Analysis of the risk factors associated with the severe
complications in these patients
We next analyzed the risk factors associated with these severe outcomes
using binary non-conditional logistic regression analysis with single
factor. Eight factors potentially associated with the consequence of
hysterectomy (p<0.05), including placenta implantation
detected by ultrasound and/or MRI, placenta previa type, application of
vascular occlusion and type of vascular blocking, postpartum hemorrhage
(≥1000ml), bleeding volume during the delivery, and confirmed placenta
implantation by the surgery. These factors were further analyzed by the
non-conditional binary logistic regression with multi-factors. As shown
in Table S1, 5.319 times higher potential of hysterectomy was found in
the patients with uterine artery embolization than those with abdominal
aortic balloon occlusion. This probability was almost doubled (1.002
times) with every 1ml increase of bleeding during the delivery, relevant
to that before bleeding. It suggested that the mode of vascular
embolization and the volume of hemorrhage during delivery were the main
risk factors associated with hysterectomy.
Similarly, 7 factors were identified by single factor analysis, which
were associated with severe bleeding during the delivery, as shown in
Table 1. Evaluation of these risks by multifactor regression analysis
showed that the pregnancies with ≥2 cesarean deliveries had higher
possibility (3.562 times) of severe bleeding than those with only once.
Compared to women without placenta implantation, patients diagnosed or
suspected placenta implantation by ultrasound had higher potential
(1.631 times, 95% CI:1.000-2.658 or 1.794 times, 95% CI:1.110-2.899)
of severe bleeding. Particularly, this probability increased 6.839 and
1.964 times in the pregnancy with confirmed or suspected placenta
implantation by surgery than those without implantation (95%
CI:4.508-10.377 or 1.259-3.064), respectively. Complete placenta previa
also resulted in 1.814 times severe bleeding in patients than marginal
one (95% CI:1.219-2.698). These results suggested that cesarean
numbers, ultrasound-detected or surgery-confirmed placenta implantation,
and type of placenta previa were associated tightly with massive
bleeding during the delivery.
Our analysis also demonstrated that under ultrasound every 0.1cm
increase of the thickness of uterine scar, promoted the likelihood of
placenta implantation to 1.559 times (95% CI:1.205-2.017). Suspected
placenta implantation under ultrasound than normal placenta had 7.79
times high likelihood of implantation confirmed later by surgery (95%
CI:1.526-39.765). The subjects without MRI examination also had 3.565
times higher probability of placenta implantation than those with
non-implantation diagnosed by MRI (95% CI:1.622-7.837). A higher
potential of placenta implantation (3.704 times) was also observed in
the women with complete placenta previa than marginal one (95%
CI:1.699-8.076). These evidences indicated that ultrasound-detected
thickness of uterus scar, ultrasound and/or MRI-detected or suspected
implantation of placenta, and complete placenta previa, were closely
associated with placenta implantation.