File
Handlery R1, Fritz S1
1University of South Carolina, Exercise Science, Physical Therapy Program, Columbia, United States
Background: Over one in three individuals with stroke (IWS) will fall annually. Falls often lead to injury or death, therefore it is crucial to identify those most at risk and subsequently intervene. Proposed cutpoints for the Timed Up & Go (TUG) attempt to dichotomize IWS into risk categories (e.g. high vs. low). While a singular cutpoint is simple to use, it does not provide the magnitude of fall risk and mistakenly implies that individuals close to, but on opposite sides of the cutpoint are clinically different.
Purpose: Determine if the TUG can be used to quantify the probability of reported falls in IWS, using pre and post-test probabilities (PrTP and PoTP) to provide an in-depth look into falls risk.
Methods: Cross-sectional retrospective study used baseline TUG and reported falls data from three studies which included individuals >6 months post-stroke who could ambulate >10 meters with or without an assistive device (AD). The sample (n=83) included 26 females and 29 individuals who used an AD. Mean age of sample was 66 (SD 12) years and median time since stroke 5 (range 0.5 - 23) years. Based on falls history, individuals were classified as “Fallers” and “Non-Fallers”. Regression determined the association between TUG and falls. Receiver operating characteristic (ROC) curves, including area under the curve (AUC), were used to calculate sensitivity, specificity, Youden's index (j), positive and negative likelihood ratios, PrTP and PoTP.
Results: Prevalence of >1 reported fall in the previous year (PrTP) was 43% (36/83). Between Fallers and Non-Fallers, no significant differences in age, sex, time since stroke or AD use were found. TUG was associated with reported falls (p = 0.021) and correctly categorized 69% of individuals. Fallers had increased TUG times (p = 0.002) and TUG AUC was 0.70 (95% CI: 0.59, 0.81). Using j, TUG times of 10.6 and 27.4 seconds (s) were identified as dual cutpoints. TUG times 10.6s yielded a 16% PoTP of a fall, whereas TUG times >27.4s yielded a 78% PoTP.
Conclusion(s): PrTP of an individual reporting a fall was 43%. Performing the TUG in 10.6s reduced this probability to 16% whereas a TUG >27.4s increased the probability to 78%. While dichotomizing IWS into risk categories based off TUG cutpoints is common, our findings demonstrate that this oversimplifies falls risk. TUG AUC (.70) indicates a low accuracy diagnostic test and nearly half (41/83) of our sample had TUG times between our dual cutpoints, making TUG times between 10.6s and 27.4s difficult to interpret. To improve the usability of cutpoints, PoTP can quantify the likelihood of reported falls. Nevertheless, the TUG and associated cutpoints should not be used in isolation. Falls risk is multifactorial and multiple measures (both performance and participant-reported) should be utilized to provide a comprehensive look into falls risk.
Implications: Identifying individuals at increased risk for falling is crucial, but falls risk exists on a continuum and is not simply present or absent. By utilizing PoTP, clinicians can go beyond dichotomizing and quantify falls risk. This may aid in assessments, goal writing and patient education.
Keywords: stroke, TUG, falls
Funding acknowledgements: Health Games Research-national program of the Robert Wood Johnson Foundation grant-64450, American Heart Association grant-0835160N, American Physical Therapy Association
Purpose: Determine if the TUG can be used to quantify the probability of reported falls in IWS, using pre and post-test probabilities (PrTP and PoTP) to provide an in-depth look into falls risk.
Methods: Cross-sectional retrospective study used baseline TUG and reported falls data from three studies which included individuals >6 months post-stroke who could ambulate >10 meters with or without an assistive device (AD). The sample (n=83) included 26 females and 29 individuals who used an AD. Mean age of sample was 66 (SD 12) years and median time since stroke 5 (range 0.5 - 23) years. Based on falls history, individuals were classified as “Fallers” and “Non-Fallers”. Regression determined the association between TUG and falls. Receiver operating characteristic (ROC) curves, including area under the curve (AUC), were used to calculate sensitivity, specificity, Youden's index (j), positive and negative likelihood ratios, PrTP and PoTP.
Results: Prevalence of >1 reported fall in the previous year (PrTP) was 43% (36/83). Between Fallers and Non-Fallers, no significant differences in age, sex, time since stroke or AD use were found. TUG was associated with reported falls (p = 0.021) and correctly categorized 69% of individuals. Fallers had increased TUG times (p = 0.002) and TUG AUC was 0.70 (95% CI: 0.59, 0.81). Using j, TUG times of 10.6 and 27.4 seconds (s) were identified as dual cutpoints. TUG times 10.6s yielded a 16% PoTP of a fall, whereas TUG times >27.4s yielded a 78% PoTP.
Conclusion(s): PrTP of an individual reporting a fall was 43%. Performing the TUG in 10.6s reduced this probability to 16% whereas a TUG >27.4s increased the probability to 78%. While dichotomizing IWS into risk categories based off TUG cutpoints is common, our findings demonstrate that this oversimplifies falls risk. TUG AUC (.70) indicates a low accuracy diagnostic test and nearly half (41/83) of our sample had TUG times between our dual cutpoints, making TUG times between 10.6s and 27.4s difficult to interpret. To improve the usability of cutpoints, PoTP can quantify the likelihood of reported falls. Nevertheless, the TUG and associated cutpoints should not be used in isolation. Falls risk is multifactorial and multiple measures (both performance and participant-reported) should be utilized to provide a comprehensive look into falls risk.
Implications: Identifying individuals at increased risk for falling is crucial, but falls risk exists on a continuum and is not simply present or absent. By utilizing PoTP, clinicians can go beyond dichotomizing and quantify falls risk. This may aid in assessments, goal writing and patient education.
Keywords: stroke, TUG, falls
Funding acknowledgements: Health Games Research-national program of the Robert Wood Johnson Foundation grant-64450, American Heart Association grant-0835160N, American Physical Therapy Association
Topic: Neurology: stroke; Outcome measurement
Ethics approval required: Yes
Institution: University of South Carolina
Ethics committee: University of South Carolina's Institutional Review Board
Ethics number: Pro00001904, Pro00000038
All authors, affiliations and abstracts have been published as submitted.
