Cryptosporidium Prevalence and Risk Factors among Mothers and Infants 0 to 6 Months in Rural and Semi-Rural Northwest Tanzania: A Prospective Cohort Study

Background Cryptosporidium epidemiology is poorly understood, but infection is suspected of contributing to childhood malnutrition and diarrhea-related mortality worldwide. Methods/Findings A prospective cohort of 108 women and their infants in rural/semi-rural Tanzania were followed from delivery through six months. Cryptosporidium infection was determined in feces using modified Ziehl-Neelsen staining. Breastfeeding/infant feeding practices were queried and anthropometry measured. Maternal Cryptosporidium infection remained high throughout the study (monthly proportion = 44 to 63%). Infection did not differ during lactation or by HIV-serostatus, except that a greater proportion of HIV-positive mothers were infected at Month 1. Infant Cryptosporidium infection remained undetected until Month 2 and uncommon through Month 3 however, by Month 6, 33% of infants were infected. There were no differences in infant infection by HIV-exposure. Overall, exclusive breastfeeding (EBF) was limited, but as the proportion of infants exclusively breastfed declined from 32% at Month 1 to 4% at Month 6, infant infection increased from 0% at Month 1 to 33% at Month 6. Maternal Cryptosporidium infection was associated with increased odds of infant infection (unadjusted OR = 3.18, 95% CI 1.01 to 9.99), while maternal hand washing prior to infant feeding was counterintuitively also associated with increased odds of infant infection (adjusted OR = 5.02, 95% CI = 1.11 to 22.78). Conclusions Both mothers and infants living in this setting suffer a high burden of Cryptosporidium infection, and the timing of first infant infection coincides with changes in breastfeeding practices. It is unknown whether this is due to breastfeeding practices reducing pathogen exposure through avoidance of contaminated food/water consumption; and/or breast milk providing important protective immune factors. Without a Cryptosporidium vaccine, and facing considerable diagnostic challenges and ineffective treatment in young infants, minimizing the overall environmental burden (e.g. contaminated water) and particularly, maternal Cryptosporidium infection burden as a means to protect against early infant infection needs prioritization.


Introduction
The World Health Organization reports that the most common diarrhea-causing protozoan parasite worldwide is Cryptosporidium [1], and a recent, large, multi-country investigation reported Cryptosporidium as the second most common pathogen indentified among care-seeking African and Asian infants 0 to 11 months [2]. The significance of this infection was underscored as this study revealed that infection was associated with a greater than two-fold increase in mortality of children 12 to 23 months [2]. Despite these indications of the potential global scope and impact of Cryptosporidium infection, a full understanding of the epidemiology of infant and early childhood infection remains limited due to logistical and methodological difficulties in conducting such research in impoverished high-burden urban and rural settings [2].
Cryptosporidium is a pathogen transmitted via the oral-fecal route from human (C. hominis) and animal (predominately C. parvum) reservoirs. Infection risk factors include a contaminated environment with elements such as: unsafe water, poor sanitation and hygiene, and close proximity to infected livestock, while severe clinical disease risk factors include: malnutrition and compromised immunity, particularly HIV-associated immunosuppression [1,3]. Symptoms include: nausea, vomiting, voluminous and watery diarrhea, dehydration, abdominal discomfort, anorexia, fever, fatigue, and respiratory problems [4,5], with chronic and lifethreatening symptoms possible amongst immunocompromised individuals due to the increased duration and severity of illness [4]. However an unknown number of individuals experience asymptomatic Cryptosporidium infection [6]. This clinically silent infection may remain undetected and untreated and therefore may contribute to malnutrition, growth impairment, and longterm cognitive and functional deficits in infants and children [7,8].
The primary aim of this research was to determine the prevalence of Cryptosporidium in young infants living in rural and semi-rural Tanzania by indentifying the timing of the first and subsequent Cryptosporidium events in both symptomatic and asymptomatic infections. Secondarily, we aimed to evaluate potential infant infection risk factors including: infant nutritional status, infant feeding practices, infant HIV-exposure, maternal nutritional status, maternal HIV infection, and, uniquely, maternal post-partum Cryptosporidium infection.

Methods
This study was a prospective birth cohort enrolling newborns and their HIV-seropositive or -negative mothers living in the rural and semi-rural areas of Kisesa Ward (population 30,000) [9] in northwestern Tanzania. Pregnant women receiving antenatal care at Kisesa Health Centre (KHC), a Tanzanian governmentadministered, publically accessible primary care facility were recruited from March through December, 2012, a period that included both the dry and rainy seasons. Women gave birth between April, 2012 and January, 2013; the study follow-up appointments for mothers and infants were conducted between May, 2012 and July, 2013. Eligibility criteria were gestation ,37 weeks at consent, singleton birth, known maternal HIV serostatus (screening with Determine HIV-1/2 [Inverness Medical], confirmation with Uni-Gold HIV-1/2 [Trinity Biotech]), maternal ability to speak and understand the local language of Kiswahili, and stated intention to reside within the clinic catchment at delivery and through six months post-partum. The study was advertised through health workers at KHC as well as rural government-run health dispensaries in the region. All HIVpositive women were receiving anti-retroviral treatment (ART) for their own care or for prevention of mother-to-child transmission by the time of delivery. Infants born to HIV-positive women were given nevirapine daily for six weeks and tested for HIV-infection by dried blood spot DNA-PCR at the regional hospital laboratory at the Month 3 follow-up visit. The study protocol was approved by the ethics review committees of the Tanzania National Health Research Ethics Review Committee and Cornell University. Written informed consent was obtained from mothers for themselves and on behalf of their infants at enrolment with verbal assent re-confirmed at follow-up.
All women were encouraged to deliver at KHC unless otherwise medically advised. As many women in this region do not deliver at health clinics, and preliminary research revealed that transportation expenses were the primary barriers to accessing healthcare [10], the study provided transportation compensation and other clinical expenses typically borne by mothers for delivery and follow-up visits. For women who delivered elsewhere, including home births, mothers and infants were requested to attend a follow-up clinic visit within three days of delivery. The study flow chart is summarized in Figure 1. If a mother-infant pair did not return for a regularly scheduled follow-up visit, a field worker traveled to their last known address to invite them to return to the clinic for a follow-up appointment.
At each follow-up, the research nurse, under supervision of the study coordinator, administered the Infant Feeding and Health Questionnaire to mothers. This questionnaire was designed to obtain data on a range of feeding, health, and environmental risk factors. Exclusive breastfeeding (EBF-WHO) was defined according to the WHO definition where ''the infant receives breast milk (including expressed breast milk or breast milk from a wet nurse) and allows the infant to receive oral rehydration solution (ORS), drops, syrups (vitamins, minerals, medicines), but nothing else'' [11]. Duration of EBF-WHO was defined as the time from birth until an infant first received food or liquids other than breast milk or medicines. Diarrhea was defined as loose or watery stools $ three times per day that represented a pattern atypical for that individual [2]. The questionnaire included: 1) infant nutrition: breastfeeding and complementary feeding practices; 2) motherreported infant morbidity: cough, difficulty breathing, fever, convulsions, vomiting, skin rash, anorexia, unscheduled clinic/ hospital visits, and episodes of diarrhea; and 3) environment: food security, using an index composed of questions relating to the mother's food consumption pattern, and sanitation and hygiene practices, such as hand-washing behavior, access to safe water, and toilet facilities. Infants exhibiting symptoms of illness were referred to the clinical officer at KHC for follow-up.
Anthropometric assessments were collected at each follow-up visit. Maternal height and weight were measured using a standard stadiometer (Health O Meter, Inc., Bridgeview, IL) to the nearest 0.2 kg and nearest 0.1 cm, respectively. Maternal mid-upper arm circumference (MUAC) and triceps skinfold thickness (TSF) were measured to the nearest 0.1 cm and 0.5 mm, respectively. Infant weight and length were measured using a calibrated digital infant scale (Seca 334 Digital Baby Scale) to the nearest 0.01 kg and a standard infant length board to the nearest 0.1 cm, respectively. Infant MUAC, TSF, and head circumference were measured to the nearest 0.1 cm, 0.5 mm, and 0.1 cm, respectively.
Active case detection was of interest so maternal and infant fecal samples were collected irrespective of self-reported intestinal symptoms at each follow-up visit. Cryptosporidium infection was detected using fresh stool samples that were stored in a cooler with ice packs for #5 hours before being transferred and stored at 4uC

Author Summary
Early infancy and childhood Cryptosporidium infection is associated with poor nutritional status, stunted growth, and cognitive deficits, yet minimal research is available regarding the burden and risk factors worldwide. Since there is no vaccine available, and because diagnostic challenges exist and treatment for children younger than one year is ineffective, prevention of early infancy infection through a better understanding of basic epidemiology is critical. This study was designed to investigate symptomatic and clinically silent infection amongst HIV-seropositive and HIV-seronegative mothers and their infants in a longitudinal cohort, and to indentify potential risk factors. Findings indicate that infants are living in a Cryptosporidium environment as demonstrated by the chronically high level of maternal infection throughout the 6-month postpartum period. Despite this, infant infection prevalence remains low until six months of age when it dramatically rises. The increase in infant infection corresponds to a reduction in exclusive breastfeeding. As expected, maternal infection is associated with increased infant infection, but unexpectedly, so is maternal hand washing prior to infant feeding. Since prevention may indeed be the ''best medicine'' for infants, investigation of beneficial breastfeeding practices, protective correlates in breast milk, and ways to reduce the maternal and environmental Cryptosporidium burden are needed.
in the parasitology laboratory of the Tanzanian National Institute for Medical Research (NIMR), Mwanza Research Centre. Within 24 hours of collection, approximately 5 g of stool was mixed with 5 mL 10% v/v formalin and stored at 4uC until analysis. Presence of Cryptosporidium was confirmed using a modified Ziehl-Neelsen staining procedure [12], which is estimated to have a sensitivity ranging from 32 to 79% and a specificity ranging from 89 to 100% [13][14][15]. After staining, slides were examined by a single technician, without knowledge of participant clinical status, using a light microscope (Olympus model CX41RF) to detect Cryptosporidium oocysts and estimate oocyst burden. Cryptosporidium infection was defined as $1 oocyst detected in stained fecal smears. A second technician re-examined a sample (10%) of the slides and inter-observer agreement was 96%.
Data were analyzed in STATA10 (STATA Corporation, Texas, USA). Means of normally distributed continuous variables were compared using Student's t-test and proportions of categorical variables were compared using the x 2 test and Fisher's Exact test. Results were considered statistically significant at a = 0.05, twosided. Univariate and multivariate logistic regression models were used to estimate the odds ratio (OR) and 95% confidence interval (95% CI) of a priori considered potential risk factors for infant Cryptosporidium infection (HIV-exposure, exclusive breastfeeding, maternal Cryptosporidium infection, and household factors, such as animal ownership, sanitation, wealth, and maternal education).
This study is registered with ClinicalTrials.gov, number NCT01699841.
The sponsors (Cornell University and the National Science Foundation) were not involved in the design or oversight of the study. Members of the writing team had full access to the study data. The authors had final responsibility for the decision to submit for publication.

Results
During the study period, 108 infants were born, and of these, six infants exited the study because of death, migration, or withdrawal of consent prior to the Month 1 study visit ( Figure 1) and were not included in follow-up analyses. Birth anthropometrics were statistically different between HIV-exposed and HIV-unexposed infants (Table 1). A greater proportion of HIV-exposed infants had low birth weight (LBW; defined as birth weight ,2500 g) compared to HIV-unexposed infants (HIV-exposed vs HIVunexposed = 15 vs 3%, respectively; p = 0.026). Likewise, a greater proportion of HIV-exposed infants were stunted at birth (defined as birth length ,44.7 cm) compared to HIV-unexposed infants (HIV-exposed vs HIV-unexposed = 18 vs 2%, respectively; p = 0.004). No HIV-exposed infant tested positive for HIV between birth and three months of age. Maternal and household characteristics did not differ based on HIV-status of the mother, The proportion of all mothers (HIV+ and HIV2 combined) infected with Cryptosporidium ranged from a low of 44% (31/70) at Month 1 to a high of 63% (45/71) at Month 6 post-partum, and this proportion was not statistically different across time points. The majority of all mothers experienced Cryptosporidium infection at some point during the study follow-up period, with 82% experiencing Cryptosporidium infection at least once and 16% infected at every time point. Self-reported diarrhea was not related to Cryptosporidium infection and symptomatic infection ranged from a low of 0% at Month 3 to a high of 14% at Month 6. While the majority (60%) of mothers experienced self-recovery from Cryptosporidium infection between visits based on the presence/ absence of oocysts in their feces, 15% of mothers who recovered later became re-infected on a subsequent visit. All infants remained free from Cryptosporidium infection until Month 2 and infection remained uncommon through Month 3. By Month 6, the increase in infection was dramatic with 33% (23/69) of infants exhibiting evidence of infection. Statistically significant differences in maternal Cryptosporidium prevalence based on HIV-serostatus were not evident, with the exception of the Month 1 study visit (p = 0.012) (Figure 2). There were no statistically significant differences in infant Cryptosporidium infection based on HIVexposure (p = 0.284).
As overall EBF-WHO declined, the proportion of infant Cryptosporidium infection increased (Figure 3  months, none had evidence of Cryptosporidium infection even though they were living in a Cryptosporidium environment as confirmed by evidence of maternal Cryptosporidium infection in all four cases. There was a pattern of lower proportion of Cryptosporidium infection in infants with a greater proportion of the diet consisting of breast milk (EBF-WHO vs. partial/no breastfeeding) and this was significant at Month 6 (p = 0.030) (Figure 4). Neither maternal nor infant Cryptosporidium infection was associated with reported symptoms of infection that included diarrhea, anorexia, vomiting, and in mothers only, abdominal pain and nausea. Care-seeking behavior, operationalized as an unscheduled clinic or hospital visit, was uncommon for both mother (4%) and infant (8%) between each scheduled follow-up visit and was not associated with Cryptosporidium infection.  infection. Maternal food security index was negatively correlated with the practice of EBF-WHO at each visit; meaning that the more food secure a household, the less likely the infant was EBF-WHO. Similarly, the wealthier a household, the less likely the infant was EBF-WHO and the more educated a mother, the less likely the infant was EBF-WHO.

Discussion
This is the first report of maternal-infant Cryptosporidium infection in Sub-Saharan Africa and the prevalence of infection was high. Post-partum infection was detected at least once in the majority of women and, for many, on multiple occasions. The Cryptosporidium burden in infants increased dramatically between three and six months of age, a period that corresponds to changes in breast feeding practices. Our results indicate that young infants living in rural and semi-rural Tanzania are susceptible to Cryptosporidium infection in early infancy with approximately 1/ 3 of infants showing evidence of infection by six months of age.
This study confirms and extends the importance of Cryptosporidium infection in young infants reported in the GEMS study [2] that included both rural and urban settings. Our results are comparative to the findings of a sub-sample of young Tanzanian infants in urban, hospital-based studies where 25% of infants 0 to 6 months had evidence of either G. lamblia or Cryptosporidium parvum [16], though the burden of Cryptosporidium parvum was not individually reported. In studies conducted in the Tanzanian capital of Dar es Salaam, only 9% of children three months to nine years and 18.9% of children 0 to 60 months had evidence of Cryptosporidium infection [5,16] and this may represent an urbanrural difference in young infant burden in Tanzania.
Previous studies in Tanzania of HIV-positive adults report a Cryptosporidium prevalence between 7 and 17% [5,6] and HIV infection has been identified as a risk factor for Cryptosporidium and cryptosporidiosis in some studies [6,[17][18][19][20] but not others [2]. Maternal HIV infection did not appear strongly related to Cryptosporidium infection in our study and this may be explained in part because the majority of HIV-positive women were otherwise healthy and not severely immunocompromised based on their CD4 cell counts. Previous studies that identified HIV infection as a risk factor were primarily conducted in the pre-ARV era and greater immunosuppression may explain differences [6,17,20]. Likewise, HIV-exposure was not a significant risk factor for Cryptosporidium infection in infants and this might be due in part to more optimal feeding methods in the HIV-exposed infants due to infant feeding counseling for HIV-positive mothers. While HIV infection may not be a significant risk factor for infection in this setting, it remains relevant for the clinical management of cryptosporidiosis in immunocompromised individuals given the lack of effective Cryptosporidium treatment other than ARV's to improve HIV immunocompetency [21].
While maternal Cryptosporidium infection was associated with greater infant infection, previously (or even currently) infected mothers may also be providing protective passive immunity in utero or in breast milk. A recent study of Bangladeshi infants reported that protection from Cryptosporidium infection was associated with high anti-Cryptosporidium IgA in breast milk [22]. Despite possible passive immunity and/or risk elimination (from contaminated food/water), EBF-WHO was uncommon in our study population and was not sustained for the universally recommended duration of six months. In this study, using the WHO definition of EBF, only a third of mothers were practicing EBF-WHO at Month 1. Previous Tanzanian studies indicated much higher levels of ''EBF'' ranging from 49% within 3 days after birth [23], 90% at Month 1 [24], and 80% at Month 2 [25], but these large differences are likely due to the less strict non-WHO-EBF definitions and/or maternal recall methods used [24,25]. Additionally, two of these studies included HIV-positive women only and HIV maternal care includes infant feeding counseling that is typically unavailable to HIV-negative mothers in this setting [24,25]. Indeed, we found significantly higher rates of EBF-WHO in HIV-positive mothers and this may explain why infant HIV-exposure was associated with lower infant Cryptosporidium infections.
Globally, knowledge of the epidemiology of Cryptosporidium infection in early infancy is scarce and, in Tanzania, such data are unavailable. When the lack of prevalence data is combined with barriers to diagnosis, the disease rarely features on the clinician's diagnostic radar. This leads to a cycle that likely perpetuates the underestimation of the Cryptosporidium burden leading to an inappropriately lower global health and research priority. This cycle reinforces ineffective clinical and public health management of Cryptosporidium. In our study, maternal hand washing prior to infant feeding was counterintuitively associated with infant infection, although given the wide 95% confidence interval, we recommend caution in the interpretation of this finding. Previous studies have indicated that household sanitation and hygiene, including hand washing, were related to reduced Cryptosporidium infection [17]. Since Cryptosporidium has notoriously robust survival and transmissibility [26,27], and mothers may wash their hands with contaminated water and then feed their children, our result is plausible in this setting. It may also be that the practice of hand washing is a proxy indicator for women who lived in more contaminated environments. Further research could include testing water sources and/or analysis of the species of Cryptosporidium in order to determine probable transmission routes of infection. Such investigations would help interpret this finding in relation to major public health messages related to hand washing in similar settings.
Our study had a number of limitations. First, at each follow-up visit, only one stool sample was collected from each mother and infant. Due to the intermittent shedding of Cryptosporidium oocysts, collection of a single stool sample may result in an underestimate of the true Cryptosporidium prevalence [28]. Additionally, our study used modified Ziehl-Neelsen staining, the most common diagnostic technique to detect the presence of Cryptosporidium oocysts in stool samples, however, the sensitivity and specificity of this method are not 100% leading to possible misclassification [13]. Lastly, our results may not be generalizable to other geographical settings due to urban/rural differences and geographical variation in Cryptosporidium contamination.
In conclusion, there is a high prevalence of infant and maternal Cryptosporidium infection in this setting. Public health interventions promoting EBF-WHO among all women, including HIVnegative mothers should be strengthened. Modeling the message of breast milk as an immunologically protective substance to prevent certain infectious diseases common in childhood may be effective in regions where there are high rates of vaccination coverage. Additionally, further research is needed to address efforts to minimize the maternal and environmental Cryptosporidium burden as a means of protecting young infants in the absence of effective vaccines, diagnostics, and treatment for early infancy cryptosporidiosis.