Translation, contextual adaptation, and reliability of the Danish Concept of Pain Inventory (COPI-Adult (DK)) – A self-reported outcome measure

2.1 Step 1: translation and contextual adaptation

The first step was conducted in June 2021. Before initiating this study, permission to translate and contextually adapt the original English language COPI-Adult [17] was granted from the original author (JP). A dual panel approach, in which two panels translate, and contextually adapt the questionnaire, was used to turn COPI-Adult into the COPI-Adult (DK). This approach was chosen as it has been found superior to a forward–backward translation [23].

2.2 Step 2: test–retest reliability and measurement error

3.1 Step 1: translation and contextual adaptation

The first panel (n = 6) had a mean age of 32 years (range 29–41 years), and only two were healthcare professionals (both trained as physiotherapist, one had further gained a master in physiotherapy and was conducting a PhD). The others were non-healthcare professionals. The second panel (n = 5) consisted primarily of laymen, but with one healthcare professional (occupational therapist) who was a stand-in on the day of the meeting. Although omitting people with the disease in question is recommended, there was one member out of five with chronic pain (representative of the background population). The mean age was 44 years (range 30–58 years).

The first author (BE) had reviewed the questionnaire with the original author (JP), to ensure complete understanding of the questionnaire, the purpose, and the intended meaning of each item before translating the COPI-Adult into Danish and to have a basis for discussion. BE is bilingual in Danish and English. Prior to the first panel discussion, the original COPI-Adult was forwarded to the panel participants along with the suggested Danish translation. The participants were asked to make their own translation prior to reading the translation made by BE and note any interpretive issues understanding the BE translated version. To ensure that the panel had a sufficient understanding of the purpose of the questionnaire, and the thoughts behind it [24], the original author of the COPI-Adult [17] had video recorded a 7-min introduction to the COPI-Adult [32], which the panel viewed before initiating the discussion.

The participants in the second panel were all included from the same workplace and could meet in person despite COVID-19 restrictions. The participants were presented with the agreed translation and had been asked to complete the questionnaire for themselves in advance and note any difficulties in completing the questions. They did not have access to the original version and were not given the scoring key until the day of the panel discussion. BE coordinated both panels and ensured that the meaning of the original items was retained.

The translation of the introduction paragraph preceding the questionnaire was discussed, and the panel quite easily agreed on the wording. Each individual item was then discussed. For each item, 100% consensus was reached in the group on the wording, but alternative wordings were noted and later introduced to the second panel as an alternative. With item 9 “Pain usually feels better if you move your body a little bit more each day,” the panel had a problem with regard to the “bit more,” since this indicates that you must move more each following day. The introduction, where information on a shared database is introduced, was found to have confusing wording and the text crowded. Before proceeding on to panel 2, the original author was contacted for clarification on the aforementioned two issues. For item 9, the panel suggested that “bit more” was removed in the translation, which was acceptable to the original author, and the suggested rewording of the information on the shared database was likewise accepted. The reworded introduction was discussed, and no further changes were made. Each item was discussed and accepted. For each item, the alternative wordings were presented, but none of the alternative wordings were found better than the agreed version. The last item: “Pain can be too protective if it stops you getting moving again,” had to be read more than once by all participants. They understood the statement, and when given clinical examples of this, they agreed that the wording was correct and that the answer would reflect how up to date with contemporary pain science one is. Overall, no words were edited from the agreed translation of the first panel, and the panel found the items meaningful, lending evidence toward good content and face validity. See Appendix S1 for the translated Danish version of the COPI-Adult.

3.2 Step 2: test–retest

Figure 1 shows that a total of 293 people participated in the test–retest study. However, 23 failed to tick the “consent box” and were excluded from the analysis. A further 108 failed to answer the retest, resulting in 162 included in the test–retest. Nine people were excluded due to a time interval of >14 days, and further three were excluded due to missing items, resulting in 150 people included in the analysis. There were some demographic differences between those who completed both questionnaires and those who did not respond to the second questionnaire (non-responders were slightly younger and living without a partner). Population characteristics and descriptive data from step 2 (test–retest) are shown in Table 1. A good-to-excellent ICC_2,1 value was found, and SEM and SDC for the questionnaire were calculated (Table 2). A Bland–Altman plot (Figure 2) illustrating the agreement between the two measures demonstrated a slightly higher score at the follow-up compared to baseline (0.59 points 95% CI 0.024; 1.163).

Figure 1

Flowchart of participants.

Figure 2

Bland–Altman plot of agreement between measurements.

To measure internal consistency, a Cronbach’s alpha was calculated. The COPI-Adult (DK) consists of 13 items, there were 150 participants who answered the study, and the value of Cronbach’s alpha = 0.939.

4.1 Comparison to previous literature

The original English version went through a test of reliability that included a period of 1–2 weeks between each test. Similarly, we tested the reliability using 14 days between test and retest to replicate the original design. Our results regarding reliability (ICC 0.88 [0.84–0.91]) aligned with the original study showing an ICC of 0.84 (0.71–0.91). The SDC_ind in the original study was 3.4 points and the LoA was 6 points. We found that the SDC was 7.0 points and the LoA ranged from −7.6 to 6.4 points. SDC_ind and LoA are calculated to reflect the smallest change in a person’s score that with a 95% probability is a true change in the construct over and above the measurement error of the questionnaire at an individual level. Collectively, the findings suggest that changes of 7 points (corresponding to a change of roughly 13%) are needed to reflect changes larger than the measurement error at an individual level.

Other tools to examine knowledge on pain exist, among others the Neurophysiology of Pain Questionnaire [33]. However, psychometric properties have not been very convincing [33]. The Danish version of the Neurophysiology of Pain Questionnaire has been examined by Stæhr T, Rathleff MS, Baad-Hansen L, and Christiansen DH and failed to demonstrate satisfactory measurement properties (ICC of 0.39). Furthermore, the NPQ is a yes/no/undecided questionnaire, which leaves little room for engaging in conversation, and is also susceptible to guesses. In contrast, the COPI-Adult has a 5-point Likert scale for all 13 items, which provides a more nuanced assessment. Hence, the COPI was found to be a relevant alternative to explore.

4.2 Clinical implications

Understanding gaps or misconceptions about pain within individual patients is a relevant treatment target in clinical practice. How and what patients think about their pain may influence their thoughts, behaviors, and ultimately their pain experience [34]. Ideally, PSE should be individually tailored to enhance learning and give the patient an opportunity to tell their story. This approach was recommended in the review and meta-analysis from Watson et al. [15]. Based on this study, the COPI-Adult is now available in a Danish language suitable for further testing of the clinical utility. This study further strengthens the knowledge about reliability and SDC, which may be useful in a clinical setting for assessing pain science knowledge and beliefs during rehabilitation. Because overall scores are quite high, and the SEM is quite large, future research could assess over longer time periods and examine whether ceiling effects occur. The COPI-Adult takes only a few minutes to complete and is possible to complete in a self-directed manner, e.g., before being seen by a health care professional.

4.3 Future research

Other psychometric properties to look at include content, construct and criterion validity, responsiveness, and interpretability. Although COPI-Adult only takes a few minutes to complete and can be answered by patients on their own, future studies should assess the appropriateness of the tool in a clinical setting. These studies should especially focus on the extent to which the COPI-Adult can be implemented in the clinic (e.g., use of the assessments to inform treatment, and if it engages patients with patient education). It is of further importance to uncover which type of patient will benefit most from responding to the COPI-Adult and whether systematic use for all is relevant. Therefore, identifying specific indicators for applying COPI-Adult to specific patients may be both time effective and feasible in a clinical setting.

4.4 Strengths and limitations

We used established methods to conduct the translation and took care in selecting relevant members of the panel and used the recommended number of panel members. However, the panel may not be representative of the general population of patients with pain complaints, or those with low levels of health literacy and lower-than-average reading skills. Future studies need to further develop and test the questionnaire across different populations to ensure that the language is suitable across the clinical populations. A limitation is also that the original questionnaire is still undergoing further validation and testing, and a translation can arguably be premature. During the test–retest part, only 150 participants (out of the 293) ended up completing both surveys and were used in the analysis. This is a significant limitation, but our drop-out analysis showed no strong signs of a selection bias, indicating little potential impact on our overall conclusions. We discovered that the autogenerated second questionnaire from REDCap ended up in the SPAM mail for those who have Gmail and Hotmail accounts, which alone accounted for 56. We did not investigate if this is also the case for other mail accounts, but this is a point of attention for further studies. There also seemed to be an issue with the consent box, since 23 had missed ticking this off despite continuing to answer the questionnaire. It is possible that the differences in responses between the two timepoints are affected by bias such as a learning effect. Future research could use more than two timepoints to help unravel this potential issue. A limitation in the study is that the pre-registration of the trial at Aalborg University is not readily available to the public.

A significant strength is the robust study design that followed recommendations from COSMIN [25], ensuring, e.g., that subjects answer in the same setting. The number of included individuals also exceeds the minimum recommended number of 10 per items in the scale for the test–retest reliability analysis.