Study design

Following the pilot study, a further investigation was carried out to study the relationship between age and infection in 2014. Based on the published literature, an increased seroprevalence with age was expected. Thus, a hypothesis of higher seroprevalence in older pigs was to be tested. Due to the aims of the study, only intensive farms with full cycles were sampled; a full cycle farm is a farm in which animals are kept from birth to the slaughter. A second sampling strategy was carried out to investigate the relationship between environment and T. gondii prevalence. To carry this out, adult pigs (20 weeks old) were sampled in farms located within the two vegetation types, tropical deciduous low forest and tropical sub- deciduous medium forest. The tropical subdeciduous medium forest has less drastic environmental conditions as the dry season lasts for shorter and, the remained vegetation

protects the soil from overheating. This conditions are favorable for the oocyst survival and therefore, higher seroprevalence was expected in the tropical subdeciduous medium forest. Thus, a hypothesis of higher seroprevalence in the tropical subdeciduous medium forest was to be tested.

The approximate total sample size for the first study was calculated using the prevalence obtained in the pilot study, with an absolute error of 7% and a 95% of confidence level and a design effect (D) of 3.52, using the software Epi-info 7.1.3 (CDC, Atlanta, USA., 2007). Where D = 1+ rho (n-1), being n the number of animals to sample in each cluster and rho is the intra-cluster correlation coefficient with a value of 0.04 (Ortega-Pacheco et al., 2013). A multistage stratified cluster sampling was carried out in which each cluster or farm was stratified in the four age groups or strata (8 weeks, 12 weeks, 16 weeks and 20 weeks). The number of samples collected per each age group (16) was calculated with a power test using the software Epi-info 7.1.3 (CDC, Atlanta, USA, 2007) with an alpha level of 95% and a power of 0.8. In which the estimated prevalence of the uninfected animals was up to 20% and the prevalence of the infected animals was at least 70%. Therefore, 64 samples per farm were collected (16 x 4 age groups). The total number of farms was calculated dividing the total sampled size obtained with Epi-info by the samples to collect in each farm, the number of farms was rounded to six; consequently, 384 samples were obtained in total for this study.

A second sampling was completed to compare the two habitats in order to study the effect of the environment with the transmission of T. gondii. The number of samples per farm was calculated using Lot Quality Assurance Sampling (LQAS) in a LQAS calculator (http://www.brixtonhealth.com/hyperLQAS.html), where the upper threshold was considered 0.7 and the lower threshold 0.2 based on the prevalence obtained in the pilot study and previous studies in this geographical area. Type I and II errors were considered as 0.05. The total number of samples per farm calculated was 10 with a decision rule of 4. This means that if four or less animals of 10 are positive, the true prevalence is lower than 70% and the alternative hypothesis is rejected. On the contrary, if more than four of 10 animals are positive, then the prevalence is greater than 20% and the null hypothesis is rejected. Sixty animals of 20 weeks old were sampled in total. LQAS is a random sampling which was originally developed in 1920 for quality control in industry and is where its name comes from as in industry products are usually

made in lots. LQAS is a low cost methodology as it utilizes a small sample size. For this reasons the use of this sampling method in health surveillance has been growing since 1980 (Lanata and Black, 1991, Robertson and Velazquez, 2006).

To study the risk factors associated with toxoplasmosis transmission, a survey was conducted by interviewing farmers of the 12 farms (questionnaire appendix II.2). Information such as the size of the herd, presence and number of cats, presence of birds, cannibalism and mice, dimensions of the farm, the percentage of abortions, the type of feeders and drinkers, the warehouse and corral characteristics, pig movement thought the farm, pest control routine, the origin of the water and the sanitary program of the farm were collected. In addition, gender and age were recorded from all animals. The questionnaire was designed by reviewing risk assessment studies on toxoplasmosis in pigs (Lehmann et al., 2003, Meerburg et al., 2006, Lubroth et al 1983, Ortega-Pacheco et al., 2013, Dubey and Beattie, 1988, Assadi-Rad et al., 1995, Villari et al., 2009, Garcia et al., 1999, Weigel et al., 1995b, Dubey et al., 1986b, Dubey et al., 1995a, Hill et al., 2010).

Phlebotomy was performed in the external jugular vein using a vacutainer system, where blood was transferred directly in two 5 ml vacutainer tubes per animal, one with EDTA for DNA extraction and another with a serum separator gel for serological screening. The vacutainer system ensured the sterility of the blood and avoided cross contamination. Samples were transported refrigerated in a cooler to the research centre Hideyo Noguchi in Merida for subsequent examination. Serum tubes were centrifuged at 1100 g for 15 minutes to remove the clot and the serum was apportioned into two 1.5 ml clean polypropylene tubes and stored at - 20°C until they were analysed.