Inconsistencies in classification by experts of cardiotocograms and subsequent clinical decision

Inter-observer agreement in the interpretation according to the FIGO guidel ines of 33 cardiotocographic tracings by experts and subsequent clinical de cision was evaluated, using the kappa statistic (K) and the proportions of agreement (Pa). Overall agreement in the classification of tracings was fai r (K = 0.48) and was better for normal (Pa = 0.62), than for suspicious (Pa = 0.42) or pathologic tracings (Pa = 0.25). Overall agreement on clinical decision was slightly higher (K = 0.59), but mostly was centred on the deci sion to take 'no action' (Pa = 0.79). Experts especially disagreed over the decisions to 'monitor closely' (Pa = 0.14) or to 'intervene immediately' ( Pa = 0.38). These Limitations should be taken into account in clinical audi ts and in medical jurisprudence.