A computer-based method for cause of death classification in stillbirths and neonatal deaths.

BACKGROUND

In large-scale epidemiological studies of stillbirths and neonatal deaths a method is needed to replace detailed medical record audits in order to determine the cause of death.

METHODS

A computer-based method is presented for determination of the cause of death in stillbirths and in neonatal deaths. It utilizes information in the Swedish medical registries. The study comprises 6044 dead infants born in Sweden from 1983-1990. For each infant the program determines 31 basic characteristics which are important in deciding the cause of death. Based on these characteristics a modified Wigglesworth's classification is used to find the cause of death. The validity of the method was checked by comparing the computer generated information with information obtained by scrutinizing medical records for a 10% representative sample (603 infants).

RESULTS

Specificity and sensitivity for each basic characteristic varied, but for the modified Wigglesworth cause of death classification the concordance was 88%. The weakest data refer to intrauterine deaths, where pertinent information was often missing in the medical registries.

CONCLUSION

The method can be used for large-scale epidemiological studies.