Next, we tested whether type of SLE (i.e., dependent or independent) moderated the association between drinking onset and PYDD (Table 3). For dSLE in women, the full model explained 11.6% of the total variation in PYDD, with 5.1% of the variation attributable to demographic variables, 5.7% to drinking onset, and 0.4% to dSLE. The onset × dSLE interactions accounted for 0.8%, due primarily to the dSLE × ONS17 term, which was statistically significant (b = 0.31, SE = 0.11). As shown in Fig. 3A, estimates based on the full model indicate that relative to women in the same onset group with no dSLE, those reporting at least 1 dSLE were estimated to consume 0.9 additional drinks per week if they began drinking by age 14 (1.8 drinks/wk for 0 dSLE vs. 2.7 drinks/wk for 1+ dSLE for onset ≤14), and 0.4 additional drinks if they began drinking by age 17 (1.3 drinks/wk for 0 dSLE vs. 1.7 drinks/wk for 1+ dSLE), whereas those who began drinking after age 17 drank slightly less (0.3 drinks) if they reported any dSLE (1.0 drink/week for no dSLE vs. 0.7 drinks/week for 1+ dSLE).
