638Reporting of gestational diabetes and chronic conditions of pregnancy: validation of routinely collected hospital data
Abstract Background Gestational diabetes, hypertension, thyroid conditions and morbid obesity in pregnancy are associated with increased risks of adverse outcomes. Hospital data are important for research on these conditions, however, up-to-date validation of reporting is needed to understand the extent to which the data reflect the clinical situation. Methods Women giving birth to singleton infants in two tertiary hospitals in New South Wales, Australia, between 2011 and 2015 were included. Obstetric data, from the ObstetriX system, was used as the gold standard to which linked hospital data, from the Electronic Medical Record, were compared. Results There were 35,928 births included. Gestational and pre-existing diabetes had high sensitivity (83.6% and 88.2%) and positive predictive values (PPV, 92.7% and 86.0%). Pre-eclampsia and eclampsia, gestational hypertension and any hypertension had good sensitivity (80.0%, 80.1%, 81.5%), but moderate PPVs (59.7%, 65.6%, 70.4%), while for chronic hypertension sensitivity (53.5%) and PPV (53.2%) were lower. Obesity and thyroid conditions showed low sensitivity (9.8%, 12.9%; PPV 65.6%, 82.3%). Specificity and NPV were high for all conditions. Conclusions We found reliable reporting of gestational diabetes, pre-existing diabetes and all types hypertension, except for chronic hypertension which was moderately well reported. Thyroid conditions and morbid obesity were very poorly reported. Key messages Diabetes appears well reported in the hospital data, and sensitivity for hypertension may be improved by using a grouped category. Hospital data on thyroid conditions and obesity should be used with caution.