Trial Outcomes & Findings for Organic Diet Intervention in Primary School Children (NCT NCT02998203)

Unit of analysis: Schools

Race and Ethnicity were not collected from any participant.

Percent change in pesticide metabolite 3-phenoxybenzoic acid (3-PBA) levels between the last sample of the conventional treatment period (before the start of the organic treatment) and the last sample of organic treatment period. A one-sample t-test was used to assess whether the percent change was different than zero. \*Note: This test does not take into account the repeated measures for each participant and the organic treatment duration. The results of the linear-mixed effect models are the ones considered (reported in a section below) as they take into account the repeated measures for each participant and the organic treatment duration.

Overall difference in median 3-PBA levels between the conventional and organic periods. The overall differences in the medians of biomarkers between the conventional and the organic phase were assessed with the non-parametric Wilcoxon rank sum test on the creatinine-adjusted concentrations pooling all conventional samples (including the baseline) and the organic samples for all participants, regardless of the duration for which they followed the organic treatment. \*Note: This test does not take into account the repeated measures for each participant and the organic treatment duration. The results of the linear-mixed effect models are the ones considered (reported in a section below) as they take into account the repeated measures for each participant and the organic treatment duration.

A linear mixed-effect regression model was fitted for 3-PBA (log-transformed, creatinine-adjusted). The model accounted for the duration and the effect of treatment. It included student-level (repeated measures within person) and school-level (multiple students clustered within each school) random intercepts with an unstructured covariance matrix. Continuous variables, other than time (days of treatment), were centered at the population means. The model included fixed effects for treatment condition (organic or conventional) and time (days of treatment), where time = 0 was used for the start of the treatment. The model was adjusted for the baseline value (first urine sample for all children) of the outcome to account for the background participant levels. An interaction term for time and treatment was considered and subsequently dropped if it did not meet the threshold of p-value\<0. 05.

Overall difference in median 6-CN levels between the conventional and organic periods. The overall differences in the medians of biomarkers between the conventional and the organic phase were assessed with the non-parametric Wilcoxon rank sum test on the creatinine-adjusted concentrations pooling all conventional samples (including the baseline) and the organic samples for all participants, regardless of the duration for which they followed the organic treatment. \*Note: This test does not take into account the repeated measures for each participant and the organic treatment duration. The results of the linear-mixed effect models are the ones considered (reported in a section below) as they take into account the repeated measures for each participant and the organic treatment duration.

A logistic model was fitted for 6-CN (binary variable; above and below LOD) due to the high number of values below LOD. The model accounted for the duration and the effect of treatment. It included student-level (repeated measures within person) and school-level (multiple students clustered within each school) random intercepts with an unstructured covariance matrix. Continuous variables, other than time (days of treatment), were centered at the population means. The model included fixed effects for treatment condition (organic or conventional) and time (days of treatment), where time = 0 was used for the start of the treatment. The model was adjusted for the baseline value (first urine sample for all children) of the outcome to account for the background participant levels. An interaction term for time and treatment was considered and subsequently dropped if it did not meet the threshold of p-value\<0. 05.

Percent change in oxidative stress/inflammation biomarker 8-OHdG levels between the last sample of the conventional treatment period (before the start of the organic treatment) and the last sample of organic treatment period. A one-sample t-test was used to assess whether the percent change was different than zero. \*Note: This test does not take into account the repeated measures for each participant and the organic treatment duration. The results of the linear-mixed effect models are the ones considered (reported in a section below) as they take into account the repeated measures for each participant and the organic treatment duration.

Overall difference in median 8-OHdG levels between the conventional and organic periods. The overall differences in the medians of biomarkers between the conventional and the organic phase were assessed with the non-parametric Wilcoxon rank sum test on the creatinine-adjusted concentrations pooling all conventional samples (including the baseline) and the organic samples for all participants, regardless of the duration for which they followed the organic treatment. \*Note: This test does not take into account the repeated measures for each participant and the organic treatment duration. The results of the linear-mixed effect models are the ones considered (reported in a section below) as they take into account the repeated measures for each participant and the organic treatment duration.

A linear mixed-effect regression model was fitted for 8-OHdG (log-transformed, creatinine-adjusted). The model accounted for the duration and the effect of treatment. It included student-level (repeated measures within person) and school-level (multiple students clustered within each school) random intercepts with an unstructured covariance matrix. Continuous variables, other than time (days of treatment), were centered at the population means. The model included fixed effects for treatment condition (organic or conventional) and time (days of treatment), where time = 0 was used for the start of the treatment. The model was adjusted for the baseline value (first urine sample for all children) of the outcome to account for the background participant levels. An interaction term for time and treatment was considered and subsequently dropped if it did not meet the threshold of p-value\<0. 05.

Percent change in oxidative stress/inflammation biomarker 8-iso-PGF2a levels between the last sample of the conventional treatment period (before the start of the organic treatment) and the last sample of organic treatment period. A one-sample t-test was used to assess whether the percent change was different than zero. \*Note: This test does not take into account the repeated measures for each participant and the organic treatment duration. The results of the linear-mixed effect models are the ones considered (reported in a section below) as they take into account the repeated measures for each participant and the organic treatment duration.

Overall difference in median 8-iso-PGF2a levels between the conventional and organic periods. The overall differences in the medians of biomarkers between the conventional and the organic phase were assessed with the non-parametric Wilcoxon rank sum test on the creatinine-adjusted concentrations pooling all conventional samples (including the baseline) and the organic samples for all participants, regardless of the duration for which they followed the organic treatment. \*Note: This test does not take into account the repeated measures for each participant and the organic treatment duration. The results of the linear-mixed effect models are the ones considered (reported in a section below) as they take into account the repeated measures for each participant and the organic treatment duration.

A linear mixed-effect regression model was fitted for 8-iso-PGF2a (log-transformed, creatinine-adjusted). The model accounted for the duration and the effect of treatment. It included student-level (repeated measures within person) and school-level (multiple students clustered within each school) random intercepts with an unstructured covariance matrix. Continuous variables, other than time (days of treatment), were centered at the population means. The model included fixed effects for treatment condition (organic or conventional) and time (days of treatment), where time = 0 was used for the start of the treatment. The model was adjusted for the baseline value (first urine sample for all children) of the outcome to account for the background participant levels. An interaction term for time and treatment was considered and subsequently dropped if it did not meet the threshold of p-value\<0. 05.

Percent change in oxidative stress/inflammation biomarker MDA levels between the last sample of the conventional treatment period (before the start of the organic treatment) and the last sample of organic treatment period. A one-sample t-test was used to assess whether the percent change was different than zero. \*Note: This test does not take into account the repeated measures for each participant and the organic treatment duration. The results of the linear-mixed effect models are the ones considered (reported in a section below) as they take into account the repeated measures for each participant and the organic treatment duration.

Overall difference in median MDA levels between the conventional and organic periods. The overall differences in the medians of biomarkers between the conventional and the organic phase were assessed with the non-parametric Wilcoxon rank sum test on the creatinine-adjusted concentrations pooling all conventional samples (including the baseline) and the organic samples for all participants, regardless of the duration for which they followed the organic treatment. \*Note: This test does not take into account the repeated measures for each participant and the organic treatment duration. The results of the linear-mixed effect models are the ones considered (reported in a section below) as they take into account the repeated measures for each participant and the organic treatment duration.

A linear mixed-effect regression model was fitted for MDA (log-transformed, creatinine-adjusted). The model accounted for the duration and the effect of treatment. It included student-level (repeated measures within person) and school-level (multiple students clustered within each school) random intercepts with an unstructured covariance matrix. Continuous variables, other than time (days of treatment), were centered at the population means. The model included fixed effects for treatment condition (organic or conventional) and time (days of treatment), where time = 0 was used for the start of the treatment. The model was adjusted for the baseline value (first urine sample for all children) of the outcome to account for the background participant levels. An interaction term for time and treatment was considered and subsequently dropped if it did not meet the threshold of p-value\<0. 05.