Abstract
The "at birth" system which is used in Sheffield to identify children likely to die unexpectedly in infancy, was tested retrospectively in Birmingham (83 cases) and in Newcastle upon Tyne (56 cases). The discrimination between cases and age-matched controls was poor in both cities. Analysis of the 8 factors used in the system showed that only 2 maintained significant case/control differences in Birmingham and Newcastle. Further investigation showed that other factors from maternity records showed significant case/control differences in these cities. Although the system used in Sheffield would not be of use in a prospective prevention programme in either Newcastle or Birmingham, the possibility of evolving an "at risk" system which might apply more widely is discussed.