Methods

This was a retrospective, cohort study design. Women with term, singleton pregnancies with known umbilical artery pH results (n = 40) who delivered between Jan. 1, 2008, to Dec. 31, 2009, had their fetal heart tracings reviewed. Eligible patients were term (≥37 weeks’ gestation) and undergoing a trial of labor with the intention to deliver vaginally. Patients were also excluded if there were not at least 2 hours of tracing available, or if they were induced for an intrauterine fetal demise. Reviewers were blinded to the patients’ clinical circumstances. The tracings were deidentified, placed in random order, and provided to the reviewers in bulk.

Forty tracings were selected from 3 different groups of women: (1) pH >7.1, (2) pH from 7.0 to 7.1, (3) pH <7.0 and base excess <−12 mEq. To have a wide spectrum of FHR to evaluate, we chose those pH categories for eligibility. To identify potentially abnormal FHR tracing segments and particularly those that may fit into category 3, we chose a pH <7.0. This pH has association with increased adverse outcome, and is also the designation for “pathologic” acidemia. When randomly choosing a limited number of tracings, it would be very unlikely that we would identify any category 3 tracings without prespecifying an abnormal group (in our case by pH). We would not be able to accomplish our objective of identifying concordance at the extremes of the systems. We chose a pH 7.0-7.10 for an intermediary category because 7.10 represent the 5th percentile. Finally, we chose >7.10 to evaluate more normal FHR tracings because this pH is considered “normal.” Thus, cord pH was used to define the 3 study groups, but it played no further role in our analyses. The sample size, though arbitrary, was based on practicality and evaluation of other studies in the literature. The patients were identified sequentially until we identified 40 patients who met inclusion criteria. There were 3 fetal heart tracings per patient, for a total of 120 tracings. Each reviewer analyzed all 120 tracings. FHR tracings were divided into 15-20 minute epochs and random samples were selected. One segment was chosen from the 60 minutes immediately preceding birth, whereas the other 2 were randomly chosen from the last 180 minutes before birth. Because the segments were distributed to the reviewers randomly, there was no way to know which 3 segments belonged to one individual, or the order of the segments. No segment overlapped.

Three board-certified MFM specialists, after having undergone a structured education and training program in each FHR classification system, independently reviewed 120 unique FHR tracings. These MFM specialists all participated in the NICHD/SMFM workshop on establishing new criteria for fetal heart tracing interpretation. Before the review, an interactive training session was performed using sample FHR tracings (training cases). Laminated cards with definitions of NICHD categories were provided during FHR tracing review.

Reviewers were asked to classify the FHR tracing by 2-tier system (“reassuring” vs “nonreassuring”), NICHD 3-tier (category 1, 2, 3), and the Parer 5-tier system (green, blue, yellow, orange, red). In addition, all tracings were evaluated for the presence of accelerations, decelerations (early, variable, or late), and variability (absent, minimal, moderate, or marked). The baseline fetal heart rate was noted. Tracings deemed uninterpretable were excluded.