Electronic linkage and interrogation of administrative health, social care, and criminal justice datasets: feasibility concerning process and content.
Cassie HigginsKeith MatthewsPublished in: Informatics for health & social care (2020)
The objective was to test the feasibility of a novel model of electronic linkage and interrogation of large, sensitive, administrative datasets derived from health care, social care, and criminal justice. Participants comprised all individuals having completed suicide or drug-related death in Tayside between 2009 and 2014. Data were hosted, linked, and pseudo-anonymized by a Trusted Third Party and were interrogated via secure access to the HIC Scottish Government-certified Safe Haven. Several barriers were encountered concerning data access, with all but one issue (obtaining criminal justice data) ultimately soluble. However, each barrier led to a substantial delay in either obtaining the required approvals or in receiving the specified data extracts. Generally, data coverage was good but data quality was poor, with almost a fifth of the data fields (17%) being less than 10% complete. The feasibility of this novel approach was demonstrated. Critically, this was achieved because of the central involvement of a Trusted Third Party and the use of a Government-certified Safe Haven. Future studies using a similar model of data acquisition and analysis should consider the potential delays resulting from organizations' lack of familiarity with their data-sharing protocols and procedures.
Keyphrases
- healthcare
- electronic health record
- big data
- mental health
- emergency department
- palliative care
- gene expression
- public health
- machine learning
- dna methylation
- climate change
- mental illness
- risk assessment
- artificial intelligence
- human immunodeficiency virus
- genome wide
- rna seq
- deep learning
- chronic pain
- high density
- current status
- human health