An attempt to improve recognition of fetal acidemia by a remodeled intrapartum cardiotocography classification: A case–control study

INTRODUCTION: Current international (FIGO) and Swedish cardiotocography (CTG) interpretation templates exhibit limited diagnostic accuracy for the detection of fetal acidemia. The objective of this study was to develop a novel intrapartum CTG classification template incorporating CTG parameters demonstrated to be strongly associated with acidemia, and to evaluate the diagnostic performance of the models by assessing their sensitivity and specificity in identifying CTG recordings from neonates with acidemia. MATERIAL AND METHODS: A retrospective case-control study based on 1092 births in Region Skåne. The study material included 364 cases with umbilical cord blood pH <7.05, or <7.10 after 1st stage cesarean, and 728 controls with pH ≥7.15 and Apgar scores ≥9 at 5 and 10 min. CTG patterns were systematically evaluated across a set of predefined variables, and associations between specific patterns and fetal acidemia were quantified using ORs. Two models defining pathological patterns were created: Model 1 based on the presence of one of several listed criteria resulting in a positive test and Model 2 also including the presence of ≥2 B-criteria resulting in a positive test. The main outcomes, sensitivity, and specificity to identify fetal acidemia at birth were calculated and compared with the current FIGO, NICE, and SWE guidelines. RESULTS: The sensitivity to identify fetal acidemia was 86.3%, 86.5%, 50.8%, 57.1%, and 87.4% for Model 1, Model 2, FIGO, SWE, and NICE, respectively. The corresponding specificity was 77.9%, 76.4%, 92.6%, 91.2%, and 70.6%. The sensitivity to identify severe acidemia (pH <7.0) was 90.4% for Model 1 and for NICE; 64.4% for SWE, and 56.7% for FIGO. A variable not included in current classification systems, "fetal heart rate below baseline for >50% of the time during 30 minutes" was the criterion contributing most to the sensitivity. CONCLUSIONS: Our models show higher sensitivity but lower specificity for the defined outcome compared with SWE and FIGO, and similar sensitivity but higher specificity compared with NICE. Validating the model on a new material is essential before considering clinical use.