Wednesday 11th March 2015

Gareth Hagger-Johnson - Department of Epidemiology & Public Health, University College London [Edited Presentation]

Data linkage errors in hospital administrative data

BACKGROUND. Our aim was to determine the rate of data linkage error in Hospital Episode Statistics (HES) by testing the HESID pseudoanonymisation algorithm against a reference standard, in a national registry of paediatric intensive care records.

METHODS. Data were drawn from the Paediatric Intensive Care Audit Network (PICANet) database covering 33 paediatric intensive care units in England, Scotland and Wales (01/01/2004 to 21/02/2014). The study population were infants and young people (age 0 to 19 years at first admission). After applying the HESID pseudonymisation algorithm, we classified records into true matches, false matches, missed matches or true non-matches.

RESULTS. Of 166,406 admissions, there were 75,476 matched records (readmission). The HESID pseudonymisation algorithm produced a small proportion of false matches (n = 141/85,173; 0.2%) but a larger proportion of missed matches (n = 5898/81,233; 7.3%). The true readmission rate was underestimated by 7.0%. Patients from ethnic minority groups, areas of high socio-economic deprivation and those with missing data were more likely to experience data linkage errors.

CONCLUSIONS. The deterministic algorithm used to link all episodes of hospital care for the same patient in England has a high missed match rate which underestimates the true readmission rate. This will lead to biased estimates for researchers who use HES data. To reduce linkage error, errors in patient identifiers need to be addressed at source by hospitals. A switch to a probabilistic matching algorithm should be considered given the high missed match rate produced by the current deterministic HESID algorithm. The role of the Personal Demographics Service (PDS) as a possible reference standard will be discussed.