The study was conducted in three steps: (1) RAEH adaptation, (2) pilot testing, and (3) psychometric validation.
Description of samples
Three different convenience samples of consecutive RA patients were considered; patients included in each sample belonged to the outpatient clinic of the Instituto Nacional de Ciencias Médicas y Nutrición Salvador-Zubirán (INCMyN-SZ), a tertiary care level and national referral center for rheumatic diseases. All the patients had RA diagnosis, which was established according to their primary physician criteria (rheumatologist or trainee in rheumatology, no validated classification criteria were used), who additionally classified each patient as in remission or with active disease (no standard definition was used). In addition, all the patients had major comorbidity assessed according to the Charlson score [29]. In the three samples, the patients’ selection was directed to have age, gender (female preponderance), disease duration, and disease activity quotes represented.
The first sample included 50 patients and was used to evaluate feasibility; the second one included 200 patients and was used for the psychometric validation; the last sample included 20 patients and was used for face validity.
Sample size calculation for psychometric validation
Considering the number of RAEH items (8 items), a sample size of at least 80 RA outpatients was required as the exploratory factor analysis recommends 5 to 10 respondents per item [30]. However, taking into account additional published recommendations (at least 150 to 200 patients), we decided to include 200 RA patients (fair sample size) in the final sample [31].
Step (1) RAEH adaptation (see Additional file 1)
The S-HES author was contacted in order to have him involved in the process. The RAEH was adapted from the S-HES, that retained 8 items (one per subscale), with 8 subscales: satisfaction and dissatisfaction related to health, identification and achievement of personally meaningful goals, application of a systematic problem-solving process, coping with the emotional aspects of living with health, stress management, appropriate social support, self-motivation and making cost/benefit decisions about making behavior changes. Each item is scored on a 5-point Likert scale, ranging from 5 (strongly agree) to 1 (strongly disagree). In the scale, higher scores indicate stronger level of health-related empowerment, with scale scores ranging from 1 to 5.
First, in each item of the S-HES, the word “health” was substituted by “rheumatoid arthritis”. Then, each of three researchers (one rheumatologist, one trainee in rheumatology and one social worker, a PhD candidate in Health Sciences) suggested one sentence per item, in order to have different perspectives represented; researchers were blinded to each other proposals. After a consensus was obtained, the three researchers selected one sentence per item and integrated a preliminary version of the RAEH. RAEH was scored as in the original scale, but the sum of individual item-scores was provided; accordingly, RAEH score ranged from 8 to 40.
Step (2) Pilot testing
Feasibility was tested (in 50 patients) according to the following criteria: time required to fill the scale, patients’ perceived item’s clarity, and patients’ format acceptance.
Step (3) Psychometric validation
Content validity
Content validity was tested by a Validation Expert Committee (VEC) that was integrated by six rheumatologists and two psychiatrists who received relevant literature related to empowerment construct, and the author of the original S-HES (a geriatrist and psychiatrist). Content validity was examined by asking members of the VEC to rate each of the eight sentences (one per item) according to three categories: unnecessary, important but not necessary or essential. In addition, the VEC rated the item’s clarity and cultural semantic accuracy.
Reliability
Internal consistency (assessed in 200 patients) and test–retest (assessed in a subsample of 50 patients, by the same researcher, with an interval of 3 ± 1 weeks) were evaluated to determine the reliability of the scale.
Construct validity
Construct validity was determined with factor analysis.
Face validity
Previously, a brief explanation of the empowerment concept was offered to 20 patients, who were, in a second step, directed to rate each item as a valid measure of the respective empowerment dimension (Yes/No) along with each sentence’s clarity (Yes/No).
Statistical analysis
Descriptive statistics was performed to estimate the frequencies and percentages (for categorical variables), means and standard deviation (SD) or medians and 25th–75th interquartile (IQ) ranges (for continuous variables) of sociodemographic- and disease characteristics-related variables.
RAEH construct validity was evaluated by confirmatory factor analysis (maximum likelihood) with Varimax rotation. Sampling adequacy was confirmed by the Kaiser-Mayer-Olkin (KMO) measure (appropriate value ≥ 0.5); use of factor analysis was supported by the Bartlett’s test of sphericity (significant value p < 0.05), eigenvalue > 1 and correlations coefficients > 0.30 [32]. Floor and ceiling effects were determined as the percentage of patients who achieved the lowest and highest score of the scale, respectively.
Cronbach’s α and inter-item correlation for the complete scale and for each dimension was used to assess RAEH internal consistency of the questionnaire. Cronbach’s alpha interpretation was as follows: < 0.70 indicates that individual items provide an inadequate contribution to the overall scale and values of > 0.90 suggest redundancy, [33].
RAEH test–retest reliability was evaluated by the t test comparison between total test scores and by between partial test dimension score. In addition, intra-class correlation coefficients (ICC) and their 95% confident intervals (CI) were calculated based on a single measurement, absolute-agreement, two-way mixed-effects model. According to the ICC, values < 0.5 indicate poor reliability, between 0.5–0.75 moderate reliability, between 0.75 and 0.9 good reliability and values > 0.9 indicate excellent reliability. Finally, 95% CI estimates between 0.83 and 0.94 were considered as good reliability level and those between 0.95 and 0.99 estimates, as excellent reliability level [34].
All statistical analyses were performed with Statistical Package for the Social Sciences version 21.0 (IBM Corp., Armonk, NY, USA). A value of p ≤ 0.05 (two tails) was considered to be statistically significant.
Ethical considerations
The study received ethical approval from the Comité de Ética en Investigación of the INCMyN-SZ (reference number: 2226-17/ 18-1. Written informed consent was obtained from all the patients who agreed to participate.