Computer interpretation of pediatric Frank vectorcardiograms in the evaluation of congenital heart disease.

The evaluation of a new computer program for analysis and interpretation of pediatric Frank vectorcardiograms is reported. The program includes extensive age- and sex-dependent criteria based on tables of limits for numerous vectorcardiographic parameters. In 728 catheterized patients, the diagnostic performance for type A statements was tested against independent and objective evidence obtained from hemodynamic and angiographic data. The overall diagnostic accuracy ranged from 75 to 89% without difference between children less than 2 years of age and those greater than or equal to 2 years of age. Sensitivities and specificities of the various diagnoses did not differ much between the 2 age groups. In the younger children, the accuracy of a positive diagnosis of left ventricular hypertrophy, right ventricular hypertrophy, and biventricular hypertrophy was 20, 15, and 32% higher, respectively, than in the older children. The accuracy of the diagnosis "normal" was 28% lower in the younger children. These differences were explained by the higher proportion of pathologic findings in the younger children: 93% versus 74% in the older children. Given the strict methods of the evaluation, the diagnostic accuracy of this pediatric program was considered clinically satisfactory. Program performance appears to be dependent not on patient age but on prevalence of abnormalities in the population analyzed. Further improvement can be expected by making the criteria more adaptable to the composition of the population.