The inclusion criteria for participants were an age of ≥ 18 years, voluntary participation, and the use of CBM for a period of at least 6 months due to CNCP. In Germany, CBM is reimbursed by health insurance providers only if patients are severely impaired by their pain condition and other therapeutic approaches have failed. This means that, in addition to certain specific symptoms and syndromes (Table 2), all patients suffered from long-lasting chronic pain associated with the ICD-10 diagnoses R52.2 or F45.41, which involve nociplastic components of pain. Further information about our cohort is provided in Table 1, and Table S1 in the Supplement. As a rule, the CBM dose is titrated over several weeks (Bashkar et al. 2021). Sometimes it is necessary to switch to a different type of CBM. To ensure that patients had reached a stable condition and were able to provide a meaningful assessment of their treatment, we therefore decided to determine their eligibility for the study only after they had been taking CBM for at least six months.

The design of our exploration was a prospective cross-sectional study with a mixed methods approach, consisting of a qualitative part (partially structured patient interviews) and a quantitative questionnaire on medication satisfaction (German TSQM, version 1.4).

We decided on a qualitative approach since, as previously discussed, there is a gap between patient reported outcomes and systematic evidence, and we considered that a quantitative approach may not be suited (Foley and Timonen 2015) to best assess the influence of CBM on sleep quality, mood and life quality, which arguably contribute significantly to a positive overall experience for patients. We used grounded theory as an analytical method for the qualitative findings, which is an inductive process and generates hypotheses from the data, as opposed to deductive techniques, which are traditional in quantitative research and analyze data to refute or confirm hypotheses (Chapman et al. 2015). The grounded theory method was developed by Barney Glaser and Anselm Strauss in 1967, and there are three different approaches that are commonly used (Singh and Estefan 2018). The study team chose Strauss and Corbin’s perspective because it provides a clear and systematic analysis (Singh and Estefan 2018). The technique of Corbin and Strauss contains three steps: open coding, axial coding and selective coding (Singh and Estefan 2018). During open coding the interviews are read, and any meaningful text passage is labelled with a code. A code is a keyword or a catchphrase that summarizes the content without interpretation. During the analytic process text passages from different interviews are compared to each other and are given the same code, if applicable (Singh and Estefan 2018). In our study we used the four-step coding tool to sort codes into (sub-)concepts, categories and themes to present the data in a structured way (Qureshi and Ünlü 2020). In the axial coding data are reassembled to identify relationships between categories and concepts forming a framework called paradigm model. In the selective coding the insights gained from the paradigm model are incorporated into a coherent descriptive presentation about the overarching phenomenon. The goal is to develop a comprehensive theory, that still acknowledges differences in the experience (Singh and Estefan 2018).

Patients from the Pain Clinic who got prescribed CBM for at least 6 months were recruited following the principles of purposive and convenience sampling and data were collected until saturation was achieved (Moser et al. 2018). To ensure the best possible quality of evidence conducting and reporting the qualitative interviews, we used the 32-item Consolidated Criteria for Reporting Qualitative Research (COREQ)-checklist (Tong et al. 2007).

For the development of an interview guideline, the study team analyzed existent literature and decided on partially structured interviews which were designed based on the framework proposed by Kallio and colleagues (Kallio et al. 2016). Pharmacological, psychological and social factors were deemed important and the study team determined which topics should be covered in the interview: the circumstances of therapy beginning, related expectations, occurred changes since therapy beginning, the effect of CBM and its effect on everyday life as well as QOL. When defining the order of questions, the study team followed the recommendations of Moser and colleagues and decided on primarily asking “what” and “why” questions, followed by “how” questions for deepening the conversation (Moser et al. 2018).

The researchers determined a logical order of questions, starting with the beginning of therapy and moving on to the effects of CBM during the course of the interview. The question about the effect of CBM on pain was posed towards the end, as to not solely focus the questioning on this topic. It was decided that the order of the questions could be altered in the interviews, to allow an open dialog instead of a strict question-answer-catalogue (Moser et al. 2018). The team then formulated follow-up questions and key points and added them to the interview guideline. It was also possible for the interviewer to adapt the questions according to the individual situation through probes and prompts as well as verbal and non-verbal signs (Moser et al. 2018). After internal testing, the final interview guide was used with participants 1–25 and following an interim evaluation by the study team, some changes were made to the guide addressing additional questions about experiences of relaxation methods and psychotherapy. The updated version was used with participants 26–32. Both versions of the interview guideline can be found in the supplemental data.

This study was registered with the German Clinical Trials Register (DRKS00037434). After approval by the local ethics committee (No. 8391_B0-K_2019) interviews took place between September 2020 to March 2021. Prior to the interviews, written consent was obtained from the patients; patients were free to withdraw from the study at any time or refuse to answer questions if they were uncomfortable. Twenty-three interviews were conducted at the pain clinic at Hannover Medical School, one interview took place at the work venue from one participant, and eight patients were interviewed by telephone.

All interviews were recorded on tape, without systematically documenting nonverbal behavior (e.g., through video recordings), transcribed verbatim, and anonymized; the transcripts were not corrected or commented by the patients. After transcription, the transcripts were reviewed sentence by sentence using MAXDA^® and analyzed for the literal meaning of the text (Hussy et al. 2013). Care was taken to ensure that the code accurately reflects the content of the statement and does not involve any interpretation. In vitro and in vivo coding was then performed (Harris 2015). In vivo coding uses purely descriptive and content-oriented terminology, while in vitro coding uses more abstract terms, but still without interpretation (Harris 2015). Examples of codes from our study are:

In vivo coding: “Well, I can definitely fall asleep better.” (UE118) ◊ Falling asleep.

In vitro coding: “Well, but… ultimately… mentally it also had many positive effects, which in turn means that, as I said, I want to have goals again relatively quickly. So I said: ‘I want to work again, I want to tackle this at some point. I’m not so old that I’m going to give up now and say: Here, I’ve lost.’ So that had a big impact on me.” (II017)◊ Self-efficacy (Fig. 1).

The four step coding tool proposed by Qureshi and Ünlü (Qureshi and Ünlü 2020) was adapted to the present study and used to analyze the data. As some topics were discussed more intensively than others in our study, in some cases there were more or less than four levels. If our data induced five steps, a subordinate concept was introduced. If only three steps were evident, concepts were excluded and categories directly extracted (Fig. 2).

Initially, open coding was employed to classify statements from different patients which were similar in terms of content and label them with the same code. Thematically related codes were then summarized into (subordinate) concepts. Concepts were in turn grouped into categories and categories into themes. Axial coding was then applied by using the paradigm model from Corbin and Strauss (Singh and Estefan 2018). This model helps to relate data to one another by defining the overarching phenomenon as “pain perception during CBM therapy” and dividing the themes into underlying conditions (in our study: prescripton, initial state), context (in our study: healthcare system, factors that have influenced patients’ attitudes toward CBM, coping strategies), intervening conditions (in our study: user experiences of CBM therapy), effect (in our study: effects of CBM therapy on different domains), and consequences (in our study: impact of CBM on QOL) (Table S2 in the supplement). At last, selective coding was applied to a comprehensive theory that characterizes the subjective pain experience during CBM-therapy, while still considering the differences in experiences. Codes were only applied once for each patient.

MAXQDA^® (2022) software was used for data management (data organization and visualization). After careful analysis of the themes, the codes attributed to each patient were considered and four groups were formed. Since each code was used only once for each patient, the only relevant factor was whether the code was used, not how often. It is also important to note that some patients did not experience any specific effects, and therefore certain codes are missing (see the Results section).

All participants were interviewed by the same researcher (FraF), who also transcribed and analyzed the interviews. In a further step, three selected interviews were analyzed independently by three researchers (MK, JMF, F-IK) who were not involved in the interviews. Results were examined and discussed by the entire research team to achieve higher intercoder reliability.

Additionally, to the interview, each patient received a questionnaire about satisfaction with medication (German Treatment Satisfaction Questionnaire for Medication, TSQM, version 1.4), comprising questions about effectiveness, side effects, convenience and global satisfaction ( 2015, Atkinson et al. 2005, Atkinson et al. 2004). Each section was individually assessed and given a value between 0 and 100 (100 representing total satisfaction). The questionnaire included 14 questions. The areas effectiveness, convenience and overall satisfaction include three questions each. Regarding side effects, the patients were first asked if they suffered any side effects; if the answer was no, it was attributed a value of 100; if the answer was yes, there were four follow-up questions posed. Analysis has been done according to the formula that is described elsewhere (Atkinson et al. 2005, Atkinson et al. 2004, Bharmal et al. 2009).

Finally, the mixed methods approach was achieved by combining qualitative and quantitative data. Regarding effectiveness, side effects, convenience, and overall satisfaction, which were determined using the German TSQM, differences were sought between the 4 groups that had been formed after analysis of the qualitative data.