Development and Validation of Predictive Models for Transition from Acute to Persistent Pain After Major Surgery
Abstract
Nearly 300 million surgeries are performed across the world every year, and 10%-30% of patients develop persistent postsurgical pain (PPSP) 6-12 months after surgery. PPSP is typically defined as pain at the site of surgery lasting for more than 3 months following a surgical procedure, after the tissue injury has healed. PPSP adversely impacts quality of life, return to work and service, causes disability, and has major economic consequences. Moreover, 6%-9% of opioid-naïve patients presenting for surgery end up using chronic opioids to treat their PPSP. As the number of surgeries increase, both nationally and internationally, PPSP has become a major public health burden. Although PPSP is very common after certain procedures such as breast and lung surgeries, with ~35% reported average occurrence in each, reported incidence of PPSP varies considerably across surgeries and individual studies. Such variations are often related to a combination of differences in patient characteristics, surgical techniques and expertise, perioperative interventions, as well as methods of study design and data collection. Risk factors associated with PPSP are multifactorial, but despite substantial research, no reliable and integrated approach exists for determining the individual risk of developing PPSP. Although there are multiple predisposing factors, as well as intraoperative and postoperative events contributing to PPSP, most studies have attempted PPSP prediction based only on patients’ pre-operative demographic and phenotypic factors. Prior research efforts investigating the causal risk factors to PPSP have typically compared baseline (primarily preoperative) parameters among patients who developed PPSP versus those who did not, or constructed multivariate models to characterize the adjusted contribution of the preoperative risk factors for PPSP development. Although such approaches provide population-level information regarding the development of PPSP, they are unlikely to help in clinical decision-making for an individual patient. A key deficit in predictive models incorporating only demographic or static factors is the lack of consideration of unexpected events that can distinctly change the postoperative care trajectory of an individual patient. The transition from acute to pathological persistent pain is complex and is dependent on multiple biological, psychological, cognitive, and social factors that change across the surgical care continuum. Moving beyond a one-time baseline assessment to repetitive multifactorial measurements across the surgical care continuum is a relatively unexplored avenue that has a substantial potential to achieve a better prediction of PPSP. Such an approach affords pragmatic opportunities for ascertaining the impact of time-varying patient characteristics (e.g., events during surgery). Our central hypothesis is that advanced machine learning models that account for individual biological, cognitive, and psychological factors across the surgical care continuum will allow personalized prediction of PPSP. Towards this end we have two aims: Aim 1: Collect longitudinal prospective data for a comprehensive biological, psychological, cognitive, and psychophysical characterization of a surgical patient cohort. Multifactorial data, including perioperative events and dynamic changes in patient-reported factors and mobile device-based ecological momentary assessments (EMAs) will be captured in a cohort of 2750 patients undergoing major surgery. Follow-up data will be collected at 3 and 6 months after surgery for the assessment of PPSP and other secondary postoperative outcomes with focus on functioning, quality of life, mental health, and cognitive function. Aim 2: Develop, validate, and test advanced machine learning models for predicting PPSP. Using multi-scaled data collected from 2750 patients, we will utilize machine learning techniques that integrate perioperative risk factors to develop ind
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Dec 05, 2021
- Source ID
- W81XWH2110736
Entities
People
- Simon Haroutounian
Organizations
- United States Army
- Washington University in St. Louis