Annie Ross1, Alison J Yarnall2, Lynn Rochester3, Sue Lord4. 1. Institute of Neuroscience, Newcastle University Institute for Aging, Newcastle upon Tyne, UK. Electronic address: annie.ross@nuth.nhs.uk. 2. Institute of Neuroscience, Newcastle University Institute for Aging, Newcastle upon Tyne, UK. Electronic address: alison.yarnall@newcastle.ac.uk. 3. Institute of Neuroscience, Newcastle University Institute for Aging, Newcastle upon Tyne, UK. Electronic address: lynn.rochester@newcastle.ac.uk. 4. Institute of Neuroscience, Newcastle University Institute for Aging, Newcastle upon Tyne, UK. Electronic address: sue.lord@ncl.ac.uk.
Abstract
BACKGROUND: Falls are a major problem for people with Parkinson's disease (PD). Despite years of focused research knowledge of falls aetiology is poor. This may be partly due to classification approaches which conventionally report fall frequency. This nosology is blunt, and does not take into account causality or the circumstances in which the fall occurred. For example, it is likely that people who fall from a postural transition are phenotypically different to those who fall during high level activities. Recent evidence supports the use of a novel falls classification based on fall related activity, however its clinimetric properties have not yet been tested. OBJECTIVE: This study describes further development of the Fall-Related Activity Classification (FRAC) and reports on its inter-rater reliability (IRR). METHOD: Descriptors of the FRAC were refined through an iterative process with a multidisciplinary team. Three categories based on the activity preceding the fall were identified. PD fallers were categorised as: (1) advanced (2) combined or (3) transitional. Fifty-five fall scenarios were rated by 23 raters using a standardised process. Raters comprised 3 clinical subgroups: (1) physiotherapists, (2) physicians, (3) non-medical researchers. IRR analysis was performed using weighted kappa coefficients and included sub group analysis based on clinical speciality. RESULTS: Excellent agreement was reached for all clinicians, κ=0.807 (95% CI 0.732 to 0.870). Clinical subgroups performed similarly well (range of κ=0.780 to 0.822). CONCLUSION: The FRAC can be reliably used to classify falls. This may discriminate between phenotypically different fallers and subsequently strengthen falls predictors in future studies. Crown
BACKGROUND: Falls are a major problem for people with Parkinson's disease (PD). Despite years of focused research knowledge of falls aetiology is poor. This may be partly due to classification approaches which conventionally report fall frequency. This nosology is blunt, and does not take into account causality or the circumstances in which the fall occurred. For example, it is likely that people who fall from a postural transition are phenotypically different to those who fall during high level activities. Recent evidence supports the use of a novel falls classification based on fall related activity, however its clinimetric properties have not yet been tested. OBJECTIVE: This study describes further development of the Fall-Related Activity Classification (FRAC) and reports on its inter-rater reliability (IRR). METHOD: Descriptors of the FRAC were refined through an iterative process with a multidisciplinary team. Three categories based on the activity preceding the fall were identified. PD fallers were categorised as: (1) advanced (2) combined or (3) transitional. Fifty-five fall scenarios were rated by 23 raters using a standardised process. Raters comprised 3 clinical subgroups: (1) physiotherapists, (2) physicians, (3) non-medical researchers. IRR analysis was performed using weighted kappa coefficients and included sub group analysis based on clinical speciality. RESULTS: Excellent agreement was reached for all clinicians, κ=0.807 (95% CI 0.732 to 0.870). Clinical subgroups performed similarly well (range of κ=0.780 to 0.822). CONCLUSION: The FRAC can be reliably used to classify falls. This may discriminate between phenotypically different fallers and subsequently strengthen falls predictors in future studies. Crown