); whether the respondent completed primary education; whether the household had a finished floor (concrete, tiles), obtained water from a piped source, or reported treating stored water; and whether the stool of at least one under-5 child in the household was safely disposed the last time the child passed stool. We retained covariates with p ≤ 0.1 in the models. We modeled E. coli prevalence using Poisson regression and log₁₀ E. coli concentration using a generalized linear model. All analyses were conducted in R (version 3.6.2).

Forty-nine percent (n = 59) of households owned poultry (primarily chickens); ownership of other animals (ruminants, dogs, cats) was less common (Table 1). Poultry-owning households owned 8 poultry on average (median: 4 poultry). Households owning and not owning poultry were similar across multiple sociodemographic indicators. Most households had a concrete floor (88.3%), obtained water from a piped source (74.2%), and had access to an improved latrine (95.0%). About half of households reported safely disposing the stool of young children. Alternatively, poultry-owning households had higher ownership of assets compared to non-poultry-owning households. Additionally, poultry-owning households more commonly had feces present around the household soil sampling area—although feces were still observed around some non-poultry-owning households.

The following statistics pertain only to households in poultry-owning compounds (55.0%; n = 66) (Table 2). The most common purposes of poultry rearing were meat, eggs, and income generation. Poultry were most commonly free-ranging and/or fed food scraps and maize; poultry were less commonly fed commercial feed. About half of households reported giving poultry medicine in the past, with antibiotics and vaccines being the most common types of medications administered. Poultry entered more households in Kibera (63.9%; n = 23) than Dagoretti South (53.3%; n = 16), which is less densely populated than Kibera. Among these households, more households in Kibera (69.6%; n = 16) kept poultry in the sleeping quarters of under-5 children compared to Dagoretti South (31.3%; n = 5). Rearing practices of allowing chickens to roam freely around both the household and compound were similar in peri-urban Kisumu, Kenya⁹, suggesting that our study site was representative of other urban sites in Kenya. The household respondent cared for poultry in 84.9% of compounds; other adults and older children from within and outside the household also participated in poultry caretaking. Other studies also observed that women and children often care for poultry^10,11, suggesting that these two populations may experience a higher risk of exposure to poultry feces.

E. coli was detected in 36.7% (n = 44) of stored water samples (range: 100–50,119 CFU/100 ml) (Figure 1). The prevalence of E. coli in water was not higher in poultry-owning households than non-poultry-owning households (40.7% vs 32.8%; χ² = 0.3; p = 0.6). E. coli was detected in 85.0% (n = 102) of household soil samples (range: 50– 501,187 CFU/dry g soil), and the prevalence of E. coli in soil was significantly higher in poultry-owning than non-poultry-owning households (94.9% vs 75.4%; Fisher’s exact; p < 0.01).

• Download figure
• Open in new tab

Figure 1.

E. coli presence in (a) household stored water and (b) household soil samples stratified by household poultry ownership. Forty-nine percent of households (n = 59) owned poultry (n = 31 in Kibera, n = 28 in Dagoretti South). Overall, E. coli was detected in 36.7% (n = 44) of household stored water samples; 3.3% of households (n = 4) did not have stored water E. coli data. In household soil samples, E. coli was detected in 85.0% (n = 102) of households overall.

Adjusting for significant covariates, household poultry ownership was not associated with E. coli prevalence in household stored water but was associated with increased log₁₀ E. coli concentration in household soil (Table S1). Poultry-owning households had a 0.67 (95% confidence interval: 0.27–1.07) log₁₀ CFU/dry g increase in soil E. coli compared to non-poultry-owning households. The number of poultry owned by a household was also positively associated with E. coli concentration in household soil; there was a 0.03 (95% confidence interval: 0.00–0.05) log₁₀ CFU/dry g increase in soil E. coli for each additional poultry animal owned. Similarly, a previous study in Kibera found that ownership or presence of chickens in the compound was associated with increased Campylobacter jejuni, an enteric pathogen, in household soil.¹² Researchers also observed a correlation between animal ownership and soil E. coli in dense communities of Harare, Zimbabwe.¹³

Our study has limitations. Chlorine was detected in some household stored water, but we could not include free chlorine residual as a covariate in our models because chlorine measurements were only collected from a small subset of households. Additionally, we were limited by sample size. Future studies with a greater sample size should investigate the relationship between E. coli contamination and indicators other than poultry ownership (e.g., day and nighttime corralling practices, animal feces disposal practices). Finally, we only measured FIB instead of actual pathogens. In urban settings where animals live near humans, investigating animal management practices may be especially important for understanding household- and community-level transmission of zoonotic enteropathogens.

Poultry ownership in dense settings may promote zoonotic disease transmission. In urban Kenya, household soil may be an important transmission pathway for poultry-associated fecal contamination. Finished flooring (concrete, tiles) is one intervention that has been hypothesized to disrupt transmission of fecal contamination from soil. However, the prevalence of finished flooring was high in our study, suggesting that areas immediately outside of the living quarters where people spend a lot of time (e.g., cleaning, playing) could still be an important place for human exposure to poultry fecal contamination. Additional research is needed to understand how animal husbandry practices influence environmental contamination in urban settlements in order to inform potential interventions for disrupting zoonotic disease transmission in low-income households.