(927) Predicting delivery for patients in threatened preterm labor by counting synchronized contractions

For patients in threatened preterm labor, current methods for diagnosing preterm birth (PTB) within 7 days are inadequate. Preterm contractions vary in strength, and predicting PTB based on expression of only strong contractions may improve sensitivity. New analytic methods, combined with recording local uterine wall electromyography (EMG) using directional sensors, can characterize each contraction as synchronized or unsynchronized and differentiate strong versus weak contractions. Additionally, the proportion of synchronized contractions may vary over short periods, so an adequate duration of evaluation is needed to reduce sampling errors. We hypothesize that a high sensitivity for predicting PTB within 7 days can be achieved by counting only synchronized contractions observed over 90 minutes.

Study Design:

Multicenter, prospective cohort study of pregnancies between 24-36 weeks GA, with cervix £ 5 cm and ³ 1 contraction in 10 min by self-report or toco. 6-channel recordings were obtained using PreTeL’s directional uterine EMG sensors for 1.5 to 4 hours. EMG signals observed in ≥ 5 channels were classified as synchronized; frequency was the number of synchronized contractions observed over 90 minutes. Outcome was delivery < 7 days of testing. A receiver operator characteristic (ROC) was calculated to evaluate performance. 

Results:

104 subjects were consented for study; 78 provided EMG data without being lost to follow-up or clinical interventions such as discharge before completion of data acquisition. The ROC revealed an AUC of 0.77. For a threshold set at 75% specificity (the approximate specificity of current methods), a sensitivity of 80% for predicting PTB within 7 days was observed.

Conclusion:

Counting synchronized contractions over 90 minutes correctly identifies 4 out of 5 patients in threatened preterm labor who will deliver within 7 days, without increasing the number incorrectly identified above current methods. This technology may improve antenatal corticosteroid administration and guide hospitalization recommendations for this population.