Macquarie University is working with an aged care provider to develop a predictive falls risk model for the aged care sector to help minimise falls among senior Australians, this week’s national conference on gerontology research heard.
The project is in response to a recent systematic review that identified a lack of suitable predictive falls risk models for aged care.
The model is being developed in partnership with New South Wales aged care provider Anglicare with funding from a national research council grant.
It will use data that is routinely collected data in residential aged care, said Dr Karla Seaman, a research fellow at Macquarie University’s Australian Institute of Health Innovation.
“We are designing and developing a dashboard for this predictive model to sit integrated within client data. And there’ll be two risk models that we’re exploring; falls and quality of life,” Dr Seaman told the 2021 Australian Association of Gerontology Conference on Wednesday.
“We’re co-designing with aged care clients, family members, aged care staff and GPs,” said Dr Seaman, a co-lead investigator on the study.
Fellow co-lead investigator Dr Kristiana Ludlow said the university conducted the study to identify models for predicting falls in residential and home aged care services using routinely collected electronic health record data.
“We were interested to know how falls risk models have been developed in these settings, what was their accuracy and use in falls prediction, and then how they’ve been implemented to prevent falls,” said Dr Ludlow, a honorary postdoctoral fellow at the Australian Institute of Health Innovation.
Falls are the single largest contributor and behind just over two in five injury-related hospitalisations among older Australians, she said.
“Six out of seven people who suffer fall-related injuries live in residential aged care homes or receive care services from home-based or community providers,” Dr Ludlow said.
The study involved screening more than 7,000 papers but only four met the inclusion criteria. They included two each for residential and home aged care settings.
It identified nine predictive fall models and seven fall predictors including demographics, assessments conducted with residents or clients, fall history, medication, health condition, physical abilities and environmental factors.
However, Dr Seaman said there were limitations on the usefulness of predictive performance of the identified models.
“This really limits the utility of using these predictive performance models for other organisations and replicating these models,” said Dr Seaman.
“None of the models identified had been implemented and evaluated within practice. And it’s critical to determine their true effectiveness and cost effectiveness for health and wellbeing outcomes,” Dr Seaman said.
“There’s a large amount of data collected and stored in routine practice in residential aged care… but there’s limited evidence from predictive models for falls within aged care services. More research is needed and more… statistical methods are needed as well,” she said.
The 2021 AAG Conference takes place 9 -12 November.
Australian Ageing Agenda is a media partner of the AAG.
Comment on the story below. Follow Australian Ageing Agenda on Facebook, Twitter and LinkedIn, sign up to our twice-weekly newsletter and subscribe to AAA magazine for the complete aged care picture.
Really sorry but there is already a system that has been developed in the UK for this called – ARMED and has been highly successful in reducing the falls risk. The researchers might like to check this out.
As an established developer in this field of technology, we have spent many hours reviewing the causal/predictive effects of a fall. The frequency figures from age 60 onwards are scary and we have witnessed some suffering by elderly people. For example who were supposedly wearing the traditional call button but left it hanging on the towel rail while they were incapacitated in the bath.
The main criteria for falls detection has to assume that that they do not wear such a device. Given that as a starting point, how does one deduce they have had a fall?
Deduction is the critical factor. If you are monitoring daily routines then one has to assume it is possible to monitor a pattern of daily routines where minimal activity is a warning flag. We went through this process with MimoCare a while back and it is quite challenging to engineer an umbrella of care (as we call it) that can get concerned about a lack of activity.
We have now evolved a standard network of six normal low cost sensors that can recognise a negative situation, give it a doubt factor, before registering an alert. Bear in mind that these sensors are also logging all sorts of minor incidents re cooking stoves, activity levels, door exits etc
In a live field trial MimoCare Activity view, with a resident Elderly person, such an incident took place literally by accident at 4am on a trip to the bathroom. The Activity view showed that alert sequence quite dramatically.
In summary it is all a question of deduction, not a one off push button.
John Williams CEO
Hi Mark, thank you for your comment and for pointing us in the direction of ARMED. From what I can see, ARMED uses real-time data from a wearable device. Our review excluded wearable devices and sensory data as we were specifically interested in routinely collected data from electronic health records. I do think systems like ARMED could be of use in informing our future research, so again, thank you for sharing this.
Thanks for your comments Kristiana and I am trialing the system at one of my clients (a residential care facility) currently and besides the falls prevention they will also be using it for general health monitoring and providing independent data on someones health condition. Happy to discuss the outcomes on this.