Traditional/Clinical Mental Health Data:
While analytics techniques have scaled up to accommodate the volume, velocity and variety that came with the evolution of health data, traditional data remains important. Its evolution from paper-form to digitisation has been beneficial not only for storing data but for research purposes. In the UK, for example, the NHS has been able to centralise the data to some extent by incorporating similar IT across the country. This has been highly beneficial for practitioners and researchers, though there are security concerns here.
Traditional mental health data is quite extensive; however, the analysis will focus on surveys and Electronic Health Records (EHRs).
Research interviews and surveys are limited by the mental health practitioner’s inability to interview a large group to obtain enough data. However, the data thus collected can be the most accurate and insightful data (assuming the participants are truthful). If interviews and surveys are conducted by large organisations it is possible to gather enough data. For example, over 500,000 participants were involved in the investigation of genetic and non-genetic risk factors for mood disorders – an investigation by the UK Biobank. The findings, supported by cross-sectional analysis, is likely to prove useful as a framework for future genetic and non-genetic studies.
Alongside the UK Biobank is the World Health Organisation (WHO) Global Burden of Disease Programme. Having some of the largest samples in the mental health field, the programme comes closest to being a “big data” survey. However, the accuracy of the data remains to be questioned. These projects are often unable to capture variabilities in the status of an individual over time. As many with mental disorders can agree, recollection of episodes can be poor at times.
Furthermore, how able are patients to identify and explain exactly how they feel, and how likely are interviewers able to notice all signs and symptoms that a clinician would be better placed to identify? In such circumstances, electronic health records are more attractive for research purposes.
Electronic Health Records (EHRs):
Traditional forms of data collection such as EHRs still matter, even more so in an age where data is in “multimedia format” and is “unstructured”. Whereas traditional structured data is “data that can be easily stored, queried, recalled, analysed and manipulated by machine”, traditional unstructured data is “office medical records, handwritten nurse and doctor notes, hospital admissions and discharge records, paper prescriptions, radiograph films, MRI, CT and other images”. The structured data in EHRs, such as the name of a patient, their age, the visiting hospital, and other information that is easy to code and input into a database has its advantages.
EHRs used by primary care doctors are useful for researching the wider effect an individual’s mental illness has on their physical health. For example, researching how likely people with severe mental illnesses are of having cardiovascular disease by using a UK general practice database, and deriving a predictive model based on this was made possible through the use of EHRs.
Nevertheless, the disparity between structured data and unstructured data – notably, databases using free text – has hindered faster progress. While accepting the importance of EHRs and emphasising that clinical documentation is central to patients, healthcare providers refused to adopt computer-based documentation systems that prioritise direct structured documents. Arguing that health providers be able to choose between expressivity and structured clinical documentation, healthcare providers’ desire to continue with handwritten notes is supported by the fact that some healthcare providers are able to write better notes in this way.
(Analysis on EHRs are continued in the next blog post).
Written by Rodney
If interested in the references, or would like more information, contact us via email: email@example.com