Errors in the classification of pregnant women according to Robson ten-group classification system

Objectives: The Robson Ten-Group Classification System (TGCS) is widely used as a classification system for perinatal analyses such as Caesarean section (CS) rates. In Germany, standardised data sets on deliveries are classified by quality assurance institutions using the TGCS. This observational study aims to evaluate potential errors in the TCGS classification of deliveries. Study design: Manual TGCS classification of all 1370 deliveries in an obstetric unit in 2018 and comparison with semi-automatic TGCS classifications of the quality assurance institution. Results: In the manual classification, 259 out of 1370 births (18.9 %) were assigned to a different Robson group than in the semi-automatic classification. The proportions of births by Robson group were significantly different in TGCS group 1 (32.2 % vs. 37.6 %, p = 0.0034) and group 2 (18.4 % vs. 14.4 %, p = 0.0053). Concordance between manual and semi-automatic classifications ranged from 59.5 % in group 2 to 100.0 % in groups 6, 7, 8, and 9. The most frequent mismatches were for the parameters "onset of labour" in 184 cases (13.4 %), "parity" in 42 cases (3.1 %) and "previous uterine scars" in 23 cases (1.7 %). In the manual classification, there were significant differences in the CS rate in group 1 (7.9 % vs. 2.5 %, p < 0.0001), group 2 (30.2 % vs. 48.2 %, p < 0.0001), and group 4 (14.1 % vs. 37.4 %, p = 0.0004), compared to the semi-automatic classification. Conclusions: Due to incorrect data entry and unclear definitions of criteria, quality assurance data in obstetric databases may contain a relevant proportion of errors, which could influence statistics with TGCS in context of CS rates in international comparisons.