Drinking water source and human Toxoplasma gondii infection in the United States: a cross-sectional analysis of NHANES data

Toxoplasma gondii, an obligate intracellular protozoan, is one of the most widespread zoonotic parasites in the world. T. gondii infects all warm-blooded animals, but utilizes domestic and wild cats in the Felidae family as its definitive host [1–3]. T. gondii infects humans via ingestion of undercooked meat containing trophozoites or tissue cysts, or through ingestion of sporulated oocysts found in food, soil, and water recently contaminated with cat feces [1, 4, 5]. When oocysts are ingested by an intermediate host such as humans [6], they migrate in the host’s body and form permanent cysts in various organs and tissues, but most commonly in the brain [7]. Toxoplasmosis is a lifelong infection that has the potential to reactivate and cause severe disease and even death in immunocompromised individuals.

Though primary toxoplasmosis is usually subclinical, sequelae can range from mild flu-like symptoms to seizures and confusion. Severity of symptoms may depend upon the parasite stage ingested, with infections from oocysts believed to elicit more clinically severe manifestations than ingestion of tissue cysts [4]. Immunocompromised individuals can experience more severe symptoms, either during primary infection or following reactivation of latent cysts [8, 9]. Likely attributed to cysts in the brain, psychiatric disorders including mood disorders, personality changes, and schizophrenia have been associated with chronic T. gondii infections; however, the underlying etiology is not entirely clear [10, 11]. Of special public health concern, fetuses can be infected via vertical transmission when a woman experiences a primary infection during pregnancy. Disseminating tachyzoites colonize placental tissues and infect the fetus, potentially causing severe injury, stillbirth, or spontaneous abortion. Congenital toxoplasmosis results in the highest burden of disease for human T. gondii infections [6].

T. gondii has infected an estimated 25-30% of the world’s population [9]; however prevalence varies across countries/geographical regions [12]. North America has a lower prevalence (10-30%) compared to Latin America and tropical regions of Africa, which have a 50-80% prevalence. In addition to felid presence and density, climatic, anthropogenic, economic, social, and cultural factors (e.g. hygienic practices, dietary habits, cooking methods), as well as water quality and sanitation coverage all play important roles in T. gondii environmental persistence, zoonotic transmission, and prevalence in humans [1]. In the United States, toxoplasmosis is responsible for one of the highest direct costs to health care among diseases partially associated with water-borne transmission [13].

Because recent epidemiological studies could not attribute the majority of T. gondii infections to contaminated food consumption or soil-related exposures [14, 15], other sources of contamination are likely starting to play a key role in transmitting T. gondii to humans, especially in the United States [16]. In the environment, oocysts can resist freezing and moderately high temperatures to remain viable for years in soil and water [4, 17]. Water chemical and physical treatments, including chlorination and ozone treatment, are not effective in killing oocysts; however, oocysts are relatively susceptible to ultraviolet radiation [18, 19]. Although modern municipal water treatment systems (with filtering, coagulation, flocculation, and settling) are generally effective at removing oocysts [4], many populations do not have access to drinking water that has undergone this level of treatment. Drinking water is an endemic source of T. gondii in Brazil and Guatemala, and drinking water has been implicated as the source of toxoplasmosis outbreaks in several countries [4], including Panama [20] and Canada [21, 22]. Oocysts have not been directly detected in US drinking water, but assessing environmental contamination of T. gondii is technically difficult [23]. Researchers have hypothesized that ingesting inadequately treated drinking water may be a significant source for human infections [4, 24].

The National Health and Nutrition Examination Survey (NHANES), within the National Center for Health Statistics of the Centers for Disease Control and Prevention, has conducted large scale surveys designed to assess changes over time of the health and nutritional status of adults and children in the United States. Descriptive epidemiology reports have previously reported T. gondii seroprevalence trends for past NHANES cohorts (1988–1994 and 1999–2004) [25–28]. Increased odds of seropositivity were reported with increasing age, poverty, foreign country of birth, lower level of education, living in crowded conditions, and working in soil-related occupations, which varied by race/ethnicity [25]. To assess the disease burden in the United States, a comparative subpopulation analysis of US-born adults aged 12–49 years revealed that T. gondii seroprevalence declined from 14.1% during the Third NHANES (1988–1994) to 9.0% during the Continuous NHANES cycles from 1999–2004 [27].

Because the CDC recently released seroprevalence data for T. gondii among eligible participants in the Continuous NHANES 2009–10 cycle, we examined all available Continuous NHANES data (1999–2004 and 2009–10) to expand on the previous literature and look for evidence of environmental risk factors associated with T. gondii seroprevalence among a nationally representative population [3].

Four 2-year cycles of Continuous NHANES data (1999–2000, 2001–02, 2003–04, and 2009–10) were examined [3]. Available serum from eligible participants 6-49 yrs old (1999–2004) and ≥6 yrs old (2009–2010) were tested by the CDC for immunoglobulin G (IgG) antibodies against Toxoplasma gondii by an enzyme immunoassay (EIA). A dichotomous seropositivity cut-off was provided by the CDC (Continuous NHANES 1999–2004: IgG ≥10 IU/mL = positive; IgG <10 IU/mL = negative; Continuous NHANES 2009–2010: IgG ≥33 IU/mL = positive; IgG <33 IU/mL = negative).

Data were collected using a complex survey design; therefore, seroprevalence estimates were weighted to represent the total US civilian non-institutionalized population, and CDC’s recommended approaches for SAS survey analysis procedures were followed [3]. All analyses were conducted using SAS v9.3 (Cary, NC). Using all of the available NHANES survey data, the authors a priori selected potential covariates of interest to examine associations with toxoplasmosis seropositivity. Of special interest were blood biomarkers (using complete blood count results) to examine clinical severity and exposure to drinking water, because these factors had not yet been thoroughly examined in previous reports. Smoking status was based on serum cotinine level (ng/mL) to delineate light smokers and second-hand exposure (10-99 ng/mL) from smokers (≥100 ng/mL). The NHANES variable of ‘household food security’ was determined using the US Food Security Survey Module. Affirmative responses were counted from 18 items for households with children and 10 items for households without children. Categorical data were derived: no affirmative responses = full food security; 1–2 = marginal food security; 3–5 (without children) or 3–7 (with children) = low food security; 6–10 (without children) and 8–18 (with children) = very low food security. Items included: worried will run out of food, food didn’t last, couldn’t afford meals, adults cut size or skip meals, eat less than should, hungry but didn’t eat, adults don’t eat whole day, and use of emergency foods from food banks, soup kitchens, or other organizations. For our analyses, we combined all categories below ‘full food security’ to create a dichotomous variable.

In the Housing Characteristics (HOQ) questionnaire participants were asked: “What is the source of tap water in this home? Is it a private or public water company, a private or public well, or something else?” and “Are any of the water treatment devices listed on this card used in your home?” Treatment devices listed were: Brita® or other pitcher water filter, ceramic or charcoal filter, water softener, aerator, and reverse osmosis. In order to better understand their interaction, the two variables associated with tap water were combined into one covariate with 4 categorical responses (home-treated well water, home-untreated well water, home-treated water company and home-untreated water company).

Seroprevalence of IgG antibodies against T. gondii was assessed with multivariable logistic regression to examine associations with key demographic characteristics and potential risk factors associated with seropositivity. Wald chi-square odds ratio (OR) estimates and 95% confidence intervals (CI) were ascertained for unadjusted and adjusted regression models. Covariates with p values <0.1 were considered for inclusion in multivariable models. Multivariable models were selected using stepwise backwards elimination of covariates with p values > 0.1. Selected models included all covariates with p values < 0.05 as well as any a priori selected demographic characteristics that confounded observed associations between independent variables and the outcome.

We observed statistically significant increased odds of T. gondii seropositivity among US-born NHANES participants whose tap water source was well water (both home treated and untreated), as well as home-untreated company water, compared to participants who used home-treated water from a private or public water company. The absence of significant associations between tap water source/treatment and T. gondii seropositivity among those not born in the United States provides additional evidence that drinking water may be an important source of toxoplasmosis in the United States.

Although T. gondii oocysts are resistant to chlorine and other types of disinfection [4], because of their relatively large size (10–12 μm), they should be removed by standard municipal treatment consisting of coagulation, flocculation, and settling prior to filtration [29]. While outbreaks of T. gondii have only been attributed to municipal waters using unfiltered water systems (Brazil and Canada), outbreaks could also occur following a water treatment breakdown or operation error [30]. In addition, T. gondii oocysts have been found to be widespread in wells located on farms [31] and can survive for long periods in soil and water [24]. A study conducted among farms in Poland reported an association between well water consumption and toxoplasmosis IgG seropositivity [31]. Not only was a positive correlation observed between drinking well water and T. gondii seropositivity among persons living on the farms (P < 0.05), but T. gondii DNA was found significantly more frequently in water samples collected from windlass-operated wells (37.5%) than water from deep wells with a pump (6.2%) and from water supply systems (0%). Well water, especially private or from a small shared source, is probably much less likely to be filtered adequately to remove T. gondii oocysts. Although we lacked much necessary detail to draw definitive conclusions, our results are consistent with an association between well water and T. gondii infection.

Because the information collected on the type of water treatment was limited, it is difficult to interpret the effect of additional home treatment. Some treatments included in the question such as water softeners, would not have any effect on T. gondii oocysts, and even most water filters are simply designed to reduce taste and odor or reduce some chemicals such as lead and chlorine. Because oocysts are 10–12 μm in size [32], some standard Brita® filters, such as faucet (0.5 micron) and pitcher (1 micron) filters are expected to effectively remove oocysts, but other filters are only designed for taste and odor improvement and would have no effect. Moreover, the effectiveness of home treatments is impacted by maintenance and upkeep of the filter system. The use of additional treatment may also signal increased awareness and concern of water contamination and such participants may be more careful about general hygiene and the water they choose to drink.

Our analyses confirmed previous reports that those born outside the United States had higher odds for T. gondii seropositivity, supporting the evidence that other countries have a higher endemicity of human toxoplasmosis. We further demonstrated that the percentage of one’s lifetime lived in the United States was inversely associated with T. gondii infection. Due to the NHANES classification of country of birth origin, we were unable to conduct meaningful analyses of which countries represent a higher risk of T. gondii infection. Seroprevalence varied among different race/ethnicity groups when stratified by country of origin, which is consistent with previous reports [25, 28]. Stratification by country of origin also suggested blood eosinophils above the normal range and reported treatment for anemia in the past 3 months were associated with T. gondii seropositivity for foreign-born participants, but not for US-born participants. While these factors may simply be correlates of infection and not clinically associated with T. gondii infections, molecular and experimental infection studies have discovered that T. gondii genotypes vary in virulence, and these genetic lineages have evolved in different regions of the world [1]. Highly clonal genotypes II and III, the primary lineages in North America and Europe are considered to be avirulent in immunocompetent persons. In contrast, atypical genotypes resulting from frequent genetic exchanges and genetic drift have been associated with greater clinical severity. High genetic diversity among T. gondii isolates has been found in South America, Africa, and Asia. Because the clinical symptoms of anemia and inflammation were observed only among seropositive foreign-born NHANES participants, perhaps those born outside the United States acquired toxoplasmosis from more virulent strains. Following ingestion, T. gondii parasites invade the gastrointestinal wall, disseminate, and elicit a strong immune response that triggers significant ancillary pathologies; anemia and eosinophilia have been reported in clinically severe cases of toxoplasmosis [33–37].

While the Continuous NHANES collects a wealth of information for a national, representative sample of the United States population, we do recognize its limitations. The cross-sectional survey design prevents us from inferring causality. In addition, combining datasets across multiple survey years can be a limitation, depending on the covariates of interest. Questions and medical exams are not constant across the survey years: response options change, questions/medical tests are added, and questions/medical tests are dropped. In addition, several questionnaires and medical exams are administered to only subgroups (e.g. adults ≥20 years old). Because previous papers of T. gondii seroprevalence among NHANES participants analyzed subgroups that had additional NHANES data, e.g. education level and occupation among adults ≥20 years old [26], our goal was to maintain the largest cohort as possible for consistency and increased power.

Unfortunately, there was very limited data for other risk factors of interest, including the presence of T. gondii in drinking water sources, occupations with soil and water exposure, mental health status (e.g. depression and schizophrenia), and overall dietary behaviors (e.g. eating undercooked meat). In addition, as a health and nutrition survey, NHANES does not collect data regarding exposure to recreational waters, a key interest in environmental T. gondii transmission. Antibodies against T. gondii have been found in several species of marine mammals off of both US coastlines [23], suggesting that freshwater runoff is contaminating sea water [38]. This raises the question about the risk to humans from accidentally swallowing contaminated water during recreational exposure to rivers, lakes, and ocean waters [4]. Despite these limitations, this study provides risk factor data that can inform future environmental assessments and prospective human studies to better link cause and effect for water-associated toxoplasmosis.

Our findings are consistent with national trends observed in previous studies of T. gondii seroprevalence among NHANES participants. Seroprevalence is declining in the United States, with only 6.6% of the US-born population 12–49 years of age having serological evidence of T. gondii infection during 2009–10, compared to 9.0% during 1999–2004 and 14.1% during 1988–1994. We similarly observed differences in seropositivity by country of birth origin and race/ethnicity. There was also a significant association with less than excellent self-reported general health condition and seropositivity. Furthermore, this is the first paper to report associations between tap water source and T. gondii among US-born residents; although, a casual association cannot be demonstrated due to the cross-sectional design of NHANES. In addition, because standard T. gondii serological assays cannot differentiate environmentally-acquired oocyst infections from foodborne tissue cyst infections, recently developed oocyst antigen-specific assays, including the enzyme-linked immunoassay developed by Hill et al. [39], are valuable tools to more accurately assess the disease burden of environmental exposures to T. gondii. In addition, further investigations are warranted to examine how well water may be contaminated with oocysts, what role well systems may play in maintaining and spreading oocysts, and which home water treatment devices are effective in eliminating oocysts from tap water.

The views expressed in this paper are those of the authors and do not necessarily reflect the views or policies of the U.S. Environmental Protection Agency. Mention of trade names or commercial products does not constitute endorsement or recommendation for use.