Spanish Validation of the Recovery Assessment Scale (RAS-24)

Javier Saavedra, Antonio J. Vázquez-Morejón, Raquel Vázquez-Morejón, Samuel Arias-Sánchez, Sergio González-Álvarez, and Patrick Corrigan 500 Recovery beyond symptom remission has become the goal of most healthcare systems, especially with regard to patients diagnosed with severe mental illnesses (SMI) (Department of Health, 2011; Slade et al., 2012). It is a complex construct which includes several components for which different conceptual frameworks have been proposed (Corrigan et al., 1999; Leamy et al., 2011; Slade, 2009). According to the classic defi nition by Anthony (1993), recovery is a single process requiring the reconstruction of identity for the goal of developing, in spite of the disease, new meanings and purposes to achieve a full life. In this sense, despite the traditional negative psychiatric view, it has been confi rmed that people diagnosed with SMI retain their capacity to coherently narrate their needs, desires and expectations (Saavedra et al., 2009). This has led to their voices being considered a source of knowledge and the recovery model being essential for organizing services and defi ning professional practice. Keeping in mind the relevance of the voices of patients in the defi nition of their recovery, and paraphrasing Robert McNamara (cited by Slade & Longden, 2015), the challenge is to be able to measure what is important, not to make what is measurable important. Evaluation of personal recovery in the scope of mental health is challenging because the construct spans subjective factors that have usually been ignored in the health sciences, such as hope and meaning, which necessarily requires considering elaboration of the identity of the person diagnosed with SMI (Slade & Longden, 2015). Therefore, although variables such as clinical symptomatology, general wellbeing or social functioning can be used as partial indicators, they do not span the complexity of the ISSN 0214 - 9915 CODEN PSOTEG Copyright © 2021 Psicothema www.psicothema.com Spanish Validation of the Recovery Assessment Scale (RAS-24) Javier Saavedra1, Antonio J. Vázquez-Morejón1,2, Raquel Vázquez-Morejón1, Samuel Arias-Sánchez1, Sergio González-Álvarez3, and Patrick Corrigan4 1 Universidad de Sevilla, 2 Unidad de Salud Mental Comunitaria Guadalquivir, Hospital Universitario Virgen del Rocío (Sevilla), 3 Fundación Pública Andaluza para la Integración Social de Personas con Enfermedad Mental (Sevilla), 4 Illinois Institute of Technology (USA) Abstract Resumen Background: Recovery is an essential construct in healthcare treatment for patients diagnosed with severe mental illnesses (SMI). Of all the psychometric instruments available for measuring recovery, the 41-item Recovery Assessment Scale (RAS) is the most widely used. Several brief versions have been proposed, including the 24-item version. In this study, the RAS-24 was adapted to European Spanish and validated in a clinical sample. Method: Participants (N = 309) diagnosed with SMI were recruited from a community mental health center and a work guidance center. The participants completed the RAS-24 and the Social Functioning Scale (SFS), both self-reported and family-reported versions. Results: The results showed good indices of fi t for the original fi ve-factor structure, acceptable internal consistency ( α = .93; ω = .95), temporal reliability (ICC = .89, p <.01), and signifi cant correlation with most of the SFS scales (total SFS self-report r = .50, p < .01; total SFS family reported r =.49, p < .01). Conclusions: These data support the use of this Spanish version as a measure of recovery in the Spanish clinical population. Keywords: Recovery; assessment; RAS-24; severe mental illness; psychometric characteristics. Validación Española de la Recovery Assessment Scale (RAS-24). Antecedentes: la recuperación es un constructo fundamental en la atención sanitaria de los pacientes diagnosticados de trastornos mentales graves (TMG). De todos los instrumentos psicométricos disponibles, la Recovery Assessment Scale (Escala de Evaluación de la Recuperación, RAS) de 41 ítems es el más utilizado. Se han propuesto varias versiones breves, entre ellas la versión de 24 ítems. En este estudio, el RAS-24 se adaptó al español europeo y se validó en una muestra clínica. Método: los participantes (N=309), diagnosticados con TMG, fueron reclutados en un centro comunitario de salud mental y un centro de orientación laboral. Todos cumplimentaron el RAS-24 y la Escala de Funcionamiento Social (SFS) en ambas versiones, autoinformada y heteroinformada. Resultados: los resultados mostraron buenos índices de ajuste para la estructura original de cinco factores, consistencia interna aceptable ( α = .93; ω = .95), fi abilidad temporal (ICC = .89, p < .01) y correlación signifi cativa con la mayoría de las escalas SFS (total SFS autoevaluado r = .50, p < .01; total SFS heteroevaluado r = .49, p <.01). Conclusiones: estos datos apoyan el uso de esta versión en español como medida de recuperación en población clínica española. Palabras clave: recuperación; evaluación; RAS-24; trastorno mental grave; características psicométricas. Psicothema 2021, Vol. 33, No. 3, 500-508 doi: 10.7334/psicothema2021.52 Received: February 11, 2021 • Accepted: April 28, 2021 Corresponding author: Samuel Arias-Sánchez Departamento de Psicología Experimental Universidad de Sevilla 41018 Sevilla (Spain) e-mail: [email protected] Spanish Validation of the Recovery Assessment Scale (RAS-24) 501 recovery construct. As recovery is an essential objective of mental health intervention, it is fundamental to have adequate instruments for its measurement. Although there is no recognized gold standard, the Recovery Assessment Scale (RAS), originally comprised of 41 items (Corrigan et al., 1999; Giffort et al., 1995), is the most widely used instrument, with over 200 publications already in 2012, 77 of them on its psychometric data (Salzer & Brusilovskiy, 2014). In factor analysis, fi ve factors have been identifi ed: “personal confi dence and hope, willingness to ask for help, goal and success orientation, reliance on others and no domination by symptoms” (Corrigan, et al., 2004; McNaught et al., 2007). The fi rst factor, “personal confi dence and hope”, combines items related to hope for the future and capacity for stress management. The second factor, “willingness to ask for help”, confi rms affi rmations related to the disposition to seek help. The third, “goal and success orientation”, evaluate the desire for success and the ability to meet goals. The fourth focuses on relationships with signifi cant others on the way to recovery, “reliance on others.” Finally, the fi fth factor, “no domination by symptoms”, includes items that evaluate whether psychiatric symptoms are not the center of the patient’s life. The reliability and validity of this model has been confi rmed in very diverse studies (Leamy et al., 2011; Salzer & Brusilovskiy, 2014). In a systematic review of instruments for measuring recovery, Sklar et al. (2015) considered the 41-item RAS the most psychometrically robust instrument of a total of 13 scales and inventories analyzed. However, these same authors warned that of the three best instruments, the RAS was the one that took the longest to complete. The need for effi ciently including the evaluation of recovery in routine clinical practice motivated the development of shorter versions of this instrument. Corrigan et al. (2004), with a sample of almost 2000 participants and using exploratory (EFA) and confi rmatory factor analysis (CFA), found a 24-item version which retained the original fi ve-factor structure. Its psychometric characteristics have been studied in other countries, such as the one done in Norway (Biringer & Tjofl at, 2018), in which, with a 24-item version, the original fi ve-factor structure was confi rmed, as well as its adequate convergent validity. The Japanese (Chiba et al., 2010) and Australian (McNaught et al., 2007) versions also had a fi ve-factor structure with good internal consistency. The recent adaptation to German using CFA of the RAS-24 led to a 14-item instrument, also with a fi ve-factor structure (Cavelti et al., 2017). The Spanish adaptation of the RAS- 24 for Latin America recently produced a 21-item scale with a fi vefactor structure, adequate psychometric properties and fi t indices (Zalazar et al., 2017). A few adaptations have found other factor structures. The Brief 20-item RAS in Hebrew found results that confi rmed adequate psychometric characteristics (Roe et al., 2012). However, CFA identifi ed four factors with only 14 of the 20 original items. The factor that was eliminated from the original fi ve was “goal and success orientation”. The Portuguese version of the RAS-24 also had a four-factor structure with 22 items, although with poor fi t indices (Jorge-Monteiro & Ornelas, 2016). Taking into account all these results, Cavelti et al., (2017) stated that the recovery conceptualization and factor structures from the different versions of the RAS did not vary much between countries. As far as we know, although there is a Spanish adaptation for Latin America, as mentioned above, there is no adaptation of the RAS-24 for Spain, which is an important defi cit for including this useful and effi cient recovery evaluation instrument in clinical practice. Its adaptation to the Spanish language in a European context and its validation in a clinical sample was undertaken to cover this defi cit. Method Participants The scale was applied to 350 patients, 170 men (55.7%) and 135 women (44.3%) with a mean age of 41.65 years (SD= 12.05, range 18-71), all of them cared for in a community mental health unit or users of a job orientation and inclusion service for persons with SMI. Table 1 presents their sociodemographic characteristics. Participants were diagnosed by clinicians of public health systems. Their clinical characteristics were: 65% diagnosed with schizophrenia spectrum disorder, 18.07% diagnosed with bipolar disorder, 11.15% in the category of personality disorder, and 5.77% of the sample was diagnosed with delusional disorder. All of the participants were informed of the objective of the study, of the legislation on data protection and the voluntary nature of their participation, in compliance with the pertinent legal requirements of Law 14/2007 on Biomedical Research and the Helsinki Declaration (Fortaleza, Brazil, October 2013). This research project was previously approved by the Andalusian Biomedical Research Ethics Portal (Code: 0339-N-17). Instruments Recovery Assessment Scale (RAS-24; Corrigan et al., 2004). This scale measures recovery in patients diagnosed with mental disorders. The 24-item version was constructed from the original 41-items (Giffort et al., 1995) and is distributed in fi ve factors: “Personal confi dence and hope” (nine items), “willingness to Table 1 Sociodemographic information (N=305) Variables N% Mental health services Community mental health unit Employment Assistance Services Marital status Single Married Separated/Divorced With stable partner Widowed Other 164 141 224 28 31 16 2 4 53.7 46.3 73.4 9.2 10.1 5.2 0.7 1.4 Educative level Primary school incomplete Primary school Secondary school certifi cate/ Professional training University Other 11 41 166 84 3 3.6 13.4 54.5 27.5 1 Employment situation Employed Unemployed Other 25 196 84 8.2 64.3 27.5 Javier Saavedra, Antonio J. Vázquez-Morejón, Raquel Vázquez-Morejón, Samuel Arias-Sánchez, Sergio González-Álvarez, and Patrick Corrigan 502 ask for help” (three items), “goal and success orientation” (fi ve items), “reliance on others”, (four items), and “no domination by symptoms” (three items). Each item is rated on a fi ve-point Likerttype scale from 1 “Strongly disagree” to 5 “Strongly agree”. The sum of the items corresponding to each of the subscales provides the score on each of them, while the sum of the 24 items is the total scale score. The fi t indices for the fi ve factors on the original English scale were adequate. The factor loadings of the items were always over 0.4 and up to 0.8 on two items in the second factor (Corrigan et al., 2004). Social Functioning Scale (SFS; Birchwood et al., 1990). This scale was designed to evaluate the areas of social functioning most specifi c and crucial to performance in the community of patients with schizophrenia. It evaluates functioning in seven specifi c areas: social engagement/withdrawal, interpersonal behavior, prosocial activities, recreation, independence-performance, autonomycompetence, employment/occupation. Based on this evaluation, a global score on social functioning and seven scores, one for each area, are found, where higher scores show higher levels of social functioning. The scale can be applied in two versions according to the information source: the other-report version from data provided by key informants, normally family members, and the self-report, based on the patient’s answers. The adequate psychometric characteristics of the scale have been confi rmed in the original English version (Birchwood et al., 1990) as well as the Spanish version (Vázquez-Morejón & Jiménez-García-Bóveda, 2000) that was used in this study. Procedure The English version of the RAS-24 was translated into Spanish and back-translated into English by two different people working independently. The fi rst, a bilingual (Spanish and English) researcher, and the second, a native US student in the last year of Psychology. The original version was translated into Spanish by the fi rst researcher, and this Spanish version was then translated back into English by the native English speaker. The research team analyzed the fi nal translation to resolve any discrepancies from the original. Following the recommendations of Hernández et al. (2020), when supervising the translation and back-translation process, this team of experts also considered the cultural, linguistic, and psychological differences between both populations. The defi nitive version of the instrument was tested in a pilot application to two people diagnosed with SMI who were users of the FAISEM (Public Andalusian Foundation for Social Integration of the Mentally Ill) employment program, who helped clarify some concepts that might seem ambiguous to the participants. The recruitment procedure followed differed for participants cared for in the community mental health unit and those who were users of an integration and occupational training program. In the fi rst case, the procedure included the RAS-24 in the battery of routine evaluation instruments, which also included the SFS self-report and other-report, in the fi rst interviews and review evaluations of patients diagnosed with schizophrenia and related disorders. A group of 35 patients selected at random was asked to fi ll out the scale again three-to-four weeks later to evaluate the temporal reliability of the scale. The participants were recruited in 2017 and 2018. In the case of the participants included in the employment reinsertion programs, the staff of the orientation and job support services were told to give the questionnaires (SFS self-report version and RAS-24) to all those with SMI who came to the service and volunteered to take it. Participants in the FAISEM trainee program were also recruited. The staff assigned to this project were told to select as many participants in it as possible, administering the questionnaires at the beginning of training (in the fi rst week before starting or in the fi rst week of training). The information was collected from this group from June 2019 to June 2020. When the scales had been completed and the sociodemographic data collected, the information was digitized, and statistical analysis was performed using SPSS v21 and MPLUS 8. Data Analysis First, the descriptive statistics were calculated for each of the 24 items on the RAS-24, as well as the dimensions and total scale score. The scale’s factor structure was analyzed using CFA with the MPLUS 8 program. Missing data were computed by Full Information Maximum Likelihood (FIML). As there was no severe violation of normality and following the recommendations of Curran et al. (1996), the MLR estimation method was used to compute the model fi t indices based on chi-square (X2), the normed X2, the root mean squared error of approximation (RMSEA), the standardized root mean squared residual (SRMR) and the comparative fi t index (CFI). The normed X2 was also calculated as it is less sensitive to sample size than the X2. The cutoff points were < 5.0 for normed X2, < .08 for the RMSEA and SRMR (Browne & Cudeck, 1992), and >.90 for the CFI (Bollen, 1989). Internal consistency was explored using the Omega coeffi cient, both for the dimensions and for the complete scale. The Cronbach’s alpha was also calculated to facilitate comparison with other studies which used that coeffi cient. The correlations between factors and between factors and the total score was analyzed by Pearson´s rank correlation coeffi cient. Concurrent validity of the RAS-24 was explored based on its correlation with the total SFS (self-report and other-report) score and its dimensions. Temporal reliability of the RAS-24 was found using intraclass correlation. Results Descriptive Statistics The descriptive data for each of the items, dimensions and total RAS-24 score are presented in Table 2. A higher score on the scale was interpreted as better patient recovery. The mean score varied from 3.16 on Item 7 and 4.14 on Item 19. Factor results showed a mean score of 32.42 for the fi rst factor, “personal confi dence and hope”. The second, “willingness to ask for help”, was 12.00, and the third factor, called “goal and success orientation”, was 18.99. “Reliance on others” was 15.90, and the “no domination by symptoms” factor had a mean score of 10.50. Showing middle-to- high recovery, the score for the total scale was 89.83. The highest possible scores on the “personal confi dence and hope”, “willingness to ask for help”, “goal and success orientation”, “reliance on others”, and “no domination by symptoms” factors and total scale were obtained by 4.7%, 26.6%, 13.2%, 14.7%, 11.0% and 1.0% of the participants, respectively. On the contrary, 0.3%, 2.4%, 1.0%, 1.3%, 2.3%, and 0.3% of the participants, respectively, scored the minimum possible. Although the distribution of the data showed Spanish Validation of the Recovery Assessment Scale (RAS-24) 503 a certain tendency to a higher concentration of higher scores, the analyses demonstrated that they were distributed throughout the range of scores of the different items and subscales. Confi rmatory Factor Analysis The results showed that the distribution of the variables was not normal (Mardia skewness = 6513.73, p < .01 and Mardia kurtosis = 49.93, p < .01). However, although the distribution was not normal, the skewness and kurtosis indices were below 2 in absolute values. The factor solution of the original model [X2(242) = 479.828, p < .01; Normed X2 = 1.98; RMSEA = .057; SRMR = .055; CFI = .904] fi t well according to the recommended cutoff points. Table 3 presents the fi t indexes obtained in this validation compared to other adaptations of this scale in fi ve different countries. These results supported the fi ve-factor structure from the original RAS-24. All the items showed standardized factor loadings from .247 to .888 as observed in Figure 1. The parameter estimates with standard errors from the CFA are presented in Table 4. Internal Consistency Internal consistency was measured with the Cronbach’s alpha and the Omega coeffi cient. The latter, unlike the more traditional Cronbach’s alpha, works with factor loadings, is more appropriate for psychometric validations and offers more stable calculations. The internal consistency of the total scale, measured with the Cronbach’s alpha, was .93. The Omega coeffi cient was .95. The Cronbach’s alpha and the Omega coeffi cient for the instrument’s dimensions are shown in Table 4. Between-Factor Correlations The correlation of each of the factors with the total score varied from .70 on “reliance on others” and .93 on “personal confi dence and hope”. Signifi cant between-factor correlations (Table 5), mostly moderate, varied from .33 (“reliance on others”, and “no Table 2 Descriptive statistics of the RAS-R items, subscales and total (N=305) Item nº RAS-R Range Mean SD Skewness Kurtosis 1 1-5 3.86 1.22 -1.09 0.28 2 1-5 3.65 1.18 -0.69 -0.33 3 1-5 4.02 1.14 -1.26 0.89 4 1-5 3.54 1.18 -0.60 -0.39 5 1-5 3.94 1.14 -1.14 0.62 6 1-5 3.76 1.23 -.098 -0.01 7 1-5 3.16 1.42 -0.25 -1.27 8 1-5 3.40 1.19 -0.32 -0.69 9 1-5 3.49 1.27 -0.53 -0.77 10 1-5 3.82 1.08 -0.81 0.20 11 1-5 3.76 1.13 -0.80 0.05 12 1-5 3.87 1.02 -0.74 0.23 13 1-5 3.91 1.07 -0.94 0.39 14 1-5 3.78 1.14 -0.93 0.20 15 1-5 3.47 1.33 -0.56 -0.81 16 1-5 3.50 1.25 -0.56 -0.65 17 1-5 3.54 1.27 -0.64 -0.61 18 1-5 3.88 1.07 -1.06 0.72 19 1-5 4.14 0.99 -1.50 2.21 20 1-5 3.98 1.06 -1.19 1.03 21 1-5 3.23 1.22 -0.25 -0.78 22 1-5 4.12 0.99 -1.49 2.14 23 1-5 3.89 1.08 -1.00 0.54 24 1-5 4.13 1.01 -1.48 2.09 PCH 11-45 32.42 7.17 -0.31 -0.20 WAH 3-15 12.00 2.78 -1.16 1.46 GSO 5-25 18.99 4.46 -0.80 0.61 RO 4-20 15.90 3.24 -1.06 1.68 NDS 3-15 10.50 2.99 -0.43 -0.31 Total 36-120 89.83 16.80 -0.50 0.36 PCH: Personal confi dence and hope; WAH: Willingness to ask for help; GSO: Goal and success orientation; RO: Reliance on others; NDS: No domination by symptoms Table 3 Fit indices of Spanish validation and other adaptations Spanish Validation Corrigan (2004) Original Cavelti et al., (2017) Germany McNaught et al., (2007) Australia Biringer & Tjofl åt, (2018) Norway Zalazar et al., (2017) Argentina Jorge-Monteiro, et al., (2015) Portugal N305 1824 156 168 231 337 213 Items 24 24 14 24 21 22 Number of factors 5 555554 Cronbach’s alpha F1 .85 F2 .87 F3 .82 F4 .67 F5 .74 Total .93 .87 .84 .82 .74 .74 .79 .65 .88 .60 .60 >.70 .83 .85 .77 .65 .76 .90 .75 .82 .88 .66 .76 .88 .88 .77 .75 .78 .90 RMSEA .057 .059 .060 .046 .066 .068 SRMR .055 CFI .904 .93 .950 .087 .924 .905 .881 TLI .890 .900 .913 .865 Javier Saavedra, Antonio J. Vázquez-Morejón, Raquel Vázquez-Morejón, Samuel Arias-Sánchez, Sergio González-Álvarez, and Patrick Corrigan 504 domination by symptoms”) and .74 (“personal confi dence and hope” and “goal and success orientation”). Concurrent Validity The overall scores on the RAS-24 and on each of its factors showed signifi cant relationships with most of the scores on the SFS, both on the self-report version and the one reported by family members (Table 6). This relationship was particularly intense in the global scores of both versions. However, no signifi cant correlation was observed of the RAS- 24 with the score on the employment dimension of either version. Temporal Reliability The intraclass correlation of scores on the fi rst and second application of the RAS-24 with an interval of three-to-four weeks showed signifi cant correlations (p < .01, bilateral) which were intense for all factors: .89 for RAS 24 Global, .88 for “personal confi dence and hope”, .83 for “willingness to ask for help”, .77 for “goal and success orientation”, .80 for “reliance on others”, and .74 for “no domination by symptoms”. Discussion In general, the results of this study show adequate psychometric properties of the Spanish adaptation of the RAS-24 in a clinical population diagnosed with SMI. The mean scores on the dimensions and the global scale of the Spanish validation of the RAS-24 are comparable to the Norwegian (Biringer & Tjofl at, 2018) and German (Cavelti et al., 2017) adaptations. Notable differences were found only in the “goal and success orientation” factor, where this sample, as in the Japanese version (Fukui et al., 2012), had a lower score than the Norwegian and German adaptations. The asymmetry of the items is always below -1.5 and in the vast majority of cases below -1. This shows a moderate upward bias in the distribution of the data. However, this bias is not specifi ed in a ceiling effect on the factor and total scale scores according to the criterion of over 15% of participants with the maximum score (Terwee et al., 2007). In other words, the concentration of the scores is not extreme in the factors or the total RAS-24 scale, except for the “willingness to ask for help” factor with 23%, since it does not exceed 15% with the highest possible score. In addition, all the categories of the scale are represented in the scores on the items and factors. This concentration of the scores in the upper range of the scale suggests that a possible ceiling effect should be explored in other samples. All the participants in this study were patients in a mental health unit under professional care and supervision of pharmacological treatment, and in addition, most were in training, job orientation, and employment programs. Considering these particularities of the sample, most scores would be expected to be above average, as in fact has been demonstrated in other studies with similar populations (Biringer & Tjofl at, 2018; Cavelti et al., 2017; Jorge- Monteiro et al., 2016). The fi t indices and parameters of the fi ve-factor model of the Spanish validation of the RAS-24 estimated with CFA equaled or surpassed those shown by other validations (see Table 3). These results support the adequate fi t to the fi ve-factor structure found in the original US version (Corrigan et al., 2004). Although the majority of studies opt for the fi ve-factor structure as proposed by the scale’s designers, some have proposed four factors. This option is based on the important overlap between the personal confi dence and hope and goal and success orientation factors. In fact, although the correlations observed between the subscales in the Spanish validation are, in all cases, below the recommended criterion of .80 (Kline, 2005), the correlation between the two factors mentioned in this study and in others such as the one done in Norway (Biringer & Tjofl at, 2018), are near 0.80. In spite of this high correlation, as it is backed by a robust theoretical model, the fi ve-factor structure is the one most supported by the various validations, and as the indices of fi t found were over the criteria established in the literature, it was decided to confi rm it in this Spanish adaptation. Furthermore, Table 4 Standardized factor loadings ( β ), communalities, unstandardized factor loadings ( B ) with standard errors ( SE ), and Cronbach’s alpha and Omega coeffi cient No domination by symptoms factor Cronbach’s alpha = .67 Omega Coeffi cient =.68 Item βCommunality BSE 15 0.68 0.47 1.00 0.00 16 0.79 0.63 1.09 0.11 17 0.44 0.20 0.62 0.10 Reliance on others factor α =.74 ω =.76 Item βCommunality BSE 6 0.59 0.35 1.00 0.00 22 0.85 0.72 1.14 0.12 23 0.76 0.58 1.12 0.13 24 0.45 0.21 0.63 0.14 Goal and success orientation factor α = .82 ω = .82 Item βCommunality BSE 1 0.52 0.27 1.00 0.00 2 0.65 0.42 1.21 0.17 3 0.75 0.56 1.34 0.17 4 0.75 0.56 1.40 0.20 5 0.79 0.62 1.40 0.20 Willingness to ask for help factor α =.87 ω =.87 Item βCommunality BSE 18 0.75 0.56 1.00 0.00 19 0.86 0.74 1.07 0.10 20 0.89 0.79 1.18 0.10 Personal confi dence and hope factor α = .85 ω =.87 Item βCommunality BSE 7 0.25 0.06 1.00 0.00 8 0.65 0.42 2.19 0.63 9 0.71 0.50 2.56 0.73 10 0.61 0.37 1.87 0.56 11 0.67 0.45 2.17 0.66 12 0.76 0.57 2.20 0.66 13 0.82 0.68 2.48 0.74 14 0.76 0.58 2.46 0.74 21 0.57 0.33 1.98 0.57 Spanish Validation of the Recovery Assessment Scale (RAS-24) 505 Fukui et al. (2012), in the only study in which RAS-24 invariance was analyzed between two cultures, the USA and Japan, concluded that in spite of the cultural differences, the fi ve-factor structure was stable in both, and showed acceptable fi t indices similar to the one in this study. Although it is not discarded that there may be items with culturally biased response patterns in this validation, the Figure 1. Final Confi rmatory Factor Analysis model Javier Saavedra, Antonio J. Vázquez-Morejón, Raquel Vázquez-Morejón, Samuel Arias-Sánchez, Sergio González-Álvarez, and Patrick Corrigan 506 results with the Spanish sample are similar to those corresponding to Western societies. The scale’s internal consistency was very similar to the Norwegian adaptation (Biringer & Tjofl at, 2018) in each of the dimensions. In this study, the Cronbach’s alpha varied from .67 to .87, similar to the .65 to .85 for the Norwegian version. It was also observed that the dimension with the least consistency was the same factor in both versions, “no domination by symptoms”, with a Cronbach’s alpha of .67 in this version and .65 in the one done in Norway. The correlation of the Spanish validation with other instruments that measure constructs compatible with recovery is adequate. That is, it shows a signifi cant association, but independent of each other. For example, the RAS-24 showed a strong association with social functioning, both in the self-report version and in the other-report, but below .60. Other studies have found associations with general wellbeing, social functioning, empowerment and self-esteem (Cavelti et al., 2017; Corrigan, 2006; Jorge-Monteiro & Ornelas, 2016). Negative correlations have also been found, although not intense, with instruments that evaluate psychiatric symptomatology (Cavelti et al., 2017; McNaught et al., 2007). Concerning the temporal reliability of the Spanish adaptation, the correlations observed support its reliability, both for each of the dimensions and the total scale score, more so if the three-to-four- week application interval is taken into consideration. With respect to the Argentinean adaptation (Zalazar et al., 2017), the numerous cultural differences that are translated into linguistic variations between the Spanish language in Argentina and Spain should be emphasized, so a Spanish adaptation in the European context is necessary. Five factors were found in both adaptations to Spanish. However, in the Argentinian adaptation, the number of items was reduced to 21. The Argentinian version administered the 41-item version and afterwards selected 24 items for validation. This change in context of completion could affect the results. Furthermore, they did not fi nd any relationship between the RAS-24 and Global Assessment Functioning, which could be explained by the simplicity of that instrument compared to a multidimensional instrument such as the SFS used in this study. Lastly, in the adaptation to Spanish in Argentina, no temporal reliability results were offered. In this sense, the present validation in Spanish, apart from its usefulness, contributes to better understanding of the psychometric properties of this instrument. It should be underlined that of the validations of the RAS in other languages and except for the study by Fukui et al. (2012) and Zalazar et al. (2017), this is the one with the largest sample (see Table 3). We might also add that those diagnosed with SMI in this study came from two different public mental health services, which contributes to a wider diversity in the data and, therefore, is advisable for validating results. In this respect, we should emphasize that the factorial analyses with samples of fewer than 200 observations are very risky (Hair et al., 2014; Stevens, 2009). Among the limitations of this study is our sample, which although it is large and diverse, does not represent the whole population of patients diagnosed with SMI. For example, the psychopathological situation of this sample, which is from a community, may be less disabled than other groups of institutionalized patients. In any case, invariance studies of the questionnaire taking this variable and others, such as culture, already discussed, or sex into account are necessary. Table 5 Pearson correlations between factors and global RAS-24 (N=305) RAS-R PCH WAH GSO RO NDS Global PCH 1 .50** .74** .52** .66** .93** WAH 1 .53** .45** .38** .67** GSO 1 .56** .52** .88** RO 1 .33** .70** NDS 1 .72** Global 1 ** The correlation is signifi cant at .01 (bilateral) PCH: Personal confi dence and hope; WAH: Willingness to ask for help; GSO: Goal and success orientation; RO: Reliance on others; NDS: No domination by symptoms Table 6 Correlations between RAS-R subscales and SFS dimensions (N=305) PCH WAH GSO RO NDS Global Self-reported SFS Engagement/social isolation Interpersonal behavior Independence-performance Recreation Prosocial activities Independence-competence Employment/occupation Total Social Functioning .40** .33** .25** .24** .30** .37** .13** .45** .19** .19** .33** .11** .18** .36** .06** .25** .33** .45** .28** .30** .30** .38** .12** .47** .24** .35** .12** .24** .23** .09** .03** .33** .13** .29** .26** .28** .14** .31** .04** .30** .38** .43** .31** .30** .31** .40** .09** .50** Family reported SFS Engagement/social isolation Interpersonal behavior Independence-performance Recreation Prosocial activities Independence-competence Employment/occupation Total Social Functioning .23** .39** .37** .34** .45** .42** .00** .47** .08** .23** .10** .14** .20** .25** .07** .21** .07** .20** .33** .26** .36** .32** -.16** .34** .18** .32** .34** .33** .32** .34** .11** .41** .29** .28** .31** .19** .32** .37** .00** .36** .22** .35** .40** .35** .45** .46** -.01** .49** ** The correlation is signifi cant at .01 (bilateral) * The correlation is signifi cant at .05 (bilateral) PCH: Personal confi dence and hope; WAH: Willingness to ask for help; GSO: Goal and success orientation; RO: Reliance on others; NDS: No domination by symptoms Spanish Validation of the Recovery Assessment Scale (RAS-24) 507 It would be interesting for future studies, with larger and more diverse samples, to analyze using Item Response Theory whether there is a bias in the scores, and if so, whether it is due to the characteristics of the sample or to problems with the scale itself. The objective of this study was to adapt a specifi c instrument using the same methodology as in other adaptations so the results would be comparable. Therefore, our results did not contemplate any modifi cation of the instrument’s structure. However, we cannot assure that future analyses might not suggest a modifi cation of the Likert scale to optimize the questionnaire. Considering the subjective nature of the recovery construct, the results of the RAS-24 should be compared with qualitative narrative-type analyses. Thus, in addition, the concurrent validity of the construct would be confi rmed more emphatically. It should also be warned that, although all the participants had assistance available for fi lling out the questionnaires, some of them did so alone in the presence of a researcher and others fi lled out the form in a group, also with a professional present. Therefore, although it was attempted to control data collection as much as possible, we cannot discard the possibility that the different contexts may have somehow infl uenced questionnaire completion. In spite of these limitations and after the exhaustive analysis of the different sources of validity of this scale, its adequate psychometric characteristics, including the fi ve-factor structure with 24 items, are adequate with good fi t indices. This Spanish version of the RAS- 24 is an instrument with a stable factor structure that expresses the essential dimensions of the recovery construct. These data support the use of the RAS-24 as a measure of recovery in this population. Its brevity and simplicity, which make the RAS-24 a very useful instrument in the scope of mental health, should be emphasized. Its practical character and psychometric robustness make it especially relevant for its application in diverse contexts, especially in the clinic. In this adaptation, Spanish researchers and clinicians can fi nd an international-quality tool, which has repeatedly been confi rmed for evaluating personal recovery, in their own language. Acknowledgements This research was carried out within the framework of the research project funded by the Gobierno de Espana [PSI2016- 80112-P] entitled “Challenges of the self: Identity reconstruction in situations of inequality and social exclusion”. References Anthony, W. A. (1993). Recovery from mental illness: The guiding vision of the mental health system in the 1990s. Psychosocial Rehabilitation Journal, 16(4), 11-23. https://doi.org/10.1037/h0095655 Birchwood, M., Smith, J. O., Cochrane, R., Wetton, S., & Copestake, S. (1990). The social functioning scale the development and validation of a new scale of social adjustment for use in family intervention programmes with schizophrenic patients. The British Journal of Psychiatry, 157(6), 853-859. https://doi.org/10.1192/bjp.157.6.853 Biringer, E., & Tjofl åt, M. (2018). Validation of the 24-item recovery assessment scale-revised (RAS-R) in the Norwegian language and context: A multicenter study. Health and Quality of Life Outcomes, 16(23), 1-12. https://doi.org/10.1186/s12955-018-0849-3 Bollen, K. A. (1989). A new incremental fi t index for general structural equation models. Sociological Methods & Research, 17(3), 303-316. https://doi.org/10.1177/0049124189017003004 Browne, M. W., & Cudeck, R. (1992). Alternative ways of assessing model fi t. Sociological Methods & Research, 21(2), 230-258. https://doi.org/1 0.1177/0049124192021002005. Cavelti, M., Wirtz, M., Corrigan, P., & Vauth, R. (2017). Recovery assessment scale: Examining the factor structure of the German version (RAS-G) in people with schizophrenia spectrum disorders. European Psychiatry, 41(1), 60-67. https://doi.org/10.1016/j. eurpsy.2016.10.006 Chiba, R., Miyamoto, Y., & Kawakami, N. (2010). Reliability and validity of the Japanese version of the Recovery Assessment Scale (RAS) for people with chronic mental illness: Scale development. International Journal of Nursing Studies, 47(3), 314-322. https://doi.org/10.1016/j. ijnurstu.2009.07.006 Corrigan, P. W. (2006). Impact of consumer-operated services on empowerment and recovery of people with psychiatric disabilities. Psychiatric Services, 57(10), 1493-1496. http://dx.doi.org/10.1176/ ps.2006.57.10.1493 Corrigan, P. W., Giffort, D., Rashid, F., Leary, M., & Okele, I. (1999). Recovery as a psychological construct. Community Mental Health Journal, 35, 231-239. https://doi.org/10.1023/A:1018741302682 Corrigan, P. W., Salzer, M., Ralph, R. O., Sangster, Y., & Lorraine, K. (2004). Examining the factor structure of the Recovery Assessment Scale. Schizophrenia Bulletin, 30(4), 1035-1041. https://doi.org/10.1093/ oxfordjournals.schbul.a007118 Curran, P. J., West, S. G., & Finch, J. F. (1996). The robustness of test statistics to nonnormality and specifi cation error in confi rmatory factor analysis. Psychological Methods, 1(1), 16-29. https://doi. org/10.1037/1082-989X.1.1.16 Department of Health (2011). No Health without Mental Health: Delivering Better Mental Health Outcomes for people of All Ages. Department of Health and Social Care of the United Kingdom. https:// assets.publishing.service.gov.uk/government/uploads/system/uploads/ attachment_data/fi le/215811/dh_124057.pdf Fukui, S., Shimizu, Y., & Rapp, C. A. (2012). A cross-cultural study of recovery for people with psychiatric disabilities between US and Japan. Community Mental Health Journal, 48(6), 804-812. https://doi. org/10.1007/s10597-012-9513-2 Giffort, D., Schmook, A., Woody, C., Valendorf, C., & Germain, M. (1995). Construction of a scale to measure consumer recovery. Illinois Offi ce of Mental Health. Hair, J., Black, W., Babin, B., & Anderson, R. (2014). Multivariate data analysis: Pearson new international edition (7th ed.). Pearson. Hernández, A., Hidalgo, M. D., Hambleton, R. K., & Gómez-Benito, J. (2020). International Test Commission guidelines for test adaptation: A criterion checklist. Psicothema, 32(3), 390-398. https://doi.org/10.7334/ psicothema2019.306 Jorge-Monteiro, M. F., & Ornelas J. H. (2016). Recovery Assessment Scale: Testing validity with Portuguese community-based mental health organization users. Psychological Assessment, 28(3), e1-e11. https:// doi.org/10.1037/pas0000176 Kline, R. B. (2005). Principles and practices of structural equation modeling (2nd ed.). Guilford Press. Leamy, M., Bird, V., Le Boutillier, C., Williams, J., & Slade, M. (2011). Conceptual framework for personal recovery in mental health: Systematic review and narrative synthesis. British Journal of Psychiatry, 199(6), 445-452. https://doi.org/10.1192/bjp. bp.110.083733 McNaught, M., Caputi, P., Oades, L. G., & Deanes, F. P. (2007). Testing the validity of the Recovery Assessment Scale using an Australian sample. Australian and New Zealand Journal of Psychiatry, 41(5), 450-457. https://doi.org/10.1080/00048670701264792 Roe, D., Maschiach-Eizenberg, M., & Corrigan, P. (2012). Confi rmatory factor analysis of the Brief Version of the Recovery Assessment Scale. Javier Saavedra, Antonio J. Vázquez-Morejón, Raquel Vázquez-Morejón, Samuel Arias-Sánchez, Sergio González-Álvarez, and Patrick Corrigan 508 The Journal of Nervous and Mental Disease, 200(10), 847-851. https:// doi.org/10.1097/nmd.0b013e31826b6d4a Saavedra, J., Cubero, M., & Crawford, P. (2009). Incomprehensibility in the narratives of individuals with a diagnosis of schizophrenia. Qualitative Health Research, 19(11), 1548-1558. https://doi. org/10.1177/1049732309351110 Salzer, M. S., & Brusilovskiy, E. (2014). Advancing recovery science: Reliability and validity properties of the Recovery Assessment Scale. Psychiatric Service, 65(4), 442-453. https://doi.org/10.1176/appi. ps.201300089https://doi.org/10.1176/appi.ps.201300089 Sklar, M., Groessl, E. J., O’Connell, M., Davidson, L., & Aarons, A. (2015). Instruments for measuring mental health recovery: A systematic review. Clinical Psychology Review, 33(8), 1082-1095. https://doi. org/10.1016/j.cpr.2013.08.002 Slade, M. (2009). Personal recovery and mental illness. A guide for mental health Professionals. Cambridge University Press. Slade, M., Adams, N., & O’Hagan, M. (2012). Recovery: Past progress and future challenge. International Review of Psychiatry, 24(1), 1-4. https:// doi.org/10.3109/09540261.2011.644847 Slade, M., & Longden, E. (2015). Empirical evidence about recovery and mental health. BMC Psychiatry, 15(285), 1-14. https://doi.org/10.1186/ s12888-015-0678-4 Stevens, J. (2009). Applied multivariate statistics for the social sciences (5th ed.). Routledge. Terwee, C. B., Bot, S. D. M, de Boer, M. R., van der Windt, A. W. M., Knol, D. L., Bouter, L. M., & de Vet, H. C. W. (2007). Quality criteria were proposed for measurement properties of health status questionnaires. Journal Clinical Epidemiology, 60(1), 34-42. https://doi.org/10.1016/j. jclinepi.2006.03.012 Vázquez-Morejón, A. J., & Jiménez-García-Bóveda, R. (2000). Social Functioning Scale: New contribution concerning its psychometric characteristics in a Spanish adaptation. Psychiatry Research, 93(3), 247-256. https://doi.org/10.1016/s0165-1781(99)00123-7 Zalazar, V., Castro-Valdez, J., Mascayano, F., Vera, N., Scorza, P., & Agrest, M. (2017). Feasibility and psychometric properties of the Recovery Assessment Scale (RAS) in people with mental illness for its use in Argentina. Journal of Psychosocial Rehabilitation and Mental Health, 4(2), 179-188. https://doi.org/10.1007/s40737-017-0098-5