|Identifying patients at risk from falling not a perfect process.|
Image: Expert Institute
Older people admitted to hospital face the risk of several adverse events such as falls, pressure injuries and malnutrition, which can cause substantial harm to them, distress to family and carers, and increased burden on inpatient services.
To prevent such harmful events hospitals are increasingly using predictive scores to try to identify individual patients mostly likely to be at risk. But clinicians at Alfred Health have for some time been concerned that the tools used to predict this risk are not accurate.
Dr Chris Moran, geriatrician at Caulfield Hospital and Research Fellow with Monash University’s Department of Medicine, Peninsula Clinical School, decided to investigate.
A study conducted at The Alfred’s Department of Aged Care at Caulfield Hospital, published in Internal Medicine, examined ways to improve predictive accuracy – and came up with some unexpected findings.
“The tools usually require nurses to put in a fairly substantial amount of time to complete risk scores,” Dr Moran said. “There are a whole mix of different tools used around Australia which is a sign that there was no clear winner.”
The scores are based on questions asked by nurses about factors such as whether the patient has a history of falls, a poor diet or appetite.
The researchers examined 4487 admissions (median age 83 years) from January 2014 to June 2016 from electronic medical records, extracting data on hospital risk scores and incidence of falls, pressure injury and malnutrition.
“It turned out that the current tools really aren’t very good at predicting the risk of harm. They were right in about 70 per cent of cases.”
The researchers tried to combine the best tools and questions to arrive at an improved tool then stripped the questions down to two (e.g. age and previous falls).
“I was surprised,” Dr Moran said. “I thought that by combining these risk tools, we’d get a better predictive value but the improvement was so small that they didn’t really translate to any meaningful improvement in predictive capability.
“The study let us know that these scores don’t guide us the way we thought they would,” he said.
It showed the limitations of harm prediction tools, a likely growth area in clinical practice in the future, he said.
Dr Moran said that not using the tools would free up nursing time, aid in efficiency and allocating resources better. Time could be better spent on interventions such as making sure the patient had their glasses, helping them access the toilet or had food that was more appropriate to them culturally or in terms of consistency, he said.
First author on the paper was Dr Stephanie Than, an Aged Care registrar at Caulfield Hospital and now a PhD candidate at Monash University’s Department of Medicine, Peninsula Clinical School.
Than S, Crabtree A, Moran C. Examination of risk scores to better predict hospital-related harms. Intern Med J. 2018 Oct 1. doi: 10.1111/imj.14121. [Epub ahead of print]