Would gestational age and presence of brain anomalies affect interobserver reliability of fetal head biometry? Using...

The objective was to assess interobserver reliability of fetal head biometry using archived three-dimensional (3-D) volumes and the impact of gestational age and presence of brain anomalies on examiners' performance. Seventy nine 3-D volume datasets of fetal head were examined: 27 were normal and 52 had brain abnormalities. Off-line analysis was done by three fetal medicine experts (E1, E2 and E2), all were blinded to history and patient details. Measurements of the biparietal diameter (BPD), head circumference (HC), lateral ventricle (Vp) and transcerebellar diameter (TCD) were compared between examiners and to two-dimensional (2-D) measurements. Comparisons were made at two gestational age groups (≤22 and >22 weeks) and in presence and absence of brain anomalies. The intraclass coefficient showed a significantly high level of measurement agreement between 3-D examiners and 2-D, with values >0.9 throughout (p < 0.001). Bias was evident between 3-D examiners. E2 produced smaller measurements. The mean percentage difference between this examiner and the other two in BPD, HC, Vp and TCD measurements was significant, of 1.6%, 1%, 4.9% and 1.8%, respectively. E1 measured statistically larger for HC and TCD. E3 measured significantly larger for only BPD. The presence of anomalies was of no influence on the 3-D examiners' performance except for E3 who showed bias in BPD measurements only in cases with brain anomalies. Unlike other examiners, bias of E2 was only seen at gestational age group ≤22 weeks. Limits of agreement in measurements between observers were narrow for all parameters but were widest for the Vp measurements, being ±23% of the mean difference. Despite the above bias, the actual mean difference between examiners was small and unlikely to be of any clinical significance. Off-line measurement of fetal head biometry using 3-D volumes is reliable. In our study, presence of brain anomalies was unlikely to influence the reproducibility of measurements. Gestational age seemed to be of an impact on examiners' bias. Among experts this bias may be of no clinical significance.