Search

Additional file 1: Table S1. Distribution of various tissue elements by the type of benign breast disease. Table S2. Associations of reproductive factors with tissue types in benign breast biopsy samples using proc glimmix procedure (Odds Ratios and 95% Confidence Intervals). Table S3. Associations of reproductive variables with tissue types in benign breast biopsy samples of premenopausal women using proc glimmix procedure (Odds Ratios and 95% Confidence Intervals). Table S4. Associations of reproductive variables with percentage of different tissue types (log-transformed) in benign breast biopsy samples (β coefficients and 95% Confidence Intervals), additionally adjusted for BBD subtype. Table S5. Associations of reproductive variables with percentage of different tissue types (log-transformed) in benign breast biopsy samples of premenopausal women (β coefficients and 95% Confidence Intervals), additionally adjusted for BBD subtype.

Additional file 1: S1. Structural nested cumulative failure time models with a time-varying instrumental variable. Supplementary Figure 1. Distributions of �� estimates across 1,000 iterations using g-estimation with an instrumental variable under different data-generating mechanisms, different sample sizes (n = 10,000; n = 25,000 or n = 50,000), and different instrument-exposure strengths (��ZA = 0.10; �� = 0.25; or �� = 0.45). The lower and upper hinges correspond to the 25th and 75th percentile. The lower and upper whiskers extend from the hinge to the smallest and largest values no further than 1.5*IQR from the hinge, where IQR is the interquartile range. The median is represented by the line between the hinges, and the mean is represented by the diamond point symbol. The percentages provided above each box plot represents the percentage of iterations in which the model did not converge. Supplementary Figure 2. A. Distributions of marginal risk differences across 1,000 iterations using different g-estimation approaches under different data-generating mechanisms with �� = 5%. The lower and upper hinges correspond to the 25th and 75th percentile. The lower and upper whiskers extend from the hinge to the smallest and largest values no further than 1.5*IQR from the hinge, where IQR is the interquartile range. The median is represented by the line between the hinges, and the mean is represented by the diamond point symbol. B. Distributions of marginal risk ratios across 1,000 iterations using different g-estimation approaches under different data-generating mechanisms with �� = 5%. The lower and upper hinges correspond to the 25th and 75th percentile. The lower and upper whiskers extend from the hinge to the smallest and largest values no further than 1.5*IQR from the hinge, where IQR is the interquartile range. The median is represented by the line between the hinges, and the mean is represented by the diamond point symbol. Supplementary Figure 3. A. Distributions of marginal risk differences across 1,000 iterations using different g-estimation approaches under different data-generating mechanisms with baseline hazards of 5%, 10% and 25%. The lower and upper hinges correspond to the 25th and 75th percentile. The lower and upper whiskers extend from the hinge to the smallest and largest values no further than 1.5*IQR from the hinge, where IQR is the interquartile range. The median is represented by the line between the hinges, and the mean is represented by the diamond point symbol. B. Distributions of marginal risk ratios across 1,000 iterations using different g-estimation approaches under different data-generating mechanisms with baseline hazards of 5%, 10% and 25%. The lower and upper hinges correspond to the 25th and 75th percentile. The lower and upper whiskers extend from the hinge to the smallest and largest values no further than 1.5*IQR from the hinge, where IQR is the interquartile range. The median is represented by the line between the hinges, and the mean is represented by the diamond point symbol. Supplementary Figure 4. Flowchart of inclusion and exclusion of Nurses��� Health Study participants. Supplementary Table 1. Characteristics of Nurses��� Health Study I participants at start of follow-up. Supplementary Table 2. SNPs identified as proposed instruments for alcohol intake. Supplementary Table 3. A. Associations of proposed instrument (weighted allele score) with alcohol intake across baseline five-year age groups. B. Distributions of alcohol intake and proportion of heavy drinkers across quartiles of the proposed instrument (weighted allele score) across baseline five-year age groups. Supplementary Table 4. Associations of the proposed instrument (weighted allele score) with hazard of endometrial cancer across baseline five-year age groups.