Falls remain a major source of morbidity and mortality for people living in residential aged care
(RAC) despite an increasing focus on fall prevention research, practice and policy over the past
decades. Fall-related hospital presentations have not decreased, highlighting the need for
ongoing evaluation of existing prevention practice methods and exploration of novel methods
which may offer greater efficacy. Falls risk assessment has been advocated as a fundamental
component of falls prevention interventions. Despite a strong culture of “evidence-based
practice”, quality evidence to support the reliability, validity and “added value” of falls risk
assessment tools (FRATs) in RAC is sparse.
Four FRATs are currently used in RAC in Australia, and recommended in best practice guidelines.
However the reliability and validity of these tools has not previously been reported in the peerreviewed
literature. Studies to provide an evidence base for these tools is urgently required to
ensure that staff time in RAC is being optimally utilised for practices which offer the greatest
opportunity to prevent falls. This thesis aimed to address this need by investigating the clinimetric
properties of these four tools.
The four tools studied were the Peninsula Health Falls Risk Assessment Tool (PHFRAT), Queensland
Health Falls Risk Assessment Tool (QFRAT), Melbourne Health Falls Risk Assessment Tool (MFRAT) and
the Falls Assessment Risk and Management Tool (FARAM). An evaluation of the literature revealed
no information on the clinimetric properties of these four tools. The six studies of this thesis were
formed from data collected from 99 residents from four RAC facilities in a retrospective chart audit,
and a retrospective audit and prospective cohort study of 87 residents from six RAC facilities. Initial
studies for this thesis (studies 1 & 2) involved the examination of the reliability, predictive, internal
construct, evaluative and discriminative validity of each of the four tools. The investigation of
predictive validity included traditional and novel analyses. A new screening tool for identifying
residents most at risk of falling was developed and validated (study 3). Subsequent studies
investigated the clinimetric properties of a commonly-used mobility assessment, the Physical
Mobility Scale (PMS) (study 4), and the relationship between resident mobility and fall and fracture
risk (study 5). A study on the high fall and fracture risk locations, activities and times of day in RAC
was also completed, providing a population-based risk model as an alternative to the assessment
of individual intrinsic fall and fracture risk factors (study 6).
The four commonly-used and recommended tools in best practice guidelines were found to have
significant clinimetric flaws. These included low levels of agreement between raters on scoring of
risk factors and inclusion of many risk factors which had no significant predictive relationship with
falls. The predictive validity of the tools was also limited, with no tool offering significantly greater
predictive ability than a simple screening question: “Has the resident fallen in the last 12 months?”.
The Rasch analysis results indicated that the tools are multidimensional, providing empiricalevidence that the tools are not valid measures of falls risk and therefore have poor evaluative and
Regression analysis found that two out of a possible sixty-seven risk factors included on the four tools
examined in the previous studies were reliable and independent predictors of future falls. These
were item 3 of the FARAM: “presence of cognitive, emotional and behavioural impairment and the
resident attempts to mobilise without assistance”, and item 17 of the MFRAT: “signs of an acute
illness”. These two risk factors were combined to form a new falls risk screen which was found to
have greater predictive accuracy than the previously studied four tools. As the new falls screen
includes only two items it is a more parsimonious tool than the previously studied tools which include
between six and thirty items. External validation studies including examination of therapeutic
impact are required of this new two-item tool before widespread use and dissemination is
The commonly-used mobility assessment, the PMS, was found to have good clinimetric properties.
Agreement between raters on scoring items of the PMS was high, and the measurement error small.
The PMS was found to have a unidimensional structure demonstrated by fit to the Rasch model,
validating the appropriateness of the summing of item scores to yield a total score representing an
overall measure of mobility. No floor or ceiling effects were detected. The study provides Rasch
transformed PMS scores, yielding an interval measure of mobility suitable to a range of clinical and
research applications in RAC.
The study of the relationship between mobility, as measured by the PMS, and falls risk identified a
non-linear relationship between risk of falling and mobility. People with mild to moderate mobility
impairment (PMS total score between 28 and 36) were at greater risk of falling than people who
were non-ambulant or independent. These findings indicate that falls prevention strategies in RAC
that focus on increasing resident mobility may indeed increase the risk of falling in some residents.
This non-linear relationship creates a paradox for common goals of physiotherapists in RAC: to
improve resident mobility and to reduce fall risk. This non-linear relationship also calls into question
the representation of mobility on FRATs, which commonly score greater mobility impairment as
greater falls risk. A similar relationship between mobility and fracture risk as mobility and falls risk was
found. People with a mild to moderate mobility impairment (PMS total score between 28 and 36)
were at greatest risk of suffering a fracture from a fall. Due to the small number of fractures
included in the fracture risk analysis, further validation of this relationship in a larger data set is
The population model study developed and validated a fall and fracture fall activity and
environment interaction risk profile. An increased fracture risk was found with falls occurring outside
or while walking, indicating a target for fracture prevention strategies. It was also found that manyfalls occurred in the bedroom while walking, providing a target for fall prevention strategies.
Population-based models have the potential to offer several advantages for falls prevention over
individual intrinsic risk models such as: efficiency, simplicity, and broader health benefits for other
geriatric syndromes. Consequently, population-based models offer a new direction for future fall
and fracture prevention research and practice. Due to the small number of fractures included in
this study, further validation of the profile in a larger data set is required.
The results of this thesis provide new information about FRATs recommended in Australian Best
Practice guidelines and commonly used in Australian RAC facilities. Current tools have unsound
reliability and validity and thus their use and promotion is not evidence based. Two alternative
individual intrinsic screening methods affording parsimony and sound clinimetic properties, and a
population-based risk model were proposed. The six studies of this thesis represent a substantial
contribution towards the knowledge of clinimetric properties of FRATs for RAC, the analyses
techniques for evaluating FRATs and also provide a basis for future investigations of fall prevention
practice in RAC. The highest priority of future research should be the investigation of the “added
value” of falls risk assessment in the RAC setting.