Search

ObjectivesElectronic Health Record (EHR) data have created unique opportunities for research. However, these data are: not curated, siloed and poorly integrated. We describe linkage of EHR data from an entire health service with government datasets to establish a linked geographic cohort within the Australian National Centre for Healthy Ageing (NCHA). ApproachResearch suitable EHR items were identified from Peninsula Health (NCHA partner) data systems based on: published research, availability and quality. Items underwent end-user Delphi processes to identify core research items (consensus=70%). Approvals were obtained from the Australian Institute of Health and Welfare (AIHW) for linkage with: Medicare, medication dispensings, Aged Care and death registry data through the AIHW spine, created using identifiers from the Medicare Consumer Directory (MCD); and from the Centre for Victorian Data Linkage for linkage to state-wide hospital data. Identifiers for local residents aged ≥60 years who attended Peninsula Health were submitted for probabilistic data linkage. ResultsDelphi participants included 10 researchers from 8 fields/departments and 13 clinicians from 11 clinical areas. To date 7 of the 11 datasets have been reviewed. N=107 potentially suitable data items were identified and 96 gained consensus for inclusion in the core dataset. Of the 49,767 Health Service users (episodes: Jan 2010-Dec May 2021) submitted for linkage, 98.4% were successfully linked to the MCD (Median age 72.2 years, 52.2% female, 1.8% regional residence). An additional 172,290 individuals living within the geographic region but not contained within the EHR dataset were identified in the MCD for linkage to the government datasets. Linkage accuracy was impacted by inaccurate/incomplete address fields (~30%) and lack of adherence to naming conventions within the EHR data. ConclusionLinking with EHR data is complex. Having an established EHR research dataset will improve the feasibility of data linkage and potential for future expansion of linkages within the NCHA. Once merged, the data will be used to underpin a range of research activities related to ageing and dementia.

ObjectivesTo develop a flexible platform for creating, reviewing and adjudicating annotation of unstructured text. Natural Language Processing models and statistical classifiers use the results for analysis of large databases of text, such as electronic health records, that are curated by the National Centre for Healthy Ageing (NCHA) Data Platform. ApproachAutomated approaches are essential for large scale extraction of structured data from unstructured documents. We applied the CogStack suite to annotate clinical text from hospital inpatient records based on the Unified Medical Language System (UMLS) for classifying dementia status. We trained a logistic regression classifier to determine dementia/non-dementia status within two cohorts based on frequency of occurrence of a set of terms provided by experts - one with confirmed dementia based on clinical assessment and the other confirmed non-dementia based on telephone cognitive interview. We used our annotation platform to review the accuracy of concepts assigned by CogStack. ResultsThere were 368 people with clinically confirmed dementia and 218 screen-negative for dementia. Of these, 259 with dementia and 195 without dementia had documents in the inpatient electronic health record system, 84045 inpatient documents 16950 for the dementia and non-dementia cohort respectively. A set of key words pertaining to dementia was generated by a specialist neurologist and a health information manager, and matched to UMLS concepts. The NCHA data platform holds a copy of the inpatient text records (>13million documents) that has been annotated using CogStack. Annotated documents corresponding to the study cohort were extracted. We tested true positive rates of annotation against 50 concepts judged by a neurologist and health information manager to be relevant to dementia patients by manually review of 100 documents. ConclusionAutomated annotations must be validated. The platform we have developed allows efficient review and correction of annotations to allow models to be trained further or provide confidence that accuracy is sufficient for subsequent analysis. Implementation within our linked NCHA data platform will allow incorporation of text based data at scale.