Multidimensional Perioperative Recovery Trajectories in a Mixed Surgical Cohort: A Longitudinal Cluster Analysis Utilizing National Institutes of Health Patient-Reported Outcome Measurement Information System Measures.

Background Pain trajectories have been described in numerous surgical settings where preoperative characteristics have been used to predict trajectory membership. Suboptimal pain intensity trajectories have been linked to poor longitudinal outcomes. However, numerous biopsychosocial modulators of postoperative pain may also have distinct longitudinal trajectories that may inform additional targets to improve postoperative recovery. Methods Patients undergoing total joint arthroplasty, thoracic surgery, spine surgery, major abdominal surgery, or mastectomy completed Patient Reported Outcome Measurement Information System (PROMIS) measures and additional scales preoperatively and at 1 week, 2 weeks, 1 month, 3 months, and 6 months postoperatively. A k-means clustering for longitudinal data was utilized to explore and describe distinct pain impact (PROMIS Pain Interference and Physical Function) trajectories and associated changes in additional biopsychosocial measures. Follow-up analyses examined participant demographics and clinical characteristics associated with trajectory memberships. Results Three postoperative biopsychosocial symptom clusters were identified across all patients (n = 402): low (35%), average (47%), and high (18%) performance cluster trajectories. Participants undergoing total knee arthroplasty (TKA), spinal surgery, reporting presurgical opioid use, and higher pain catastrophizing scale scores were found to be associated with the low performance trajectory. Patients within the low performance trajectory, while demonstrating small improvements by 6 months, remained mild to moderately impaired in both pain impact and physical health outcomes. Alternatively, participants in the average performance trajectory demonstrated improvement in pain impact to population norms compared to baseline and demonstrated continued improvement across physical and psychological outcomes. Patients within the high performance cluster started within population norms across all measures at baseline and returned to baseline or exceeded baseline values by 6 months postoperatively. Self-reported opioid utilization was significantly higher in the low performance cluster across all time points. While a larger proportion of average performance patients reported opioid utilization during the first postoperative month compared to the high performance cluster, no differences were detected at 6 months postoperatively between these 2 clusters. Conclusions These pain impact trajectories build upon previous unidimensional pain intensity trajectories and suggest that additional distinct biopsychosocial measures may have unique trajectories related to cluster assignment. Additionally, these findings highlight the importance of continued pain impact surveillance through the perioperative recovery period to detect patients at risk of experiencing a poor trajectory and subsequently poor longitudinal health outcomes.