Effects of psychological distress on the general health to self-reported pain and function outcome relationship in knee arthroplasty: A causal mediation study

Abstract

Objectives: We examined two potential causal pathways that could be intervention targets to enhance knee arthroplasty outcomes. Data from a no-effect trial of persons with moderate to high pain catastrophizing were used to determined whether pain catastrophizing, depressive symptoms causally mediate the effect of preoperative general health on postoperative knee pain and functional difficulty.

Methods: We used natural-effects models to conduct causal mediation analyses using the preoperative dichotomized EQ-5D-5L general health measure as the exposure, 2-month postoperative pain catastrophizing, depressive symptoms, and localized knee pain as potential mediators, and 12-month dichotomized Western Ontario and McMaster's University Osteoarthritis Index (WOMAC) Pain and Function scores reflecting good versus poor outcome as the outcomes.

Results: Estimates of the indirect (mediating) effect suggested that pain catastrophizing mediated the effect of preoperative general health on 12-month WOMAC pain score by increasing odds of a good outcome by 8% (natural indirect effect odds ratio ​= ​1.08, 95% CI: 0.88, 1.29). The direction of mediating effects and their magnitude were similar for depressive symptoms; Sensitivity analyses suggested similar magnitudes and mediating effects to those reported for the main analyses.

Conclusions: Our findings suggested that pain catastrophizing and depressive symptoms have a mediating role on the effect of baseline general health on self-reported pain and function outcomes. These findings support the continued treatment of pain catastrophizing and depressive symptoms as viable targets for interventions to potentially enhance pain and function outcomes for patients with moderate to high levels of psychological distress prior to surgery.

Keywords: Arthroplasty; Knee; Mediation; Outcome.

Fig. 1

Directed acyclic graph representing the natural indirect effect, mediated by 2-month postoperative pain catastrophizing, depressive symptoms and localized knee pain severity; and the natural direct non-mediated effect of general health on 12-month WOMAC Pain or Function (pathway not passing through the mediators). Potential confounders and their association with exposure, mediators and outcomes are included. A directed acyclic graph denotes hypothesized causal associations via solid directional arrows. The arrow from the exposure to the outcomes denotes a direct causal path and the arrow from the exposure to the mediator and a subsequent arrow from the mediator to the outcome indicates an indirect causal path in the figure. These solid arrows form a causal loop from exposure to outcome and from exposure to mediators to outcome.