Methods
Design
The design was a descriptive secondary analysis of patient and practitioner discussions during ambulatory medical visits. The parent study was a randomized controlled pilot study that examined the effects of a virtual pain coach intervention on subsequent elderly patient discussions of pain information (McDonald, Walsh, Vergara, & Gifford, in review).
Sample
The sample consisted of transcripts from 22 community-dwelling older adult medical ambulatory medical visits. All of the older adults were aged ≥60 years, had osteoarthritis, spoke and understood English, and had self-identified their pain level at ≥4 on a 0–10 pain intensity scale within the last month. Older adults with rheumatoid arthritis, pain from cancer, or who planned a total joint replacement within 1 month of the time of study participation were excluded.
Procedure
Institutional Review Board approval was secured for the parent study and the present study. A brief description of the parent study, a randomized controlled clinical pilot study, provides context for the current study. Immediately before a scheduled ambulatory medical visit, older adults were randomly assigned to watch a videotape teaching them important osteoarthritis pain information to discuss with their practitioner and then to practice talking with a virtual pain coach, or to watch only the videotape. The ambulatory medical visits were audiotaped and transcribed verbatim. Data reported in the present study included one practice setting and visits with five practitioners. The five practitioners that participated were all physicians.
The a priori pain content criteria used in the parent study analysis were from the American Pain Society (2002) pain guidelines and had been used in previous studies (McDonald, Gifford, & Walsh, 2013; McDonald, Shea, Rose, & Fedo, 2009). The previously coded transcripts were used in the present study to focus the secondary analysis only on sections of the conversation where clinically significant pain information was exchanged.
Krippendorff's (2004) content analysis methodology was used for the coding and analysis of data. Krippendorff defined content analysis as "a research technique for making replicable and valid inferences from texts … to the contexts of their use" (p. 18). Using this methodology, it was hoped to infer that specific elements of CAT were occurring in the conversations between practitioner and patient during pain content discussions. The steps in Krippendorff's method are unitizing, sampling, recording, reducing, inferring, and narrating.
Focus was maintained on only the segments of the ambulatory medical visit in which pain content was discussed. The units to be analyzed were any word or phrase that described a component of one of the six a priori CAT attuning strategies (communication strategies) identified for analysis that occurred during discussion of the previously coded pain information content. A priori communication strategy criteria included patient selecting the pain topic (patient select), patient taking a turn (patient turn), patient maintaining focus on the pain topic (patient focus), practitioner using an open-ended question without social desirability to start the pain discussion (practitioner open-ended start), practitioner encouraging the patient to take a turn by asking open-ended questions without social desirability (practitioner open-ended encourage), and practitioner interrupting the patient during pain content discussion (practitioner interruption). Social desirability is a type of bias that causes people to respond in a more socially acceptable way, rather than to respond accurately. For example when greeting someone and saying "How are you?" one expects the answer to be "Fine." Table 1 contains definitions and scoring of criteria for patients, and Table 2 contains definitions and scoring of criteria for practitioners.
Transcripts were examined by two trained independent raters. Disagreements were resolved through discussion and recorded to calculate reliability scores. Data for patient focus and practitioner open-ended start were coded as present (1) or absent (0). To confirm the presence or absence of practitioner interruption, the tapes were reviewed and data audibly collected. Data for the remaining items (patient select, patient turn, and practitioner open-ended encourage) were each summed for a total number present in each ambulatory medical visit. Length of ambulatory medical visit was timed as the tapes were listened to and included only the amount of time when the practitioner and patient were in the room together to accurately record the total patient-practitioner communication during the ambulatory medical visit.
The final steps in Krippendorff's (2004) technique are reducing, inferring, and narrating. In the present study, results from coding were summed for each transcript and entered into SPSS statistical software.
Analysis
Frequencies, means, and standard deviations were conducted to describe older adult characteristics for age, sex, ethnicity, marital status, and education. The parent study consisted of two experimental conditions: A) a pain communication education plus coaching intervention to help older adults practice discussing their pain; and B) a pain communication education video only. The participant's group membership (A or B) and the content analysis items were statistically examined to look for large differences using cross-tabulation analysis with the chi-square statistic on the items patient maintaining focus and practitioner open-ended start, even though small group differences would not be detectable in our small sample. These tests revealed no significant differences between the two groups of participants. The remaining content analysis items and length of ambulatory medical visit were examined between the two groups (A or B) using independent t tests and were also found to have no large differences based on group membership. For the sake of the descriptive analysis, the groups were therefore combined for the content analysis. Interrater reliability was calculated using percentage of agreement and the more conservative Krippendorff alpha. Results from the CAT strategy content analysis were summarized with frequencies.