Big data can spot psychiatric inpatients who risk being subjected to mechanical restraint

Last year in Denmark, a total of 1,342 psychiatric patients were restrained to their bed using belts (mechanical restraint) and although this is fewer than previously, the figure is still far too high according to experts. Researchers have now identified patients who are at risk for mechanical restraint.

"We’ve used the Danish health registers to identify patients who have been admitted to a psychiatric department in the Central Denmark Region and who have not previously been subjected to mechanical restraint" explains Andreas Aalkjær Danielsen, PhD student at Aarhus University and medical doctor at Aarhus University Hospital, Psychiatry, who is behind the study.

Based on the data in the medical records and the health registers, a range of mathematical models known as algorithms were trained to recognise patterns that increase the risk of a patient being subjected to mechanical restraint during the first three days after admission to the unit. One of these were then validated in an independent dataset.

“The validated algorithm identified patients that have a high risk of being subjected to mechanical restraint with a level of precision that means it can be utilised by the psychiatric units,” says Andreas Aalkjær Danielsen.

Early warning of personnel

He hopes that the results can be used to develop an automatic warning system that can help medical doctors and nurses identify those patients who could benefit from more intensive treatment.

“Mechanical restraint are a drastic intervention and can be connected with a number of adverse events such as psychological stress or shoulder injuries for both patient and personnel. For many years the aim has therefore been to avoid the use of mechanical restraint as far as possible," says Andreas Aalkjær Danielsen.

The study included 5,050 patients with a total of 8,869 admissions. The results have been published in the scientific journal Acta Psychiatrica Scandinavica.

Previous studies had already identified some risk factors for the use of mechanical restraint, but not with sufficient precision. There remains a lack of knowledge about how these risk factors interact.

“Mental health professionals still face a daily challenge in quickly finding out who is at risk and how to prevent the use of mechanical restraint. We hope that by identifying risk factors in large amounts of data, these algorithms can be used to predict these kinds of incidents," says Andreas Aalkjær Danielsen.

• The study was carried out with the help of statistical analyses of snippets of data from the medical records of patients in the Central Denmark Region, among other sources.
• The study is funded by Aarhus University, the Central Denmark Region and the Lundbeck Foundation.
• The scientific article Predicting mechanical restraint of psychiatric inpatients by applying machine learning on electronic health data can be read in Acta Psychiatrica Scandinavica.

Contact:

PhD student & MD Andreas Aalkjær Danielsen
Aarhus University, Department of Clinical Medicine and
Aarhus University Hospital, Psychosis Research Unit
Email: andreas.aalkjar.danielsen@clin.au.dk
Mobile: (+45) 5136 4890