[Reproducibility of data in a study of the risk factors for insulin-dependent diabetes mellitus]

[Consistency between data sources and inter-observer reliability in the Study on Neonatal and Perinatal Morbidity and Mortality and Care in the City of Rio de Janeiro]

The objective of this study was to evaluate the quality of data in the research project entitled "Study on Neonatal and Perinatal Morbidity and Mortality and Care in the City of Rio de Janeiro", analyzing the completeness of patient records, inter-observer reliability, and concordance of collected data. The study interviewed a sample of 10,072 post-partum women, corresponding to 10.0% of the deliveries in the City of Rio de Janeiro. This article analyzed the concordance between data on patient records and the reproducibility of questionnaires by mean of rest/ retest, using common and prevalence-adjusted Kappa as well as the intra-class correlation coefficient. Losses totaled 4.5%, and the proportion of unknown data on patient records varied from 3.0 to 90.0%. Lower proportions were concentrated in the neonatal assessment and higher ones in the data on maternal hospital admission. There was a high concordance between the data reported by the mother and that written on the patient record, with the Kappa varying from 0.77 to 0.96. In the test/retest verification, Kappa varied from 0.61 to 0.94. This study demonstrated a high inter-observer reliability as well as reliability between the different data sources (interviewee and patient record).