Research protocol - Evaluating data quality in the Netherlands Perinatal Registry (Perined): A data comparison study using hospital records from the IUGR Risk Selection (IRIS) study

Abstract

Background The quality of registry based studies depends largely on the data accuracy of the registries. The Dutch Perinatal Registry (Perined) is a nationwide database comprising perinatal data digitally provided by different healthcare providers. Perined data are used for comparing outcomes across regions and healthcare institutions as well as for quality analyses and research purposes. However, little is known about the data quality of the Perined database. Therefore, the aim of this study is to assess the quality of Perined data compared to hospital records and case report forms (CRFs) that were part of the IUGR Risk Selection (IRIS) study. Methods In this comparison study data from Perined and the IRIS Study will be used. The IRIS study was a large cluster-randomized trial investigating the effectiveness of routine third trimester ultrasonography in reducing severe adverse perinatal outcomes among Dutch low-risk pregnant women. A subsample of the IRIS study of neonates being at risk of severe adverse perinatal outcomes and their mothers will be used. Baseline demographic data were collected by midwives from participating women at inclusion (around 22 weeks’ gestation) using CRFs, and in-depth neonatal and maternal clinical data were retrieved from hospital records by trained research assistants. These latter IRIS study data were linked and compared to Perined data. Completeness of Perined data will be calculated for every variable. The reliability will be assessed as the percent of agreement between Perined and hospital record data or the CRF-based data. Additionally, intra-class correlation coefficients will be calculated for continuous variables, and Kappa and ‘Prevalence-and-Bias-Adjusted Kappa’ will be calculated for categorical variables. Discussion The results of the proposed comparison study will provide users of Perined data insight in its data quality. This will serve as an example of the accuracy of registry based data used in maternal and neonatal care research.