Differences between the original English-language version and the back-translated version of the PSPS-Spanish can be seen as Additional file 1
. Most of the twenty-four items of the PSPS-Spanish were back-translated identically to or with minor differences with the original questionnaire. The final Spanish version is shown in Table .
Final Spanish version of the ESCALA DE AUTOPERCEPCIÓN DEL DOLOR
Characteristics of the sample
Sixteen patients were ruled out from the study because CFS symptoms predominate. Of 253 potential subjects, three (1.1%) declined to participate. None of the participants were ruled out because of the exclusion criteria. The final study sample consisted of 250 patients, 229 (91.6%) women and 21 (8.4%) men, aged 24-61 (mean 44.9, SD: 7.2 years), all self-described as White European. Ratio women: men are quite more frequent in the sample reflecting a similar ratio in the prevalence of FM in either gender. On average, the patients who participated in the study had suffered from FM for 7.9 years (range 1-20; SD: 2.3 years), and 122 (48.8%) had been granted an invalidity pension. Most of the patients (231; 92.4%) were taking one or more prescription drugs. More than half of the patients (N = 131; 52.4%) suffered from some form of psychiatric morbidity assessed with SPPI, mainly depression and anxiety. A group of 21 patients (8.4%) were also diagnosed with PTSD.
Distribution of total scores
The distribution of PSPS-Spanish total score (M = 33.4, SD = 26.7, Minimum = 0, Maximum = 94) did not differ significantly from a normal symmetric distribution (skewness = 0.66, SE = 0.19; kurtosis = - 0.74; SE = 0.35), although there was a slight positive skew. There were no significant differences in scoring between men (M = 32.4, SD = 27.8) and women (M = 34.2, SD = 25.9). However, the subsample of patients with FM and PTSD showed significantly higher scores in PSPS (M = 48.3, SD = 18.5) compared with the subgroup of patients with FM but without PTSD (M = 31.2, SD = 28.3) (t = 2.98, df = 248, p < 0.01). Mean and SD scores of the Spanish versions of the instruments used are summarized in Table . There was not significant association between PSPS total score and most demographic characteristics including gender, age, marital status, duration of pain, education level or work status, as can be seen in Table .
Mean and SD scores of the Spanish versions of the instruments used
Association between the Spanish version of PSPS and demographic parameters
For assessing face validity a subsample (N = 150) of the validation study sample were randomly selected. They were asked whether they thought that the test could adequately measure the effects of pain as an assault on the person's life and sense of identity. A total of 95.3% patients (143 out of 150) agreed.
Cronbach's α calculation for the 24 items in the PSPS-Spanish was 0.90 (95% CI: 0.87-0.93), indicating a high degree of internal consistency. Item-total r correlation coefficients ranged between 0.68 and 0.86; median = 0.80) (Table ).
One-factor solution from Principal Component Analysis of the PSPS-Spanish
The response to the PSPS-Spanish provided by a random subsample of 75 patients with fibromyalgia (gender female: 70, 93.3%; age: mean 43.8 years, SD: 7.4 years; duration of the disorders: mean 7.6 years SD: 2.5 years; 39 (52%) granted an invalidity pension) showed satisfactory temporal stability of the scale over a 1-2 week interval, during which the patients did not change baseline treatment. The test-retest correlation assessed with the intraclass correlation coefficient was 0.78.
Principal components analysis
The KMO was found to be 0.81, which exceeds the recommended minimum value of 0.60 [26
]. Bartlett's Test of Sphericity was highly significant (χ2 = 753, p
< 0.001), supporting the suitability of the data for principal components analysis [26
]. A one-component solution was extracted using the Kaiser-Guttman rule (eigenvalues > 1.0) [28
]. Results of the principal components analysis are shown in Table . It yielded two factors with eigenvalues greater than 1. The first factor, with an eigenvalue of 13.8, explained 61.4% of the variance; the second factor, with an eigenvalue of 1.9, explained only 7.3% of the total variance. All 24 items of the PSPS had high loadings on the first factor, ranging from 0.69 to 0.86. When factor loading smaller than 0.4 was suppressed, there was only 1 item loading on both factors (item 22). Therefore, item 22 is considered a "crossloading" item because it loaded high on more than one factor. With these results and a visual inspection of the scree plot, a one-factor solution is considered the most appropriate.
Intercorrelations between PSPS-Spanish, depression, anxiety, pain, global function and pain catastrophizing
The total PSPS-Spanish score was significantly associated with all the questionnaires assessed (Table ): HADS-dep (r = 0.63, p < 0.001), HADS-anx (r = 0.57, p < 0.001), pain assessed with PVAS (r = 0.42, p < 0.001), global function assessed with FIQ (r = 0.41, p < 0.001), and pain catastrophizing (r = 0.40, p <0.001). Total SPSS in the subsample of patients with fibromyalgia and PTSD also show a significant association (p < 0.001) with all the questionnaires having similar Pearson's r values to those of the whole sample: HADS-dep (r = 0.61), HADS-anx (r = 0.55), PVAS (r = 0.44) and pain catastrophizing (r = 0.42). However, PSPS correlation with FIQ is much more important in this subsample (r = 0.63).
Interrelationship between mental defeat (PSPS), Pain, Anxiety, Depression, Catastrophizing and FIQ in patients with fibromyalgia
To confirm that pain catastrophizing and mental defeat are different constructs, a partial correlation analysis was performed between PSPS-Spanish and the other scales controlling pain catastrophizing. The PSPS-Spanish score was significantly associated with all the questionnaires assessed, even those controlling pain catastrophizing: HADS-dep (r = 0.51, p < 0.001), HADS-anx (r = 0.48, p < 0.001), pain assessed with PVAS (r = 0.39, p < 0.001) and global function assessed with FIQ (r = 0.38, p < 0.001).