Journal of Rehabilitation Medicine 51-8 | Page 58

P. Enthoven et al. 594 Fig. 1. Receiver operating characteristic (ROC) curve for different cut-off points for the Patient Enablement Instrument (PEI) in different groups. (a) All participants, n  = 504, area under the curve (AUC) 0.897 (95% confidence interval (95%CI) 0.867–0.928). (b) Whiplash-associated disorders (WAD) group (n  = 116), AUC 0.914 (CI 0.858–0.970). (c) Cervical radiculopathy (CR) group, n  = 112, AUC 0.862 (CI 0.794–0.929). (d) Mixed chronic pain (MixCP) group, n  = 276, AUC 0.914 (CI 0.858–0.970). small (< 0.10, with 2 exceptions that were close to 0.10) (24). This supports the internal construct vali- dity of the PEI. The original authors investigated the construct validity by adding 3 items to the instrument and found the construct validity of the original 6 items to be satisfactory (3). To our knowledge, only 1 study, conducted in Japanese patients with chronic illnesses, found that the PEI consisted of 2 principal factors (6). The first factor comprised questions 1–4, and the second factor comprised questions 5 and 6. However, other studies using factor analysis support the finding that the PEI is unidimensional (10, 11). Internal consistency of the Patient Enablement Instrument Cronbach’s alpha coefficient for the PEI varied bet- ween 0.878 and 0.907, indicating good internal con- sistency (27). For the original PEI, Cronbach’s alpha www.medicaljournals.se/jrm coefficient was 0.925, and it decreased each time an item from 2 different satisfaction scales was added or when any of the 6 PEI items was removed. This sug- gests that the 6 original PEI items comprise a unified group of questions that differ from other concepts, such as patient satisfaction (3, 4). Other studies conducted in primary care found Cronbach’s alpha values between 0.86 and 0.93 (2, 6, 10, 19), while studies with a PEI that was modified to fit patients with asthma reported values between 0.87 and 0.92 (5, 11), all within sug- gested alpha limits (27). Relationship between the Patient Enablement Instrument and other measures As hypothesized, higher PEI score showed a fair to moderate relationship with better function and men- tal and general health in all groups with chronic pain after treatment (Table VI). Furthermore, a higher PEI