Recovery in alcohol use disorder (AUD) has been increasingly focusing on wellbeing aspects over time (Kaskutas et al., Reference Kaskutas, Borkman, Laudet, Ritter, Witbrodt, Subbaraman, Stunz and Bond2014, Reference Kaskutas, Witbrodt and Grella2015), by comprehending the different life dimensions that are compromised for individuals who suffer from it. In this way, the process of recovery itself becomes relevant, in terms of lifestyle changes, psychological wellbeing and personal resources (Kaskutas et al., Reference Kaskutas, Borkman, Laudet, Ritter, Witbrodt, Subbaraman, Stunz and Bond2014; Kelly et al., 2018; Laudet, Reference Laudet2008; Slade et al., Reference Slade, Leamy, Bacon, Janosik, Le Boutillier, Williams and Bird2012). In this direction, biopsychosocial models aim to extend the concept of recovery in substance dependence beyond the concept of abstinence. This approach also aims to address environmental, social, personal, and cultural factors that interact in a reciprocal and dynamic manner with recovery (Kelly & Hoeppner, Reference Kelly and Hoeppner2015).
In dependence processes, recovery is defined as the voluntary process of control over the substance use, with positive influences on health, well-being, and social participation, following the UK Drug Policy Commission (2008), and it can be divided in several “recovery” stages, related to abstinence, as it follows: Early sobriety (first year), sustained sobriety (1–5 years) and stable sobriety (≥ 5 years). The relation between psychological dimensions related to recovery and abstinence length has been studied before, and the results indicate that quality of life, a variable strongly related to recovery (Laudet & White, Reference Laudet and White2008), significantly predicts abstinence length at 1 and even 2 years after patients’ assessments (Laudet et al., Reference Laudet, Becker and White2009).
At this juncture, the recovery capital (RC) framework is gaining momentum. Cloud and Granfield (Reference Cloud and Granfield2008), and Granfield and Cloud (Reference Granfield and Cloud1999), defines it as the amount and scope of resources that can be tapped to initiate and sustain recovery from substance use problems. The several domains to which they refer are: Physical or economic capital; human capital, related to individual’s abilities to function in society (education, physical and psychological health); social capital, related to group belonging and resources, obligations and benefits from it; cultural capital, associated with norms and the capacity to act in a correspondent manner to them, in order to satisfy needs and maximize opportunities (Hennessy, Reference Hennessy2017). The accumulation of this capital is fundamental, since a greater quantity and availability of actives influences the resilience and coping strategies (Kelly & Hoeppner, Reference Kelly and Hoeppner2015); apart from helping to mitigate stress associated to abstinence adaptation and to enhance satisfaction with life ( Laudet & White, Reference Laudet and White2008). All this could contribute to the prognosis of treatment results, since individuals who show a greater capital of recovery are those who find themselves in better positions to solve substance use problems, such as alcohol dependence. In this way, these factors, distributed in an unequal manner through society, could differentiate the capacity of individuals to put an end to these issues, once they have been produced (Cloud & Granfield, Reference Cloud and Granfield2008).
Given the evidence regarding the importance of recovery capital, the demand for its correct measurement has increased. Hennessy’s systematic review (Reference Hennessy2017) presents three scales aimed at measuring RC. Sterling et al. scale (Reference Sterling, Slusher and Weinstein2008) was the first to attempt to measure RC; however, it did not present adequate predictive validity of time in abstinence and severity of addiction. The Burns & Marks scale (Reference Burns and Marks2013), which presents four domains (physical capital; human; social; cultural and community), showed good predictive validity of the severity of addiction for physical capital, but the rest of the domains did not present such good results. Finally, the Groshkova et al. scale (Reference Groshkova, Best and White2013) seems to be a good predictor of recovery outcomes based on personal and social RC, and it was used in several studies (Best et al., Reference Best, Lubman, Savic, Wilson, Dingle, Haslam, Haslam and Jetten2014, Reference Best, McKitterick, Beswick and Savic2015, Reference Best, Beckwith, Haslam, Haslam, Jetten, Mawson and Lubman2016; Brown et al., Reference Brown, Ashford, Figley, Courson, Curtis and Kimball2019; Chen & Gueta, Reference Chen and Gueta2020; Honess et al., Reference Honess, Andrianjazalahatra, Fernandez and Griffiths2012; Mawson et al., Reference Mawson, Best, Beckwith, Dingle and Lubman2015; Rettie et al., Reference Rettie, Hogan and Cox2019; Best et al., Reference Best, Haslam, Staiger, Dingle, Savic, Bathish, Mackenzie, Beckwith and Lubman2016). This scale is named as Assessment of Recovery Capital (ARC), and it shows a single factor of recovery, that explains 57% of variance. It is developed based on 10 dimensions, related to psychological and physical health, meaningful activities, social support and participation, house safety, life coping skills and risk taking, together with substance use control and recovery experience. ARC shows good reliability values and convergent validity with quality of life (Groshkova et al., Reference Groshkova, Best and White2013), an aspect that is also strongly related to recovery efforts and remission in alcohol dependent individuals (Laudet, Reference Laudet2008; Laudet et al., Reference Laudet, Becker and White2009; Laudet & White, Reference Laudet and White2008). The one-dimensionality and predictive validity of this scale have also been confirmed in other studies (Arndt et al., Reference Arndt, Sahker and Hedden2017; Basu et al., Reference Basu, Mattoo, Basu, Subodh, Sharma and Roub2019; Cano et al., Reference Cano, Best, Edwards and Lehman2017; Sánchez et al., Reference Sánchez, Sahker and Arndt2020). Sánchez et al. (Reference Sánchez, Sahker and Arndt2020) prove that the original ARC predicts successful completion of treatment, and Basu et al. (Reference Basu, Mattoo, Basu, Subodh, Sharma and Roub2019) report that the Hindi version of the ARC scale predicts 1 year of abstinence.
This instrument, far from being a diagnostic tool, has the purpose to evaluate positive measures of personal and social resources, trying to approach strengths and means of the individual to satisfy his needs and aspirations in the recovery process. Thus, the use of this scale is interesting for the study of recovery and for clinical practice. It improves the understanding of how recovery capital can be leveraged to help improve people’s ability to overcome alcohol use disorder problems and it provides new guidance for interventions. In addition, among the scales for measuring recovery capital, it is the most widely used in alcohol and other substance dependence in population in recovery, therapeutic communities and/or in treatment (Hennessy, Reference Hennessy2017), hence its use in alcohol dependent individuals in recovery can be adequate.
Considering that alcohol is the main substance of admissions to treatment for substance use in Spain (Observatorio Español de las Drogas y las Adicciones [OEDA]; 2019) the validation of the scale in this specific population is considered necessary. This work has the aim to translate and adapt to the Spanish language the ARC scale (Groshkova et al., Reference Groshkova, Best and White2013) and validate it in abstinent clinical population with alcohol use disorders, that is to say, in severe patients attending several treatment programs for this disorder. This population has been chosen since the ARC is not a diagnostic tool (these patients already have the diagnosis), but rather it is aimed at identifying strengths and points to be reinforced. Thus, the VCR could be highly beneficial for the treatment these patients are already receiving.
Method
Participants
The participants included in this study were patients diagnosed with alcohol use disorder in abstinence. These patients attended group and individual therapy programs at the Psychiatry Service of the 12 de Octubre Hospital or mutual aid group therapies, either at centres of the Federation of Former Alcoholics of the Community of Madrid (FACOMA) or at Alcoholics Anonymous groups in the hospital’s area of influence.
The participants had at least one month of abstinence and no active or recent use of other substances (at least 5 years of abstinence), except for tobacco and coffee. Those with psychiatric and/or neurological comorbidities were excluded from the study. All participants gave their written consent to participate in the study. The final sample included in the analysis was 184 participants, aged 27 to 75 (mean = 54.51; SD = 9.42), of whom 151 were men and 33 women.
Patients had mostly primary studies (35.3%), followed by college-level (22.4%) and secondary ones (21.9%), as well as professional training (20.4%). 35% of participants had an active employment situation, while 31.7% were unemployed or under work leave and 31.2% were retired. Clinical variables associated to dependence are described in Table 1.
Clinical Variables Related to Alcohol Use Disorder
Note. Means, standard deviations (SD), medians (Mdn), and frequency (valid percentages) of clinical data related to alcohol and other substances consumption, as well as the past attendance to treatments and mutual-help groups.
Materials
Recovery sources were assessed through the Assessment Recovery Capital (ARC), developed by Groshkova et al. (Reference Groshkova, Best and White2013). This scale has 50 dichotomous items, and it is organized in 10 subscales with 5 items each: Abstinence, psychological global health, physical global health, community involvement, social support, meaningful activities, house safety, risk taking, coping and life functioning, and experience with recovery. ARC is a one-dimensional scale, where the only factor explains 57% of variance (the weights for each variable are in the range .54–.78) (Groshkova et al., Reference Groshkova, Best and White2013). ARC scores show an intraclass correlation coefficient between .50–.73 and a convergent validity with WHOQOL-BREF scores of .80) (Groshkova et al., Reference Groshkova, Best and White2013).
The WHOQOL-BREF (World Health Organization Quality of Life) scale (The WHOQOL Group, 1998) is a brief version of 27 items from the original WHOQOL (World Health Organization [WHO], 1998). This instrument measures several aspects of quality of life, such as physical and psychological health and social relations, together with the environment of the individual. The range of scores for each domain is between 4 and 20 points. Its metrical data is good, with an internal consistency that varies between .68 and .8 for its subscales (Benítez-Borrego et al., Reference Benítez-Borrego, Guàrdia-Olmos and Urzúa-Morales2014; Skevington et al., Reference Skevington, Lotfy and O’Connell2004). In the present work, the Spanish version of WHOQOL-BREF scores have a Cronbach alfa of .84.
Procedure
Firstly, a translation and adaptation to Spanish language of the recovery capital scale (in Spanish: Valoración del Capital de la Recuperación, VCR) (Groshkova et al., Reference Groshkova, Best and White2013) was carried out, in the most accurate and close manner to the original (See the final Spanish version of VCR in the Appendix). Since the original scale was validated in population with a predominant consumption of several substances, some items have been slightly modified, by changing the word “substances” for “alcohol”. Additionally, VCR was administrated individually to 5 additional participants to check and improve the level of comprehension of the items. Further, VCR was inversely translated to English by a bilingual expert, to ensure the similarity to the original version.
A selection phase was carried out, in which patients who attended the outpatient therapeutic program of the Psychiatric Service of the 12 de Octubre Hospital were recruited. These patients had periods of abstinence between 1 month and 2 years (the program has a duration of 2 years). We also contacted the Federation of Ex-Alcoholics of the Community of Madrid (FACOMA) and Alcoholics Anonymous (AA), with the aim of increasing the sample and recruiting patients with longer periods of abstinence.
Following the signature of the informed consent, a semi-structured interview was carried out individually, where sociodemographic and clinical data were recorded, in addition to self-informed measures administration. The questionnaires were completed in digital format with the support of the researchers. The participants of the Hospital 12 de Octubre filled out the questionnaires in the hospital; whereas, the rest of the participants were evaluated by researchers in the different installations of Community of Madrid associations. The number of subjects was determined by factor analysis criteria: Minimum sample size of 100 subjects when there are less than 2 factors and at least 10 subjects per dimension (Kline, Reference Kline1986, Reference Kline, Petscher, Schatschneider and Compton2013).
All procedures carried out in this study meet ethic criteria of the committee of the Biomedical Institute of Research of 12 de Octubre Hospital. The study was conducted prior to the Covid–19 pandemic.
Statistical Analysis
Before analysing the psychometric properties, an exploratory analysis was performed and the distribution of scores was checked. Descriptive and frequency data were computed for quantitative and nominal variables, respectively.
The following analyses focused on the psychometric properties of the VCR. Firstly, reliability was calculated through Cronbach’s alpha for the VCR and for the WHOQOL-BREF. Secondly, the convergent validity with WHOQOL-BREF was examined, similar to previous studies (Basu et al., Reference Basu, Mattoo, Basu, Subodh, Sharma and Roub2019; Groshkova et al., Reference Groshkova, Best and White2013). The rationale for this step involves the importance of quality of life in the recovery journey (Laudet et al., Reference Laudet, Becker and White2009), since WHOQOL-BREF scores are related to overall functioning, social network recovery and engaging in meaningful activities, among other aspects of recovery (Best et al., Reference Best, Gow, Knox, Taylor, Groshkova and White2012). It was calculated through Spearman’s correlations (since the items of the VCR were dichotomous and their distribution asymmetrical) of VCR and the Spanish version of WHOQOL-BREF scores.
Third, we explored the differences in VCR scores between stages of recovery defined by the UK Drug Policy Commission (2008): “Early sobriety” (first year), “sustained sobriety” (1–5 years) and “stable sobriety” (more than 5 years). For this purpose, a comparison of means was performed using nonparametric tests (Kruskal-Wallis and pairwise comparisons) between the abstinence groups.
Although VCR is not a diagnostic instrument, we consider that it would be interesting to provide data on the sensitivity and specificity of the questionnaire. For this purpose, we calculated the ROC curve, considering 5 years of abstinence as stable recovery (De Soto et al., Reference De Soto, O’Donnell and De Soto1989; Sobell et al., Reference Sobell, Ellingstad and Sobell2000; UK Drug Policy Commission, 2008), as in the original validation by Groshkova et al. (Reference Groshkova, Best and White2013).
For the structure validity, we carried out a confirmatory factor analysis (CFA) driven by previous results indicating the one-dimensionality of ARC (Arndt et al., Reference Arndt, Sahker and Hedden2017; Cano et al., Reference Cano, Best, Edwards and Lehman2017; Groshkova et al., Reference Groshkova, Best and White2013). Taking into consideration the possible spurious results obtained with non-continuous data, we comprised the dichotomous items into the 10 VCR susbcales and performed the factor analyses on the scores of each of them, similar to the strategy carried out in the original scale (Groshkova et al., Reference Groshkova, Best and White2013) and as previous literature indicates (Bandalos & Finney, Reference Bandalos, Finney, Marcoulides and Schumacker2001; Nasser & Wisenbaker, Reference Nasser and Wisenbaker2003). This would allow obtaining more interpretable models. Further, CFA analysis was carried out through Unweighted Least Squares (ULS) parameter estimation method, which minimizes the possible residuals (differences between the observed and estimated correlations), can be robust against asymmetric data, and obtains optimal solutions for factor analysis (Ferrando & Anguiano-Carrasco, Reference Ferrando and Anguiano-Carrasco2010; Lee et al., Reference Lee, Zhang and Edwards2012; Sellbom & Tellegen, Reference Sellbom and Tellegen2019).
Additionally, descriptive, reliability, convergent validity and predictive validity were computed through SPSS v.22, whereas CFA was calculated by AMOS v.26.
Results
VCR scores do not have a normal distribution, following Kolmogorov-Smirnov test (p < .05). Patients obtained a mean VCR total score of 43.62 (6.71). Descriptive data related to VCR subscales and WHOQOL-BREF can be observed in Table 2.
Descriptive Data for VCR and WHOQOL-BREF Scores
Note. Mean, Standard Deviation (SD) and Asymmetry of recovery capital (VCR) and quality of life (WHOQOL-BREF) scores.
Psychometric Properties for VCR
With respect to reliability, Cronbach’s alpha value is .902 for the 50 items scores of VCR. Whereas the convergent validity results show positive correlations between VCR scores and WHOQOL-BREF ones, with Rho Spearman values of .51 for psychological health, .27 for physical health and .54 for social relations or support (p < .01), among others (Table 3).
Convergent Validity of VCR with WHOQOL-BREF
Note. Spearman correlations values between VCR and WHOQOL-BREF subscales and total scores.
The asterisk indicates the significance level * p < .05, ** p < .001.
VCR Differences regarding Early, Sustained, and stable Sobriety
Kruskal-Wallis and subsequent pairwise comparisons indicate statistically significant differences (p range < .001 –.05) between early (1–12 months of abstinence) and stable sobriety (> 5 years of abstinence), the latter obtaining higher scores for total VCR and all dimensions (except for Meaningful Activities). In addition, patients with sustained sobriety also had significantly higher scores (p range < .01–.05) on Substance use control and Life Functioning comparing to early sobriety; stable sobriety patients had more increased (p range < .01–.05) scores on Total VCR and Recovery Experience compared to sustained sobriety (see Table 4 for descriptive data of VCR among the three groups of abstinence and Table 5 for comparative analyses).
Descriptive Data for VCR regarding Early, Sustained, and Stable Recovery
Note. Mean, Standard Deviation (SD) of recovery capital (VCR) subscales and total score, and sample size (N) for patients with early (1–12 months), sustained (1–5 years) and stable (≥ 5 years) sobriety.
VCR Differences regarding Early, Sustained, and Stable Recovery
Note. Table 3 represents VCR differences between early, sustained, and stable recovery, through non-parametric Kruskal-Wallis and pair comparison tests (Mann Whitney U; pair comparison tests were corrected through Dunn-Bonferonni). 1, 2 and 3 represent groups of patients with early (1–12 months), sustained (1–5 years) and stable (more than 5 years) recovery.
The asterisk indicates the significance level * p < .05. ** p < .01. *** p < .001.
Sensitivity and Specificity of VCR
The ROC curve presents an area under the curve of .683, 95% CI [.602, .764]. Although it does not approach values of perfect discrimination, the CI does not cover values of .50 (non-discrimination) and the p value is < .001, which means that the ROC curve could have discriminatory capacity between patients who are in stable recovery (considered at 5 years of abstinence) and those who are not. The Youden Index is .564, which corresponds to a score of 42.5 (Sensitivity = 85.2%, Specificity = 71.2%).
Evidence of Validity based on the Internal Structure of the VCR
Table 6 summarizes the main characteristics of CFA analysis through ULS parameter estimation method. Results indicate that measures of fit have acceptable values, such as goodness of fit (GFI) and adjusted goodness of fit (AGFI), slightly superior to the proposed criteria (.95) (Jöreskog & Sörbom, Reference Jöreskog and Sörbom1989; Tanaka & Huba, Reference Tanaka and Huba1989). Further, incremental (Normative fit index, NFI; Relative fit index, RFI) (Bentler & Bonett, Reference Bentler and Bonett1980; Bollen, Reference Bollen1987) and parsimony tests (Parsimony ratio, PRATIO; Parsimony normative fit index, PNFI) show good values (Mulaik et al., Reference Mulaik, James, van Alstine, Bennett, Lind and Stilwell1989). In this way, it seems like ULS estimation method shows a good fit for the CFA single factor model.
CFA Properties for VCR for ULS Parameter Estimation Method
Note. Absolute measures of fit: goodness of fit (GFI) and adjusted goodness of fit (AGFI). Incremental measures of fit: Normative fit index (NFI) and relative fit index (RFI). Parsimony fit measures (PRATIO, PNFI) for and Unweighted Least Squares (ULS) parameter estimation method.
Considering the previous results, standardized regression weights and multiple squared correlations of variables that load on the Recovery factor by using ULS parameter estimation method, are shown in Figure 1. Variables have acceptable factor weights (between .40 and .77) and communalities (.21–.59), achieving the minimum acceptable (Child, Reference Child2006). Nonetheless, the variable use of substances had a lower communality (0.16).
CFA model using ULS
Note. AMOS graphical representation of CFA using ULS parameter estimation method, where the observed variables appear (10 VCR subscales, in squares) and their relation (standardized regression weights) with the latent variable (Recovery, in the circle). Communalities are also represented for each observed variable (values next to the squares) and their uniqueness (circles with arrows).
Discussion
The aim of this work was to translate and validate the ARC scale in the clinical population with alcohol use disorder. The clinical context and outpatient programs allow us to evaluate patients through different phases of the recovery journey, with various periods of abstinence, that gain progressively in recovery. To have an instrument capable of measuring recovery capital in Spanish population is beneficial, given the importance of recovery capital accumulation for the prognosis and resolution of alcohol use disorders.
The main results of this study point that VCR could be an adequate instrument for measuring recovery in abstinent alcohol dependent individuals, showing certain proper psychometric properties, in line with previous findings (Arndt et al., Reference Arndt, Sahker and Hedden2017; Cano et al., Reference Cano, Best, Edwards and Lehman2017; Groshkova et al., Reference Groshkova, Best and White2013; Sánchez et al., Reference Sánchez, Sahker and Arndt2020).
In the present study, VCR scores present good values of reliability, in the same way as previous studies (Arndt et al., Reference Arndt, Sahker and Hedden2017; Cano et al., Reference Cano, Best, Edwards and Lehman2017; Groshkova et al., Reference Groshkova, Best and White2013). It shows a somewhat low convergent validity with quality of life measured by WHOQOL-BREF (World Health Organization, 1996), inferior to the one found by Groshkova et al (Reference Groshkova, Best and White2013), yet similar to other findings (Basu et al., Reference Basu, Mattoo, Basu, Subodh, Sharma and Roub2019). Despite this result, previous literature supports a strong relationship between quality of life and recovery in terms of remission and clinical improvement (Best et al., Reference Best, Gow, Knox, Taylor, Groshkova and White2012; Laudet et al., Reference Laudet, Becker and White2009; UK Drug Policy Comission, 2008) and our findings also show a moderate relationship between VCR and WHOQOL-BREF psychological health domains, social and environmental support ones.
VCR distribution is asymmetrical, and a ceiling effect is observed. This outcome is similar to other studies (Bowen et al., Reference Bowen, Scott, Irish and Nochajski2020; Cano et al., Reference Cano, Best, Edwards and Lehman2017) and it could be due to the dichotomous features of the items, in addition to clinical, treatment and abstinence characteristics of patients, that could be affecting the variability of scores (Bowen et al., Reference Bowen, Scott, Irish and Nochajski2020).
With respect to the differences in recovery capital between different periods of sobriety, non-parametrical tests revealed significant differences between patients with early, sustained, and stable recovery in VCR dimensions, as defined by The UK Drug Policy Commission (2008). Specifically, our data reveals that recovery (understood as psychological, physical health, community involvement, home safety, life functioning, etc) is increased in patients who are in a more advanced stage of sobriety (≥ 5 years of abstinence), compared to those in early stages, understood as the first year of abstinence. Moreover, patients with stable sobriety seem to gain in psychological recovery in comparison to those in sustained sobriety periods (1–5 years of abstinence). And the patients with sustained sobriety also seem to have a gain in life functioning and substance use control comparing to those in early stages. Hence, it may seem like VCR could be beneficial in studying changes that might occur in the recovery process related do sobriety periods. And this might be in line with previous findings regarding VCR’s predictive ability with respect to long-term recovery and abstinence (Basu et al., Reference Basu, Mattoo, Basu, Subodh, Sharma and Roub2019; Groshkova et al., Reference Groshkova, Best and White2013). Moreover, our results regarding specificity and sensitivity features of VCR seem to indicate that statistically the VCR has the capacity to predict stable recovery, although the area under the curve (AUC) is somewhat low (.68). These values are similar to those found by Sánchez et al. (Reference Sánchez, Sahker and Arndt2020), AUC of .67, for predicting successful completion of treatment. Other studies, such as those by Basu et al. (Reference Basu, Mattoo, Basu, Subodh, Sharma and Roub2019) and by Groshkova et al. (Reference Groshkova, Best and White2013), have found more appropriate values (.82 for 1 year of abstinence and .89 for 5 years of abstinence, respectively), although both studies used participants with active alcohol consumption, which may contribute to the differences between those who are recovered and those who are not.
With respect to VCR structure, the CFA shows that the 10 VCR subscales load acceptably on the recovery dimension, indicating acceptable regression weights values (0.40–.77), similar to previous research (Arndt et al., Reference Arndt, Sahker and Hedden2017; Cano et al., Reference Cano, Best, Edwards and Lehman2017; Groshkova et al., Reference Groshkova, Best and White2013). Notwithstanding, the variable use of substances and sobriety had a lower factor weight, likely indicating a lesser contribution to the recovery dimension. Possibly, maintaining sobriety or the perception of controlling the substance use has less impact than other resources that are gained through the recovery process, which is also indicated by Bowen et al. (Reference Bowen, Scott, Irish and Nochajski2020) and the results of Arndt et al.(Reference Arndt, Sahker and Hedden2017). Or, perhaps, it is because the entire sample in this study was abstinent, giving place to more homogenous results in this dimension.
Regarding the parameter estimation method, ULS showed proper goodness of fit indexes and simplicity values and it showed an adequate proportion of covariance between VCR subscales. Therefore, it seems that the ULS parameter estimation method would be a right choice for VCR factorial solution, in line with previous literature on its suitability with non-continuous and non-normal distributions (Ferrando & Anguiano-Carrasco, Reference Ferrando and Anguiano-Carrasco2010; Sellbom & Tellegen, Reference Sellbom and Tellegen2019).
Thus, the results suggest the existence of a single factor, since the VCR subscales could be adjusted to a one-dimensional structure of recovery capital in alcohol dependence. However factor weights and communalities values could be interpreted as somewhat low and VCR could benefit from further exploration for its psychometric properties, as noted in Bowen’s study (Reference Bowen, Scott, Irish and Nochajski2020), in which CFA failed. This could indicate, in line with other studies, the need to consider the multiple dimensions of recovery (Cloud & Granfield, Reference Cloud and Granfield2008), revealing the need to improve the structure and items characteristics of the VCR scale (Arndt et al., Reference Arndt, Sahker and Hedden2017; Bowen et al., Reference Bowen, Scott, Irish and Nochajski2020). That is, although the VCR can fit into a one-dimensional model, it would be necessary to assess whether the unique information provided by each dimension is also relevant to clinical practice.
As to the limitations of this study, we would have to mention the sample size. Despite its properness for factor analysis (N > 100), the literature indicates that some of the aspects of factor models could be influenced by the sample size (e.g., goodness of fit measures, communalities, etc.), (Mundfrom et al., Reference Mundfrom, Shaw and Ke2005). Moreover, it would be of interest to analyse the characteristics and structure of VCR by using tetrachoric correlation matrixes, given the dichotomous features of the items (Arndt et al., Reference Arndt, Sahker and Hedden2017; Bowen et al., Reference Bowen, Scott, Irish and Nochajski2020). This type of analysis has better results with greater sample sizes (> 250, 500 or 1,000) (Lee et al., Reference Lee, Zhang and Edwards2012; Lorenzo-Seva & Ferrando, Reference Lorenzo-Seva and Ferrando2012), hence, future studies should take this into account.
Another limitation may be given by the clinical characteristics of the sample, since all patients attended therapeutic resources and remained under total abstinence, which implies a lower heterogeneity of the data, which may have an impact on the robustness of the statistical analyses used. In addition, people in treatment seem to gain in values and resources and have more opportunities than other types of population, which may have led to the ceiling effect. Therefore, in the future it could be interesting to compare this sample in treatment with others without these opportunities, in which some measures collected by the VCR can be better captured. For example, it would be of interest to study VCR behaviour in alcohol dependence in the general Spanish population or in individuals in different type of treatments or facilities.
Considering the importance of measuring recovery and personal resources, as well as social and environmental ones along the dependence and its impact on the therapeutic success, it becomes necessary to dispose from an instrument able to measure properly all these aspects. In this way, the VCR scale could show its usefulness in the assessment of capital recovery in alcohol use disorder, although it would require more detailed and precise exploration in further research.
Appendix
Recovery Capital Assessment (ARC) Spanish Version
Versión española de la escala de Valoración del Capital de Recuperación (VCR)
Open access
