AutoMal: Automating malnutrition identification

The AutoMal is a new, quick, and accurate way to identify residents within aged care facilities that are at risk of malnutrition allowing for earlier identification and action. The AutoMal is unique as it draws from resident information that is routinely collected and documented as part of usual care. The AutoMal is similar to existing malnutrition screening tools at correctly identifying people who are at risk of being malnourished.

Why is this needed?

While there are several existing screening tools, these all require staffing to implement which is a substantial burden on time and resources. We know that in practice malnutrition screening is not routinely conducted and as such the nutrition status of most residents in aged care is unknown. Early detection and management are essential for avoiding the negative consequences of malnutrition such as increased risk of falls, infections, and pressure injuries.

The details:

Development: AutoMal was developed based on data from 539 residents across 10 residential aged care facilities across Queensland, South Australia, and Victoria.
Calculation: AutoMal screens for malnutrition using Body Mass Index (BMI) and weight change % over 6-months.
Accuracy: The accuracy of a test is often described in terms of sensitivity and specificity:
– Sensitivity: AutoMal is 89% sensitive, meaning that AutoMal detects 89% of people who are malnourished. It is important to have high sensitivity to minimise missing anyone who may require additional nutrition support.
– Specificity: AutoMal is 50% sensitive, meaning that AutoMal incorrectly identifies 50% of non-malnourished people as malnourished. As AutoMal is a screening tool, having these false-positive results is acceptable as further follow-up will detect those who do not require additional nutrition support.
Customisation: Depending on your facility requirements, AutoMal accuracy can be adjusted through a trade-off. Either sensitivity is increased while specificity is decreased; or sensitivity is decreased while specificity is increased.

Publication: Foo, J., Roberts, M., Williams, L., Osadnik, C., Bauer, J., & O’Shea, M.-C. (in press). An automated malnutrition screening tool using routinely collected data for older adults in long-term care: development and internal validation of AutoMal. Journal of the American Medical Directors Association.

How to get involved:

Residential aged care facilities provide a de-identified list of residents and their clinical measurements. These measurements are outlined in an template excel provided by the research team. Our research team use the AutoMal to quickly and easily identify residents who are at risk of malnutrition and alert the facility manager. No resident identifying information is available to the research team, only the facility staff or manager.

We are looking to further validate the AutoMal with more facilities and more residents.