9% at T1 to 44.7% at T2 (p>0.1). There were no statistically significant changes in the use of Pesticides Ib and II between the two periods (p>0.1). At T1, the mean number of years of schooling was 6.1 (SD 2.4), and the mean age was 41 years (SD 13.0; table 1). In the lost-to-follow-up analysis

(data are not shown in tables), at T2, 19 individuals (8.3% of the total at T1) were either ALK kinase inhibitor lost-to-follow-up (14) or excluded for other reasons (5). In this latter group, the percentage of individuals (57.9%) who did not participate in organisations was significantly higher (p<0.1), as was the percentage of individuals who did not use IPM practices (47.5%). The mean value for neurocognitive performance at T1 was found to be insignificantly higher (4.8, SD 1.3, p>0.1). No statistically significant differences between these groups and the final study population were found (p>0.1) with respect to the mean number of years of schooling (6.4, SD 2.2) or age (44.2, SE 15.6) at T1. Multivariable analysis Results from the multivariate analysis (table 2) showed that for greater implementation of IPM practices (good/very good and slight/medium vs not applicable), the value of the neurocognitive performance index increased significantly (p<0.05; models A, B and C). In

these models, the magnitude of the coefficient of association was almost twice as high (β=0.71, SE 0.19) where the application of these practices was good/very good compared to when the application was slight/medium (β=0.41, SE 0.15), indicating a dose–response relationship. The coefficient of association between the implementation of IPM practices and neurocognitive performance was similar in all three models when people participated in organisations (A: β=0.24; B: β=0.27; and C: β=0.26). However, the values were significantly different (p<0.1) in models B and C when the association of the use of pesticides and product terms was adjusted. Table 2 Multivariable linear regression coefficients† (SE)‡ for the association between the use of IPM practices and neurocognitive performance (n=416 observations) In Batimastat Model

D, when the product term Application of good/very good IPM× Participation in organizations” was removed, the coefficient of association for the implementation of IPM practices good/very good decreased and the association was no longer significant (β=0.08, SE 0.15). Based on ‘QICu’ values and the existing literature,27 29 model B was chosen for the stratified analysis according to the presence or absence of participation in organisations. To more closely examine these relationships, we stratified table 1 data on use of IPM by participation in organisations (table 3). With stratification, we can clearly see: that Good/very good use of IPM was more prevalent among those participating in organisations at both times.