We read with much interest the study by Kunzier et al, which concluded that measurement of cervical length by sonography, for women at term, could assist in decision making, delineating between false and true labor.
As clinicians, we share the need to find factors to better predict the progression from latent phase to active labor. However, such factors should have an additive benefit to the common practice based on painful uterine contractions and Bishop score. We are concerned that the inclusion criteria used by Kunzier and colleagues (cervical dilatation <4 cm, 80% effacement, and regular uterine contractions ≥4/20 min) are broad and nonspecific, forming a heterogeneous study population with diverse pretest probabilities for progressing to active labor. For example, both parturients, one with closed cervix, Bishop 2, and painless contractions and the other with dilated cervix, Bishop 7, with 8 painful contractions in 20 minutes, are eligible to participate in the study. Obviously these 2 parturients have a different pretest probability for progressing to active labor and we would expect these factors to be characterized and presented in Table 1, titled: Comparison of demographic and obstetric variables for true vs false labor patients. The different distribution of these specific clinical characteristics, painful frequent contraction and Bishop score, between the false and true labor groups could theoretically predict better the progress to active labor than measurement of cervical length by sonography. In such case, measurement of cervical length by sonography is pointless.
Furthermore, there are some basic methodological flaws in the study that could affect the study’s conclusion. First, Kunzier and colleagues enrolled 47 primiparous and 30 multiparous participants and were underpowered to conclude that the difference of the area under the receiver operating characteristic curves between primiparous and multiparous groups was nonsignificant. The sample size needed to detect a significant difference of 0.88 vs 0.76 requires 90 participants in each curve. Combining the 2 curves to one is inappropriate and can lead to misleading results. Second, the study enrolled 101 nonconsecutive cases in 2013 through 1016. According to Winthrop University Hospital World Wide Web site there are 5000 annual deliveries, ie, 15,000 deliveries during the study period. Although not all parturients who ultimately delivered were eligible for enrollment, a significant number were qualified and not enrolled. There is a need to compare between the latter and the study group to verify lack of selection bias issues. We are concerned that these methodologic limitations impact the validity of the study’s findings and conclusions.