Cross-cultural adaptation of the GCPS involved the use of the 5-phase structured protocol proposed by Beaton et al.14 (Fig. 1). The initial phases of the translation were carried out by 2 linguistics experts and 2 native English speakers with full proficiency in Spanish. The final draft was subjected to an evaluation of the content validity before a panel of 6 expert judges, which included specialists in psychometry, linguistics and chronic pain. The criteria for the selection of the experts were as follows15: (a) experience in issuing rulings and evidence-based decision-making (research, publications and experience); (b) availability and motivation to participate; and (c) impartiality and flexibility. The members of the panel of expert judges were asked to provide a qualitative assessment of each of the items (degree of comprehensibility, suitability of the wording and phrasing), as well as a quantitative evaluation of the items, taking into account the following points: (1) appropriateness (the item applies to the established theoretical factors); (2) clarity (the item is easy to understand and has the proper semantics and syntax); (3) coherence (the item is related to the factor that it measures); and (4) relevance (the item is essential and should be included). The aspects to be evaluated for each item were scored using a 3-point Likert scale (1: Agree; 2: Neither agree nor disagree; 3: Disagree). Once the information provided by the panel of expert judges had been established and analyzed, a pilot test of the comprehensibility of the instrument was performed with 10 patients.

Fig. 1.

Algorithm showing the process of translation and cross-cultural adaptation of the Graded Chronic Pain Scale 2.0 to the version in Spanish. BT, back-translation; T translation.

(0.19MB).

A nonprobabilistic sample of 75 patients with chronic low back pain was obtained by consecutive recruitment in 3 health centers of the Community of Madrid, Spain. The sample size was calculated taking into account the proportion of individuals to items of 10:1 recommended by Kline.16 All the recruited participants accepted voluntarily to participate in the study and gave their written informed consent.

The patients received a questionnaire on sociodemographic data, the GCPS and a battery of self-report instruments to check the convergent validity; these instruments were the Visual Analog Scale (VAS) to evaluate pain intensity, the Spanish version of the Pain Catastrophizing Scale (PCS) to analyze the level of catastrophism produced by the pain experience,17 the Spanish version of the Tampa Scale for Kinesiophobia (TSK-11) to evaluate fear of injury and of movement,18 the Spanish version of the Roland Morris Disability Questionnaire (RMDQ), which evaluates the physical disability caused by low back pain,19 the Chronic Pain Self-efficacy Scale (CPSS), which was used to evaluate the degree of self-efficacy,20 and the Spanish version of the Fear-Avoidance Beliefs Questionnaire (FABQ).21

The data from this study were analyzed using the SPSS software package (version 22.0). The descriptive statistics are expressed as the mean±standard deviation (SD), range and 95% confidence interval (CI). The statistical inference utilized to verify the psychometric properties involved the use of the following tests:

The content validity of the final draft of the GCPS was calculated using Aiken's V coefficient,22,23 the value of which ranges between 0 and 1; in principle, values lower than 0.70 should be rejected and should ultimately not be included.24

The internal consistency of the instrument was analyzed using Cronbach's α coefficient and item-total correlation. Internal consistency is considered to be adequate when the α coefficient is greater than or equal to 0.70.25

The factor structure was analyzed by exploratory factor analyses of principal components. Prior to this, the adequacy for these analyses was tested using the Kaiser–Meyer–Olkin (KMO) measure and Bartlett's test of sphericity. The number of factors to be extracted is based on Kaiser's eigenvalue criterion (eigenvalue≥1) and the evaluation of the scree plot.

The ceiling and floor effects were analyzed on the basis of the percentage of patients with the lowest (floor) and highest (ceiling) scores on each item in the scale. The ceiling and floor effects are considered to occur when over 15% of those surveyed achieve the highest possible and lowest possible scores, respectively.26

The convergent validity was evaluated using Pearson's correlation coefficient measuring the association between the GCPS and the other instruments: VAS, TSK-11, PCS, CPSS, FABQ and RMDQ.

The test/retest reliability (reproducibility) was analyzed with the intraclass correlation coefficient (ICC). For this purpose, the GCPS was run twice at an interval of 7–10 days in a subsample of 46 patients. An ICC value over 0.70 is considered acceptable.26

The measurement error is expressed as a standard error of measurement (SEM), which is calculated using the formula SEM=SD×√1−ICC. The error of measurement is the systematic error and the random error in a patient's score that are not attributable to real changes in the construct to be measured.27

The responsiveness was evaluated according to the minimal detectable change (MDC). The MDC expresses the minimal change necessary to be sure, with a 90% confidence level, that a change observed between 2 measurements reflects a real change and not a measurement error. It is calculated using the formula MCD=SEM×√2×1.96.28

The present version of the GCPS consists of 8 items (Table 4). However, item 1 (“On how many days have you felt pain in the past 6 months?”) was excluded from the statistical analysis because the range of the scores for this item differed widely (18-fold greater) from those of the other 7 items, and its inclusion reduced the internal consistency of the questionnaire.

Table 4.

Spanish Version of the Graded Chronic Pain Scale.

Cronbach's α coefficient was 0.87 (95% CI, 0.83–0.92), indicating a high internal consistency for a scale of only 7 items. The item-total correlation coefficients ranged from 0.47 to 0.82, where item 3 had the lowest correlation and item 8 the highest. The complete results of the 2 subscales can be found in Table 5.

Exploratory factor analysis of the principal components was performed without rotation. In addition, for theoretical reasons, we tested a 2-factor structure using oblimin rotation, which converged in 4 iterations; finally, we chose a 2-factor solution supported by the scree plot (Fig. 2).

Fig. 2.

Findings of the scree test.

(0.05MB).

The KMO measure of sampling adequacy of 0.82 was higher than the recommended value of 0.60, and the results of Bartlett's test were significant (chi-squared=293.44, df=21, P<.001), indicating the adequacy of the data for performing principal components analysis.

The first factor, composed of 4 items, can be referred to as “pain-related disability”; the second, “pain intensity”, consists of 3 items. The correlation between the 2 rotated factors was 0.55. These factors explain the total variance of 72.37%; the distribution of the items is shown in Table 5.

The overall value on the GCPS showed a moderately positive correlation with all the questionnaires used except the TSK-11 injury subscale, and moderately negative with the CPSS (Table 6).

The ICC was 0.81 (95% CI, 0.684–0.891), which showed a satisfactory temporal stability after 10 days, when analyzing the response to the GCPS of a randomized subsample of 46 participants (sex: women, n=31, 67.4%; age: 54.8±14.9 years; with a history of chronic pain of 74±84.13 months).

We observed neither a ceiling nor a floor effect, as none of the participants scored the minimum (0 points) or the maximum (70 points) on the GCPS, in which the scores for the sample ranged from 6 to 64 points.

The 2 factors that were obtained are commonly dealt with in studies on chronic pain that separate the disability associated with the pain, the more relevant of the 2, from the intensity of that pain. In the different adaptations, these factors have been referred to as the “disability score” and “characteristic pain intensity”, as is the case in the validation of the German version of the questionnaire,11 the validation of the Italian version concerning musculoskeletal disorders10 and the validation of the Portuguese version.12 On the other hand, in a report from the United Kingdom, only 1 factor appeared.30 The latter finding was in line with the original version by von Korff, the Chronic Pain Grade Questionnaire (1992), as it presents a categorical proposal with 4 levels pertaining to the same pain-disability continuum.

The results of our study should be examined with caution until the 1-factor or 2-factor structure of the Spanish adaptation of the GCPS can be validated, in a sample of more than 200 patients and using confirmatory factor analysis. Future studies should also assess the function of the GCPS in populations with different diseases and of different ages.

The validation of the GCPS, like the versions in German and in Brazilian Portuguese, was performed in primary care patients with musculoskeletal pain,11,12 which predisposes it for use mainly at that level of health care.

Moreover, the 2-factor model used in this scale is commonly employed in the study of chronic pain.18,33,34

One recommendation at the present time is the development of international collaborative networks, which improve the tools we have for patient evaluation and treatment.35 The presentation of a new version of this scale in the Spanish language will facilitate the development of joint international research in English, German, Italian, Portuguese and Spanish-speaking populations.

The study has a number of limitations that must be taken into account in future research. One of them is the fact that the patients were recruited consecutively, without randomization, a circumstance that may have modified the results of the investigation. Another limitation is the inclusion only of patients with a diagnosis of chronic low back pain, which may affect the external validity of the questionnaire for use in other populations, depending on age and diagnosis. We also consider the use of a 2-dimensional questionnaire to be a possible limitation of the study, taking into account that an important group of experts in methods, measurements and evaluation of pain recommend multidimensional tools (up to 6 dimensions) for its assessment,36 a modification that would require considerably longer questionnaires.

We think that future studies, using prospective, longitudinal intervention designs, will need to examine responsiveness and other analyses complementary to tests of significance.

With respect to the sample size, the inclusion of 75 patients may be a limitation. Some authors recommend approximately 10 individuals for each item in the scale,18 which would have meant raising the number of participants to a total of 80; however, other authors have reported that the study could be carried out in a smaller population without affecting the results.37

The authors declare that no experiments were performed on humans or animals for this study.

The authors declare that no patient data appear in this article.

The authors declare that no patient data appear in this article.