Methods
The study design and protocol are described in detail elsewhere. Briefly, the study used a 2-group pragmatic pilot randomized-controlled–trial design; 13 clinicians were randomized to receive the intervention; patients (n = 319) were the primary unit of analysis. For practical and logistical reasons, the clinicians were randomized so that Group 1 clinicians (n = 6) participated in the intervention first; Group 2 (n = 7) participated in the intervention 8 months later. The intervention consisted of a clinician training program designed to increase use of the 5As for physical activity counseling with patients. Outcomes were patient and clinician assessments of the effectiveness of the intervention in increasing use of the 5As for physical activity counseling. Other than the timing, there was no difference in the intervention or the assessment for the 2 groups. The University of Rochester Review Board approved the study protocol. The intervention period started in June 2009 and ended with the final follow-up data collection in October 2011.
Recruitment, Enrollment, Randomization of Clinicians and Patients
Clinicians and patients were recruited through 2 federally qualified health centers in Rochester, New York, serving a predominantly low-income, racially/ethnically diverse population of 14,000 patients. Details on inclusion and exclusion criteria for participants are described elsewhere. Clinicians were eligible if they practiced at one of the aforementioned centers. Clinicians were recruited, enrolled, and randomized to Group 1 or 2 from January through March 2009 before patient recruitment began. Of 15 clinicians eligible, 13 consented to participate (9 family physicians, 2 family nurse practitioners, and 2 physician assistants) and were randomized. Of the 13 clinicians randomized, 2 clinicians relocated and one retired, leaving 10 for analysis (5 in Group 1 and 5 in Group 2). Neither clinicians nor research staff was blinded to the assignment of clinicians. Clinicians were paid up to $365 for participation in all study activities.
Patients were eligible if they were currently enrolled at one of the health centers, were aged 18 years or older, were able to provide informed consent, and made a health maintenance visit or follow-up visit for a chronic condition for which physical activity counseling would be appropriate. Nurse assistants mentioned the study to the patients when they were in the examination room. If the patient expressed interest, a research assistant entered the room to provide additional details about the study; patients who agreed to participate signed an informed consent form. For ethical and regulatory reasons, patients were not blinded to study topic and objective; as part of the informed consent process, they were made aware of these. However, they were blinded to time point (baseline, postintervention, 6-month follow-up) and did not know whether or when their clinicians took part in the intervention; patient participants were paid $20 for participation.
During the study period, 1,029 patient participants were assessed for eligibility. Of these, 481 (or 46.7% of those assessed) were ineligible. Among the remaining 548 patients, 325 consented to participate (Figure). Of the 325 patients, 319 completed surveys for analysis.
(Enlarge Image)
Figure.
Assessment of patient participant eligibility and reasons for exclusions from study. Within each group, patients at baseline, postintervention, and 6-month follow-up were assessed independently of one another (ie, patients participating at baseline did not necessarily participate at postintervention or 6-months postintervention). [A text description of this figure is also available.]
Intervention
The intervention consisted of 4 one-hour training sessions for clinicians at the health center sites. Clinicians were trained through techniques known to be effective for communication training: didactic materials, skills and competency checklists, role-play, and cognitive rehearsal. Training topics consisted of introduction to the 5As, current recommended guidelines for physical activity, techniques to elicit patient motivation, strategies to problem-solve, tips on documentation in electronic health records, and a list of community resources for physical activity. As part of the Assist and Arrange steps, clinicians were taught to refer patients to a community exercise program (the Healthy Living Program), which had partnered with the federally qualified health center. The training period was interactive, and at its conclusion, each clinician received a competency checklist noting their accomplishment of the 5As completed by a standardized patient (a person who realistically portrays a patient with a certain health condition). No booster sessions or other follow-up training activities were provided. All training sessions were audio-recorded. Two students listened to the recordings and completed a checklist of predefined training goals and activities to ensure fidelity.
Patient-reported Outcome Measures
For each clinician, patients completed a survey at 3 points: baseline, immediately postintervention, and at 6-month follow-up. Each patient completed 1 survey, so different groups of patients provided surveys at each point. Patients completed the survey, a modified version of the Physical Activity Exit Interview (PAEI), after their office visit. The PAEI is a validated 12-item measure asking patients to report clinician use of 5As for physical activity counseling. Patients dichotomously (yes/no) answered questions such as, "Did your doctor advise you to become more physically active?"; "Did your doctor discuss difficult situations you might encounter or problems you might have in trying to become more physically active?"; and "Did you and your doctor put the plan to become more physically active in writing?" The PAEI was modified for this intervention by adding 3 questions: "Has your provider ever asked you about your confidence to change your exercise habits?"; "Has your provider ever asked you about your willingness to change your exercise habits?"; and "Has your provider ever referred you to other programs, resources, consultants, etc., to help you with physical activity?" The score range for the modified PAEI was 0 to 15 (ie, 1 point for each yes). The patient survey also asked about sociodemographic characteristics, barriers to physical activity, and support and resources for physical activity.
Clinician Baseline and Follow-up Assessments
Each clinician completed baseline and 6-month follow-up surveys. The surveys asked about demographic characteristics, current practice of physical activity counseling, confidence in using 5As in counseling on physical activity, and knowledge of resources that could meet their patients' needs. Their confidence in using counseling skills was self-rated on a 5-point Likert scale (1 = not confident, 5 = very confident). At follow-up, each clinician also completed a 30-minute interview to provide feedback about the overall experience in the intervention, changes in skills, problems or difficulties, training techniques, satisfaction with the intervention, and suggestions for improvement.
Statistical Analysis
All analyses were based on a significance level of .05. Two-sample t tests for continuous variables and χ tests for categorical variables were used to evaluate the success of the randomization in balancing baseline covariates between groups. From the power and sample-size analysis, given our sample size, we could detect a difference of 1.5 or greater in PAEI scores between baseline and follow-up with 80% power. Data were analyzed with generalized estimating equations (GEE) models with patients nested within clinician using a negative-binomial link to account for the distribution of PAEI scores. We chose to use GEE models to account for nesting of patients within clinician and a possible unknown correlation between outcomes and to obtain robust standard errors. Also, GEE models provide a practical method for analyzing conditional responses that may depart from the normality assumption required for general linear models. Finally, we determined effect sizes and categorized them as small, medium, or large. For the qualitative analysis of interviews, we conducted interviews of patients and clinicians after the patient visit. All interviews were audio-recorded, transcribed, coded, and analyzed by using a constant comparative analysis technique.