Maximising data value and avoiding data waste: a validation study in stroke research | JoVE Visualize

Area of Science:

Health Informatics
Clinical Epidemiology
Public Health Data Linkage

Background:

The Australian Stroke Clinical Registry (AuSCR) collects vital data on stroke patients.
Linking AuSCR data with other health datasets can provide a more comprehensive understanding of stroke outcomes.
Current data linkage feasibility and accuracy between AuSCR and national/state databases are not well-established.

Purpose of the Study:

To assess the feasibility of linking Australian Stroke Clinical Registry (AuSCR) data with the National Death Index (NDI) and state hospital admission/emergency presentation databases.
To evaluate the completeness and concordance of common variables across these linked datasets.
To determine the accuracy of AuSCR in-hospital death data in predicting National Death Index (NDI) registrations.

Main Methods:

A cohort design was employed, linking records of patients treated for stroke or transient ischaemic attack in Australian hospitals (2009-2013).
Probabilistic and deterministic data linkage methods were utilized to merge records from AuSCR, NDI, and state-managed health databases.
Descriptive statistics were used to analyze data matching success, variable concordance, and the predictive accuracy of AuSCR death data against NDI records.

Main Results:

Successful data linkage was achieved for 95% of patients with hospital admissions data and 80% with emergency department presentations data.
Concordance for key variables like sex, age, in-hospital death, and Indigenous status between AuSCR and hospital admissions data exceeded 99% (κ=0.99) and 83% (κ=0.83), respectively.
AuSCR in-hospital death data demonstrated high sensitivity (98.7%) and specificity (99.6%) in predicting NDI-registered in-hospital deaths.

Conclusions:

This study demonstrates the successful linkage of a national clinical registry with routine government health datasets for stroke patients in Australia.
Data linkage significantly enriches clinical registry data, providing a more holistic view of patient outcomes beyond acute care.
Comprehensive assessment of stroke patient mortality and hospital contacts is enabled through this integrated data approach, improving clinical outcome evaluation.