Prevalence, incidence estimations, and risk factors of Toxoplasma gondii infection in Germany: a representative, cross-sectional, serological study

Representative data on the extent of endemicity, burden, and risk of human toxoplasmosis are scarce. We assessed the prevalence and determinants of seropositivity of Toxoplasma gondii among adult participants of a nationwide representative cross-sectional survey in Germany. Sera collected from a representative cohort of adults (age 18–79; n = 6,663) in Germany were tested for anti-T. gondii IgG antibodies. Interview-derived data were used to evaluate associated factors. Multivariable logistic regression was applied using sampling weights and accounting for survey design cluster effects. Seroprevalence increased from 20% (95%-CI:17–23%) in the 18–29 age group to 77% (95%-CI:73–81%) in the 70–79 age group. Male gender, keeping cats and BMI ≥30 were independent risk factors for seropositivity, while being vegetarian and high socio-economic status were negatively associated. Based on these data, we estimate 1.1% of adults and 1.3% of women aged 18–49 to seroconvert each year. This implies 6,393 seroconversions annually during pregnancies. We conclude that T. gondii infection in Germany is highly prevalent and that eating habits (consuming raw meat) appear to be of high epidemiological relevance. High numbers of seroconversions during pregnancies pose substantial risks for unborn children. Efforts to raise awareness of toxoplasmosis in public health programs targeting to T. gondii transmission control are therefore strongly advocated.


Results
Seroprevalence and associated factors. Altogether 6,663 participants with available blood samples were included in this study; 99 (1.5%) thereof had equivocal anti-T. gondii IgG titers of 4-7 and were thus excluded from the analysis. Out of 6,564 individuals, 3,602 (55%) were seropositive (Table 1). Seroprevalence increased from 20% (95%-confidence interval (95%-CI):17.1-23.1%) in the age group 18-29 to 77% (95%-CI:72.7-80.5%) in the 70-79 age group. Seroprevalence in participants older than 79 years (n = 116), not included in further analyses, was 84%. The observed seroprevalence increased at a rate of 1.09% with each year of age ( Fig. 1). A significant interaction was noted between age and sex (p-value = 0.023), since higher seroprevalences were observed among younger males and older females. Male gender, although not associated in univariable analysis, was a significant risk factor in multivariable analysis (odds ratio (OR): 1.8; 95%-CI1.1-2.9). Age-related increase of seroprevalence was stronger amongst residents of East Germany when compared to the West (p-value of interaction term: < 0.001) (Fig. 2). Prevalence in East Germany is more than 20% higher than in West Germany in the 40-69 age group.
Incidence estimates of seroconversions. Using a regression model we estimated an annual incidence of 1,099 (95% CI:1,016-1,181) seroconversions per 100,000 adult inhabitants of Germany (Table 2). In women aged 18-49 (considered here as main age group potentially bearing children) this corresponds to 1,325 (95% CI: 1,007-1,642) seroconversions. Taking the age-dependent distribution of births among women (data from 2011) and an age-related increase in seroprevalence into account, it can be assumed that 74.1% of births were delivered by mothers susceptible to primary infection ("pregnancies at risk"). The susceptibility decreased from 90.8% in the age group 15-19 years to 52.6% in those 45-49 years old. The incidence of 1,325 in 100,000 women implied 6,393 predicted infections during the annual 662, 485 pregnancies in Germany (Table 3). These are 1.0% of all pregnant women and 1.3% of seronegative pregnant women (pregnancies at risk).
Estimates of incidence of congenital toxoplasmosis. In a recent meta-analysis among patient cohorts, the rate of mother-to-child transmission among seroconverting pregnant women was reported to be 20% (95%-CI:15-26%) 9 . Extrapolated to the numbers derived here this would result in 1,279 (95% CI:959-1,662) annual cases of fetal infections in Germany. Based on estimates that 27% of infected neonates manifest with T. gondii-specific symptoms 10

Discussion
This first nationwide representative seroprevalence study for T. gondii antibodies in Germany finds a high seroprevalence compared to other countries. Comparing our findings to previous data from Germany is hampered by the older studies' lack of representativeness, but higher prevalences were found in a serosurvey among blood donors between 1994 and 1996 in North-East Germany (59%; n = 4,854; age 20-40) 27 , as well as in a study among pregnant women from South-West Germany (39%; n = 5,670; age 15-47) 28 31 . These differences show that improvements for T. gondii infection prevention in Germany are necessary and also possible. The case numbers from the German notification system are restricted to congenital toxoplasmosis and range between 23 cases in 2008 and 6 cases in 2014, without obvious time trends or regional clustering 32 . This is only a small fraction of cases we calculated in this study. Reasons for this discrepancy could be a high degree of under-ascertainment and under-reporting, as not all cases of congenital toxoplasmosis are laboratory confirmed, and laboratories do not have the clinical or demographic information to carry out the notification. As a result, current notification rates do not reflect the full picture of disease burden in Germany. Our extrapolations used  manifestation rates from Li et al. 9 and Dunn et al. 10 for the calculation of incidence of congenital toxoplasmosis,which is independent of the influence of disease surveillance processes and therefore might provide a more accurate picture.
Calculations of the incidence of ocular and cerebral toxoplasmosis based on seroconversion rates are difficult, since information from surveys on manifestation rates is scarce, and the few sources available are outdated. Jones and Holland 33 estimated, based on North American surveys in the 1970s, that 2% of T. gondii-infected individuals develop ocular toxoplasmosis. Extrapolated to incidence of seroconversion in Germany, this results in 22 patients of clinical ocular toxoplasmosis per 100,000 inhabitants annually. Due to an increase of immunosuppression in the aging population it can be assumed that incidence of cerebral toxoplasmosis is also increasing, although data on the population level are missing.
We found no evidence for a role for T. gondii in depression. This supports previous evidence from the population-based United States National Health and Nutrition Survey (NHANES III) 25 . Altogether, evidence on manifestation rates of all health outcomes of T. gondii infections is scarce and therefore studies have to be initiated to eliminate these limitations and to fully understand the disease burden of toxoplasmosis.
Seropositivity strongly increases in all age groups, indicating a strong force of infection also in the elderly. Frequent seroconversion in this group is particularly problematic since immunosuppression becomes more    Table 4. Stratified seroprevalence of IgG antibodies against T. gondii detected by ELFA in adults aged 18 to 79 years, and results of weighted logistic regression analysis of seropositivity on depression as an outcome, 2008-2011. a unweighted, lifetime incidence of depression (self-reported) was assessed by asking the closed question "Have you ever been diagnosed with depression by a physician or a psychotherapist"? b adjusted for age, sex and interaction between both. c adjusted for variables in multivariable model of Table 1.
prominent with age. The observed increase is particularly strong in East Germany, arguing that the risk pattern changed after unification of Germany in 1990. Alternatively, this could be explained by a birth cohort effect of a previous time frame, when a higher risk of seroconversion resulted in higher seroprevalence today in older age groups. Although not associated in univariable analysis, men have a 1.76-times higher chance of being seropositive in multivariable analysis. This can be explained by effect modification of the factors higher socio-economic status, cat holding and male gender included in the final model. Since male Germans eat about twice as much meat and meat products as females 34 , the higher seroprevalence we observe in men can be reconciled with these eating habits. In our study, seroprevalences for T. gondii vary between regions in Germany. The consumption of freshly prepared raw minced meat (beef and pork), known in Germany as "Hackepeter" or "Mett", is more widespread in East compared to West Germany 35,36 . This may explain the higher seroprevalence in East Germany and additionally confirms that raw meat poses a substantial risk for infection with T. gondii 4,37 . Furthermore, our results show that overweight and obese study participants have an increased chance of becoming seropositive, consistent with recent data from a smaller study in Germany 38 . There is good evidence that higher meat consumption is connected to an increase in body weight 39,40 ,which in turn leads to a higher chance to ingest contaminated meat. Eating vegetarian, on the other hand, is negatively associated with seropositivity, arguing that consumption of oocyst-contaminated vegetables is not a major driver of seroconversion. Altogether, these data are consistent with the majority of infections being food-borne 4,37,41 .
Some previous epidemiological studies did not observe an elevated risk for cat owners 37,42 , while representative serosurveys in The Netherlands have reported such an association 30,43 . A case-control study in the USA reported a higher risk for cat owners with 3 or more cats 44 . Cats are biologically essential in the life cycle of T. gondii, but cat exposure seems to be a less important factor compared to exposure to contaminated food, in particular meat. Nevertheless, prevention of infection through contact with cats is possible since they shed oocysts only for 1-3 periods during lifetime, and sporulation can be avoided by daily removal of cat litter 45 .
The strengths of our study include a large sample size with a sampling process leading to a representative study population, with an appropriate adjustment for demographic variables. As the DEGS1 study recruited only the adult population, we have no data on infants and children. This deficiency should be addressed in the future. The analyses in our study are based on seroprevalence data, but seroconversion is not equivalent to clinical manifestation of disease. The associated factors in our study are mostly contextual factors. Specific and proximal factors like lifetime cat exposures or lifetime raw meat consumption were not available from the survey. The same is true for the association to lifetime depression, which is only one mood disorder hypothesized to be connected to T. gondii infections.
Toxoplasmosis is hard to combat from the public health point of view and often neglected in programs targeting food-borne diseases 46 . Primary prevention of infections is important and can prevent a significant proportion of disease cases of all clinical manifestations 47 . Since a T. gondii infection can reactivate upon immune suppression, it should be avoided in all population groups to prevent subsequent ocular or cerebral disease later in life.
Secondary prevention in the form of a repeated serological screening during pregnancy might be successful to prevent further pregnancy-associated cases. Arguments for serological screening have to consider the positive predictive value of the screening tests and the risk-benefit ratio of medication used to treat infection. The effectiveness of prenatal screening and treatment is debated, largely because of different opinions on cost-benefit 48,49 , health consequences of the screening for the mother 50 , and the effectiveness of the treatment 16,51 . Nevertheless, early serological screening of pregnant women to detect maternal seroconversion, followed by rapid treatment in utero (as is done in France and Austria) may prevent transplacental transmission and thus neonatal infection and clinical manifestation [11][12][13][14] . In these countries, the key for success seems to be frequent re-testing using highly sensitive PCR analysis of appropriate maternal samples and rapid treatment 2,52 . This practice might explain the striking differences in percentages of observed hydrocephalus cases in newborns of congenital infections seen between France and the US (0.3-0.8% vs. 31%) [53][54][55] . In the latter country, this regimen is not practiced. In Austria, with its systematic prenatal screening program, a 6-fold decrease in transmission compared to untreated women was recently reported 56 . At the rates of congenital infection reported here, screening is considered to be cost-saving 57 . In Germany, screening tests to detect T. gondii infection during pregnancy are currently not covered by statutory health insurance, after scientific evaluation of its pros and cons by their medical services 58 . As a result, pregnant women have to self-pay for these tests. This significantly affects the testing rate, especially in women of lower socioeconomic status 59 .
Patient numbers and preventable cases are expected to be significantly higher than can be inferred from the notification system. It should be re-evaluated whether serological testing of pregnant women should be offered by physicians and paid for by health insurance in Germany. Moreover, women of childbearing age and those immunocompromised should be more specifically targeted and informed about potential risks, with emphasis on the prevention of food-borne infection pathways. Finally, veterinary services, the meat industry and agriculture should continue their efforts to reduce T. gondii in meat, especially pork, as this is frequently eaten raw in Germany.

Methods
The German health interview and examination survey of adults (DEGS1) was a representative, nationwide cross-sectional survey conducted between 2008 and 2011. Detailed study descriptions have been presented previously 26 . Briefly, it followed a stratified two-stage cluster sampling strategy to assess the health status of adults between 18 and 79 years of age in Germany. Three professionally trained teams each visited a total of 180 sample points. The sample points are distributed across Germany according to federal state and municipality size in order to reflect the distribution of the German population. After written informed consent 7,988 persons (age 18-79) participated in DEGS1. These were interviewed with a standardized questionnaire and blood samples were given. The response rate among those approached was 48.4% and the analysis of non-responder questionnaires revealed high population representativeness among participants. Interview data were used to assess potential risk factors for positive outcome 26 . DEGS1 was approved by the ethical review board of the Charité Medical School, Berlin, Germany (No. EA2/047/08) and all methods were carried out in accordance with the approved guidelines.
Aliquots of sera (thawed twice) from DEGS1 were analyzed for the presence of anti-T. gondii IgG antibodies by a commercial automatic and quantitative enzyme-linked fluorescence assay (ELFA; VIDAS TOXO IgG II; Biomérieux SA, France) of the same batch according to the manufacturer's instructions. In comparative studies, sensitivity of the assay was shown to be above 99% and specificity above 98% 60 . Titers of 0-3 were considered negative, 4-7 equivocal, and ≥8 positive. We excluded all samples with equivocal test results (1.5%) from the analysis.
Univariable and multivariable logistic regression used sampling weights, accounted for the cluster structure of the survey design and checked for effect modifications. Survey weights based on age, sex, residence in West or East Germany and nationality (German vs. non-German) were calculated to correct for deviations from the German population statistics (December 31, 2010) and used throughout the analysis.
We investigated associations between explanatory variables and seropositivity by univariable logistic regression analyses. The most frequent category was selected as the reference category except for variables where a norm category was available. All variables below p-value of 0.15 were included in a multivariable logistic regression model. As seropositivity regarding T. gondii is cumulative and therefore strongly related to age, all two-way interactions involving age were included. Age-related prevalence was smoothed by using the Lowess procedure of the program Stata 12.1 and graphed stratified for sex and East-West geographical origin. The annual incidence of seroconversions was calculated from linear regression using the one-year age group as independent and weighted prevalence as dependent variable. The annual number of seroconversions among pregnant women between 15 and 49 years of age in Germany was calculated by using the number of births in each one-year age group of mothers multiplied by the expected proportion of susceptibles, considering an annual incidence of 1,325 in 100,000 women, and corrected for the mean duration of pregnancies (266 days). For the age group 15 to 17 years, prevalence estimates were derived from the study data, assuming a constant increase as in the age group 18 to 25 years.
For the calculation of seroconversions among pregnancies, data on number of women and number of childbirths by age of the mother for 2011 were retrieved from the German Federal statistical office (https://www.destatis.de/). In the calculations we regarded multiple births by looking at how many children each respondent had.