Table 3.
Sensitivity of association between smoking > 5 pack-years (versus 0 pack-years) and miscarriage among those who received > 2.5 Gy uterine radiation (N=121)a because of non-participation.
| Assumed prevalence of miscarriage among non-participants | Observed unadjusted odds ratio of 25.9b among participants (N=121) | Assumed odds ratio among non-participants (N=15)c | Summary odds ratio (N=136)d |
|---|---|---|---|
| 30% | 25.9 | 1.0 | 6.3 |
| 25% | 25.9 | 1.0 | 5.5 |
| 20% | 25.9 | 1.0 | 3.7 |
| 15% | 25.9 | 1.0 | 4.4 |
| 30% | 25.9 | 0.5 | 4.4 |
| 25% | 25.9 | 0.5 | 4.0 |
| 20% | 25.9 | 0.5 | 4.2 |
| 15% | 25.9 | 0.5 | 3.0 |
Out of 149 pregnancies with > 2.5 Gy uterine radiation (see Table 2), 8 had history of > 5 pack-years and 113 had no history of smoking (N=121).
Out of the 8 with history of > 5 pack-years 7 had miscarriages and out of the 113 with no history of smoking 24 had miscarriages. Thus, the observed unadjusted risk of miscarriage among those with history of smoking > 5 pack-years was 25.9.
We assumed a non-participation rate of 11% (15 of 136) based on the observed 558 (11%) non-participants out of 5,050 eligible participants
If exposure-outcome distribution of non-participants were combined with exposure-outcome distribution of participants