Linking data from a large clinical trial with the Australian Cerebral Palsy Register.

<bold>Aim: </bold>To link data from a large maternal perinatal trial with the Australian Cerebral Palsy Register (ACPR) to identify children with cerebral palsy (CP).<bold>Method: </bold>Deidentified data from the Australasian Collaborative Trial of Magnesium Sulphate (ACTOMgSO4 ) and the ACPR were linked. Children born from 1996 to 2000 at Australian hospitals who survived and had 2-year paediatric assessments were included. Children identified with CP in: (1) both the ACTOMgSO4 (2y) and the ACPR (5y), (2) the ACTOMgSO4 only, and (3) the ACPR only were compared.<bold>Results: </bold>We included 913 children (492 males, 421 females; mean gestational age at birth 27.8wks [standard deviation 2.1wks]; range 23.0-40.0wks). Eighty-four children received a CP diagnosis: 35 by the ACTOMgSO4 and the ACPR, 29 by the ACTOMgSO4 only, and 20 by the ACPR only. The ACTOMgSO4 diagnosed 76.2% (95% confidence interval [CI] 65.9-84.1) and the ACPR identified 65.5% (95% CI 54.7-74.9). Children born in states/territories with long-standing versus more recently established registers were more likely to be included on the ACPR (p<0.05).<bold>Interpretation: </bold>Linking deidentified perinatal trial data with the ACPR was achieved. Limitations of both strategies for identifying children with CP in this era (late 1990s and early 2000s) probably explain many of the differences observed, and inform future linkage studies and evaluations of CP-preventive interventions.<bold>What This Paper Adds: </bold>Randomized trial data were linked with the Australian Cerebral Palsy Register. Trial (2y) and register (up to 5y) diagnoses of cerebral palsy (CP) differed. States with long-standing registers were more likely to include children with CP. [ABSTRACT FROM AUTHOR]