Surgeons Receiving Information About Patient Language Reflecting Unhelpful Thoughts or Distress About Their Symptoms Identify Such Language More Often Than Those Who Do Not Receive This Information.

Abstract

Unhelpful thoughts and feelings of distress regarding symptoms account for a large proportion of variation in a patient's symptom intensity and magnitude of capability. Clinicians vary in their awareness of this association, their ability to identify unhelpful thoughts or feelings of distress regarding symptoms, and the skills to help address them. These nontechnical skills are important because they can improve treatment outcomes, increase patient agency, and foster self-efficacy without diminishing patient experience. In this survey-based study, we asked: (1) Are there any factors, including exposure of surgeons to information about language reflecting unhelpful thoughts about symptoms, associated with the total number of identified instances of language rated as reflecting unhelpful thoughts or feelings of distress regarding symptoms in transcripts of patient encounters? (2) Are there any factors, including exposure of surgeons to information about language reflecting unhelpful thoughts about symptoms, associated with the interobserver reliability of a surgeon's identification of language rated as reflecting unhelpful thoughts or feelings of distress regarding symptoms in transcripts of patient encounters? Surgeons from an international collaborative consisting of mostly academic surgeons (Science of Variation Group) were invited to participate in a survey-based experiment. Among approximately 200 surgeons who participate in at least one experiment per year, 127 surgeons reviewed portions of transcripts of actual new musculoskeletal specialty encounters with English-speaking patients (who reported pain and paresthesia as primary symptoms) and were asked to identify language believed to reflect unhelpful thoughts or feelings of distress regarding symptoms. The included transcripts were selected based on the rated presence of language reflecting unhelpful thinking as assessed by four independent researchers and confirmed by the senior author. We did not study accuracy because there is no reference standard for language reflecting unhelpful thoughts or feelings of distress regarding symptoms. Observers were randomized 1:1 to receive supportive information or not regarding definitions and examples of unhelpful thoughts or feelings of distress regarding symptoms (referred to herein as "priming") once at the beginning of the survey, and were not aware that this randomization was occurring. By priming, we mean the paragraph was intended to increase awareness of and attunement to these aspects of human illness behavior immediately before participation in the experiment. Most of the participants practiced in the United States (primed: 48% [29 of 60] versus not primed: 46% [31 of 67]) or Europe (33% [20 of 60] versus 36% [24 of 67]) and specialized in hand and wrist surgery (40% [24 of 60] versus 37% [25 of 67]) or fracture surgery (35% [21 of 60] versus 28% [19 of 67]). A multivariable negative binomial regression model was constructed to seek factors associated with the total number of identified instances of language believed to reflect unhelpful thoughts or feelings of distress regarding symptoms. To determine the interobserver agreement, Fleiss kappa was calculated with bootstrapped 95% confidence intervals (resamples = 1000) and standard errors. After controlling for potential confounding factors such as location of practice, years of experience, and subspecialty, we found surgeons who were primed with supportive information and surgeons who had 11 to 20 years of experience (compared with 0 to 5 years) identified slightly more instances of language believed to reflect unhelpful thoughts or feelings of distress regarding symptoms (regression coefficient 0.15 [95% CI 0.020 to 0.28]; p = 0.02 and regression coefficient 0.19 [95% CI 0.017 to 0.37]; p = 0.03). Fracture surgeons identified slightly fewer instances than hand and wrist surgeons did (regression coefficient -0.19 [95% CI -0.35 to -0.017]; p = 0.03). There was limited agreement among surgeons in their ratings of language as indicating unhelpful thoughts or feelings of distress regarding symptoms, and priming surgeons with supportive information had no influence on reliability (kappa primed: 0.25 versus not primed: 0.22; categorically fair agreement). The observation that surgeons with brief exposure to supportive information about language associated with unhelpful thoughts and feelings of distress regarding symptoms identified slightly more instances of such language demonstrates the potential of training and practice to increase attunement to these important aspects of musculoskeletal health. The finding that supportive information did not improve reliability underlines the complexity, relative subjectivity, and imprecision of these mental health concepts. Level I, therapeutic study.