Abstract
Introduction
Ambulation in the postoperative period following TKR is a marker of speed of recovery and, potentally, longer term outcomes. However, patient lifestyle factors are a major confounder. This study sought to develop a model of expected patient step count taking into account preoperative condition and demographics in order to benchmark recovery at a patient specific level.
Method
94 patients were recruited to the study. BMI, demographics, the Short Form 12 (SF-12) and the Knee injury and Osteoarthritis Outcome Score (KOOS) were all captured preoperatively. Step count was measured using commercially available Fitbit devices preoperatively, immediately postoperatively and at 6 weeks postoperatively. Stepwise multiple linear regression models were developed using the preoperative information to define a predictive model of the postoperative step count levels. Spearman's Rho correlations for all relevant data series were also calculated.
Results
Of the personal and clinical characteristics, BMI and the SF-12 physical component score had the strongest correlations with outcome. Prior step count periods all had significant correlations with later step count periods. The most significant correlations occurred between the 6 week postoperative step count period and the preoperative period (0.709), while correlations with the period immediately following surgery were weaker (0.389 and 0.536 for preoperative and 6 week postoperative step counts respectively.) All are significantly different from 0 (to p < 0.01.) Likewise, BMI had a significantly negative relationship with step count (−0.526, −0.346 and −0.553 for the preoperative, immediate postoperative and 6 week postoperative periods, see Figure 2), as did the KOOS activities of daily living score and the SF-12 physical health component score. Males were significantly less mobile than females during recovery.
A multiple linear regression model of 6 week step count using prior data had an adjusted R2 of 0.754, explaining much of the variation, but the immediate postoperative period performed poorly. Predictors in the 6 week model were gender, preoperative SF-12 score, preoperative and immediate postoperative step count.
Conclusions
Patient specific factors, including but not limited to that from prior step count periods need to be considered if using step count as a means of benchmarking patient recovery after surgery. The variation in recovery at 6 weeks is more readily explained with the data collected than in the immediate postoperative period, where variations in specific care received, anaesthetic response or surgical outcomes might be more expected to have an impact. Reporting patient performance against customised goals on an individual patient basis could provide a means to drive greater patient mobility and appropriate activity levels during postoperative recovery.
