From single-case analysis of neonatal deaths toward a further reduction of the neonatal mortality rate.

Background The infant mortality rate (IMR), a key indicator of the quality of a healthcare system, has remained at approximately 3.5‰ for the past 10 years in Germany. Generic quality indicators (QIs), as used in Germany since 2010, greatly help in ensuring such a good value but do not seem to be able to further reduce the IMR. The neonatal mortality rate (NMR) contributes to 65-70% of the IMR. We therefore propose single-case analysis of neonatal deaths as an additional method and show an efficient way to implement this approach. Methods We used the Nordic-Baltic classification (NBC) to detect avoidable neonatal deaths. We applied this classification to a sample of 1968 neonatal death records, which represent over 90% of all neonatal deaths in East Berlin from 1973 to 1989. All cases were analyzed as to their preventability based on the complete perinatal and clinical data by a special commission of different experts. The NBC was automatically applied through natural language processing and an ontology-based terminology server. Results The NBC was used to select the group of cases that had a high potential of avoidance. The selected group represented 6.0% of all cases, and 60.4% of the cases within that group were judged avoidable or conditionally avoidable. The automatic detection of malformations showed an F1 score of 0.94. Conclusion The results show that our method can be applied automatically and is a powerful and highly specific tool for selecting potentially avoidable neonatal deaths and thus for supporting efficient single-case analysis.