AIDS Care Psychological and Socio-medical Aspects of AIDS/HIV ISSN: 0954-0121 (Print) 1360-0451 (Online) Journal homepage: http://www.tandfonline.com/loi/caic20 Posttraumatic growth inventory: factor structure in Spanish-speaking people living with HIV Helena Garrido-Hernansaiz, Rocío Rodríguez-Rey & Jesús Alonso-Tapia To cite this article: Helena Garrido-Hernansaiz, Rocío Rodríguez-Rey & Jesús Alonso-Tapia (2017): Posttraumatic growth inventory: factor structure in Spanish-speaking people living with HIV, AIDS Care, DOI: 10.1080/09540121.2017.1291900 To link to this article: http://dx.doi.org/10.1080/09540121.2017.1291900 View supplementary material Published online: 14 Feb 2017. Submit your article to this journal Article views: 8 View related articles View Crossmark data Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=caic20 Download by: [University of Newcastle, Australia] Date: 17 February 2017, At: 17:20 AIDS CARE, 2017 http://dx.doi.org/10.1080/09540121.2017.1291900 Posttraumatic growth inventory: factor structure in Spanish-speaking people living with HIV Helena Garrido-Hernansaiz a , Rocío Rodríguez-Rey b and Jesús Alonso-Tapia a a Department of Biological and Health Psychology, Psychology Faculty, Universidad Autónoma de Madrid, Madrid, Spain; bDepartment of Psychology, Health Sciences Faculty, Universidad Internacional Isabel I de Castilla, Burgos, Spain ABSTRACT ARTICLE HISTORY This cross-sectional study analyzed the factorial structure of the Posttraumatic Growth Inventory (PTGI) in a sample of 304 Spanish-speaking HIV-positive adults. Participants completed the PTGI and a socio-demographic questionnaire. Exploratory factor analysis (EFA) was carried out through structural equations modeling, with a Varimax rotation. Factors with eigenvalues greater than 1 were extracted, and items with loadings higher than .5 on a factor and lower than .4 on the rest were retained. Two confirmatory factor analyses (CFA) were performed to test a hierarchical model and a bifactor model. Reliability analyses were conducted. EFA suggested a three-factor model keeping 11 of the original 21 items. The three factors that emerged were changes in philosophy of life, in the self and in interpersonal relationships. CFAs suggested that only the bifactor model fitted the data. The three factors as well as the global scale showed good reliability. The factor structure of PTGI’s scores in our data is consistent with the three dimensions theorized by Tedeschi and Calhoun, which speaks in favor of the construct validity of this measure. Received 10 August 2016 Accepted 1 February 2017 Introduction Posttraumatic growth (PTG) represents the positive psychological changes that occur as the result of one’s struggle with a potentially traumatic event. Such positive changes may happen in the philosophy of life (e.g., how the traumatic event may have changed people’s life priorities), the perception of the self (e.g., how this experience may have improved their self-reliance) and interpersonal relationships (e.g., how it may have improved their relationships with others; Tedeschi & Calhoun, 1995, 1996). Thus, living through life’s adverse experiences can have a positive impact. HIV diagnosis is considered a traumatic experience, and although it has been less explored than posttraumatic stress disorder, evidence of PTG has been found in people living with HIV (PLHIV), which in turn has been related to better mental and physical outcomes (Barskova & Oesterreich, 2009; Milam, 2004). The most widely-used instrument for PTG assessment is the Posttraumatic Growth Inventory (PTGI; Tedeschi & Calhoun, 1996). Although originally developed to account for the three above-mentioned dimensions, the validation study found a five-dimensional structure which is often used in research without KEYWORDS Posttraumatic growth inventory; factor structure; structural validity; HIV; Spanish conducting further analyses (Morris, ShakespeareFinch, Rieck, & Newbery, 2005). While some studies have indeed supported this structure (Lee, Luxton, Reger, & Gahm, 2010; Morris et al., 2005), others have found one- (Milam, 2004), three- (Powell, Rosner, Butollo, Tedeschi, & Calhoun, 2003; Rodríguez-Rey, Alonso-Tapia, Kassam-Adams, & Garrido-Hernansaiz, 2016; Weiss & Berger, 2006) and four-factor solutions (Ho, Chan, & Ho, 2004; Taku et al., 2007). Moreover, a recent study which found a five-factor solution explored the possibility of a bifactor model versus a hierarchical one, finding that the former explained the data better (Konkolÿ Thege, Kovács, & Balog, 2014). Therefore, it does not seem justifiable to assume that a five-factor structure of the PTGI is optimal and will hold across different trauma-exposed populations such as PLHIV, and research should also consider complex solutions beyond the number of factors (i.e., hierarchical or bifactor models). Regarding the HIV context, extant literature does not provide sufficient evidence regarding the PTGI dimensions. For instance, Milam (2004) reported the unitary character of the PTGI, but he only used 11 items of the original 21-item PTGI and altered the response format. CONTACT Helena Garrido-Hernansaiz helenagarrido42@gmail.com Departamento de Psicología Biológica y de la Salud, Facultad de Psicología, Universidad Autónoma de Madrid, C/Ivan Pavlov, 6, Madrid 28049, Spain Supplemental data for this article can be accessed here. 10.1080/09540121.2017.1291900 © 2017 Informa UK Limited, trading as Taylor & Francis Group 2 H. GARRIDO-HERNANSAIZ ET AL. Another recent study in PLHIV used the PTGI but did not test its factorial structure (Murphy & Hevey, 2013). Consequently, this study aimed to examine the factor structure of the PTGI in a sample of Spanish-speaking PLHIV so as to contribute to the understanding of this construct in this population. Methods Procedures The present cross-sectional study was approved by the Institutional Review Board at the first author’s University. Participants were either referred to the study by the staff of a healthcare center in Spain (n = 86) or recruited through HIV non-profit organizations which shared information about the study through their online social networks (n = 231). The sample was composed of 304 PLHIV with a mean age of 35.51 years and a mean of 55.75 months since diagnosis. It was predominantly male and homosexual. More details are given in Table S1 in the supplemental material. Instruments Participants reported their age, gender, sexual orientation, country of origin, relationship status, educational level, employment status and time since diagnosis. They also completed the PTGI Spanish version (Weiss & Berger, 2006), a 21-item self-report measure of positive changes after having experienced traumatic events which showed good reliability (Cronbach’s alpha of global scale = .92, philosophy of life = .85, the self = .80, interpersonal relationships = .87). Participants rated each item on a 6-point Likert-scale (0 = I did not experience this change as a result of my crisis; 5 = I experienced this change to a very great degree as a result of my crisis). We substituted “as a result of my crisis” for “as a result of my HIV diagnosis” to ensure that reported PTG was related to HIV diagnosis. Statistical analyses We performed an exploratory factor analysis (EFA) through structural equations modeling (SEM) and we used MLMV as the estimation method, which is adequate for ordinal variables (DiStefano, 2002). Congruently with previous PTGI studies (Tedeschi & Calhoun, 1996; Weiss & Berger, 2006), varimax rotation was applied, factors were extracted if eigenvalues > 1, and items were retained if their loading was > .5 on a factor and < .4 on the rest. We then tested two models in confirmatory factor analyses (CFA) through SEM. Both included the 11 items and three factors resulting from EFA. The hierarchical model had an additional second-order general factor on which the three first-order factors loaded. The bifactor model had an additional general factor on which all the items loaded. The fit of both models was assessed through fit indexes (RMSEA, SRMR, CFI, TLI), following standard criteria (SRMR ≤ .08; RMSEA ≤ .06; CFI, TLI ≥ .95) (Hu & Bentler, 1999). The 90% confidence interval (CI) of RMSEA was also examined for model comparison (Preacher, Zhang, Kim, & Mels, 2013). Proportions of common variance explained by each factor were obtained with the explained common variance index (Rodriguez, Reise, & Haviland, 2016) for the retained model. MPlus 7 was used for all these analyses (Muthén & Muthén, 2012). Reliability was assessed by Cronbach’s alpha (as in previous studies), using SPSS 23. Results The EFA suggested a three-factor solution which explained 59% of the common variance. Table 1 shows the item loadings on each factor and indicates the factor to which each item pertained in the original PTGI validation study (Tedeschi & Calhoun, 1996). Eleven items Table 1. Factor loadings of the three-factor model. Factor loadings in present study New factor and item number Factor in original PTGI I II III Factor I: Philosophy of life #1 AL .78 .17 .21 #2 AL .69 .38 .22 Factor II: Self #4 PS .27 .58 .38 #10 PS .26 .68 .38 #12 PS .38 .73 .32 #19 PS .33 .56 .32 Factor III: Interpersonal relationships #8 RO .23 .36 .68 #9 RO .21 .31 .67 #14 NP .18 .38 .53 #15 RO .39 .12 .65 #16 RO .25 .24 .76 Items failing to load differentially #3 NP .57 .45 .32 #5 SC .43 .32 .40 #6 RO .18 .40 .47 #7 NP .51 .43 .40 #11 NP .45 .64 .40 #13 AL .54 .60 .35 #17 NP .46 .33 .52 #18 SC .22 .25 .38 #20 RO .21 .43 .56 #21 RO .18 .44 .55 Note: Factor loadings > .5 are highlighted in boldface when the item loaded < .4 on the other factors. PTGI = Posttraumatic Growth Inventory. AL = appreciation of life. NP = new possibilities. PS = personal strength. RO = relating to others. SC = spiritual change. AIDS CARE Table 2. Model fit statistics for two models tested with confirmatory factor analysis. Model type RMSEA (90% CI) SRMR CFI TLI Hierarchical Bifactor .10 (.08–.11) .05 (.02–.07) .08 .03 .91 .98 .88 .97 Note: df = degrees of freedom. p = level of significance. CI = confidence interval. were retained, their content was inspected, and factor labels were generated: Factor I = positive changes in philosophy of life, Factor II = positive changes in the self, and Factor III = positive changes in interpersonal relationships. Pearson’s correlation between the 21-item and the 11-item versions of the PTGI was .98 (p < .001), indicating that there was no significant loss of information. Confirmatory factor analyses were then conducted. Table 2 shows the fit indices of the hierarchical and bifactor models. Those of the former fell short of the standard limits of acceptance while those of the latter were excellent. Moreover, there was no overlapping between the two models concerning the 90% CI of RMSEA. Thus, the bifactor model was retained and is depicted in Figure 1 along with the factor loadings and squared multiple correlations. Of the 100% common variance, the general factor explained 72% and the three specific factors explained 28%: 9% was explained by Factor I, 5% by Factor II and 14% by Factor III. Cronbach’s alpha for the whole 11-item scale was .92, and was as follows for the factors: Factor I = .79; Factor II = .87; and Factor III = .87. Discussion A three-factor structure of the PTGI emerged as the one with the best fit in PLHIV. Eleven items were retained, which is similar to the number of items retained in validation studies for different languages (Ho et al., 2004; Powell et al., 2003; Weiss & Berger, 2006). Three dimensions of PTG emerged in our sample – philosophy of life, the self and interpersonal relationships – which are congruent with the three PTG components originally theorized (Tedeschi & Calhoun, 1995), thus supporting the construct validity of the instrument. Moreover, a bifactor structure explained data better than a hierarchical one (Konkolÿ Thege et al., 2014), which supports the idea of a common underlying theoretical model of PTG (Tedeschi & Calhoun, 1996; Weiss & Berger, 2006). The global scale and the factors had good to excellent reliability. Nevertheless, individuals not using online social networks or attending the healthcare center had little opportunity to be recruited, and the sample was Spanish-speaking and mostly composed of males, so findings should not be generalized to female PLHIV or nonSpanish speakers. Research should aim to overcome these limitations, replicate our findings, and also examine whether there are cultural differences among Spanish-speakers and whether other growth dimensions not currently reflected in the PTGI may emerge after HIV diagnosis. Our study has shown the importance of studying the latent structure of the PTGI before computing and interpreting its scores, as it varies across populations and a five-dimensional structure cannot always be assumed. Health caregivers interested in fostering PTG in PLHIV should do so along the three dimensions proposed by Tedeschi and Calhoun (1995) – philosophy of life, the self and interpersonal relationships – and they may do so by helping PLHIV reflect on which Figure 1. Bifactor model of the posttraumatic growth inventory. Standardized solution. Note: Squared multiple correlations are highlighted in boldface. 3 4 H. GARRIDO-HERNANSAIZ ET AL. ways such critical event could lead not only to distress, but also have a positive legacy. Geolocalization information The paper reports data concerning Spain and other Latin American countries. Acknowledgements The authors have no conflict of interest to disclose. They would like to acknowledge the HIV associations in Spain and Latin America and the healthcare providers at Centro Sandoval, in Madrid, whose help was fundamental for the data collection. The first author would like to acknowledge the financial support given by the Spanish Ministerio de Educación, Cultura y Deporte through a FPU fellowship. Disclosure statement No potential conflict of interest was reported by the authors. ORCID Helena Garrido-Hernansaiz http://orcid.org/0000-00018715-0842 Rocío Rodríguez-Rey http://orcid.org/0000-0001-8006-5012 Jesús Alonso-Tapia http://orcid.org/0000-0001-6544-0224 References Barskova, T., & Oesterreich, R. (2009). Post-traumatic growth in people living with a serious medical condition and its relations to physical and mental health: A systematic review. Disability and Rehabilitation, 31(21), 1709–1733. doi:10. 1080/09638280902738441 DiStefano, C. (2002). The impact of categorization with confirmatory factor analysis. Structural Equation Modeling: A Multidisciplinary Journal, 9(3), 327–346. doi:10.1207/ S15328007SEM0903_2 Ho, S. M. Y., Chan, C. L. W., & Ho, R. T. H. (2004). Posttraumatic growth in Chinese cancer survivors. PsychoOncology, 13(6), 377–389. doi:10.1002/pon.758 Hu, L., & Bentler, P. M. (1999). Cutoff criteria for fit indexes in covariance structure analysis: Conventional criteria versus new alternatives. Structural Equation Modeling: A Multidisciplinary Journal, 6(1), 1–55. doi:10.1080/ 10705519909540118 Konkolÿ Thege, B., Kovács, É., & Balog, P. (2014). A bifactor model of the posttraumatic growth inventory. Health Psychology and Behavioral Medicine, 2(1), 529–540. doi:10.1080/21642850.2014.905208 Lee, J. A., Luxton, D. D., Reger, G. M., & Gahm, G. A. (2010). Confirmatory factor analysis of the posttraumatic growth inventory with a sample of soldiers previously deployed in support of the Iraq and Afghanistan wars. Journal of Clinical Psychology, 66(7), 813–819. doi:10.1002/jclp.20692 Milam, J. E. (2004). Posttraumatic growth among HIV/AIDS patients. Journal of Applied Social Psychology, 34(11), 2353–2376. doi:10.1111/j.1559-1816.2004.tb01981.x Morris, B. A., Shakespeare-Finch, J., Rieck, M., & Newbery, J. (2005). Multidimensional nature of posttraumatic growth in an Australian population. Journal of Traumatic Stress, 18(5), 575–585. doi:10.1002/jts.20067 Murphy, P. J., & Hevey, D. (2013). The relationship between internalised HIV-related stigma and posttraumatic growth. AIDS and Behavior, 17(5), 1809–1818. doi:10.1007/s10461013-0482-4 Muthén, L. K., & Muthén, B. O. (2012). Mplus user’s guide (7th ed.). Los Ángeles, CA: Author. Powell, S., Rosner, R., Butollo, W., Tedeschi, R. G., & Calhoun, L. G. (2003). Posttraumatic growth after war: A study with former refugees and displaced people in Sarajevo. Journal of Clinical Psychology, 59(1), 71–83. doi:10.1002/jclp.10117 Preacher, K. J., Zhang, G., Kim, C., & Mels, G. (2013). Choosing the optimal number of factors in exploratory factor analysis: A model selection perspective. Multivariate Behavioral Research, 48(1), 28–56. doi:10.1080/00273171. 2012.710386 Rodriguez, A., Reise, S. P., & Haviland, M. G. (2016). Evaluating bifactor models: Calculating and interpreting statistical indices. Psychological Methods, 21(2), 137–150. doi:10.1037/met0000045 Rodríguez-Rey, R., Alonso-Tapia, J., Kassam-Adams, N., & Garrido-Hernansaiz, H. (2016). The factor structure of the posttraumatic growth inventory in parents of critically ill children. Psicothema, 28(4), 495–503. doi:10.7334/ psicothema2016.162 Taku, K., Calhoun, L. G., Tedeschi, R. G., Gil-Rivas, V., Kilmer, R. P., & Cann, A. (2007). Examining posttraumatic growth among Japanese university students. Anxiety, Stress, and Coping, 20(4), 353–367. doi:10.1080/10615800701295007 Tedeschi, R. G., & Calhoun, L. G. (1995). Trauma and transformation: Growing in the aftermath of suffering. Thousand Oaks, CA: Sage. Tedeschi, R. G., & Calhoun, L. G. (1996). The posttraumatic growth inventory: Measuring the positive legacy of trauma. Journal of Traumatic Stress, 9(3), 455–471. doi:10.1002/jts. 2490090305 Weiss, T., & Berger, R. (2006). Reliability and validity of a Spanish version of the posttraumatic growth inventory. Research on Social Work Practice, 16(2), 191–199. doi:10. 1177/1049731505281374