Associations Between Walking Ability Questionnaires and Performance-based Walking Speed in Survivors of Critical Illness

Résumé fourni par la source

Abstract RATIONALE: Slower pre-critical illness walking speed is associated with greater mortality from critical illness. Assessing walking speed is difficult during emergent critical illness, suggesting alternative approaches are needed. We compared patient and surrogate responses on a walking ability questionnaire to performance-based walking tests completed by patients. METHODS: We included patients age ≥60 years, who were treated for respiratory failure, shock, and/or sepsis along with surrogates who spent ≥1 hour/day or ≥8 hours/week with the patient. We excluded patients who could not walk independently or who did not have an eligible surrogate. At 10 weeks after hospital discharge, patients and surrogates completed a questionnaire on the patient's walking abilities including transferring from a chair/bed, problems due to walking or balance, falls in the last year, difficulty walking across a room, crossing a room to answer the phone, difficulty climbing a flight of stairs, crossing the street before the light changed, difficulty walking a quarter mile, ability to take 2 steps per second, estimated walking speed, walking speed compared to peers, and satisfaction with walking speed. Patients then completed a timed 4-meter walk test. We used univariable regression, Least Absolute Shrinkage and Selection Operator (LASSO) regression, and Random Forest models to determine questionnaires’ ability to predict walking time. RESULTS: We enrolled 75 patient-surrogate dyads. Patients were a median (interquartile range) of 66 (64-70) years old, 33% were female, with a baseline ADL score of 1 (1-3). Most surrogates were spouses (68%) or adult children (12%). On univariate analysis of patient responses, crossing a room to answer the phone (p=0.35), ability to take 2 steps per second (p=0.12), and falls in the last year (p=0.64) were not significantly associated with walking time. On surrogate responses, only falls in the last year was not significantly associated with walking time (p=0.31). All other responses by patients and surrogates were significantly associated with walking time. For patient and surrogate responses, LASSO models demonstrated an R2 of 0.44 and 0.71 and Spearman's rho of 0.69 and 0.70, respectively (Figure). Random Forest models demonstrated an R2 of 0.71 and 0.78 and Spearman's rho of 0.79 and 0.79, for patient and surrogate responses, respectively (Figure). CONCLUSION: Predictive models using a walking ability questionnaire completed by patients and well-chosen surrogates showed good fit and strong correlation with performance-based walking time. Future work is needed to understand differences between patient and surrogate responses and to refine and externally validate these models.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.