Factors That Influence Changes to Existing Chronic Pain Management Plans

Julie Diiulio, MS; Laura G. Militello, MA; Barbara T. Andraka-Christou, JD, PhD; Robert L. Cook, MD, MPH; Robert W. Hurley, MD, PhD; Sarah M. Downs, MPH; Shilo Anders, PhD; Burke W. Mamlin, MD; Elizabeth C. Danielson, MA; Christopher A. Harle, PhD


J Am Board Fam Med. 2020;33(1):42-50. 

In This Article


This method section describes a qualitative study in which a sample of 20 PCCs took part in a series of interviews about 1 or more of their patients with chronic pain. Patients did not participate in the interviews, but they were asked to consent to the use of their records during the interviews. The research team performed a thematic analysis of the interview transcripts, with a focus on the events that led to changes in pain management plans. The study was approved by the Indiana University Institutional Review Board.


Our sample was purposive with the aim to recruit PCCs who were able to prescribe opioids and had patients with chronic noncancer pain. We sought to recruit PCCs from urban, suburban, and rural health clinics within different health systems in Indiana. Recruitment took place between April 2016 and July 2018. We initially obtained resources to conduct up to 5 interviews with each of 25 PCCs for a total of 125 interviews. Analysis began as interview transcripts became available. The research team observed thematic saturation and stopped data collection after analyzing 89 interviews with 20 PCCs. Some PCCs participated in fewer than 5 interviews due to limited availability to participate or because thematic saturation had been reached and data collection stopped. A member of the study team, a public health researcher, recruited the participants through word-of-mouth, e-mail invitations, and in-person presentations at clinic staff meetings. Participants were given a $100 gift card for each interview.

To identify the patients who would be discussed in the interviews, we worked with the PCCs. Typically, an interview would be scheduled, and then the provider would look at their patient schedule to find a patient who fit the eligibility criteria and had an upcoming appointment within 3 days before the scheduled interview. Patients were eligible if they were over 18 years, had chronic pain, and had no history of cancer within the past 3 years. To obtain patient permission to be used as the subject of the interview, a member of the research team met with eligible patients immediately before their appointments, confirmed eligibility, and invited the patients to participate in the study. Patients who agreed to be the subject of the PCC interview provided written informed consent and received a $25 gift card.


Each interview focused on a unique patient with chronic noncancer pain. Interviews lasted approximately 60 minutes and took place in the PCCs' clinics. The interviewers included public health researchers and cognitive scientists from the research team. Three interviewers were present for the first 6 interviews. This was done to ensure common ground across the team and to facilitate refining the interview guide. The remaining interviews were conducted by 1 or 2 interviewers. The interviews occurred within 3 days following the patient appointment to facilitate accurate recall. PCCs had access to the patient's electronic medical record during the interview as a memory aid. All interviews were audio recorded and transcribed. Resulting transcriptions were deidentified.

At the initial interview, PCCs completed a demographic questionnaire and answered questions about their use of tools and general approach to chronic pain care. For the remainder of the first interview, and all subsequent interviews, we used an adapted Critical Decision Method (CDM) technique[22] to explore the major events that occurred over the course of treating the patient's chronic pain. CDM is an approach to cognitive task analysis that is aimed at understanding cognitive aspects of work, specifically tacit aspects of expertise. During the interview, PCCs were encouraged to describe key information that influenced their decision making for each major event. Questions related to changes in the patient's condition, treatment plan, and goals (eg, What new information did you get during the visit that impacted your view of the patient or treatment plan?).


We used an inductive thematic analysis approach, wherein we focused on exploring differences in interpretation of the data, and then built consensus.[23] First, a multidisciplinary team (3 public health researchers, 2 cognitive scientists, 2 PCCs, and 1 pain specialist physician) reviewed 2 transcripts and highlighted themes of interest. The topics of interest were used to create a preliminary codebook. The codes were then applied to other transcripts, with codes modified and added inductively overtime if additional topics of interest emerged in the data. The final codebook included 55 codes. After the codebook was finalized, remaining transcripts were coded by 2 members of the team who met regularly to reach coding consensus. Agreement involved a negotiation process during which discrepancies were discussed and a final code was applied. The code, "change in pain management plan," is the subject of this article. We labeled transcript excerpts with this code if they included any change that occurred in the PCC's plan to manage the patient's pain. We chose to focus on this code because less is understood about the factors that influence changes to existing pain management plans as opposed to other clinical decisions in pain care, such as the decision to initiate an opioid prescription.

The code, "change in pain management plan," was applied to 217 transcript excerpts. For each excerpt, we identified the event or information that triggered the change and the resulting clinical action that was taken (eg, medication change, referral, etc.). Excerpts that did not have a clear precipitating event or resulting action were removed from the data set. We also removed excerpts that did not involve a change made by the interviewee. For example, some PCCs described actions taken by other providers. These excerpts were excluded because they did not reflect the interviewee's own decision making. In sum, 86 event and action pairs were identified. The event and action pairs were categorized by theme, resulting in 7 themes, as described in the results. The themes that emerged were vetted by the larger research team before finalization.