Cost per fall: a potentially misleading indicator of burden of disease in health and residential care settings
Haines, TP and Nitz, J and Grieve, JB and Barker, A and Moore, K and Hill, K and Haralambous, B and Robinson, A, Cost per fall: a potentially misleading indicator of burden of disease in health and residential care settings, Journal of Evaluation in Clinical Practice, 19, (1) pp. 153-161. ISSN 1365-2753 (2013) [Refereed Article]
Copyright 2011 Blackwell Publishing Ltd, Journal of Evaluation in Clinical Practice
Rationale Little scrutiny has been applied to how ‘cost per fall’ values have been calculated
and applied. This paper presents discourse discussing how the cost of fall statistic
could potentially be misleading when applied to inpatient health or residential care settings
and provides estimates of the cost of falls and cost of falls per person.
Method Burden of disease was estimated using a decision tree approach. Data informing
the decision tree were drawn from a retrospective audit of 545 falls in a residential care
facility, a visual observation study of 46 residents from the same facility and a cohort study
of 186 residents from nine different facilities in Australia. Acute care and transportation
costs were extracted from the retrospective audit using incident reports and care note
review. The distribution of falls per person and associations between falls, fractures and
change in resident mobility were extracted from the cohort study. The association between
resident mobility and the amount of time required to perform toileting, transfer and
dressing activities was extracted from the visual observational study.
Results The minimum ‘cost per fall’ was estimated to be $AUD 841 and the maximum
was $AUD 1024. The ‘cost of falls per person’ estimate was $AUD 1887 (2008 base year).
Conclusions This cost per fall estimate was substantially lower than three previous internationally
derived estimates possibly as a consequence of how fall data were collected and
modelled in these studies. Cost of falls per person may be a preferable statistic for future