Introduction
Both preoperative and postoperative patient expectations have been shown to be correlated with better postoperative outcomes. While previous literature has focused on either domain in isolation, there may be utility in combining both to stratify patients. Although preoperative expectations are associated with postoperative met expectations based on past literature, some patients with high preoperative expectations may ultimately have low postoperative met expectations or vice versa. The purpose of this study was to determine the prognostic value of clustering knee surgery patients into groups based on preoperative and postoperative met expectations.
Methods
Patients undergoing knee surgery between June 2015 to April 2018 were prospectively enrolled. Charts were reviewed for relevant demographic/medical information. 401 patients completed both baseline and two-year follow-up questionnaires, consisting of PROMIS in six domains, IKDC, knee and whole-body Numeric Pain Scale (NPS), Marx Activity Rating Scale (MARS), and surgical satisfaction. Preoperative and postoperative met expectations were measured with Musculoskeletal Outcomes Data Evaluation and Management System (MODEMS) domains. Patients were clustered into four distinct profiles using k-means cluster analysis of preoperative and postoperative met expectations, and difference between values. Chi-square or Kruskal-Wallis tests were conducted for bivariate analyses. Multivariate analysis for relevant two-year outcome metrics identified if expectation clusters were independent predictors.
Results
Cluster analysis revealed four distinct groups: High Preoperative to High Postoperative met expectations (HIGH-HIGH), High Preoperative to Low Postoperative met expectations (HIGH-LOW), Low Preoperative to High Postoperative met expectations (LOW-HIGH), Low Preoperative to Low Postoperative met expectations (LOW-LOW). HIGH-HIGH was the largest, accounting for 62% of patients. Multiple sociodemographic/operative factors were associated with cluster profiles including age, number of prior knee surgeries, education level, income, and arthroplasty vs arthroscopy. All preoperative PROs were associated with grouping except PROMIS Physical Function. HIGH-HIGH and LOW-HIGH had the best preoperative scores, with no significant difference between the two groups. Clusters with high met expectations (HIGH-HIGH and LOW-HIGH) had better two-year scores and improvement than those with low met expectations (LOW-LOW and HIGH-LOW). HIGH-HIGH patients scored significantly better on PROMIS Physical Function and surgical satisfaction than LOW-HIGH at two years postoperatively (p=0.033 and p=0.026, respectively). HIGH-LOW patients scored significantly worse than LOW-LOW on IKDC (p=0.025) and SSQ8 (p=0.002) at two-year postoperatively. HIGH-HIGH patients improved significantly more from baseline to two years in Pain Interference (p=0.031) and IKDC (p=0.007) compared to LOW-HIGH, and LOW-LOW patients improved more than HIGH-LOW patients on Fatigue (p=0.013), MARS Lower Body (p=0.01) and NPS Whole Body (p=0.004). Clusters were independent predictors of two-year scores and change for all PROs. HIGH-HIGH cluster independently predicted the greatest two-year scores and improvements while HIGH-LOW predicted the lowest.
Conclusion
Although greater preoperative expectations have been shown to predict better outcomes, these results suggest that this is the case only if those high preoperative expectations are met postoperatively.