The association between Toxoplasma gondii infection and asthma in the United States: A cross-sectional survey analysis

METHODS

We used a cross-sectional study design to examine a public database of U.S. residents aged 6 to 80 years or older from the 2012-2014 survey cycles of the American National Health and Nutrition Examination Survey (NHANES). This is a continuous survey based on a stratified multistage clustered probability design to select a representative sample of the civilian, noninstitutionalized U.S. population (9). Identification of individuals is not possible from these data. Only subjects 6 years of age or older were tested for T. gondii. Demographic, socioeconomic, and health-related information were taken from the NHANES Questionnaire, and serological data from the NHANES surplus sera program (2012-2014). The data were accessed on October 16^th, 2023.

The University of British Columbia’s Policy LR9, item 7.10.3, on studies involving human participants and Article 2.2 of the Tri-Council Policy Statement on Ethical Conduct for Research Involving Humans provide ethical support for this study (10,11). The NHANES protocols were approved by the National Center for Health Statistics institutional review board, and written informed consent was obtained from all participants (9).

The response rates for in-home interviews in 2011-2012 and 2013-2014 were 72.6 and 71.0%, respectively (9). Of the 19,931 persons interviewed, 16,733 (84%) were ≥ 6 years old. Of these 16,733 interviewed subjects, 12,677 (76%) had sera tested for T. gondii antibodies. Three of these subjects were excluded due to missing IgG results, and an additional 45 were excluded because they most likely had acute infections (exposure occurred within the last two weeks); the cross-sectional NHANES data would probably not capture the effects of an acute infection (12). Nine subjects were excluded because the outcome of asthma was missing, resulting in an analytic sample of n = 12,620 (Fig 1, Table 1).

• Download figure
• Open in new tab

Fig 1.

Flowchart representing the process for selecting the analytic sample from the NHANES 2012-2014 survey cycle data to evaluate the association between Toxoplasma gondii infection and asthma in the U.S.

The exposure variable was infection with T. gondii, defined as having IgG antibodies to T. gondii above ≥33 IU/mL. IgG antibodies were measured with the Toxoplasma IgG Enzyme Immunoassay (Bio-Rad, Redmond, WA) and IgM antibodies with the Platelia Toxo IgM (Bio-Rad). Test results were expressed as international units (IU/mL) or sample ratios, respectively. If IgG assay results were missing, the subject was excluded from the analysis (n = 3). If IgG was positive and IgM was missing, the exposure was counted as “positive”. If IgG was negative (<33 IU/mL), and IgM was missing, the exposure was counted as “negative”. If IgG was negative, and IgM was positive (Sample Ratio ≥1) the exposure was categorized as “acute” and excluded from the analysis (n = 45) (Fig 1)(12).

We created a proxy variable representing five levels of asthma severity from five variables in the Questionnaire. Level 0 indicates “never been told you have asthma”, level 1, “told you have asthma”, level 2, “told you have asthma and still have asthma”, level 3 “still have asthma and had an asthma attack in the past year” and level 4 “had an emergency care visit for asthma in the past year”. For comparison, we created a binary outcome variable for asthma based on the question, “Has a doctor or other health professional ever told you that you have asthma?” “Yes” or “No”.

We selected the minimum sufficient adjusted set of confounders using the modified disjunctive cause criterion, informed by evidence in the current literature and a directed acyclic graph (DAG) (Fig 2)(1,7,13–17). As cat ownership is not an important risk factor for T. gondii infection, and pet ownership information was not collected in the 2012-2014 NHANES, cat ownership was not included as a covariate (18–23). Because the NHANES categorizes subjects over the age of 80 years as “80+”, we grouped the age variable into decades as follows: “6-9”; “10-19”; “20-29”; “30-39”; “40-49”; “50-59”; “60-69”; “70-79”; and “80+”. This grouping is consistent with other analyses of T. gondii infection and is a clinically relevant interval for assessing trends in asthma severity and T. gondii seropositivity (7,24,25).

• Download figure
• Open in new tab

Fig 2.

Directed acyclic graph illustrating the relationships amongst the exposure variable “Toxoplasma gondii”, the outcome variable “Asthma severity”, and adjusted confounders. T. gondii infection is characterized as positive (pos) or negative (neg) and Asthma severity is categorized from 0 to 4, where 0 indicates no asthma and 4 indicates the subject has asthma that caused at least one emergency room visit in the last year. Age is age in years at last birthday. Education refers to the highest level of education attained by the household reference person. Poverty index refers to the poverty index assigned to the household reference person and # Children is the number of children <18 years of age living in the household.

Gender was “male” or “female”. Race was categorized as “Hispanic”, “White”, “Black”, “Asian”, or “other” according to the respondent’s self-assessment. Country of birth was “U.S.” or “other”. The education level of the household reference person was categorized as “<9^th grade”, “9^th-11^th grade”, “high school”, “some college”, or “college completed”. Economic status was defined according to the poverty index of the household reference person and expressed as a continuous numeric variable (0-5), where a poverty index of ≥3.5 indicates a high level of poverty (26). Number of children in the household was 0-3 or greater than 4 (9).

It is not clear from the literature if level of education is associated with asthma after adjusting for income (13,24). We calculated Akaike information criteria (AICs) for the ordinal and logistic regression models with and without the variable “education of the household reference person” to evaluate model fit. In both cases, model fit was improved by including the level of education in the model.

We tested for interaction effect between: “age” and “poverty index”, “race” and “poverty index”; and “poverty index” and “number of children in the household.” There was no significant interaction between “age” and “poverty index”. We detected significant interaction (p<0.05) for the latter two combinations and included these interaction terms in the final model.

Because T. gondii infection and asthma prevalence vary with age (Fig 3), and the relationship between asthma and atopy could also differ with age (2), we conducted additional ordinal regression analyses stratified by age category to check for effect modification and calculated the AIC values to compare model fit (Table 3).

• Download figure
• Open in new tab

Fig 3.

Toxoplasma gondii and asthma prevalence by age category in the U.S. non-institutionalized civilian population age 6 years and older. Data is from the NHANES 2011-2014 survey cycles (n=12,620).

Analyses were performed in R (version 4.3.1)(27). We used a design-based analysis approach to account for survey weights and complex survey design using the CRAN survey package (version 4.2-1)(28). Descriptive statistics are presented in Table 1; reported frequencies are based on the unweighted analytic sample while the percentages account for complex survey design.

Among all subjects, country of birth was missing for 0.03%, education level was missing for 3.29%, poverty index was missing for 6.86%, and age was missing for 19.0%. We assumed the data were missing at random (MAR) and imputed the missing covariate data using five imputations (29). All analyses were conducted on each imputed data set and results were pooled using Rubin’s rules (30). We compared results using imputed data with those obtained from a complete case analysis.

We fit an ordinal logistic regression model to estimate odds ratios (OR) and 95% confidence intervals (CI) of the unadjusted association between T. gondii infection and asthma severity (0-4), and the association after adjusting for all confounders (Table 2). For comparison, we fit a logistic regression model to estimate the unadjusted and fully adjusted association between T. gondii infection and asthma as a bivariate outcome, “Yes” or “No”.

RESULTS

Of the 12,620 subjects 6 years of age or older tested for T. gondii infection, 10.8% were seropositive and 89.2% were seronegative. Most subjects (83.5%) reported that they had never been told that they had asthma (level 0), while 16.5% reported varying degrees of asthma severity (levels 1-4). Asthma prevalence varied by age category (Fig 3), with the highest prevalence found in 10–29-year-olds and the lowest in subjects 80 years and older. Prevalence of T. gondii infection increased with age (Fig 3). No statistically significant difference was detected for seroprevalence of T. gondii across the five levels of asthma severity (p = 0.140). Significant unadjusted associations between asthma and the covariates gender (p = 0.036), age category (p = 0.019), poverty index (p = 0.003), and race (p = 0.024) are presented in Table 1.

No significant relationship was detected between T. gondii infection and asthma. A complete case analysis using the unadjusted ordinal regression model returned an OR of 0.89 (95% CI = 0.79-1.01), suggesting an insignificant protective effect of T. gondii infection. Adjusting for all confounders, the complete case model returned an OR of 1.02 (95% CI 0.87-1.19). Using multiple imputation to account for missing covariate data, the unadjusted regression model returned an OR of 0.90 (95% CI = 0.83-0.97), suggesting a protective effect of T. gondii infection, but the fully adjusted model returned an OR of 1.00 (95% CI = 0.91-1.10), confirming there was no detected association between T. gondii infection and asthma (Table 2). The unadjusted and adjusted logistic regression models evaluating asthma as a bivariate outcome variable (Yes/No) yielded similar results (Table 2). The ordinal model fit the data better than the logistic model according to the AIC values.

Regression analyses stratified by age category suggested that there is some effect modification by age. Model fit, evaluated using AIC, was much better within age categories when compared to the full model using all age categories (Table 3). The association between T. gondii infection and asthma severity also varied across age categories, but within each category, the association remained insignificant (Table 3).

DISCUSSION

The NHANES data has some limitations. An estimated 20% of asthma cases are undiagnosed in the U.S. and therefore not captured by the NHANES (31), which could bias our results.

Additionally, the data are cross-sectional and cannot capture temporal relationships between exposure and outcome. If T. gondii exerts an effect on asthma over only a short period post-infection, our analysis might fail to detect this. Furthermore, asthma is a syndrome comprised of several diseases, not all of which are associated with atopy (4). Because T. gondii’s potential impact on asthma is postulated to occur via immune system modulation (6,32), the inability to distinguish atopy-related asthma from other types might reduce the sensitivity of our analysis.

Model fit was improved when stratified by age category, and ORs appeared to vary across age categories but remained statistically insignificant (Table 3). The non-linear relationship between asthma and age (Fig 3) might contribute to this effect.

For all adjusted regression models, no significant relationship was detected between T. gondii infection and asthma. These findings are consistent with two other large cross-sectional studies in Scotland and Australia that found no association between T. gondii seropositivity and asthma (14,33). While our results appear to contradict several case-control studies that have found a protective effect of T. gondii infection on allergy symptoms and asthma, these studies did not consider confounders which could impact exposure to a diversity of microorganisms and their products, such as socioeconomic status or number of siblings (34,35). Toxoplasma gondii is, therefore, likely a marker for the protective effect of exposure to a diversity of microbial organisms and their products early in life, but may not have any causal effect on this phenomenon, or may not be important when considered in isolation (1,7,15,36).