? DESCRIPTION (provided by applicant): In the United States, total knee arthroplasty (TKA) is an elective procedure-the most common inpatient elective procedure performed (~700,000 per year)-but it is also a serious surgery with an unpredictable course of recovery. The current standard for postoperative planning and rehabilitation monitoring relies on the average patient prognosis (APP)-or the estimated trajectory of recovery for the average patient. Thus, rehabilitation strategies remain protocol-based (e.g. 3 times per week for 6 weeks), assuming the same general recovery for all patients. However, given the heterogeneity in the population, this standard is inadequate. A more appropriate approach would be to estimate the individual patient prognosis (IPP), or the anticipated trajectory of recovery for this patient, at this point n time. Patients and providers could use such estimates to weigh the appropriateness of surgery and develop individualized rehabilitation plans. Moreover, physical therapists could better detect deviations from expected recovery throughout rehabilitation, thereby improving rehabilitation monitoring. The focus of this application is to develop IPPs through use of novel methodologies that allow for the prediction of a new patient's post-TKA prognosis on the basis of previous patients with similar characteristics. This approach-often referred to as the nearest neighbors or patients-like-me approach-has yet to be applied in rehabilitation science. Furthermore, we plan to utilize a database of measures collected in routine clinical practice, developed in partnership with clinicians at Proaxis Physical Therapy (see letter of support), to enable the estimation of IPPs in a clinical population that is not exposed to the eligibility criteria (and associated selection bias) of research studies. Thus, we also anticipate an improved representation of minority populations in this study. Our approach involves a two-phase process for estimating prognosis in 3 key areas: 1) pain; 2) self-reported function; and 3) physical performance. For Aim 1, we will develop predictive models for postoperative (6-month) outcomes, using a comprehensive battery of preoperative measures, spanning domains of physical function, psychological distress, and health status. For Aim 2, we will utilize a combination of the most robust predictors (identified in Aim 1) to develop IPPs via multiple imputation techniques, for the following outcomes: a) pain, b) self-reported function, and c) physical performance. We hypothesize that postoperative prognoses can be made significantly more precise through the use of IPPs, thereby aiding AHRQ efforts to develop and evaluate strategies for incorporating evidence into decision-making for patients and providers. The IPP paradigm proposed in this application represents a shift from traditional evidence-based thinking and will set a new standard for personalized monitoring following TKA.

R03HS024316

2016-04-01

2018-03-31