Prevalence and Correlates of Cryptosporidium Infections in Kenyan Children With Diarrhea and Their Primary Caregivers

Abstract Background Cryptosporidium is a leading cause of diarrhea in Sub-Saharan Africa and is associated with substantial morbidity and mortality in young children. Methods  We analyzed data from children aged 6–71 months presenting to 2 public hospitals in Western Kenya with acute diarrhea and their primary caregivers, including detection of Cryptosporidium by quantitative polymerase chain reaction (PCR) and immunoassay analysis in stool samples from both children and their caregivers. Associations between potential transmission sources and child/caregiver Cryptosporidium infection were evaluated using prevalence ratios (PRs). Secondary analyses evaluated host and clinical risk factors of child/caregiver Cryptosporidium infection. Results Among 243 child–caregiver pairs enrolled, 77 children (32%) and 57 caregivers (23%) had Cryptosporidium identified by either immunoassay or PCR. Twenty-six of the 243 child–caregiver pairs (11%) had concordant detection of Cryptosporidium. Cryptosporidium infection in children was associated with detection of Cryptosporidium in caregivers (adjusted PR [aPR], 1.8; 95% CI, 1.2 to 2.6; P = .002) and unprotected water source (aPR, 2.0; 95% CI, 1.3 to 3.2; P = .003). Risk factors for Cryptosporidium detection in caregivers included child Cryptosporidium infection (aPR, 2.0; 95% CI, 1.3 to 3.0; P = .002) as well as cow (aPR, 3.1; 95% CI, 1.4 to 7.0; P = .02) and other livestock ownership (aPR, 2.6; 95% CI, 1.1 to 6.3; P = .03) vs no livestock ownership. Recent diarrhea in caregivers and children was independently associated with child and caregiver Cryptosporidium infections, respectively. Conclusions Our results are consistent with the hypothesis that Cryptosporidium transmission can occur directly between child–caregiver dyads as well as through other pathways involving water and livestock. Additional research into caregivers as a source of childhood Cryptosporidium infection is warranted.

Cryptosporidium is a leading cause of diarrhea among children in many resource-limited settings [1,2]. Cryptosporidium is responsible for >60 000 child deaths per year [3] and is also associated with linear growth faltering [4][5][6]. No vaccine is available to prevent Cryptosporidium, and current treatment options are limited, particularly for children with malnutrition or HIV, conditions common in Sub-Saharan Africa [7,8]. An improved understanding of Cryptosporidium epidemiology and transmission dynamics may illuminate opportunities for interventions to reduce Cryptosporidium-associated morbidity and mortality in young children.
Cryptosporidium is transmitted person-to-person (anthroponotic transmission) [9,10], through contact with infected animals (zoonotic) [10], and through ingestion of contaminated water [11]. The predominance of anthroponotically transmitted subtypes of Cryptosporidium observed in Sub-Saharan Africa suggests that person-to-person transmission may be a predominant transmission pathway in this region [10,12].
Caregivers may be a source of Cryptosporidium infection in young children or may have secondary infections from children or other close contacts. In Western Kenya, caregiver HIV infection, a known risk factor of Cryptosporidium susceptibility and prolonged oocyst shedding [13,14], was associated with child Cryptosporidium infection [15], even in the absence of child HIV infection. In addition, other factors such as childhood malnutrition, breastfeeding history, and environmental factors may dramatically affect risk of child Cryptosporidium infection [5,6,16].
Among children presenting with acute diarrhea at 2 hospitals in Western Kenya, we sought to determine the prevalence of Cryptosporidium in accompanying primary caregivers and to identify risk factors for infections in both the children and their caregivers.

Study Design
Between March and December 2015, children aged 6 to 71 months presenting with acute diarrhea (2 or more loose stools per 24-hour period and lasting <7 consecutive days) to 2 public hospitals in Western Kenya (Kisii Teaching and Referral Hospital and Homa Bay County Referral Hospital) were enrolled in this cross-sectional study. Children were excluded if they were not accompanied by a biological parent or legal guardian, if they were unable to provide a stool sample, or if the primary caregiver elected not to receive HIV counseling and testing (if indicated). Sociodemographic characteristics, breastfeeding history, and clinical history of the child and caregiver were collected by a standardized questionnaire, and a brief physical exam of the child was performed by study clinicians. Study staff measured height (or length if <2 years) and weight and assessed danger and dehydration signs according to the World Health Organization (WHO) Integrated Management of Childhood Illness (IMCI) algorithm. Children were managed according to Kenyan Ministry of Health guidelines for diarrhea [17]. Height/length for age and weight for height/length z-scores (HAZ and WHZ) were calculated for children using WHO reference populations, and stunting and wasting were defined as HAZ and WHZ <-2, respectively [18,19]. A body mass index (BMI) <18.5 kg/m 2 in adult caregivers was defined as underweight. Moderate to severe diarrhea (MSD) was defined as diarrhea with signs of dehydration (sunken eyes, loss of skin turgor, intravenous hydration administered or prescribed), visible blood in stool, or hospital admission based on diarrhea or dysentery, as defined elsewhere [20].
A whole stool sample was collected from enrolled children before administration of antibiotic therapy or hospital admission (if applicable). Caregivers were asked to provide a stool sample at the hospital before returning home. Stool samples were accepted from caregivers up to 72 hours after child enrollment. After collection, stool samples were immediately placed in a cool box and maintained at 2-8°C until further processing.

Laboratory Methods
Stool was processed within 2 hours of receipt of the sample. A small portion of the stool sample was used for immediate Cryptosporidium testing using a point-of-care immunoassay (Quik Chek, Alere). The Quik Chek enzyme-linked immunosorbent assay (ELISA) was used in accordance with the manufacturer's instructions [21]. The remaining stool sample was aliquoted into 2-mL cryovial storage containers and frozen to -80°C within 36 hours of enrollment. DNA extraction of stool samples was performed using a QiaAmp stool DNA extraction protocol that included bead-beating for oocyst lysing, and extracted DNA was shipped to the University of Virginia for quantitative PCR using previously described methods [22].
A small amount of blood (<0.5 mL) was collected from caregivers if there was no documentation of HIV status in the last 2 months. If the biological mother was HIV-infected or had an unknown HIV status, blood was collected either by heel or finger prick from the child. Adults and children over 18 months of age were tested for HIV using antibody testing (Abbott Determine rapid test kit) and confirmed using First Response (Premier Medical Corporation). Both tests were performed according to the manufacturer's instructions. HIV DNA PCR assays were performed in children <18 months of age. Children of known HIV-infected biological mothers who were HIV-uninfected themselves were classified as HIVexposed uninfected (HEU).

Statistical Analysis
Cryptosporidium infection was defined as detection of Cryptosporidium antibodies (by immunoassay) or DNA (by PCR at the lower limit of detection [cycle threshold {CT} values ≤40]).
Univariate prevalence ratios (PRs) estimating the association between child and caregiver Cryptosporidium infections were estimated using Poisson regression with robust variance estimates [23]. Multivariable models included detection of Cryptosporidium in children and caregivers, as defined above, as well as other potential transmission sources (water, livestock, inadequate sanitation). As the primary aim of the analysis was to determine potential sources of Cryptosporidium infection, host and clinical characteristics were not included in the primary analyses. However, in exploratory analyses, we examined general host and clinical risk factors for Cryptosporidium infection, separately in children and caregivers, using Poisson regression with robust variance estimates, respectively. Separate multivariable models were constructed for each risk factor, each adjusting for recruitment site as well as child and caregiver age (in child and caregiver models, respectively). Student t tests were used to assess mean CT values by immunoassay result. All analyses were performed using Stata 15.1 (College Station, TX, USA), with an alpha of .05 used to determine statistical significance.

Minimum Detectable Effect
We calculated the minimal detectable effect for analyses examining risk factors for child Cryptosporidium infection. Using our sample of 243 participants, we assumed that 31% of children presenting to a health care facility with diarrhea would have Cryptosporidium detected by ELISA or PCR. Assuming 10%, 25%, or 50% of participants without Cryptosporidium detection have the risk factor, we had 80% power to detect a prevalence ratio of 2.4, 1.8, and 1.4, respectively.
Cryptosporidium in Children With Diarrhea and Their Caregivers • ofid • 3

Patient Consent Statement
The study was approved by the Kenya Medical Research Institute Ethics and Research Committee and the University of Washington Institutional Review Board. Written informed consent was obtained from all primary caregivers for their and their child's participation in the study.

Study Population
Two hundred forty-three children with acute diarrhea and their caregivers were enrolled in the study ( Figure 1). Fifty-three percent (129) of the children were enrolled from Kisii County Hospital, and 94% (227) were accompanied by their biological mother (Table 1). Forty-six percent (112) of the enrolled children were under 2 years of age. Of the enrolled children, 28% (68) had MSD and 27% (66) reported fever at the time of presentation. Two children (0.8%) were HIV-infected, and 26 (11%) were HEU. Caregivers were a median (interquartile range [IQR]) of 28 (23-32) years of age, 12% (28) were HIV-infected, and 3% (7) were underweight. Eight percent (20) of caregivers reported diarrhea in the previous 14 days.
Seventy-three percent (178) owned some type of livestock, with chickens being the most common (168), followed by cows (117). Ninety percent (219) also reported access to pit latrines (vs flush toilet or open defecation). Eighteen (7%) households reported having access to only unprotected water sources (unprotected well, tubewell, borehole, rainwater, surface water). Most caregivers did not report treating their drinking water with methods effective against Cryptosporidium oocyst contamination, boiling or filtering water (33% of caregivers reported boiling, and 2% reported filtering).

Risk Factors for Child Cryptosporidium
Risk factors for child Cryptosporidium infection included detection of Cryptosporidium among their caregiver and using an unprotected water source (Table 3). Children with a Cryptosporidium-infected caregiver were nearly 2 times more likely to have Cryptosporidium themselves (adjusted prevalence ratio [aPR], 1.8; 95% CI, 1.2 to 2.6; P = .002). In addition, living in a household with an unprotected water source was significantly associated with detection of Cryptosporidium in a child (aPR, 2.0; 95% CI, 1.3 to 3.2; P = .003). Livestock ownership did not appear to be an important predictor for child Cryptosporidium infections.
Age, HIV infection and HIV exposure status, malnutrition, breastfeeding history, and severity of diarrhea were not associated with child Cryptosporidium infection in adjusted analyses     (Table 4). Two children in the study were HIV positive without detection of a Cryptosporidium infection, and both caregivers were positive for PCR detection of Cryptosporidium and were HIV-infected and not on HAART.

DISCUSSION
In this cross-sectional study of children presenting to Kenya hospitals with acute diarrhea and their accompanying caregivers, we found that Cryptosporidium prevalence was high. The presence of Cryptosporidium infection in caregivers was a risk factor for infection in their children, and similarly the presence of Cryptosporidium in children was a risk factor for infection in their caregiver. Diarrhea and diarrhea severity were also associated with child and caregiver Cryptosporidium infections, respectively, consistent with possible person-to-person transmission within the child-caregiver pairs. We describe a prevalence (~30%) of Cryptosporidium infection among children with diarrhea on the higher end of the range reported in other studies conducted in Sub-Saharan Africa (13-32%) [12,[24][25][26]. In this study, over a third of caregivers of Cryptosporidium-infected children with diarrhea also had Cryptosporidium. This prevalence (35%) is within the range of 2 recent studies evaluating Cryptosporidium infections in household contacts of children with Cryptosporidium in Bangladesh and in a multicountry study in Sub-Saharan Africa (Gabon, Ghana, Madagascar, and Tanzania) reporting prevalence rates of 51% and 31%, respectively [12,27]. Taken together, these studies highlight that Cryptosporidium is likely present in 1 or more additional household members during episodes of Cryptosporidium diarrhea in a child, which presents challenges for prevention and control strategies.
Studies conducted in Norway [9], Brazil [28], Bangladesh [27], and, most recently, in a multicountry study in Gabon, Ghana, Madagascar, and Tanzania [12] have observed evidence   .81 1.6 (1.0 to 2.5) .04 1.6 (1.0 to 2.5) .04 No. of loose stools in the last 24 h 1.0 (0.9 to 1.1) .40 1.0 (0.9 to 1.1) .77 1.0 (0.9 to 1.1) .63 1.0 (0.9 to 1.1) .87 Duration of diarrhea, d of person-to-person transmission, further supporting our associative findings. Cryptosporidium is characteristically highly infectious and associated with persistent diarrhea [5,28,29] and continued Cryptosporidium oocyst shedding after diarrhea ceases [30,31], providing a long duration of infectivity. We found diarrhea and diarrhea severity to be risk factors for child and caregiver Cryptosporidium infection, respectively, further substantiating a possible person-to-person transmission. Contact with a person with diarrhea has previously been noted as a risk factor for Cryptosporidium infection [32], particularly in outbreak investigations [9,33]. While as much as 50% of Cryptosporidium infection may be asymptomatic, Cryptosporidium infection with diarrhea is more infectious, with increased exposure to the parasite through symptoms and higher parasitic burden relative to asymptomatic infection [5]. Person-to-person transmission is unlikely to be the only source of Cryptosporidium infection. We found livestock ownership and unprotected water source to be important risk factors for caregiver and child Cryptosporidium infection, respectively. Livestock ownership and contaminated water source are well documented as sources of transmission for Cryptosporidium in low-resource settings, although they are inconsistently identified as significant risk factors for Cryptosporidium infection in individual studies [6,25,29,34]. Contaminated water is often associated with Cryptosporidium outbreaks [11,32,35,36]; however, the source of sustained endemic transmission remains unclear and may include a combination of contaminated water, zoonotic, and person-to-person transmission depending on the setting and population.
Cryptosporidium infection is more common and more severe among immunocompromised hosts, particularly adults with HIV and young children with malnutrition [5,37]. However, in this study, neither HIV infection in caregivers nor malnutrition among children was significantly associated with detection of Cryptosporidium. The lack of association may, in part, be due to the relatively small number of HIV-infected caregivers not on antiretroviral therapy (ART) and children with acute malnutrition, which limited our statistical power to detect an association. The use of antiretrovirals for treatment and improved immune function of people with HIV appears to have reduced the prevalence of Cryptosporidium infection among persons with HIV [38,39]. Our results did show that 100% of the 2 HIV-infected children had caregivers with PCR-detected Cryptosporidium but, little can be drawn from such small numbers.
The concordance between the point-of-care ELISA test and PCR results was consistent with previous research. PCR is known to be more sensitive than most immunoassays, often detecting small quantities of Cryptosporidium DNA in diarrheal episodes likely caused by another pathogen [40]. In our study, Cryptosporidium was detected in samples from 2 asymptomatic caregivers by immunoassay and not by PCR. The 2 immunoassay positives may have been false positives or false negatives by PCR, both of which have been reported, although not frequently, in the literature [21,40,41].
This study has several important limitations. Due to the cross-sectional nature of the study design and lack of genotyping, we were unable to confirm transmission of Cryptosporidium infection within the child-caregiver pair, or the direction of the transmission. Further, household members other than the child and caregiver were not included in the study; as such, we are unable to make inferences about the caregiver-child pair relationship in context of other persons living in the household. Additional research should assess the importance of the primary caregiver relative to other household members as a source of Cryptosporidium infection for young children. Lower numbers of children with acute malnutrition and immunocompromised caregivers would explain our inability to detect a significant association between these host factors and Cryptosporidium infection.
The results of this study suggest that the child-caregiver dyad should be further explored as a source of household person-to-person transmission. If transmission between young children and their primary caregivers, which in this context is mostly biological mothers, is common, effective disease control strategies may need to focus on prevention and treatment of infection among caregivers to effectively reduce Cryptosporidium-associated disease, malnutrition, and mortality in children.