MONTEREY, CALIF. – A new model may help take the guesswork out of identifying placenta accreta before cesarean delivery, possibly sparing some women unnecessary interventions such as general anesthesia, placement of large-bore IV lines, and elective hysterectomy.
A study team led by Dr. Carolyn F. Weiniger, an obstetrical anesthesiologist at the Hadassah Hebrew University Medical Center in Jerusalem, reviewed data for 46,623 women who were delivered at the center over a 9-year period.
They identified 92 women with antenatally suspected placenta accreta, of whom only 52 (56%) were confirmed to have had the condition at the time of surgery, she reported at the annual meeting of the Society for Obstetric Anesthesia and Perinatology.
At a maximal-sensitivity cut-off, a predictive model based on just three clinical factors that would be assessed in any woman evaluated for suspected placenta accreta – the presence of placenta previa (yes vs. no), the number of previous cesarean deliveries, and ultrasound suspicion of accreta (high vs. low) – identified 100% of the women who were confirmed to have the condition. The trade-off was a low specificity of 25%.
However, at the optimal cut-off, the model still identified 94% of the women who were confirmed to have placenta accreta, and specificity was better (53%).
The investigators developed a nomogram based on the model that should be easy to apply clinically, according to Dr. Weiniger. "This nomogram could be a laminated card which a physician could sit with at his desk when faced with a suspected placenta accreta patient, to decide whether she is above or below our cut point, whether we think she actually does or does not have placenta accreta," she explained.
Although the study involves one of the largest series of patients with placenta accreta, it still had a fairly small sample size, she acknowledged. "In order to validate such data, a multicenter study would be appropriate."
"Have you changed your practice ... based on your nomogram?" asked Dr. Vilma E. Ortiz, session moderator and an anesthesiologist at the Massachusetts General Hospital in Boston.
"We have just created the nomogram, just finished analyzing the statistics. We haven’t yet had an accreta patient," Dr. Weiniger replied. "But I think as we become much more experienced at managing these patients, in patients who we feel comfortable [that they] have got no risk or low suspicion [of accreta], we will do regional anesthesia."
During the study period, her center’s management in cases of suspected accreta typically included general anesthesia, the booking of an ICU bed, the insertion of two large-bore IVs, and the placement of loops on the internal iliac vessels to aid with hemostasis. It did not include planned cesarean hysterectomy.
"Surgeons were not aware of our primary study outcome [that is, the rate of surgically diagnosed accreta among all cases of antenatally suspected accreta] in order to avoid bias," she noted.
A comparison of the 52 women with and 40 women without surgically confirmed accreta showed that they were statistically indistinguishable with respect to many clinical factors. But they differed significantly on the number of previous cesarean deliveries, the presence of placenta previa, and the presence of ultrasound signs of accreta.
The predictive model based on these three factors had an area under the receiver operating characteristic curve of 0.846, according to Dr. Weiniger. "When we tried to add maternal age and antenatal bleeding, because [these factors have been found to be predictive] in previous studies, it did not improve the area under the curve."
Giving some examples using the new nomogram, she noted that a woman having two previous cesarean deliveries who has placenta previa and a high-suspicion ultrasound would have a probability score approaching 80%, which falls well above cut points based on both 100% and 94% model sensitivity. "So she would be classified as having placenta accreta and would require massive hemorrhage preparation for surgery."
In contrast, a woman having two previous cesarean deliveries who does not have placenta previa and has a low-suspicion ultrasound would have a probability score of almost 0%, which falls well below the cut points, and she would be classified as not having placenta accreta. "So we could manage her with regional anesthesia and maybe not do preparations for massive hemorrhage," Dr. Weiniger commented.
Finally, a woman having two previous cesarean deliveries who has a low-suspicion ultrasound but does have placenta previa would have a probability score of about 20%. This score falls above the cut point corresponding to 100% sensitivity, "so if that was the clinician’s practice, that patient would have massive hemorrhage preparations," she said. "But if we were using the 94% sensitivity, the optimal cut point, she would be right on that, so it would be equivocal as to whether we would classify her as having massive transfusion requirements or not. Possibly, the state of her airway would define how we would manage such a patient."