Objective. To describe the development and initial application of the Chronic Pain Coding System.
Design. Secondary analysis of data from a randomized clinical trial.
Setting. Six primary care clinics in northern California.
Subjects. Forty-five primary care visits involving 33 clinicians and 45 patients on opioids for chronic noncancer pain.
Methods. The authors developed a structured coding system to accurately and objectively characterize discussions about pain and opioids. Two coders applied the final system to visit transcripts. Intercoder agreement for major coding categories was moderate to substantial (kappa = 0.5–0.7). Mixed effects regression was used to test six hypotheses to assess preliminary construct validity.
Results. Greater baseline pain interference was associated with longer pain discussions (P = 0.007) and more patient requests for clinician action (P = 0.02) but not more frequent negative patient evaluations of pain (P = 0.15). Greater clinician-reported visit difficulty was associated with more frequent disagreements with clinician recommendations (P = 0.003) and longer discussions of opioid risks (P = 0.049) but not more frequent requests for clinician action (P = 0.11). Rates of agreement versus disagreement with patient requests and clinician recommendations were similar for opioid-related and non-opioid–related utterances.
Conclusions. This coding system appears to be a reliable and valid tool for characterizing patient-clinician communication about opioids and chronic pain during clinic visits. Objective data on how patients and clinicians discuss chronic pain and opioids are necessary to identify communication patterns and strategies for improving the quality and productivity of discussions about chronic pain that may lead to more effective pain management and reduce inappropriate opioid prescribing.