[Mortality of extremely low birthweight infants - large differences between quality assurance data and the national...

To evaluate neonatal outcome, the German neonatal quality assurance dataset is often used. However, a systematic bias may occur, as not all live births are registered in this dataset. The aim of our study was to evaluate the magnitude of this systematic error by comparing this dataset to the national birth/death registry.

The summary statistics of live births and deaths with a birthweight <1 500 g from the quality assurance datasets 2007-2009 of 5 German States (Baden-Württemberg, Bavaria, Hesse, Lower-Saxony and North Rhine-Westfalia) were compared with the dataset from the national birth/death registry.

Our analysis included 68% of the German birth cohort 2007-2009. The quality assurance dataset included 1 151 of 1 716 neonatal deaths (67.1%) in infants with a birthweight <1,000 g registered within the national birth/death registry; 565 deaths were missing. A total of 277 of 303 deaths (94.1%) with a birthweight 1,000-1,499 g were registered; 26 deaths were missing. In the state of Baden-Württemberg up to 11% more infants were registered in the quality assurance dataset than in the national registry, but an average of 36 neonatal deaths/year with a birthweight <1,000 g were missing (63.4% registration rate).

This analysis shows that the quality assurance data miss more than 1/3 of deaths in extremely low birthweight infants. Transfers between hospitals may result in multiple data entries and additional bias. Comparing outcome statistics based on the neonatal quality assurance dataset may lead to a substantial systematic error. Linkage to national birth and death certificates and/or to the perinatal dataset is urgently needed.