Variations in the level of recording of common health conditions and risk factors in hospital morbidity data

Abstract: 

Administrative hospital data is commonly used in risk adjustment but the degree to which it is complete and accurate is unclear. This talk will outline the findings from a study using data linkage of a large scale survey and administrative data, examining the agreement between self-reported and coded diagnoses; quantifying the amount of between-hospital variation in this agreement and identifying patient- and hospital-level characteristics that predict higher or lower levels of agreement.

Bio: Ms Sanja Lujic is a Lecturer in Biostatistics at the newly established Centre for Big Data Research, Faculty of Medicine, UNSW. Prior to this, she worked as a Biostatistician at the University of Western Sydney and as a Biostatistical Officer Trainee at the NSW Ministry of Health. Sanja’s research interests include health services research, population health ageing, multimorbidity, data linkage methods and use of routinely linked administrative datasets in statistical modelling. Her interest in health data linkage begun in 2006, having been involved in numerous projects involving complex large-scale linked datasets spanning health, community care and cohort studies. Sanja completed her Masters of Statistics at UNSW in 2005, and Masters of Biostatistics from USyd in 2008. The seminar will be followed by drinks and finger food in the staff room RC-3082. All attendees are welcome!