In fall 2003, 347 individuals enrolled in an RCT of PACE in 18 urban and rural community sites across North Carolina. To be eligible, participants had to be 18 years or older, exercise fewer than three times per week, and have any type of self-reported arthritis or joint pain with moderate to severe limitation in joint motion, strength, or both. Individuals exercising 3 or more days per week for 20 minutes or more each day were excluded. Arthritis or joint pain and physical limitations were assessed during the enrollment process using the short-version Health Assessment Questionnaire Disability Index (HAQ-DI) (26). Participants were also asked to report on pain, fatigue, and stiffness using visual analog scales based on the Multidimensional HAQ scale (27).

The intervention group included 168 randomly assigned participants who received the basic-level PACE class in the fall of 2003. Control subjects (n = 155) received a delayed treatment, participating in PACE classes after the initial 8-week intervention was completed. Twenty-four participants were not randomized because of transportation and other personal reasons. The class met two times a week for 8 weeks. For the purpose of the qualitative evaluation, participants in the intervention arm were classified into two groups based on class attendance: completers and noncompleters. Completers were participants who attended 75% or more of all classes, and noncompleters were participants who attended fewer than 75% of all classes.

A purposive sample comprising two completers and one noncompleter randomly chosen from each of the 18 PACE sites was selected for telephone interviews, with a sampling goal of 54 interviews (36 completers and 18 noncompleters). Fifty-one participants were actually interviewed (36 completers and 15 noncompleters). Completers were oversampled to obtain more information from participants who had the most exposure to the course. Table 1 details the health status and demographic characteristics of these interviewed participants.

All interviews were conducted by one of the researchers (LD) between October 2003 and February 2004 within a month after the participant completed the 8-week PACE exercise class. Participants were first called at home during the weekday, and if they could not be reached, they were contacted in the evening.

A semistructured interview guide was developed to elicit participant views on the factors that motivated them to attend PACE exercise classes and the barriers that prevented them from attending classes, including course content, the PACE instructor, and the class location and schedule. (See the Appendix for sample interview questions.) The order of the interview questions varied slightly depending on how the conversation developed during each interview; participants were only asked questions about the topics that they had not covered in their responses to earlier questions during the course of the interview. Interviews conducted with the noncompleters tended to be shorter than those conducted with the completers because the noncompleters had participated in significantly fewer classes and so could not comment as extensively. As certain themes arose, the researcher probed to explore these themes further and recorded notes of her impressions after each interview.

Interviews lasted an average of 17 minutes (range, 4–38 minutes). Digital audio files of the interviews were saved under the participant identification number to ensure confidentiality. The University of North Carolina School of Medicine Institutional Review Board approved all methods.

In this study, the information-motivation-behavioral skills (IMB) model was used to provide the conceptual framework for analyzing the factors that lead to exercise behavior. The model's major components were exercise information, exercise motivation, and exercise behavioral skills. The IMB model was originally developed in 1992 to predict HIV-preventive behavior; however, its concepts can be broadly applied to predict positive health behavior change in a range of contexts, such as exercise behavior (28).

Two of the researchers (LD and BS) conducted the analyses, beginning with a verbatim transcription of the interviews. Once all data were transcribed, a random sample of 10 (20%) of the total transcripts was reviewed for accuracy and completeness by a member of the research team who had not been involved in the data collection process.

The data were then analyzed using NUD*IST (N6) (QSR International, Melbourne, Australia), a software program for qualitative data analysis. The initial categories were developed deductively based on the broad topics of each interview question, which measured the IMB model concepts. Subcategories were added to each of these initial base categories to further organize participant responses. After the initial round of coding was completed, the researchers reread the transcripts and condensed the list of deductive codes, retaining only those that occurred most frequently across all interviews.

The method of constant comparison was used to develop higher level themes (29). Transcripts were reread, and a series of inductive codes was created based on emerging themes. When new themes arose, all researchers were consulted to ensure consistent coding of the transcripts. Each time a new theme emerged, all transcripts were reread and recoded according to the new understandings. Key phrases used by the participants in the interviews were retained to name some of the inductive codes. The codes were eventually reduced and refined into key themes informed by the concepts from the IMB model (Table 2).

Baseline demographic and health status characteristics of the 51 qualitative participants were examined. Two-sample t tests for continuous variables and the Fisher exact test for categorical variables were used to assess differences between the completers and noncompleters.

Keep moving. Participants described the overall information they received at the PACE exercise class as helpful and felt certain that even if it was not helpful to them, that it was helpful to other class members. Participants predominately talked about the importance of keeping their bodies in motion throughout the day. They perceived this as an effective strategy for staying active and preventing or reducing arthritis pain and stiffness. Many participants spoke of the informational brochures about arthritis and physical activity that they used to guide their exercise at home. They also used information on breathing and relaxation exercises to manage pain and stress:

Practical information. Both completers and noncompleters regarded learning to move safely as an important topic. Participants' examples of safe movement included learning how to get up from a fall and transferring from one position to another. Many participants reported that these skills helped to quell their uncertainties about their arthritis and take control of their health by preparing for the future. Some participants spoke of wanting to maintain their independence as long as possible, so staying active and learning to move safely was essential. Learning such skills can enhance participants' confidence that they will be able to cope with and adjust to their arthritis. In turn, this positive attitude may also increase their self-esteem and sense of efficacy about being able to do what they want to do independently:

Class social support.A major motivator for both completers and noncompleters participating in the PACE class was social support received from other class participants and instructors. The classes provided supportive environments for the participants. Many participants mentioned that simply knowing that each week the group was expecting them to show up for class, as well as having a structured time set aside for exercise, was significant motivation to attend class:

The group structure motivated participants not only to attend class but also to challenge themselves to move their bodies in ways they may not have if they had been exercising on their own:

Participants also valued being able to exercise in a group with others who lived with arthritis. Many commented about this aspect of the class, seeing it as an opportunity to interact and empathize with other people who could truly relate to them. It was also a time for sharing practical information, such as recommendations for a rheumatologist or arthritis-appropriate devices to use in the kitchen:

Instructor support. The instructors' personality characteristics were important factors in participants' perceived sense of support. Above and beyond all other topics, completers and noncompleters talked about how much they liked their instructors. When describing their instructors' personalities, they used words such as "nice," "patient," "friendly," and "polite." Participants' high regard for their instructors increased their desire to attend the class and helped them feel safe engaging in the recommended class exercises:

Participants frequently mentioned empathy as an important characteristic of a supportive instructor. In this context, empathy connotes the ability of the instructor to truly understand what it feels like to live with arthritis:

Participants also frequently described instructor behaviors that they found supportive. These included paying personal attention to class members, skillfully demonstrating class exercises, looking up answers to participants' questions outside of class, and competently understanding and suggesting appropriate exercises for arthritis. Nearly all participants, both completers and noncompleters, talked about how the instructors paid personal attention to the class members, learning their names and calling them at home to check on them if they missed a class.

Participants appreciated being able to trust that their instructors would know what exercises were safe for them to perform. They liked being able to ask the instructors for modification suggestions when an exercise proved too difficult or uncomfortable. If a participant found that a particular exercise was too difficult or too painful to perform, the instructor would suggest modifications, such as assuming a sitting rather than a standing position:

Participants also appreciated when the instructors demonstrated the exercises and performed them along with the class:

Completers and noncompleters generally reported only positive things about their instructors. The few negative comments included a participant feeling that she knew more about arthritis than her instructor and that her instructor did not do a good job of pacing the class. Another participant did not like that her instructor consulted the PACE book of exercises while performing the exercises with the group, perceiving the instructor as ill-prepared to teach the class. However, another participant in the same class liked this teaching method, reporting that it made him feel like they "were all learning to exercise together." Lastly, one participant complained that she felt the class was more about socializing than exercising and suggested that the instructor stick to a tighter exercise schedule. These comments were made equally by completers and noncompleters.

I could do it. Knowing that they could work at their own pace and modify the activities as needed, coupled with the strong sense of trust they had in their instructors, participants felt comfortable engaging in the exercises and trying things they had believed they could not do, both at home and in class. This increased their sense of self-efficacy, allowing participants with diverse functional capacities to participate in the class, tailor the level of difficulty according to their individual needs, and incorporate these exercise skills into their daily lives:

At your own pace. Instructors encouraged participants to work at their own pace, reminding them that exercise does not have to be aerobic and fast. The ability to exercise at one's own pace was one of the most frequently reported themes and was referred to by both the completers and the noncompleters:

Completers and noncompleters differed in reporting barriers to participating in PACE.

Personal illness. A major factor affecting participants' motivation to participate in PACE was personal illness. Noncompleters often reported missing class because of arthritis-pain–related illness, whereas completers missed because of illness in general, such as being sick with the flu. Noncompleters explained that they could not predict how they would be feeling the day of the class; if their arthritis were to "flare up," it would make it nearly impossible for them to drive to class, and it would be uncomfortable to exercise. Only one noncompleter reported that exercising actually made his bodily pain worse.

Class complaints. Some noncompleters complained that their classes were not challenging enough. A few saw themselves as either significantly younger, more fit, or both, making it hard to relate to the other class participants:

A few completers also commented on lack of class challenge and therefore should have been advised to enroll in the advanced-level PACE class.

Practice at home. Participant reports support the IMB model predictions that the availability of exercise information and increased motivation to exercise affect exercise behavior either directly or indirectly through the acquisition of exercise skills. Both completers and noncompleters reported practicing the PACE exercises at home. One participant could not attend the class because she had to leave to care for an ill family member. Although she did not attend class, she used the pamphlets given to her by her instructor to guide her exercises. In this case, appropriate exercise information led directly to engagement in exercise behavior.

Continue to exercise over time. Completers and noncompleters who reported practicing the PACE exercises at home also reported that they continued to exercise after the PACE class had ended. In the class, they had developed the skills and confidence they needed to safely engage in an exercise routine at home. Noticing an improvement in level of pain and stiffness as a result of the exercise class encouraged participants to continue to exercise.