The study was conducted at the General Internal Medicine Clinic (GIMC), University of Washington Medical Center, Seattle, WA. Before the training, the protocol was reviewed by a focus group of 7 physicians. Twenty-six physicians were randomly assigned to the intervention group and 23 to the control group. Four intervention physicians were subsequently excluded because of their inability to participate in both training sessions. Our final pool consisted of 22 intervention and 23 control physicians. All physicians were assessed at 2 months preintervention and at 3 months postintervention.
Physician satisfaction was assessed with a modified version of the Physician Satisfaction Questionnaire,6
adapted to assess satisfaction with recent encounters with chronic pain patients. This 20-item scale contains 4 subscales: relationship quality, adequacy of data collection, appropriate use of time, and patient's cooperative nature. Physician patient centeredness was assessed with a 20-item measure7
, which has been validated for use with primary care physicians. The scale contains 4 subscales: doctor receptiveness, patient involvement, affective content of the relationship, and information giving. Three additional measures were developed by the authors: (1) frequency of opioids prescribed and patient treatment agreement completion, (2) changes in prescription practices and attitudes concerning caring for patients with chronic pain, and (3) the overall usefulness and effectiveness of the shared decision-making training.
In November 2004, intervention and control physicians were given an educational packet on the use of chronic opioids that included the following (1) an opioid conversion table, (2) an educational resource list, (3) a review article on opioid therapy for chronic pain,8
and (4) GIMC policy on the use of opiates for chronic pain (including recommendations concerning patient care agreements, depression assessment, and methadone use but no strategies or guidance). Intervention physicians attended 2-hour long shared decision-making training sessions held at GIMC.
Each session utilized a videotape of an office visit where the “patient” presented with low back pain of 1 year's duration and requested continuation of Percocet prescribed by her previous physician. The first training was divided into three 5-minute video segments with the following learning objectives: (1) to introduce the shared decision-making model, including negotiation and documentation of shared treatment goals with the patient; (2) to negotiate nonmedication treatments; and (3) to negotiate the prescription of methadone (as the long-acting opioid least prone to abuse).
The second training hour followed the patient into her next visit when she asks for more pain medication because of worsening pain. The learning objectives were as follows: (1) to review goals established in her previous visit, (2) to screen for depression while exploring declining health status, and (3) to negotiate from the patient's initial request for more pain medication to a prescription for antidepressant medication. Hypothetical third visit scenarios were then presented for discussion, portraying improving, stable, or deteriorating health status. Learning objectives for the training are listed in Table S1.
To explore the training's generalizability, it was repeated for resident and attending physicians (n=20) at the Seattle Providence Family Medicine Program. Assessments were administered immediately before and after training. There was no control group or 3-month posttraining assessment for this sample.
Differences in demographics and questionnaire responses at baseline were assessed using 2-sided t-tests on mean scores. To assess for a treatment effect, we performed ANCOVA statistical analysis on posttraining assessments for intervention and control groups, controlling for gender, training rank, and baseline scores. Analyses were performed using SPSS v12.0 (Chicago, IL).