Results of interaction analyses involving alcohol use, parity and their combination as predictors of the effect of omega-3 supplementation on PTB
| Any previous births ≥20 weeks | Drank alcohol in 3 months leading up to pregnancy | Omega-3 n/N (%) Preterm | Control n/N (%) Preterm | OR (95% CI) | Interaction P value |
| Overall | Overall | 169/2654 (6.37) | 209/2674 (7.82) | 0.80 (0.65, 0.99)* | |
| No | Overall | 84/1170 (7.18) | 81/1164 (6.96) | 1.03 (0.75, 1.42) | 0.04 |
| Yes | Overall | 84/1473 (5.70) | 128/1502 (8.52) | 0.65 (0.49, 0.87) | |
| Overall | No | 68/1196 (5.69) | 103/1161 (8.87) | 0.62 (0.45, 0.86) | 0.04 |
| Overall | Yes | 100/1446 (6.92) | 106/1505 (7.04) | 0.97 (0.73, 1.30) | |
| No | No | 25/454 (5.51) | 42/451 (9.31) | 0.57 (0.34, 0.95)† | 0.02† |
| No | Yes | 59/716 (8.24) | 39/713 (5.47) | 1.54 (1.01, 2.34)† | |
| Yes | No | 43/742 (5.80) | 61/710 (8.59) | 0.66 (0.44, 0.99)† | |
| Yes | Yes | 41/730 (5.62) | 67/791 (8.47) | 0.64 (0.43, 0.97)† |
ORs are comparing the omega-3 group to the control group. ORs >1 indicate the odds of preterm birth are higher in the omega-3 group, while ORs <1 indicate the odds of preterm birth are higher in the control group. Unless otherwise indicated, ORs are based on a logistic regression model including treatment group, alcohol use/parity and an interaction between treatment group and alcohol use/parity, with adjustment for enrolment centre and use of omega-3 supplements in the last 3 months.
*OR is based on a logistic regression model including treatment group, with adjustment for enrolment centre and use of omega-3 supplements in the last 3 months.
†ORs are based on a logistic regression model including treatment group, alcohol use, parity, all two-way interactions and a three-way interaction, with adjustment for enrolment centre and use of omega-3 supplements in the last 3 months. Interaction p value is for the three-way interaction.
PTB, preterm birth.