INTRODUCTION
Toxoplasma gondii infects about one third of inhabitants on Earth (Tenter et al. Reference Tenter, Heckeroth and Weiss2000; Pappas et al. Reference Pappas, Roussos and Falagas2009). In many countries, including several highly developed ones like France and Germany, more than 50% of the population acquire T. gondii infection during their lifetime (Pappas et al. Reference Pappas, Roussos and Falagas2009). Acute toxoplasmosis promoted by rapidly dividing tachyzoites has a mostly subclinical course with minor symptoms in immunocompetent subjects followed by latent stage (Montoya & Liesenfeld, Reference Montoya and Liesenfeld2004). It is, however, the most common food-borne parasitic infection requiring hospital treatment in France (Vaillant et al. Reference Vaillant, de Valk, Baron, Ancelle, Colin, Delmas, Dufour, Pouillot, Le Strat, Weinbreck, Jougla and Desenclos2005), the third most common cause of hospitalization due to food-borne infections (Mead et al. Reference Mead, Slutsker, Dietz, McCaig, Bresee, Shapiro, Griffin and Tauxe1999) and one of the leading causes of death attributed to food-borne illness (Scallan et al. Reference Scallan, Hoekstra, Angulo, Tauxe, Widdowson, Roy, Jones and Griffin2011). About one million new infections are estimated to occur each year in the USA, which result in 20 000 cases of retinal pathology (Jones & Holland, Reference Jones and Holland2010). Latent toxoplasmosis can be defined as the presence of slowly dividing bradyzoites of Toxoplasma in tissue cysts localized mostly in the immunoprivileged organs, namely the brain, eye and testes. It is generally believed that latent toxoplasmosis, accompanied by fluctuating anamnestic levels of anti-Toxoplasma IgG antibodies, is a lifelong condition (Tenter et al. Reference Tenter, Heckeroth and Weiss2000). The presence of anamnestic antibodies protects infected subjects against new infections. In immunosuppressed subjects, e.g. AIDS patients or artificially immunosuppressed transplant recipients, toxoplasmosis can reenter the acute phase and without proper treatment, can lead to life-threatening cerebral toxoplasmosis (Porter & Sande, Reference Porter and Sande1992; Akanmu et al. Reference Akanmu, Osunkalu, Ofomah and Olowoselu2010; Addebbous et al. Reference Addebbous, Adarmouch, Tali, Laboudi, Amine, Aajly, Rhajaoui, Chabaa and Zougaghi2012). For a long time, latent toxoplasmosis was considered clinically asymptomatic. About 20 years ago, behavioural manifestations of toxoplasmosis such as specific changes in personality traits or prolonged reaction times were described (Flegr et al. Reference Flegr, Zitkova, Kodym and Frynta1996; Havlíček et al. Reference Havlíček, Gašová, Smith, Zvára and Flegr2001). Some of the observed changes are believed to be the result of the manipulative activity of Toxoplasma, the evolutionary adaptation of this parasite that increases the chances of parasites transmission from an intermediate host (any warm-blooded animal) to the definitive host (any feline species) by predation (Flegr & Hrdý, Reference Flegr and Hrdý1994; Webster, Reference Webster1994). Some are probably side-effects of other activities of the parasite, such as the upregulation of the production of dopamine (Flegr et al. Reference Flegr, Preiss, Klose, Havlíček, Vitáková and Kodym2003; Gaskell et al. Reference Gaskell, Smith, Pinney, Westhead and McConkey2009; Prandovszky et al. Reference Prandovszky, Gaskell, Martin, Dubey, Webster and McConkey2011) and testosterone (Flegr et al. Reference Flegr, Lindová and Kodym2008; Kaňková et al. Reference Kaňková, Kodym and Flegr2011; Lim et al. Reference Lim, Kumar, Hari Dass and Vyas2013), or of a mild chronical stress accompanying latent infection (Lindová et al. Reference Lindová, Kuběna, Šturcová, Křivohlavá, Novotná, Rubešová, Havlíček, Kodym and Flegr2010, Reference Lindová, Novotná, Havlíček, Jozífková, Skallová, Kolbeková, Hodný, Kodym and Flegr2006).
In the past 10 years, many associations between latent toxoplasmosis and certain health disorders have been observed and published. Infected subjects have a highly increased probability of being diagnosed with schizophrenia (Torrey et al. Reference Torrey, Bartko, Lun and Yolken2007, Reference Torrey, Bartko and Yolken2012), mood disorders (Pearce et al. Reference Pearce, Kruszon-Moran and Jones2012; Radford et al. Reference Radford, Williams, Kane and Groer2012), epilepsy (Stommel et al. Reference Stommel, Seguin, Thadani, Schwartzman, Gilbert, Ryan, Tosteson and Kasper2001; Palmer, Reference Palmer2007), autism (Prandota, Reference Prandota2010; Blomstrom et al. Reference Blomstrom, Karlsson, Wicks, Yang, Yolken and Dalman2012), migraine and other headaches (Koseoglu et al. Reference Koseoglu, Yazar and Koc2009; Prandota, Reference Prandota2009), melanoma (Nagineni et al. Reference Nagineni, Detrick and Hooks2002), carcinoma of female genitals and breast cancer (Sanchis-Belenguer et al. Reference Sanchis-Belenguer, Cuadrado-Mendez and Ortiz Munoz1984; Vos, Reference Vos1987), heart diseases (Paspalaki et al. Reference Paspalaki, Mihailidou, Bitsori, Tsagkaraki and Mantzouranis2001; Yazar et al. Reference Yazar, Gur, Ozdogru, Yaman, Oguzhan and Sahin2006), inflammatory bowel disease (Prandota, Reference Prandota2012), celiac disease (Nejad et al. Reference Nejad, Rostami, Cheraghipour, Mojarad, Volta, Al Dulaimi and Zali2011; Prandota, Reference Prandota2012) impaired liver functions (Vethanyagam & Bryceson, Reference Vethanyagam and Bryceson1976; Ustun et al. Reference Ustun, Aksoy, Dagci and Ersoz2004), hematological changes (Flegr & Stříž, Reference Flegr and Stříž2011), thyroid diseases (Singh et al. Reference Singh, Singh, Pandav, Pandav and Karmarkar1994; Kankova et al. Reference Kankova, Prochazkova, Flegr, Calda, Springer and Potlukova2014), rheumatoid arthritis (Tomairek et al. Reference Tomairek, Saeid, Morsy and Michael1982; Torrey & Yolken, Reference Torrey and Yolken2001; Shapira et al. Reference Shapira, Agmon-Levin, Selmi, Petrikova, Barzilai, Ram, Bizzaro, Valentini, Matucci-Cerinic, Anaya, Katz and Shoenfeld2012), glomerulonephritis (van Velthuysen & Florquin, Reference van Velthuysen and Florquin2000; Kapoor, Reference Kapoor2012; Toporovski et al. Reference Toporovski, Romano, Hartmann, Benini and Chieffi2012), diabetes (Kaňková et al. Reference Kaňková, Flegr and Calda2015b ), and changes in lipid contents, including atherosclerosis (Coppens, Reference Coppens2006; Flegr et al. Reference Flegr, Hampl, Černochová, Preiss, Bičíkova, Sieger, Příplatová, Kaňková and Klose2012). Infected subjects also have a higher probability of suicides (Yagmur et al. Reference Yagmur, Yazar, Temel and Cavusoglu2010; Ling et al. Reference Ling, Lester, Mortensen and Postolache2011; Pedersen et al. Reference Pedersen, Mortensen, Norgaard-Pedersen and Postolache2012) and traffic or working place accidents (Alvarado-Esquivel et al. Reference Alvarado-Esquivel, Torres-Castorena, Liesenfeld, Estrada-Martinez and Urbina-Alvarez2012; Flegr et al. Reference Flegr, Havlíček, Kodym, Malý and Šmahel2002, Reference Flegr, Klose, Novotná, Berenreitterová and Havlíček2009). An exhaustive survey of such effects of toxoplasmosis is provided in a previous article by our group (Flegr et al. Reference Flegr, Prandota, Sovickova and Israili2014). In that study, the existence of most of the already known associations, as well as many new ones, was observed. That ecological study was performed on a set of 88 countries for which the necessary data were available. It showed that the prevalence of toxoplasmosis is correlated with the data on specific disease burden collected and published by the WHO. The study revealed that for example, morbidity of 23 of 128 analyzed diseases and disease categories on the WHO list showed correlations with the prevalence of toxoplasmosis and another 12 diseases had positive trends (P < 0·1). When the confounding variables like gross domestic product (GDP) per capita, geolatitude and humidity were controlled, the prevalence of toxoplasmosis explained 23% of the variability in disease burden in Europe (Flegr et al. Reference Flegr, Prandota, Sovickova and Israili2014).
The interpretations of ecological studies are sometimes complicated, especially if aggregated data are used for the estimation of the strength and direction of the influence of particular factors within a population (Guthrie & Sheppard, Reference Guthrie and Sheppard2001; Wakefield & Salway, Reference Wakefield and Salway2001). Results obtained in retrospective case-control studies are more reliable in this respect and can detect even weak associations between a particular health disorder and environmental factors like toxoplasmosis. However, the results performed on clinical patients could be biased by the sieve effect – for example by the unwillingness of psychiatric patients with certain forms of disorders to enter voluntarily into scientific study. Here the results are reported of an explorative cross-sectional study on a population of 1486 volunteers recruited from an internet community called Guinea Pigs and consisting of Czechs and Slovaks willing to participate in mostly evolutionary psychology experiments.
METHODS
Subjects and recruitment
The recruitment of subjects was undertaken by using the Facebook-based snowball method (Kankova et al. Reference Kankova, Flegr and Calda2015a ). To address potential volunteers, an invitation to participate in ‘an experimental searching for associations of a subject’ blood group and other biological factors with his/her personality, performance, morphology, and health’ was posted on the Facebook wall page ‘Guinea Pigs’ (in Czech ‘Pokusni kralici’) for Czech and Slovak nationals willing to take part in evolutionary psychology experiments (http://www.facebook.com/pokusnikralici) (Flegr & Hodny, Reference Flegr and Hodny2016). The first page of the electronic questionnaire provided information about the goal of the study. The following note was also included: ‘The questionnaire is anonymous and obtained data will be used exclusively for scientific purposes. Your cooperation with the project is voluntary and you can terminate it at any time by closing this website.’ The population of Guinea Pigs is ‘enriched’ with subjects that we serologically tested for toxoplasmosis in the past 15 years of our systematic study of behavioural effects of latent toxoplasmosis (Flegr, Reference Flegr2013b ). However, most of our recent internet studies have no relation to toxoplasmosis. Moreover, toxoplasmosis was not specifically mentioned neither during the recruitment of participants nor in the informed consent to keep the study blind, and therefore to avoid a possible bias (approved by Institutional Review Board (IRB)). The first and also the final page of the questionnaire contained the Facebook share button and the following request for the participants: ‘We need the highest possible number of responders. Therefore, please, share the link to this questionnaire with your friends, for example on Facebook’. The share button was pressed by 541 participants, which resulted in obtaining data from 6463 Czech and Slovak responders between 28/4/2014 and 16/11/2015. The study, including the method of obtaining the informed consent (by pressing the Next button on the first page) was approved by the IRB of the Faculty of Science (Eticka komise pro praci s lidmi a lidskym materialem Prirodovedecke Fakulty Univerzity Karlovy) No. 2014/21.
Questionnaire
The anamnestic questionnaire was prepared by two medical doctors, a clinician (internist/hematologist) and a researcher (molecular geneticist) and was distributed as a Czech/English Qualtrics survey (http://1url.cz/q05K). It contained two categories of questions. The first of them monitored presence and intensity of general and specific health problems of responders. The responders were asked to subjectively rate of their allergic, cancer, digestive, fertility, genitourinary, heart, haematological, immunity, mental health, metabolic including endocrine, musculoskeletal, neurological, respiratory organs, sense organs and sexual life problems using 6-points Likert scales. The second group of questions tried to collect objective information reflecting the health status of responders. We asked the responders, for example, how many drugs prescribed by doctors they currently take per day, how many of ‘different herbs, food supplements, multivitamins, superfoods etc.’ they currently take per day, how many times they used antibiotics during the past 365 days. We also provided the responders lists of about 250 disorders (separated to 15 categories) and asked them to tick which of them they were diagnosed with. The questionnaire contained, among others, also the following questions: ‘Are you infected with Toxoplasma (a parasite living in cats which is dangerous for pregnant women)?’ with three options: (a) I do not know/I am not sure, (b) no (I was tested but I was negative) (c) yes (I was tested and I was positive – I have antibodies against this parasite). Implicitly, the answer a) (I do not know/I am not sure) was checked. The responders of our questionnaires had three options: they could complete any questionnaire absolutely anonymously, they could sign the finished questionnaire by a code obtained after anonymous registration, or they could sign the finished questionnaire by a code obtained after non-anonymous registration (see http://pokusnikralici.cz). Some questionnaires are ‘signed’ by less than 1% of non anonymously registered subjects (e.g. the questionnaire about sexual behaviour), some by 15% of subjects. The present questionnaire (containing sensitive information about mental health) was ‘signed’ by 3% of the subjects. When we checked the information about the toxoplasmosis status provided in the questionnaire by participants of our past experiments with corresponding information in our records, we found a perfect (100%) agreement.
The questionnaire contained also some questions unrelated to the topic of the present study (e.g. a short personality test) and simple tests of reaction times, operational, short-term and long-term memory, psychomotor performance and intelligence. In the present paper, however, only the health status and diseases and disorder-incidence-related questions were analyzed.
Immunological tests for T. gondii infection
Most of the women and nearly all of men who know their T. gondii-infection status were tested for T. gondii infection during systematic research of behavioural effects of latent T. gondii infection, which has been running at the Faculty of Science for 20 years. All testing was performed at the National Reference Laboratory for Toxoplasmosis, National Institute of Public Health, Prague. The complement-fixation test (CFT), which determines the overall levels of IgM and IgG antibodies of particular specificity and Enzyme-Linked Immunosorbent Assays (ELISA) (IgG ELISA: SEVAC, Prague) were used to detect T. gondii infection status of the subjects. ELISA assay cut-point values were established using positive and negative standards according to the manufacturer's instructions. In CFT, the titre of antibodies against T. gondii in sera was measured in dilutions between 1:4 and 1:1024. The subjects with CFT titres between 1:8 and 1:128 were considered T. gondii infected. Only subjects with clearly positive or negative results of CFT and IgG ELISA tests were diagnosed as T. gondii-infected or T. gondii-free; subjects with different results of these tests or ambiguous results of tests were retested or excluded from the study.
Data analysis
SPSS v. 21 was used for the statistical analysis. Differences in age were tested by t-test and differences in the prevalence of toxoplasmosis between men and women by logistic regression with the age as a confounding variable. Ordinal and binary data were analyzed by partial Kendall's correlation test, which is used to measure the strength and significance of the association between binary, ordinal, or continuous data regardless of their distributions and allows the control for one confounding variable, here the age (Siegel & Castellan, Reference Siegel and Castellan1988; Kaňková et al. Reference Kaňková, Kodym and Flegr2011). The Excel spreadsheet used to compute the partial Kendall tau and the significance for variables A (diseases) and B (Toxoplasma infection), once C (age) was controlled -based on Kendall Taus AB, AC and BC- is available at: http://web.natur.cuni.cz/flegr/programy.php. When the number of subjects reporting a disorder was less than 10, the Fisher exact test was used for studying the association between toxoplasmosis and the disorder. Correction for multiple tests was performed with Benjamini-Hochberg procedure. (Benjamini & Hochberg, Reference Benjamini and Hochberg1995). In the contrast to simple Bonferroni's correction, this procedure takes into account also the distribution of p values of performed multiple tests. Therefore, when the studied factor has multiple effects, the number of significant results after the correction could be higher than before the correction.
The prevalence of certain diseases is gender specific and also many effects of latent toxoplasmosis have been shown to differ between men and women (Lindová et al. Reference Lindová, Novotná, Havlíček, Jozífková, Skallová, Kolbeková, Hodný, Kodym and Flegr2006, Reference Lindová, Kuběna, Šturcová, Křivohlavá, Novotná, Rubešová, Havlíček, Kodym and Flegr2010). Therefore, all analyses were performed separately for men and women. All raw data are available as the Supporting Information S1, available at https://figshare.com/s/64f6b0230f733e8e3aa2.
RESULTS
Descriptive statistics
Of 6463 Czech and Slovak respondents, 1486 (365 Toxoplasma-free males, age 34·8, s.d. 12·7, 69 Toxoplasma-infected males, age 34·0, s.d. 10·5, 788 Toxoplasma-free females, age 32,4, s.d. 11·0, and 264 Toxoplasma-infected females, age 36,5, s.d. 12·3) provided information about their Toxoplasma status. The difference in the age between Toxoplasma-infected and Toxoplasma-free subjects was significant for women (t 1048 = −5·10, P < 0·0001) but not for men (t 430 = 0·520, P = 0·603). The logistic regression with age and sex as independent variables showed that the prevalence of toxoplasmosis was significantly higher in women (25·1%) than in men (15·9%), odds ratio (OR) = 1·85, P < 0·0001.
Intensity of health problems
Twenty-nine dependent variables (mostly ratings of particular health problems on a 1-to-6 scale, where 1 was ‘no problems at all’ and 6 was ‘frequent or serious’) were ordinal and most of them had a highly skewed distribution. Therefore, the nonparametric partial Kendall correlation test (which enables us to control for one confounding variable, in this case, the respondent's age) was used to search for the statistical association between Toxoplasma infection and the intensity of 15 categories of health problems (allergic, cancer, digestive, fertility, genitourinary, heart, hematological, immune, mental health, metabolic including endocrine, musculoskeletal, neurological, respiratory, sense organs and sexual disorders). Furthermore, the association of Toxoplasma seropositivity with other fourteen ordinal health-related variables was analyzed: the subjectively rated physical health, the subjectively rated mental health, the number of prescription drugs currently taken by the respondent per day, the number of alternative medicines such as herbs, food supplements, multivitamins, superfoods, etc. currently taken by the respondent per day, the maximum number of times per calendar year antibiotics were taken by the respondent, the number of times acute medical attention was needed by the respondent for a serious illness (not injury) that lasted more than 3 days in the past 5 years, the number of types of specialized medical doctors the respondent saw in the past 2 years, sexual desire in comparison with age-matched peers, sexual activity in comparison with age-matched peers, the frequency of feeling tired (not after exertion, e.g. not after sports), adherence if on a diet, the frequency and severity of feeling sleepy during the day, intensity of suffering from insomnia and the frequency of headaches. The results corrected for multiple tests with Benjamini-Hochberg procedure showed that the Toxoplasma-infected subjects of both genders altogether, the male subjects and the female subjects reported higher number or increased severity of health problems in 28, 17 and 19 of the 29 analyzed variables, respectively (Table 1). Theoretically, less than 5 of 87 statistical tests could provide false positive results. This suggests that nearly all positive results reflect a real difference in health between Toxoplasma-infected and Toxoplasma-free subjects, rather than statistical artefacts of multiple comparisons.
Table 1. Correlation between toxoplasmosis and health-related variables
Effects of age on health status were controlled in present partial Kendall Tau test. Positivity of τ indicates that infected subjects have higher values of particular health related variables, i.e. a worse health status, or higher sexual desire and activity. The effect size is shown as τ. Significant results (P < 0·05) are printed in bold and results significant after Benjamini-Hochberg procedure correction for multiple tests (performed separately for all three populations, all subjects, men and women) are denoted with asterisks. P-values <0·0005 are coded as 0·000.
Infected subjects of both genders altogether, infected male subjects and infected female subjects also reported more frequent appointments with 9, 5 and 5 out of 10 medical specialists than the corresponding Toxoplasma-free controls, respectively (Table 2). Some variables were significantly correlated to toxoplasmosis exclusively in one gender. For instance, as to psychiatric disorders and sense organs problems were reported only by men while metabolic problems and insomnia problems only by women.
Table 2. Differences in number of types of medical specialists appointed in the past 2 years between Toxoplasma-infected and Toxoplasma-free subjects
Numbers of Toxoplasma-free subjects non appointed particular medical specialist, Toxoplasma-free subjects appointed particular medical specialist, Toxoplasma-infected subjects non appointed particular medical specialist, Toxoplasma-infected subjects appointed particular medical specialist, odds ratio and statistical significance, respectively, are shown in six columns of each section. The effect of age on health status was controlled by using bivariate partial Kendall's correlation (non-parametric) test. Odds ratio higher than 1·0 indicates a positive association of Toxoplasma infection with probability of visiting particular medical specialist. Significant results (P < 0·05) are printed in bold and results significant after Benjamini-Hochberg procedure correction for multiple tests (performed separately for all three populations, all subjects, men and women) are denoted with asterisks. P-values <0·0005 are coded as 0·000.
Incidence of particular diseases and disorders
Once the participants rated the intensity of the 29 health problems, they were asked to evaluate the presence or absence of specific disorders. They answered questions like ‘What kind of respiratory problems are you suffering from or did you suffer from in the past?’ by ticking the corresponding boxes on the list of disorders. Similarly, they also reported which medical specialists they saw in the past 2 years. The total numbers of disorders and medical specializations on the list were 211 and 10, respectively. The associations between these 211 binary variables and Toxoplasma infection were analyzed by the partial Kendall Tau correlation test with age as a covariate. Out of the 211 disorders, 134 were reported by at least 10 respondents. Of this subset of 134 diseases, 77 (57%) showed a significant association with toxoplasmosis in subjects of both genders altogether, 34 (24%) in men and 51 (38%) in women after the correction for multiple tests with Benjamini-Hochberg procedure (Table 3). All of these associations -except for one- were positive, i.e. the infected subject reported a higher incidence of a particular disorder. Again, some variables were significantly correlated to toxoplasmosis exclusively in one gender. For instance, as to increased frequency of migraines and anxieties were only observed in men while increased frequency of epilepsy and fasciculation only in women.
Table 3. Differences in the incidence of particular disorders between Toxoplasma-infected and Toxoplasma-free subjects
Numbers of Toxoplasma-free subjects without particular disorders, Toxoplasma-free subjects with particular disorders, Toxoplasma-infected subjects without particular disorders, Toxoplasma-infected subjects with particular disorders, odds ratios (OR), and statistical significance (P), respectively, are shown in six columns of each section. The effect of age on health status was controlled in partial Kendall's correlation (bivariate non-parametric) test when the incidence of particular disorder was higher than 9, otherwise the univariate exact Fisher test was used for computing statistical significance. ORs higher than 1 indicate positive association of Toxoplasma infection with incidence of particular disorder. Asterisks indicate results significant in two-sided tests after Benjamini-Hochberg procedure correction for multiple tests. P-values <0·0005 are coded as 0·000.
It is possible that a general population consists of two subpopulations, one healthy and one troubled with many disorders and diseases, including toxoplasmosis. This could result in a false correlation of toxoplasmosis with various disorders. If this is true, then we should see much stronger and more numerous associations between particular disorders only in the whole population than in the population consisting of Toxoplasma-free subjects only. To test this, we measured the correlation between those Toxoplasma seropositivity-associated disorders that occurred among at least in 150 subjects, both men and women (bronchitis/pneumonia, flatulence, urethral tract infection, low sexual appetency and scoliosis). As the incidence of certain disorders differs in men and women, we analyzed the associations separately for men and women. While there were strong positive correlations between all these disorders in the total population, except those between sexual appetency and bronchitis/pneumonia, urogenital infections, and scoliosis in men, we detected only certain correlations in subpopulations of Toxoplasma-negative subjects. For example, in men, we detected only the positive correlation of urogenital infections and bronchitis/pneumonia (τ = 0·10, P = 0·005), flatulence (τ = 0·10, P = 0·007) and low sexual appetency (τ = 0·13, P = 0·005). Moreover, the correlation between scoliosis and bronchitis/pneumonia in men was negative (τ = −0·09, P = 0·015).
DISCUSSION
Out of the 29 ordinary variables describing health problems, 28 showed a statistically significant association with T. gondii infection after the correction for multiple tests. For example, infected subjects reported to take higher numbers of drugs prescribed by doctors, as well as higher number of alternative medicines such as herbs, food supplements, multivitamins, superfoods, etc. per day, higher frequency of seeking medical attention for a serious illness (not injury) that lasted more than 3 days in the past 5 years, a higher number of types of medical specialists appointed in the past 2 years, to feel tired and to have a headache more often. They also reported to have more severe or more frequent allergic, immune, digestive, heart and vascular, metabolic, cancer, fertility, genitourinary, neurological, psychiatric, musculoskeletal and respiratory disorders. Of the 134 specific disorders ticked by at least 10 respondents, 77 showed a statistically significant association with toxoplasmosis in the whole population of men and women. Strong associations were observed for example with bronchitis (but not with pharyngitis), acquired immunodeficiencies except AIDS, both diarrhoea and constipation, mononucleosis, allergies, amoebiasis, coeliac disease, weight loss, recurrent abortion, hypothyroidism, leukemia, cervical uterine cancer, tics, fasciculation, learning disabilities, depression (men), autism, osteoporosis, scoliosis and asthma. For example, the observation of 77 (57%) of significant associations of 134 tests was more than ten times the 6·7 (5%) expected false significant results, indicating that most of the study results were not simply due to multiple comparisons. Again this conclusion was formally confirmed with Benjamini-Hochberg procedure.
This study confirmed the results of a previous ecological study that has shown a strong association between the incidence of various diseases and disorders and seroprevalence of toxoplasmosis in particular countries (Flegr et al. Reference Flegr, Prandota, Sovickova and Israili2014). Many of the observed associations have been already reported to exist in various, mostly small-scale, case-control studies, for review see (Flegr et al. Reference Flegr, Prandota, Sovickova and Israili2014). Some of the observed associations were, however, new or at least underreported. For example, the strong associations of toxoplasmosis with many digestive system disorders (acute and chronic diarrhoea, constipation, flatulence, chronic gastritis, maldigestion and food intolerance, weight loss, pyrosis and reflux, gall bladder attacks and coeliac disease) suggest that namely the digestive organs of infected hosts are affected by toxoplasmosis. Being a food-borne parasite, T. gondii first comes into contact with the host's intestinal lumen and cells, including enterocytes, fibroblasts and intestinal neurons. The intestine is not only the first site of infection but also a site where the first line of immune response usually occurs. As a result of the immune system's efforts to fight off the infection, cytokine levels in the nearby tissue are altered. This could result in acute inflammation and immunization of the host by various food antigens and autoimmunization. All these could be the causes of the observed toxoplasmosis-associated gastrointestinal afflictions (Prandota, Reference Prandota2012; Bhadra et al. Reference Bhadra, Cobb, Weiss and Khan2013).
Another class of toxoplasmosis-associated health problems that had been underreported in the scientific literature were immunity-related problems. The general pattern was that infected subjects expressed symptoms of immunodeficiency, allergy and also autoimmunity. Such changes could result in the observed increased risk of bronchitis, bacterial skin infections, urogenital infections, hypothyroidism, osteoporosis and scoliosis. Some of observed blood cell-related disorders, e.g. mononucleosis, could be the results of the manipulative activity of Toxoplasma, either the suppression or downregulation of the immune defence of the infected host (Flegr & Stříž, Reference Flegr and Stříž2011) or enhancement of migratory activity of the infected leukocytes to facilitate transport of Toxoplasma to other organs, including the brain (Lambert et al. Reference Lambert, Dellacasa-Lindberg and Barragan2011; Fuks et al. Reference Fuks, Arrighi, Weidner, Kumar Mendu, Jin, Wallin, Rethi, Birnir and Barragan2012).
Very strong associations were also observed between toxoplasmosis and certain neurological problems, such as tics, fasciculations, migraine and headache. The association was found also with epilepsy, which was in an agreement with previous correlation and case control studies (Ngoungou et al. Reference Ngoungou, Bhalla, Nzoghe, Darde and Preux2015). The relation between toxoplasmosis and neurological problems could be explained by the presence of tissue cysts of the parasite and associated lesions (Berenreiterova et al. Reference Berenreiterova, Flegr, Kubena and Nemec2011) in the infected brain tissue.
The present study showed the expected associations between toxoplasmosis and some neuropsychiatric disorders, for a review see (Flegr, Reference Flegr2013a ). This concerned epilepsy, migraine, tics, fasciculation, autism, learning disabilities. Some associations, e.g. cramps, panic disorder, unipolar depression, anxiety disorder, attention deficit and hyperactivity were significant only in men and some were just borderline, e.g. posttraumatic disorder in women, probably due to the low number of the affected subjects in the study population or opposite effects of toxoplasmosis in men and women. In addition to the expected positive correlations, certain negative associations between toxoplasmosis and mental health disorders were observed. For example, a negative association with unipolar depression and panic disorder in women and a negative association with alcohol use disorder in both men and women. The lower incidence of alcohol use disorder could be the result of decreased novelty seeking observed in infected subjects (Flegr et al. Reference Flegr, Preiss, Klose, Havlíček, Vitáková and Kodym2003; Skallová et al. Reference Skallová, Novotná, Kolbeková, Gašová, Veselý and Flegr2005). A lower incidence of panic disorder is in agreement with the results of earlier studies where infected women under imminent danger remained rather slow and passive and had a weak instinct for self-preservation – they claimed to stay abnormally calm in dangerous situations (Flegr, Reference Flegr2010, Reference Flegr2013b ). No explanation was found for the observed lower incidence of unipolar depression in the infected women – six of 235 (2·6%) vs 35 of 659 (5·3%) in Toxoplasma-free women (OR = 0·47, P = 0·006). A positive rather than negative association should be expected in view of the decreased concentration of tryptophan (Hsu et al. Reference Hsu, Groer and Beckie2014) as well as in view of the increased incidence of suicides in the infected subjects (Pedersen et al. Reference Pedersen, Mortensen, Norgaard-Pedersen and Postolache2012; Hsu et al. Reference Hsu, Groer and Beckie2014). The absence of association between toxoplasmosis and unipolar depression in the mixed population of men and women was in agreement with the negative results of some previous studies (Gale et al. Reference Gale, Brown, Berrett, Erickson and Hedges2014; Markovitz et al. Reference Markovitz, Simanek, Yolken, Galea, Koenen, Chen and Aiello2015).
Study limitations
The design of the present study, a cross-sectional study performed on a large cohort of internet users, has some strengths and some limitations. It is in principle the only method suitable for the assessment of the real strength of association between toxoplasmosis and other disorders, and therefore the real potential health impact of toxoplasmosis in the general population. It must be emphasized, however, that not all observed associations exist due to the effect of toxoplasmosis on human health, see the discussion about the causality problem below. Also, the results concerning the associations of an environmental factor, here a past Toxoplasma infection, with the incidence of uncommon diseases are not reliable enough because of low number of subjects with rare diseases in our population sample.
A second limitation of the present study is the possible influence of a strong sieve effect. The respondents entered the study voluntarily, without any financial reward. It is highly probable that only a certain kind of people (e.g. extreme altruists) are willing to spend 30–60 min to answer the questions of the electronic questionnaire. The Facebook community Guinea Pigs consists primarily of individuals interested in participating in evolutionary psychology studies (Flegr & Hodny, Reference Flegr and Hodny2016). However, this test was advertised as part of a study of the influence of the blood group (not toxoplasmosis) on human performance and health. It is, therefore, possible that people with special interest in their own health, e.g. the people with health problem, preferably took part in the present study. This could result in positively biased incidence rates of particular disorders, therefore this part of our results cannot by generalized on whole Czech population. However, there is no reason to expect that such a bias will differ between Toxoplasma-infected and Toxoplasma-free subjects.
The study was based on the information provided by the subjects themselves and not on objective medical records. It is highly probable that some part of the data is wrong or at least obsolete. For example, some of the subjects who tested negative for toxoplasmosis in the past could have acquired the infection in the meantime. Again, such errors increase the risk of false negative and not false positive results of statistical tests (Flegr, Reference Flegr2016).
All participants of the present study have been informed about their toxoplasmosis status in the past, mostly after finishing our previous behavioural studies. It is possible that the awareness about being infected by Toxoplasma could have a negative effect on the subjects’ health or that it motivates the infected subjects to have more medical examinations, which could reveal more health disorders. However, up to now, all medical textbooks say, and nearly all medical doctors firmly believe, that latent toxoplasmosis, i.e. the presence of low concentration of anamnestic anti-Toxoplasma IgG antibodies have no impact on the health and quality of life of a subject. Participants of our past studies have obtained this information (in a written form) as part of debriefing after finishing their participation in (double blind) behavioural experiments. Therefore, it is rather improbable that the knowledge of their latent infection could have a strong impact on their health or on the frequencies or intensity of their medical examinations. Moreover, some of the observed associations between toxoplasmosis and various disorders have already been described in literature, mostly on the basis of case-controls studies. In these studies, no effect of the participants’ knowledge of toxoplasmosis status exists. Still it would be very important to repeat our cohort study on a similar population of subjects who will be tested for the presence of anti-Toxoplasma antibodies after completing their anamnestic questionnaire.
It must be emphasized that no observational study could definitively solve the causality problem. The cross-sectional study, including our cohort study, could quantify the probability that statistical association exists between the effects A and B. However, it cannot decide whether A (e.g. toxoplasmosis) is the effect of B (impaired health status, for example impaired immunocompetence) or whether it is the cause the B. There is a relatively low probability that, for example, schizophrenia, muscular fasciculation or pneumonia could cause the Toxoplasma infection. It is, however, always possible that both A and B are caused by an unknown factor C, for example by being born in a village, rather than in a large city. It is often believed that the problem of causality could be solved by longitudinal studies. However, even this is only partly true in the real word. It is possible that the specific symptoms of a disorder, e.g. acquired immunodeficiency, evolve earlier than the humoral anti-Toxoplasma immunity and therefore the disorder could be diagnosed earlier than the presence of Toxoplasma infection in some affected subjects. In the other words, even the criterion of temporality could fail. Therefore, any conclusions regarding the causal relation between two factors based on the results of observational study must always be considered just preliminary, and must be confirmed by manipulative study, for example by the experimental infection of laboratory animals.
This study is of exploratory nature. Still, the results in tables are presented after the correction for multiple tests. It must be stressed out, however, that the number of the obtained significant results is ten times higher than the theoretical number of false significant results obtained due to multiple tests. This suggests that most of the positive results reflect the existence of real biological effects rather than being an artefact of the multiple tests.
In conclusion, our study demonstrated a very strong negative association between toxoplasmosis and the health status of infected subjects. The responders diagnosed in the past with toxoplasmosis on the basis of presence of anti-Toxoplasma antibodies expressed a higher incidence of many, but not all, disorders, suggesting that the association with toxoplasmosis is not only strong, but also specific. Unlike the previous case-control studies and anecdotal observations, this explorative non-clinical cohort based study suggests that the immune system and digestive organs could be most strongly affected by the infection.
Toxoplasma is probably the most common parasitic disease in developed countries. Its prevalence is declining in the USA and in most countries in Europe. However, it is on the rise in highly populated Asian countries like China and Korea. Therefore, the global impact of toxoplasmosis is probably increasing. Neither an effective method for the treatment of latent toxoplasmosis nor an effective human vaccine is now available. The results of recent studies, including the present one, suggest that searching for effective drugs and a safe vaccine is of utmost importance.
SUPPLEMENTARY MATERIAL
To view supplementary material for this article, please visit http://dx.doi.org/10.1017/S0031182016001785.
ACKNOWLEDGEMENTS
We would like to thank Zdeněk Hodný, Rudolf Hoffmann, and Mike Dammann, for their help in preparing the medical questionnaire and Charlie Nichols and Raymon Gongora for their help with the final version of the paper.
FINANCIAL SUPPORT
The work of J. F. was supported by project UNCE 204004 (Charles University in Prague) and the Czech Science Foundation (Grant No. P303/16/20958).
