Development of a collaborative model of low back pain: report from the 2017 NASS consensus meeting

Low back pain (LBP) is a multifactorial problem with complex interactions among many biological, psychological and social factors. It is difficult to fully appreciate this complexity because the knowledge necessary to do so is distributed over many areas of expertise that span the biopsychosocial domains.

This study describes the collaborative modeling process, undertaken among a group of participants with diverse expertise in LBP, to build a model to enhance understanding and communicate the complexity of the LBP problem.

The study involved generating individual models that represented participants' understanding of the LBP problem using fuzzy cognitive mapping (FCM), and 4 subsequent phases of consultation and consensus with the participants to characterize and refine the interpretation of the FCMs.

The phases consisted of: proposal of Categories for clustering of model Components; preliminary evaluation of structure, composition and focal areas of participant's FCMs; refinement of Categories and Components with consensus meeting; generation of final structure and composition of individual participant's FCMs. Descriptive statistics were applied to the structural and composition metrics of individual FCMs to aid interpretation.

From 38 invited contributors, 29 (76%) agreed to participate. They represented 9 disciplines and 8 countries. Participants' models included 729 Components, with an average of 25 (SD = 7) per model. After the final FCM refinement process (Components from separate FCMs that used similar terms were combined, and Components from an FCM that included multiple terms were separated), there were 147 Components allocated to ten Categories. Although individual models varied in their structure and composition, a common opinion emerged that psychological factors are particularly important in the presentation of LBP. Collectively, Components allocated to the "Psychology" Category were the most central in almost half (14/29) of the individual models.

The collaborative modeling process outlined in this paper provides a foundation upon which to build a greater understanding and to communicate the complexity of the LBP problem. The next step is to aggregate individual FCMs into a metamodel and begin disentangling the interactions among its Components. This will lead to an improved understanding of the complexity of LBP, and hopefully to improved outcomes for those suffering from this condition.