Strengths and limitations:
The principal strength of our study was that OSA was diagnosed objectively by inpatient polysomnography, the reference examination, so that the diagnosis should not be either under- or overestimated. It nonetheless has some limitations. As our population was limited to obese women with BMI >40 kg/m2, we could not calculate the prevalence and extent of OSA in a population of non-obese woman. Even though our population specifically included especially obese women, comparison of the two groups found a significant difference in their BMI: it was higher in the group with OSA (43.84 ± 6.24 kg/m2 vs 41.18 ± 5.96 kg/m2 for the group without, P =0.045). We did not adjust our results for BMI, which is an important confounding factor in studies of OSA and pregnancy.26 A high BMI by itself causes vascular complications, gestational diabetes, and both cesarean and operative vaginal deliveries.6,7 We also found in our population the usual comorbidities expected to be associated with morbid obesity (34.3% gestational diabetes, 23.9% vascular complications, 43.3% cesareans). Despite this confounding factor, we did not find a significant difference between the groups for other obstetrical data, such as pregnancy-related vascular diseases or outcomes of either the pregnancy or the infant. The only significant difference shown was a higher frequency of gestational diabetes in the OSA group.
Another limitation was associated with OAS severity. Most cases were mild or moderate, with only four severe enough to merit CPAP. Of the four women with severe OSA, only three agreed to start CPAP treatment and only one continued it to delivery. We therefore could not study its benefits for these outcomes.
Finally, our observation of trends that did not reach significance for some of the criteria studied (the composite criterion, cesarean deliveries, and labor induction) suggests a lack of power due to an insufficient number of women included.