Journal of Rehabilitation Medicine 51-8 | Page 54

590 P. Enthoven et al. Cronbach’s alpha (26) was used to measure the internal consistency or internal reliability of the PEI. If 3 studies show Cronbach’s alphas between 0.85 and 0.90 this indicates strong evidence for good internal consistency (27). The relationships between the PEI and other measures were investigated with Spearman’s rank correlation (r Spearman ), using the following coefficients: 0–0.25 none to little; 0.25–0.50 fair; 0.50–0.75 moderate to good; > 0.75 very good to excellent (28). The SRC was estimated by a measure integrating anchor- and distribution-based approaches. The SRC value is the optimal cut- off point of the receiver operating characteristic curve (SRC ROC ) (29). The GPE, dichotomized into importantly changed (very much or much improved) or not importantly changed (slightly improved, unchanged, or slightly worsened), was used as the external criterion (anchor). Six participants reported greater de- terioration, and were excluded from further analyses, since this number was too small to determine the SRC for deterioration. The Spearman’s correlation (r Spearman ) between the GPE and PEI was used to examine whether the anchor was adequate; a correlation coefficient of at least 0.5 has been recommended (27). The distribution of the PEI for participants who were im- portantly improved or not (anchor) was described. The sensitivity and specificity were calculated to determine the ROC cut-off point for each PEI score. To construct the ROC curve, the combination of sensitivity and 1-specificity was plotted for each PEI score. The SRC ROC , defined as the optimal cut-off point, is found on the ROC curve where the sum of the percentages of misclassified participants is lowest (29). Furthermore, positive predictive value (PPV) estimates the proportion of participants who actually had a high PEI score from the total number of participants classified as importantly improved. The negative predictive value (NPV) estimates the proportion of participants that actually had a low PEI score from the total number of participants that were not improved. Values of PPV and NPV that are close 1.00 suggest a higher probability of correctly classifying participants into im- proved/not improved. Statistical analyses were conducted using IBM SPSS, version 23. The level of significance was set at 0.05. No imputation for missing values was performed. RESULTS Participant characteristics A total of 738 participants answered the questionnaire at the start of the treatment (not shown in Table II). A total of 159 subjects did not answer the questionnaire after treatment (WAD study n = 16, CR study n = 25, MixCP study n = 118, of whom 30 did not complete the treatment, and for 88 participants the reason for non-completion was unknown). A total of 579 participants answered the questionnaires after treatment and 516 participants completed the Neck Disability Index/Functional Rating Index (NDI/FRI), the GPE item “Compared with before treatment, how would you describe your problems now?”, and all items of the PEI after treatment (Table II). Completers were less often living alone than non-completers (n = 108 (20.9%) vs n = 63 (28.9%), respectively, p = 0.020). Completers were less often worried about their finances than non-comple- ters (n = 217 (54.8%) vs n = 114 (65.9%), respectively, p  = 0.014). Otherwise there were no significant differences between the groups in participant characteristics (Table II). Completers had lower EuroQol visual analogue scale (EQ-VAS) scores than non-completers (mean (SD) 47.6 (21.0) vs 54.9 (20.9), respectively, p = 0.014). There were no significant differences between the groups regarding self-reported NDI/FRI, EQ-5D index, Anxiety and Depression (Hospital Anxiety and Depression Scale; HADS), and current work ability. For the participants who completed the NDI/FRI, the GPE, and all items of the PEI after treatment (Table II), the mean (SD) age was 45.1 (10.1) years, and 385 Table II. Baseline characteristics of the participants who completed the Neck Disability Index (NDI) or Functional Rating Index (FRI), the Global Perceived Effect (GPE), and all items of the Patient Enablement Instrument (PEI) after treatment All (n  = 516) n (%) Age, years, mean (SD) Sex, women BMI Overweight Obese Country of origin Sweden Other Scandinavian Non-Scandinavian Living alone Education Compulsory High school University/college Other Working status (Self-) employed/student, part- or full time Unemployed Other Sick leave (yes) n  = 116, 109, 283 Worries about finances often/quite often Expectations for treatment: full recovery/some improvement Probability of working within 6 months: very large or large, n  =  – , 102, 163 WAD (n  = 116) n (%) CR (n  = 115) n (%) 45.1 (10.1) 385 (74.6) 39.8 (11.2) 81 (69.8) 48.9 (7.4) 67 (58.3) 45.7 (9.8) 237 (83.2) 192 (38.6) 126 (25.4) 41 (36.0) 22 (19.3) 54 (47.8) 23 (20.4) 97 (35.9) 81 (30.0) – – – 108 (20.9) 105 (91.3) 5 (4.3) 5 (4.3) 26 (22.4) – – – 24 (20.9) 204 (73.6) 4 (1.4) 69 (24.9) 58 (20.4) – – – – 66 (23.3) 162 (57.2) 55 (19.4) – – – – – 379 (73.9) 86 (16.8) 48 (9.4) 239 (47.0) 217 (54.8) 356 (69.8) 208 (57.1) 7 (6.1) 61 (53.0) 41 (35.7) 6 (5.2) 108 (93.1) 5 (4.3) 3 (2.6) 13 (12.9) 36 (31.9) 83 (72.2) – 93 (81.6) 7 (6.1) 14 (12.3) 63 (57.8) – 109 (95.6) 91 (89.2) Data are reported as numbers (percentages) unless stated otherwise. WAD: whiplash-associated disorders; CR: cervical radiculopathy; MixCP: mixed chronic pain; SD: standard deviation; BMI: body mass index. www.medicaljournals.se/jrm MixCP (n  = 285) n (%) 178 (62.9) 74 (26.1) 31 (11.0) 169 (59.7) 181 (64.0) 164 (58.4) 117 (45.7)