Probing the mechanisms underpinning recovery in post‐surgical patients with cervical radiculopathy using Bayesian networks

Abstract

Eur J Pain. 2020;24:909-920. wileyonlinelibrary.com/journal/ejp | 909 Abstract Background: Rehabilitation approaches should be based on an understanding of the mechanisms underpinning functional recovery. Yet, the mediators that drive an improvement in post-surgical pain-related disability in individuals with cervical radiculopathy (CR) are unknown. The aim of the present study is to use Bayesian networks (BN) to learn the probabilistic relationships between physical and psychological factors, and pain-related disability in CR. Methods: We analysed a prospective cohort dataset of 201 post-surgical individuals withCR. In all, 15 variables were used to build a BN model: age, sex, neck muscle endurance, neck range of motion, neck proprioception, hand grip strength, self-efficacy, catastrophizing, depression, somatic perception, arm pain intensity, neck pain intensity and disability. Results: A one point increase in a change of self-efficacy at 6 months was associated with a 0.09 point decrease in a change in disability at 12 months (t = −64.09, p < .001). Two pathways led to a change in disability: a direct path leading from a change in self-efficacy at 6 months to disability, and an indirect path which was mediated by neck and arm pain intensity changes at 6 and 12 months. Conclusions: This is the first study to apply BN modelling to understand the mechanisms of recovery in post-surgical individuals with CR. Improvements in painrelated disability was directly and indirectly driven by changes in self-efficacy levels. |

Movement, posture and low back pain. How do they relate? A replicated single‐case design in 12 people with persistent, disabling low back pain

“…Psychological factors (such as pain catastrophising, fear and pain self-efficacy) have been shown to relate more to improved pain or activity limitation than physical parameters such as movement or abdominal muscle function (Mannion, Caporaso, Pulkovski, & Sprott, 2012;Mannion et al, 2001;Nordstoga, Meisingset, Vasseljen, Nilsen, & Unsgaard-Tondel, 2019). Psychological factors have also been shown to influence the embodiment of cautious and protective movement behaviours (Matheve et al, 2019;Olugbade, Bianchi-Berthouze, & Williams, 2019;Osumi et al, 2019) and mediate improvement (Lee et al, 2017;Liew et al, 2020;Mansell, Kamper, & Kent, 2013;Smeets, Vlaeyen, Kester, & Knottnerus, 2006). It may be that threat reduction, following the safe completion of previously painful, feared, or avoided activities perceived as dangerous or damaging, led to clinical improvement, irrespective of whether this was related to changes in movement or posture (Mannion et al, 2012;Steiger et al, 2012).…”

Inferring skin–brain–skin connections from infodemiology data using dynamic Bayesian networks

“…The implication is that each variable depends in probability on the variables that are its parents in the DAG: as a result, the joint multivariate distribution of the data factorises into a set of univariate distributions associated with the individual variables. This property allows automatic and computationally efficient inference and learning of BNs from data and has made them popular for analysing clinical data 79 – 81 . In particular, BN inference can automatically validate arbitrary hypotheses: in the simplest instance, a BN is used as a generative model, and hypotheses are validated by stochastic simulation.…”

Inferring Skin-Brain-Skin Connections from Infodemiology Data using Dynamic Bayesian Networks

“…The implication is that each variable depends in probability on the variables that are its parents in the DAG: as a result, the joint multivariate distribution of the data factorises into a set of univariate distributions associated with the individual variables. This property allows automatic and computationally-efficient inference and learning of BNs from data and has made them popular to analyse clinical data [61,62,63]. In particular, inference makes it possible to use BNs to automatically validate arbitrary hypotheses involving the variables it is modelling: in the simplest instance, a BNs is used as a generative model and hypotheses are validated by stochastic simulation.…”