Journal of Rehabilitation Medicine 51-8 | Page 59

Psychometric measurement of PEI in chronic musculoskeletal pain score was related to a positive change in complaints compared with baseline. Haughney et al. (5) found a fair correlation (0.30) between higher enablement and improvement in asthma-related quality of life after treatment. Ožvačić Adžić et al. (31) found an associa- tion between low enablement and poor self-perceived health in patients consulting a GP. These results suggest that a higher PEI score may be related both to better current health and improvement in health over time. Self-rated change The SRC of a measure depends on several factors, such as the patient group being studied and the method used to calculate the measure (27). In this study the SRC in perceived enablement for the WAD, CR and MixCP groups were 5, 6 and 4, respectively, illustrating that the SRC may be different for different groups. In patients in primary care, a PEI score ≥ 6 was reported to indicate clinically meaningful “enablement” (32). However, this was an arbitrary judgement by the ori- ginal authors (3). Although overall model quality was acceptable; for example, in the model with all participants ap- proximately one-third were misclassified. The GPE, a more generic measure of health effects, correlated moderately with the PEI, justifying its use as an anchor. Furthermore, the PEI was correlated with change in health and disability measures, as well as with the GPE. The PEI had high sensitivity and specificity, but low PPV. As the PPV depends on prevalence this may partly have stemmed from a low proportion of patients who improved (23.6%, n = 119) compared with those who did not (76.4%, n = 385). In summary, we suggest a study-specific self-rated important change should be applied. Since it may not be realistic to determine the cut-off for every new population, a score of at least 4 on the PEI is recommended for important self-rated change in perceived enablement in patients with chro- nic pain. However, due to participant misclassification the instrument must be administered with caution as it may be overly sensitive. Strengths and weaknesses The relatively large number of patients included from different healthcare centres, with different chronic musculoskeletal conditions, and treated with different interventions, is a strength of this study, as this might enhance the generalizability of the results. It is also a strength to have a relatively large number of patients in the context of the planned statistical analyses. In the WAD and MixCP groups the proportion of women was high, making the results less generalizable to men. In the MixCP group a considerable number of patients did not complete the self-reported ques- 595 tionnaire after the treatment period. One reason for this was that a number of patients never finished the treatment, either because they stopped participating in treatment or because the rehabilitation team found a different treatment to be more appropriate. Another reason for this was administrative issues, since the staff was not very experienced in administration of the questionnaires. There were only a few differences between those who did vs did not complete the ques- tionnaires, which might indicate that there was no significant selection bias. However, a better response rate would be preferable. The PEI measures change in enablement, which might be considered a limitation or weakness of the instrument. Patients with less experience and know- ledge of their disease might be more likely to improve in enablement (i.e. to have higher PEI scores) than patients who have experienced problems for a longer period of time. In addition, patients might be satisfied with their treatment even if there was no improvement in enablement (13). Another issue is that, since the PEI measures change in enablement, it is not known how “enabled” the patient actually is. Traditionally, the measurement of, for example, the minimal important change is based on the difference between 2 measure- ments. It is important to mention that in this study the self-rated change is based on 1 measurement with the PEI. It is considered important to investigate respon- siveness as a measurement property. Responsiveness is defined as “the ability of an instrument to detect change over time in the construct to be measured.” However, the PEI is aimed to measure self-rated change; therefore, assessment of responsiveness was judged not to be appropriate. Further development of the instrument measuring current patient enablement at different time-points might improve the possibility to measure responsiveness. For this, it would also be better to use response options with a larger range (e.g. 0–10) instead of the current limited response options 0–2, leaving little room for change. A recent publication (33) recommends 10 criteria for evaluating the content validity of patient-reported outcome measures, asking about the relevance, com- prehensiveness and comprehensibility of the items, response options, and instructions. This paper investi- gated only parts of these criteria. The PEI was not originally developed for use in patients with chronic musculoskeletal pain, and further assessment of the content validity is recommended (27, 33). In conclusion, the PEI showed fair content validity, construct validity, and internal consistency in indi- vidual patients with chronic musculoskeletal pain. The estimated SRC values could be used to indicate relevant changes in patient enablement in clinical practice and to guide interpretation of the results of J Rehabil Med 51, 2019