Perinatal Services BC, Vancouver, British Columbia, CanadaDepartment of Obstetrics & Gynaecology, University of British Columbia, Vancouver, British Columbia CanadaPerinatal Services BC, Vancouver, British Columbia, Canada
Checking for direct PDF access through Ovid
In Reply:We thank Drs. Mayo, Schacher, Stevenson, and Shaw; Drs. Roberts, Algert, and Nippita; and Drs. Ahrens, Thoma, and Rossen for their recent letters regarding our article examining interpregnancy interval and adverse pregnancy outcomes.1 We have conducted a number of supplementary analyses in response to their comments and suggestions to continue the scientific discourse on this methodologically challenging topic.Mayo et al provide useful new subgroup data in response to Dr. Klebanoff’s call for more data among women with a previous preterm birth and a previous cesarean delivery, and they report an elevated risk of preterm birth among women with a short interpregnancy interval among those with a first birth preterm or term and among those delivering their first child vaginally or by cesarean delivery. In our cohort, unconditional regression (across-mother comparisons) produced an adjusted odds ratio (OR) of 0.95 (95% CI 0.79-1.15) for a preterm birth after a first birth preterm among women with a short interpregnancy interval (less than 6 months as compared with 18-23 months) and an adjusted OR of 1.51 (95% CI 1.33-1.77) after a first birth at term. When using conditional logistic regression (within-mother comparisons), we found an adjusted OR of 1.68 (95% CI 1.12-2.52) after a first birth preterm and an adjusted OR of 0.69 (95% CI 0.56-0.86) after a first birth at term for women with short interpregnancy intervals (less than 6 months). The reasons that our subgroup analyses, which suggest a difference in risk based on whether a woman’s first birth is preterm, differ from those of Mayo et al are unclear and warrant further investigation.Our unconditional logistic regression model examining whether preterm risk is altered according to mode of delivery for the first birth yielded similar conclusions to Mayo et al. We found an adjusted OR of 1.33 (95% CI 1.08-1.63) for women with a short interpregnancy interval after a cesarean delivery in their first birth and an adjusted OR of 1.38 (95% CI 1.25-1.51) among those with a vaginal first birth. In contrast, conditional logistic regression reports an adjusted OR of 0.63 (95% CI 0.43-0.92) for preterm birth after a first cesarean birth and a short interpregnancy interval, whereas the adjusted OR for women with a vaginal first birth is 0.96 (0.78-1.19), suggesting no increased risk for preterm birth based on route of delivery in the first birth.As suggested by Drs. Ahrens, Thoma, and Rosen, we examined the generalizability of the women included in the case-crossover design employed in our study. The results of unconditional logistic regression models based on only the sample of women included in the conditional logistic regression analyses (ie, women with discordant exposure and outcome status) are shown in Table 1. The results do change from those reported among the larger population of women with two or more births, and many relationships are no longer statistically significant; however, the general trend of an elevated risk among those with short interpregnancy intervals remains, but there is no significantly elevated risk for women with long interpregnancy intervals. These results suggest that generalizability is an important consideration in case-crossover designs such as the one employed in our study. The results further suggest that researchers should analyze exposures and outcomes as continuous variables where possible (rather than categorize them) to allow estimates to be based on the full sibling cohort.In response to Drs.