Abstract
Introduction
With many stakeholders, healthcare decisions are complex. However, patient interests should be prioritized. This maximizes healthcare value (quality divided by cost), simultaneously minimizing costs (objective) and maximizing quality (subjective). Unfortunately, even ‘high value’ procedures like total knee arthroplasty (TKA) suffer from recovery assessment subjectivity (i.e. high assessment variability) and increasing costs. High TKA costs and utilization yield high annual expenditures (∼$22B), including postoperative physical therapy (PT) accounting for ∼10% of total costs (∼$2.3B annually). Post-TKA PT is typically homogenous across subjects ensuring most recover, however recent work shows outcomes unimpacted by PT. Accordingly, opportunities exist improving healthcare value by simultaneously reducing unnecessary PT expenditures and improving outcomes. However, discerning recovery completion relies on discrete ROM measures captured clinically and subjective clinician experience (i.e. intuition about recovery). Accordingly, our goal was developing objective post-TKA performance assessment methods utilizing gait knee ROM and statistical analyses to categorize patient recovery (‘accelerated,’ ‘delayed,’ or ‘normal’).
Methods
We first established statistical reasons for current post-TKA rehabilitation including risk-reward tradeoffs between incorrectly ascribing ‘poor recovery’ to well-recovering patients (T1 error) or ‘good recovery’ to poorly-recovering patients (T2 error) using methods described by Mudge et al. and known TKA volumes/rehabilitation costs. Next, previously captured gait ROM data from well-healed patients was utilized establishing standard recovery curves. These were then utilized to assess newly captured patient recovery. Following IRB approval, we prospectively captured gait ROM from 10 TKA patients (3M, 69±13 years) 1-week pre-TKA and 6-weeks immediately post-TKA. Performance was compared to recovery curves via control charts/Shewhart rules (daily performance) as well as standard deviation thresholds (weekly performance) establishing recovery as ‘accelerated,’ ‘delayed,’ or ‘normal.’ The categorization was extrapolated to US TKA population and savings/expenses quantified. Statistical analyses were performed in Minitab with statistical significance set to α<0.05.
Results
Current post-TKA approach is as much PT as possible (AMPTAP). AMPTAP was confirmed statistically. Because poor recovery costs are significant, balancing T1/T2 error minimizes risk by removing T1 error risk (α=0.00) via 27 PT sessions for equal cost to one manipulation under anaesthesia (MUA). Previously captured, well-healed subject gait ROM were always normally distributed. Assessing performance via control charts showed serial ‘accelerated’/‘delayed’ recovery and would serially under/over-prescribe PT. Establishing recovery performance via ±1SD thresholds successfully evaluated 3 clinically established “poor recoverers” as ‘delayed’ and the reaming clinically established “good recoverers” as ‘normal’ or ‘accelerated’ throughout recovery.
Discussion
Optimization for current AMPTAP approach, while effective reducing poor recovery risk, is a gross misuse of rehabilitation spending. Improved methods are necessary including those rooted with strong statistical foundation. Control charts are likely too fine an assessment as patient performance day-to-day is too variable resulting in clinical rehabilitation prescription over-reactions. In contrast, standard deviation thresholds likely provide a conservative approach that allows clinicians the opportunity to improve postoperative rehabilitation week-after-week throughout recovery. However, PT was not altered herein. Thus, the impact altering PT has on postoperative outcomes remains unknown. Future work should investigate how altering postoperative rehabilitation changes postoperative outcomes.
For any figures or tables, please contact authors directly.
