Development of a falls risk prediction tool for use with people with multiple sclerosis

People with Multiple Sclerosis (MS) present with a wide range of symptoms including sensory, motor and visual impairment as well as cognitive dysfunction and fatigue. Many of these symptoms affect mobility and balance and have been shown to be associated with falls risk among this population. It is known that falls are prevalent among people with MS, with a high rate of multiple falls and injurious falls. While much is known about the factors and serious consequences associated with falls, as of yet there is no reliable stand-alone clinical measure or multivariable model suitable to assess falls risk in a busy clinic setting. Thus, the aim of this thesis was to develop a simple falls risk screening model suitable for use in everyday clinical practice. To understand what clinical measures of balance are currently useful in identifying falls risk in MS, a systematic review and meta-analysis was carried out. There was significant heterogeneity across the included studies and discriminative ability of the measures is commonly not reported. The Timed Up and Go (TUG) did show significant difference between fallers and non-fallers in retrospective study designs, is commonly used and does not require specialist equipment, and thus was investigated in a prospective cohort that monitored falls using diaries for 3 months. The association between dual task cost and falls was explored in more depth by examining objectively measured dual task cost and subjective problems dual tasking. Different patterns of cognitive -motor interference and their association to faller status was also analysed. Results showed that objectively measured dual task cost is not associated with an increased falls risk but self-report problems of difficulty doing two things at once doubled the risk of falling with an associated risk ratio of 2.07 (CI 1.15-3.71). From the main longitudinal study multiple clinical and objective variables were analysed to determine the model with the greatest sensitivity and best discriminative ability for identifying falls risk in people with MS. Following multivariable regression analysis, the model with the greatest sensitivity (88%) and predictive validity (AUC = 0.72, 95% CI 0.62-0.82), included the variables of history of a fall, no visual problems, problems with bladder control and a slower speed on the TUG. The clinical implications arising from this research are important; firstly, all healthcare professionals working with people with MS should ask about history of falls, visual problems, problems with bladder control and difficulty dual tasking. Clinicians should not rely on a clinical measure of balance alone to identify falls risk but consider a multivariable model that would be more sensitive and provide more useful information. Future research should validate this falls risk model using a larger sample size, with a wider range of EDSS levels and disease subtypes. Following validation, implementation could be carried out and if used successfully in daily clinical practice this model could help prioritise waiting lists and enable earlier access to fall prevention interventions at the most appropriate time point for that individual