The study was conducted by ophthalmologists at Clinica Silveira and the Federal University at São Paulo, Brazil, in collaboration with researchers from the University of California, Los Angeles, California, and the Centers for Disease Control and Prevention (CDC), Atlanta, Georgia. The questionnaire was adapted from those used in previous T. gondii case-control studies (12,13), with modification and input from the study researchers. The questionnaire took ≈25 minutes to complete and inquired about demographic variables and a comprehensive set of risk factors related to meat, vegetables, food preparation, soil contact, drinking water, and animal exposure (especially cat and cat feces exposure). Information about the number of pregnancies and children was collected from women participating in the survey. Questions involving habitual behavior focused on the most recent 12 months. The questionnaire was self-administered at Clinica Silveira, the principal ophthalmology clinic in Erechim, with parental assistance for children <16 years of age.

The study was reviewed and approved by the human subjects review committee at the Federal University of São Paulo and approved for analysis at CDC. Oral consent was obtained for completion of the questionnaire. All participants with ocular disease received care and follow-up at Clinica Silveira.

Case-patients with recent T. gondii infection were defined as consecutive visitors to the clinic from June 2003 to June 2004 with positive test results for both T. gondii–specific immunoglobulin G (IgG) and IgM. For each patient, an attempt was made to identify a control as the next patient with negative T. gondii antibody test results. Antibody testing was performed at the Fleury Laboratory by using the Abbott AxSYM system (Abbott Diagnostics, Abbott Park, IL, USA).

Data were first examined with univariate analysis; then factors with p value <0.05 for association with recent T. gondii infection (and sufficient sample size) were entered into logistic regression models for the sample as a whole (N = 241), for women >18 years of age (n = 86), and children <18 years of age (n = 106). Men >18 years of age were not examined as a subgroup in logistic models because the sample size was too small (n = 49). In odds ratio (OR) calculations with a zero value in >1 cells, a correction of 0.5 was added to every cell to avoid undefined results (logit method). Attributable risks were calculated from the logistic regression results for factors that significantly (p<0.05) increased risk for T. gondii infection by using the following formula: attributable risk = proportion of cases exposed for each factor × ([OR – 1]/OR). Data analysis was conducted with SAS software (14).

Of 140 persons who had positive Toxoplasma IgG and IgM test results, 131 (93.6%) agreed to participate and completed questionnaires. Of 121 persons who had negative Toxoplasma IgG and IgM test results, 110 (90.9%) agreed to participate and completed questionnaires. The demographics of patients and controls are presented in Table 1. Controls were more likely to be in the 16- to 20- and 21- to 30-year age ranges and female and less likely to be in the <5-, 31–40-, and >41-year age groups.

In univariate analysis among the group as a whole, patients were significantly more likely than controls to eat rare meat (this question referred to meat in general); eat cured, dried, or smoked meat given by a friend or relative (purchased cured, dried, or smoked meat was not significantly associated with case status); have a garden; work in the yard or garden; eat frozen lamb; and be male. Patients were less likely to eat raw beef, eat raw ground beef, eat raw ground chicken, and be born in Erechim or upper Uruguay (Table 2). Among women >18 years of age, having had >1 child (OR 14.68, 95% confidence interval [CI] 4.10–52.50, p<0.0001) (or >1 pregnancy, colinear with number of children) was strongly associated with risk for T. gondii infection (Table 3). Women infected with T. gondii were also more likely than those not infected to have lived in their present home >12 months, work with animals (all types), work in the garden, and have >1 cat or kitten living at home. They were less likely to be infected if they ate raw ground beef, raw chicken, rare or raw ground chicken, and fed their cat dried food. In addition, the trend in risk for recent T. gondii infection increased from having no children (4 [17.4%] of 23), to 1 child (11 [68.8%] of 16), to >2 children (23 [79.3%] of 39) (p<0.0001).

Among children <18 years of age, in univariate analysis those with recent T. gondii infection were more likely than those without infection to have a cat that catches its own food; feed the cat raw food; work in the yard more than once per week; eat cured, dried, or smoked meat given by a friend or relative; eat frozen lamb; eat rare meat; eat rare pork; and be male. They were less likely to be born in Erechim or upper Uruguay or wear gloves while working in the yard (Table 4).

In multivariate analysis for the complete group, persons recently infected with T. gondii were significantly more likely than those who were not infected to work in the garden (OR 2.35, 95% CI 1.27–4.33, p = 0.006) and eat frozen lamb (OR 2.06, 95% CI 1.15–3.67, p = 0.02) but less likely to be born in Erechim or upper Uruguay (OR 0.19, 95% CI 0.09–0.39, p<0.0001) (Table 5). In multivariate analysis, for women >18 years of age, having had >1 child compared to having no children markedly increased risk for recent T. gondii infection (OR 14.94, 95% CI 3.68–60.73, p<0.0002) (Table 5). For children <18 years of age, being born in Erechim or upper Uruguay was associated with a reduced risk for recent T. gondii infection compared to those born elsewhere (OR 0.04, 95% CI 0.01–0.22, p = 0.0002), and male sex (OR 5.70, 95% CI 1.95–16.66, p = 0.002) was associated with an increased risk for T. gondii infection (Table 5).

The attributable risks for the factors that significantly increased the risk for recent T. gondii infection in logistic regression for the group as a whole (from Table 5) were working in the garden (0.25) and eating frozen lamb (0.15). Among women >18 years of age, having had >1 child was associated with a high attributable risk (0.84). Among children <18 years of age, male sex was associated with an attributable risk of 0.64.

In the study group overall (Table 2), we found that a number of meat- and soil-related factors were associated with recent T. gondii infection, which included eating cured, dried, or smoked meat given by a friend or relative; eating rare meat; eating frozen lamb; and having a garden and working in the yard or garden (soil contact). The presence of meat-related factors emphasizes the importance of cooking meat well (>67°C) and not assuming that all home freezing methods will kill T. gondii cysts. Having a garden or working in the yard increases the chances of oocyst exposure through soil contact. Men were at increased risk for recent T. gondii infection, which could also be related to increased soil contact. To minimize T. gondii exposure, gloves should be worn while gardening, and hands should be washed thoroughly afterward with soap and water. Although sample size limitations precluded us from including all the variables in multivariate analysis, working in the garden remained significant in multivariate analysis.

Although the number of persons who ate raw meat was small, eating raw ground chicken was associated with a decreased risk for T. gondii infection. Chickens grown commercially are probably not involved in the transmission of T. gondii because modern methods reduce soil exposure, chickens are usually frozen for storage before purchase, and they are thoroughly cooked (15). Eating chicken could be associated with a decrease in risk because those who eat chicken may be less likely to eat other meats that are associated with a greater risk. However, free-range chickens in Brazil can be infected with T. gondii (9), and eating fresh, undercooked meat from free-range chickens could pose a risk for infection. In addition, eating raw beef or raw ground beef decreased the risk for T. gondii infection in univariate analysis, although the sample size was small for this variable as well. Beef is rarely contaminated with T. gondii and is not believed to be a consequential source of infection (15), but the role of beef in transmission to humans has not been completely determined. Eating raw ground beef that is contaminated with pork because the grinding machine was not cleaned after grinding pork could increase the risk for T. gondii infection. Eating rare meat (all types combined) was a risk factor for T. gondii infection. Pork and lamb are the most likely meats to be contaminated with T. gondii (15). Although these data are not specifically from Erechim, researchers in southern Brazil have recently reported that 8.7% of 149 fresh pork sausage samples were positive for T. gondii by bioassay in mice in the state of Parana (16), and 17% of 286 finishing pigs tested positive for T. gondii antibodies with the modified agglutination test in the state of São Paulo (17). Dubey et al. determined that tissue cysts survive in pork for <3 minutes if heated to 64°C (18), so cooking meat (especially pork or lamb) below this temperature could lead to infection.

The strongest factor in analysis of the group overall was the reduced risk for T. gondii infection among persons born in Erechim or upper Uruguay compared to persons born elsewhere. This reduction in risk was also significant in multivariate analysis and in children and is likely due to referral of patients with ocular disease from outside the area. Risk for T. gondii infection was also reduced among women born in Erechim or upper Uruguay, but the reduction was not significant (data not shown).

For women >18 years of age, having had >1 child was strongly associated with recent T. gondii infection in both univariate and multivariate analysis (Tables 3 and 5). Having had >1 pregnancy also showed a strong association with recent T. gondii infection in univariate analysis, but we did not examine it in multivariate analysis because it is highly correlated with the number of children. Pregnancy and number of children were risk factors for T. gondii infection in other studies, including several from Brazil (19–22). Avelino et al. (20) suggest that the greater vulnerability of pregnant women to T. gondii infection is probably due to alterations in immune mechanisms associated with gestation. In some settings, pregnancy could affect culinary habits. In our study, the number of children and past pregnancies increased risk for recent T. gondii infection. Another possible explanation for this finding is that children bring, or track, contaminated soil into the house, increasing risk for infection.

Among children <18 years of age, in univariate analysis many of the meat- and soil-related risk factors were associated with an increased risk for infection that was seen in the group as a whole (Table 4). In addition, feeding cats raw food and having cats that catch their own food were associated with increased risk. To prevent cats from becoming infected with T. gondii, they should be fed only dry, canned, or well-cooked food and kept indoors when possible to discourage hunting. In multivariate analysis of the risk factors for children, male sex was the principal factor that significantly increased risk for infection, although eating rare meat and frozen lamb approached significance.

One of the strengths of our study is that serologically defined acute T. gondii infection was required for the case definition. Most T. gondii IgM- and IgG-positive persons are infected for <1 year, so their more recent behavior correlates with the time of their infection. Our study also has a number of limitations. The Clinica Silveira is known for expertise in ocular toxoplasmosis and draws patients with toxoplasmosis-related eye disease from the surrounding area. This fact is the likely reason why patients with acute T. gondii infection were more likely to come from outside the Erechim area than from Erechim. Although the sample size was adequate to identify numerous risk factors for T. gondii infection, it was not large enough to allow analysis for all subgroups (for example, multivariate analysis of adult men). Some associations in our study may have been due to confounders that we did not consider in the analysis (especially in univariate analysis); however, we inquired about a comprehensive set of variables, including those previously known to be associated with T. gondii infection, and performed multivariate analysis. In addition, our study was limited to 1 area of Brazil, and the results cannot necessarily be applied to other areas of the country.

In conclusion, our study identified a number of risk factors for T. gondii infection among persons with ocular disease. Because T. gondii infection often leads to ocular disease in this region of Brazil, avoiding infection can improve ocular health. Proper meat and soil- and water-related hygiene could reduce the risk for infection with T. gondii, especially with preventive education efforts based on risk factors identified in this case-control study. The association of T. gondii infection in women with the number of children they have had (and their pregnancies) requires further study to fully elucidate the factors that contribute to this increase in risk. In the future, type 1 T. gondii could spread to other areas of the world and increase the risk for ocular disease in those regions.