A new automated tool known as the Military Service Identification Tool (MSIT) has been developed by researchers at the King’s Centre for Military Health Research (KCMHR), part of King’s College London, and has been found to successfully identify veterans in electronic healthcare records.
Veteran status is not consistently recorded in healthcare records, making it difficult for clinicians to recognise if someone ever served in the military, whether military-related factors may be relevant to their healthcare, and if they could be entitled to military-specific support. Recording veteran status is also important for researchers to study the health needs of this population.
The Military Service Identification Tool uses machine learning to analyse healthcare records and using the text in those records to make a prediction on whether the patient has ever served in the military.
The Military Service Identification Tool was successful in accurately identifying veterans from the healthcare records of a specific secondary mental healthcare service – South London and Maudsley (SLaM) NHS Foundation Trust. Data was rigorously analysed by checking the tool’s predictions against clinical records and by asking patients themselves for their veteran/civilian status. When asking patients themselves, 84% were accurately classified as having military experience. Researchers found the tool to be substantially more accurate and less time-intensive than other possible approaches, like using military keywords in a simple manual search.
The tool is available for researchers and NHS trusts for further testing and development across other parts of the health service, with the code available in the report. Doing so will provide a better picture of veterans’ health and will help policy-makers, commissioners and other stakeholders ensure the health and support services provided to veterans are appropriate.
Use of the Military Service Identification Tool allowed the researchers to identify and then compare individuals with a military background to those without. They found that veterans accessing SLaM’s mental health services were more likely to have diagnoses of depressive, anxiety, stress, personality and psychotic disorders compared to those without military experience, and veterans were more likely to be detained under the Mental Health Act. The researchers urged caution in generalising these emerging findings, as they are specific to those accessing SLaM services and require further investigation.
Treatment pathways of SLaM patients could not be explored due to a large volume of missing data within healthcare records. Other recent research identifies similar gaps, with many healthcare and support organisations calling for improvements and consistency in data collection. The researchers advocate for veteran status to be more widely recorded, as findings suggest that they may demonstrate different health needs from the general population.
Tom McBarnet, Chief Executive (Acting) of Forces in Mind Trust, said “This tool represents a large saving in human resources, cost and time required to identify veterans using healthcare services.
“There is a clear need for healthcare services that are designed to best meet the sometimes unique health and care needs of ex-Service personnel and their families. There must be sufficient expertise, awareness and understanding of the ex-Service community and their physical and mental health needs, and that these are taken into account in any healthcare provision wherever it is delivered. The Government’s Veterans’ Strategy Action Plan recognises that there is a lack of data which hampers efforts to provide for the needs of ex-Service personnel in our healthcare systems. This research adds to the body of evidence which has identified this issue, but it also points to a possible solution.
“More work is clearly needed to improve veteran health data capture, but this is a really important step.”
Dr Laura Palmer, a researcher and lecturer at KCHMR, said
“By helping us identify veterans in electronic healthcare records, the Military Service Identification Tool can open doors to previously unexplored datasets. This will help researchers address the ever-present question of whether individuals with military experiences are similar or different to those without, and whether different treatment approaches are warranted”.
Professor Rob Stewart, Academic Lead for the Maudsley’s Clinical Record Interactive Search (CRIS) platform, said
“The Military Service Identification Tool is a useful innovation that solves the lack of marker for veteran status, and will allow a better understanding of why and how veterans use healthcare services in the NHS”.
The research was led by the King’s Centre for Military Health Research at King’s College London and Combat Stress, and was funded by Forces in Mind Trust.