If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Correspondence to: Mehdi Heidarzadeh Ph.D., RN, Department of Medical-Surgical Nursing, Nursing & Midwifery School, Ardabil University of Medical Sciences, Shohada Street, Ardabil, Iran.
This study aimed to assess the factor structure of the Persian version of posttraumatic growth inventory (P-PTGI).
Methods
Participants included 272 Iranian people of Azari ethnicity (111 women and 161 men), aged between 21 and 91 years (mean 52.65 years), who were diagnosed with cancer and were referred to the oncology department of the university hospital. The P-PTGI was assessed to determine the construct validity, using various indices of confirmatory factor analysis and standardized lambda coefficient, followed by further assessment of the discriminant and convergent validities by using the structural equation model. LISREL 8.8 for Windows and SPSS were used for data analysis.
Results
The calculated values of comparative fit index, incremental fit index, normed fit index, and non-normed fit index > .90 and the values of standardized root-mean-square residual < .08 indicate an acceptable fit for the original PTGI. Considering that the values of average variance extracted (.52–.74) were greater than the square of correlation coefficients between the five dimensions of P-PTGI, discriminant validity was approved. Convergent validity was confirmed through a high value of standardized lambda coefficient (.52–.92) between the items and their related factors.
Conclusions
This study revealed that P-PTGI has an acceptable validity and reliability for posttraumatic growth assessment in Iranian cancer patients and its factor structure is similar to that of the original form developed by Tedeschi and Calhoun.
]. In particular, although stressful events like cancer lead to adverse physical, mental, and social effects, fighting such events may cause certain positive impact on various life aspects [
] defined PTG as a set of subjective positive psychological changes experienced following the struggle with highly challenging life circumstances. PTG can be a result of the traumatic conditions evoked because of the diagnosis and treatment of cancer [
]. According to previous studies, the diagnosis and treatment of cancer have been found to be associated with certain positive psychological outcomes [
Several instruments have been designed to evaluate PTG, of which the most applicable and efficient available tools is the 21-item posttraumatic growth inventory (PTGI), designed by Tedeschi and Calhoun [
]. The five categories of psychological growth after dealing with stressful events, as assessed by PTGI are: new possibilities, relating to others, appreciation of life, personal strength, and spiritual change. The validity and reliability of the PTGI have been confirmed by various studies from numerous countries and among populations who have experienced these different stressful events [
The validity and reliability assessment of PTGI faces several challenges, of which one challenge relates to the fact that PTGI is based on a study conducted on students with an experience of traumatic events such as death of loved ones, etc. [
]. Moreover, the samples used for evaluating the validity and reliability were mainly selected from participants with an experience of nonclinical traumatic events [
]. Few studies have also investigated PTGI in participants with an experience of cancer.
Furthermore, another challenge is the disagreement among various studies on the PTGI factor structure. Although PTGI was used extensively in the past two decades, its dimensionality still remains unclear. Tedeschi and Calhoun introduced PTGI as a five-factor structural tool; however, all previous researches did not approve its five-dimensional structure [
] examined the PTGI structure in 379 undergraduate student samples reporting Diagnostic and Statistical Manual of Mental Disorders 4th edition (DSM-IV) Criterion-A potentially traumatic events; herein, confirmatory factor analysis (CFA) did not support the original five-factor structure. Therefore, they concluded that the PTGI factor structure is unclear among individuals with DSM-IV traumatic events. In a study by Powell et al [
], the factor structure of the original instrument was nonreproducible, and a three-factor model with 20 items was considered appropriate; alternatively, in a study by Weiss and Berger [
], using sample of Latina immigrants, a three-factor model with 13 items was considered more appropriate than the original five-factor model.
Considering the disagreement in dimensions of PTGI, there seems to be some potential for factor overlapping in PTGI. In particular, some studies reported profound correlations among the factors [
] noted poor discriminant validity between the new possibilities and personal strength subscales. Thus, PTGI is presumed to be efficiently represented by less than five factors.
These literatures suggest that the PTGI factor structure is unclear among individuals experiencing various traumatic events in different countries. The resulting factor structure variation may be due to the differences in traumatic events or the culture of the participants. Furthermore, meticulous exploration of the stability of individual factors and their ability to capture the unique aspects of PTG is required. Assessment of the psychometric properties of PTGI in clinical disorders such as cancer is essential considering the extensive use of PTGI with individuals reporting nonclinical traumatic events (e.g., sexual assault, war). Cultural aspects seem to be important variables in affecting the dimensions of PTG in different countries. By use of various coping strategies, culture plays a crucial role in enhancing the individual's growth that experiences the traumatic event [
in: Weiss T. Berger R. Posttraumatic Growth and Culturally Competent Practice: Lessons Learned From Around the Globe. 978-0-470-35802-3 Wiley,
New York2010
] and clinical (e.g., cancer) or nonclinical stressful events; considering the lack of appropriate research on the factor structure of PTGI in Iran, the evaluation of the psychometric properties of PTGI via a study on the clinical samples of Iranian patients seems essential. Therefore, we translated PTGI and validated its Persian version in cancer patients from the Iranian population. Considering certain challenges related to the factor structure of PTGI, we focused on the construct validity rather than other issues. In this study, we aimed to determine the psychometric properties of the Persian version of PTGI (P-PTGI) and to explore whether the original five-factor structure was replicable in the cancer patients.
Methods
This methodological study translated the PTGI and validated its Persian version among Iranian cancer patients.
Participants
The study population was comprised of Iranian patients of Azari ethnicity suffering from cancer, who were referred to the oncology department of a university hospital. A total of 530 cancer patients were admitted to the clinics over a period of 8 months (from June 2015 to January 2016); however, 210 of these patients did not have the inclusion criteria (168 of them had a cancer history of less than 1 year, 11 aged less than 21 years, and 31 of these had communication barriers). Eventually, of the 320 remaining patients, 85% consented to participate in the study. Given that minimum sample size required to CFA analysis is 5 to 20 participants per each item [
], 272 participants were included in the study. Inclusion criteria included a minimum age of 21 years, cancer diagnosis had been confirmed by an oncologist, suffered from cancer for at least 1 year and were aware of their disease.
Instruments
Instruments included a demographic questionnaire and PTGI. This inventory has 21 items that identify the five dimensions of psychological growth after a traumatic event. All items are scored based on a six-point Likert scale from 0 (no growth) to 5 (highest growth). Although the total scores of PTGI can vary between 0 and 105, higher scores reflect substantial growth [
]. Patients who met the inclusion criteria were invited to participate in the study; briefed on the objectives of the study, reassured regarding the confidentiality of their data, and their written consent was obtained; questionnaires were then filled out individually. Moreover, the researchers read the items to the illiterate individuals and marked their responses.
Translation process
We translated and validated the PTGI according to the method suggested by Wild et al [
Principles of good practice for the translation and cultural adaptation process for patient-reported outcomes (PRO) measures: report of the ISPOR task force for translation and cultural adaptation.
]. After obtaining written consent from the instrument designer, the PTGI was translated into Persian by two people who were fluent in Persian and English. Furthermore, the two translations were compared and edited, and a final copy was prepared. Subsequently, the final translated version was translated back into English by another person who was fluent in both English and Persian. Moreover, two versions of PTGI (the back translation and original scale) were assessed by a supervisor, and few minor revisions were carried out in the Persian version. There was no difference between PTGI and P-PTGI in the number and content of items.
Statistical analysis
The validation process was carried out using the content and face validity, construct validity, and test–retest (after 30 days) and internal consistency reliability. To determine the construct validity, factorial validity of the P-PTGI was initially assessed using indices of CFA and standardized λ coefficient, followed by further assessment of the discriminant and convergent validities by using the structural equation model (SEM). To determine the discriminant validity of the test, the factors in the test were supposed to be so distinct despite their associations that no two factors could be related as one or same factor. The two stages used in this study to determine the discriminant validity were: (1) unidimensionality and multidimensionality of P-PTGI were determined using the difference in fit indices of fixed and free solutions based on the assumption that the multidimensional model should be considered if the unidimensional model did not fit [
] and (2) the average variance extracted (AVE) was compared with the square of correlation between factors. The evidence for discriminant validity is known to be provided when the square of correlations between factors is less than every single AVE [
]. To determine the convergent validity of the test using the SEM method, we considered all values of AVE, and the standardized λ coefficient values exceeded .50.
LISREL 8.8 for Windows (SSI Inc., Skokie, IL, USA) was used for evaluating the fitness indices, and convergent and discriminant validities of all five dimensions of the P-PTGI. Internal consistency reliability and test–retest were performed by using SPSS for Windows version 22 (SPSS Corporation, Chicago, IL, USA). Internal consistency reliability of the scale was calculated by Cronbach α. The correlation between test–retest among 20 patients with a 30-day interval was used to assess the repeatability.
Ethical considerations
Before commencement of the study, the researchers obtained the ethical approval of the Ethics Committee of Ardabil University of Medical Sciences (Approval no. IR.ARUMS.1394.6). After being briefed by the researchers regarding the study objectives and methods, all participants filled the questionnaires and gave their consent.
Results
Sample description
The demographic and clinical characteristics of the participants are shown in Table 1. A total of 272 patients (111 men and 161 women) aged between 21 and 91 years (mean 52.65 years), suffering from cancer were evaluated in the study. The educational qualification of the participants was 40.8% illiterate, 34.9% completed primary school, 16.2% completed secondary school, and 8.1% were university graduates. Most patients (87.2%) were married, 22.8% were self-employed, 10.3% were employed elsewhere, and a majority (51.8%) were housewives. Table 1 displays the duration of postcancer diagnosis and the types of cancer.
Table 1Demographic Characteristics of Patients with Cancer (N = 272).
For determining the content validity, the content validity index (CVI) was calculated. To assess CVI, six oncology nursing instructors, one oncologist, two psychiatric nurses, five nursing professors, and one educational supervisor were asked to comment on the translated scale. These people were chosen based on their experience in handling cancer patients or working on instrument development. These 15 experts scored each item of P-PTGI, in a range of 1 to 4, for relevance, simplicity, and clarity. CVI for each item was calculated as the number of experts who gave a rating of either 3 or 4 divided by the total number of experts; CVI in the total scale was calculated as the proportion of items on the scale that achieved a rating of 3 or 4 divided by all the content experts [
]. Based on the views of all 15 experts, the CVI of the total scale was calculated as .95. These experts also confirmed the face validity of the P-PTGI. The Persian version was finalized with some minor revision in its content.
Construct validity
To determine the construct validity, initially, CFA was used to assess two models; in the first model, we considered P-PTGI as a five-factor model and 21 items that is completely similar to the original version of PTGI derived by Tedeschi and Calhoun (Model I). In the second model, 21 items of P-PTGI was evaluated as a one-factor model (Model II). The overall model fit statistics are presented in Table 2 and the standardized parameter estimates for both models are shown in Figure 1, Figure 2.
Table 2Goodness-of-fit Indices of Original and One-dimensional Models of PTGI.
Judgment of the goodness of fit was based on indices used in the SEM. Nonsignificant χ2 generally indicates that the model fits the data appropriately; however, χ2 is sensitive to large sample size effects. Consequently, we also used comparative fit index, normed fit index, non-normed fit index, incremental fit index, root-mean-square error of approximation, and standardized root-mean-square residual. In the Model I, the values of comparative fit index, incremental fit index, normed fit index, and non-normed fit index > .90 indicated acceptable fit. Value of standardized root-mean-square residual < .08 also indicated a good fit for Model I. But Model II was rejected because of unacceptable value of these indices. Nevertheless, as proving correlations between the factors (unidimensionality of the tool) is associated with the failure of the model in terms of fit coefficients, the factors can be considered to have distinct features from each other, and cannot be considered as one factor (Table 2).
Discriminant and convergent validities
Furthermore, discriminant and convergent validities were used as evidence of construct validity. The rejection of the unidimensionality model of P-PTGI is an evidence for discriminant validity and confirms the multidimensionality of P-PTGI (Table 2 and Figure 2). For a more precise assessment of the discriminant validity, the AVE was compared with the square of correlation between factors. As shown in Table 3, the AVE (diagonal values in bold) of the each construct is higher than the square of correlation coefficients between the respective constructs. Referring to Table 3, one can conclude that the discriminant validity for all five constructs is achieved.
Table 3The Discriminant Validity Index Summary for the Construct in Model I.
These are the square of correlation coefficients between the respective constructs. As the AVE of the each construct is higher than the square of correlation coefficients between the respective constructs, the discriminant validity for all five constructs is achieved.
These are the square of correlation coefficients between the respective constructs. As the AVE of the each construct is higher than the square of correlation coefficients between the respective constructs, the discriminant validity for all five constructs is achieved.
These are the square of correlation coefficients between the respective constructs. As the AVE of the each construct is higher than the square of correlation coefficients between the respective constructs, the discriminant validity for all five constructs is achieved.
These are the square of correlation coefficients between the respective constructs. As the AVE of the each construct is higher than the square of correlation coefficients between the respective constructs, the discriminant validity for all five constructs is achieved.
These are the square of correlation coefficients between the respective constructs. As the AVE of the each construct is higher than the square of correlation coefficients between the respective constructs, the discriminant validity for all five constructs is achieved.
These are the square of correlation coefficients between the respective constructs. As the AVE of the each construct is higher than the square of correlation coefficients between the respective constructs, the discriminant validity for all five constructs is achieved.
These are the square of correlation coefficients between the respective constructs. As the AVE of the each construct is higher than the square of correlation coefficients between the respective constructs, the discriminant validity for all five constructs is achieved.
These are the square of correlation coefficients between the respective constructs. As the AVE of the each construct is higher than the square of correlation coefficients between the respective constructs, the discriminant validity for all five constructs is achieved.
These are the square of correlation coefficients between the respective constructs. As the AVE of the each construct is higher than the square of correlation coefficients between the respective constructs, the discriminant validity for all five constructs is achieved.
These are the square of correlation coefficients between the respective constructs. As the AVE of the each construct is higher than the square of correlation coefficients between the respective constructs, the discriminant validity for all five constructs is achieved.
.52
Diagonal values in bold indicating AVE of the each construct.
a Model I = The five-factor PTGI including relating to others, new possibilities, personal strength, appreciation of life, and spiritual changes.
b These are the square of correlation coefficients between the respective constructs. As the AVE of the each construct is higher than the square of correlation coefficients between the respective constructs, the discriminant validity for all five constructs is achieved.
To determine the convergent validity, a high value of standardized λ coefficient between measuring indices and factors and a significance of item-loading according to t tests as relevance of each item, was considered. Path coefficients (.52–.92) and their significance suggest that items are largely associated with their measuring factor. Moreover, the convergent validity for the measurement model is achieved when all values of AVE exceed .50. Table 4) shows all AVE > .50.
Table 4Convergent Indices and Some Parameters of PTGI
Cronbach α was .93 for the whole P-PTGI and ranged r = .68–.93 for the five subscales. The test–retest correlation for the P-PTGI score using 20 patients (who were selected randomly from all the 272 participants) was r = .81 (p < .001) during 30-day interval between two tests that reported the consistency of P-PTGI. Thus, the results indicated the stability of the total scale over time (Table 4).
Discussion
The PTGI is the most applicable tool for measuring growth in people with clinical [
Confirmatory factor analysis of the posttraumatic growth inventory with a sample of soldiers previously deployed in support of the Iraq and Afghanistan wars.
]. Although most previous studies have evaluated and confirmed the validity and reliability of the PTGI, the most important challenge was related to its structural factor [
]. This study aimed to assess the psychometric properties of P-PTGI with an emphasis on exploring the factor structure in cancer patients. For determining the psychometric characteristics of P-PTGI, we assessed the content and construct validity, test–retest, and internal consistency.
This study used a clinical sample for the evaluation of the psychometric properties of P-PTGI; whereas, most studies in this context used participants who experienced nonclinical traumatic events [
Confirmatory factor analysis of the posttraumatic growth inventory with a sample of soldiers previously deployed in support of the Iraq and Afghanistan wars.
]. Another distinctive feature of this study included its profound emphasis on construct validity, and in particular, in determining the discriminant and convergent validities according to SEM. Discriminant validity is observed when measurements of different concepts or constructs are distinct (i.e., there are low correlations among the concepts) [
]. Consistently, correlations between factors and internal measures of each are chiefly used to assess the discriminant and convergent validities. Despite its simplicity, this method has certain issues, of which the most important is the lack of a specific criterion to confirm or reject its validity [
]. Using SEM provides a stronger evidence for determining the discriminant and convergent validities.
Although this study confirmed the original five-factor structure of P-PTGI, literature reviews have revealed certain challenges in the structure of PTGI; hence, some researches confirmed the five-factor structure of PTGI [
Confirmatory factor analysis of the posttraumatic growth inventory with a sample of soldiers previously deployed in support of the Iraq and Afghanistan wars.
Confirmatory factor analysis of the posttraumatic growth inventory with a sample of soldiers previously deployed in support of the Iraq and Afghanistan wars.
] confirmed the five-factor structure of PTGI using a sample of active duty soldiers exposed to combat in Iraq or Afghanistan; moreover, Alex Linley et al [
] reported good fitness of the model after evaluating a sample of 372 students who experienced the stressful life events. Nevertheless, some researchers have not confirmed the five-factor structure of PTGI, and have claimed that it should be restructurized [
]; For instance, despite the good reliability of the Chinese version of the PTGI in a sample of 188 cancer survivors, researchers reached a factor structure different from the original version with four rather than five factors [
] validated the Spanish version of PTGI using 100 people with distressing experiences of immigration. Although they found the translated version highly reliable, factor analysis failed to confirm the original five factors and a three-factor model was hence suggested.
Several factors appear to influence the difference in the factor structure considering various studies, of which is the role of culture and personal–social differences. Considerably, Tedeschi and Calhoun [
] state that “people from different cultures have different views on traumatic events and adopt different strategies to cope with the problem.” The distinct nature of PTGI in various cultures such as United Kingdom [
] samples may be attributed to their cultural differences.
Another important factor contributing to the difference in factor structure in various studies is their different approach for the PTGI factor structure assessment. CFA is generally used in assessing the factor structure of a tool whose factors have already been defined [
Confirmatory factor analysis of the posttraumatic growth inventory with a sample of soldiers previously deployed in support of the Iraq and Afghanistan wars.
] have primarily assessed and successfully confirmed the PTGI factor structure by using CFA; however, when the initial structure was not considered in assessing the factor structure of the tool, researchers studied different structures that were compatible with the demographic features [
]. This reveals that using exploratory factor analysis approach in assessing the PTGI factor structure has not been ineffective because of its distinct nature.
Most previous studies have not used discriminant validity to determine the psychometric features of PTGI. An exception to this is the study by Taku et al [
], which in contrast to the present study reported no difference between “spiritual changes” and “appreciation of life” dimensions. Therefore, Taku et al suggested a four-factor structure, in which three factors are similar to “relating to others,” “personal strength,” and “new possibilities” dimensions in the original PTGI, and the fourth factor is a combination of “spiritual changes” and “appreciation of life” dimensions in the original version. Nevertheless, our study proved that although all 21 items of PTGI and its subscales are intercorrelated (good convergent validity), these correlations are not profound enough to assume them as a one-factor structure tool; hence, the results confirmed PTGI as having five separate domains (good discriminant validity). In general, using accurate statistical techniques, the present study strongly supported the five-factor tool, and unlike previous studies showed that P-PTGI structure is perfectly compatible with the original version.
The five-factor structure of P-PTGI was confirmed repeatedly by the acceptable alpha coefficients found in the present study. Cronbach α is the most widely used method for evaluating internal consistency [
]. We could obtain acceptable Cronbach α for the five dimensions of the P-PTGI. This indicates that all items in each dimension are aligned to each other, and it provides another support for a five-dimensional structure of P-PTGI. Collectively, high test–retest correlations in the present study indicated good reliability of P-PTGI in Iranian cancer patients.
Confirmatory factor analysis of the posttraumatic growth inventory with a sample of soldiers previously deployed in support of the Iraq and Afghanistan wars.
] have confirmed the factor structure of PTGI, comparing the subscales of those studies reveals that the mean items score obtained differ with the present study. For instance, participants in the aforementioned four studies [
Confirmatory factor analysis of the posttraumatic growth inventory with a sample of soldiers previously deployed in support of the Iraq and Afghanistan wars.
] scored the highest in “appreciation of life” and “personal strengthening” dimensions and the lowest in “spiritual changes”; whereas, in the present study, the highest items mean scores were observed in “spiritual changes” and “relationship with others” and the lowest in “new opportunities.” This indicates that although PTGI factor structure in Iran is similar to that in the French, British, and American societies, and that people experience PTGI in all dimensions in stressful events, the experience of these dimensions may differ depending on factors such as cultural background and religious belief. Hence, future studies on PTG in different societies reporting differences in their dimensions and implementing these differences in their plans to facilitate growth or adjustment of patients are required.
This study had certain limitations. First, this study used cancer patients as participants, which differed from most aforementioned studies. This may affect the result comparison, which was not too evident, but should always be considered. Second, this study used a convenient sample of cancer patients whose results may not generalize to other clinical groups. Third, the study only determined the discriminant and convergent validities according to the statistical methods, and did not use other known constructs to determine the theory-based construct validity. Fourth, the researchers had to fill the questionnaires through interviews in about 40% of the participants who were illiterate.
Conclusions
The findings of the present study confirm the original five-factor structure and 21 items of PTGI developed by Tedeschi and Calhoun, and support the applicability of the P-PTGI for measuring perceived growth after cancer diagnosis. Nevertheless, further studies on PTG in cancer patients in Iran are warranted.
The findings of this study have certain applications in nursing practice and research. Considering the confirmed validity and reliability of the P-PTGI, the present study results can prove to be essential in conducting further studies on positive aspects in cancer patients in Iran. As PTG is a concept that occurs in cancer patients, introducing a standard tool can familiarize the nurses and psychology consultants with positive psychological changes and its dimensions in cancer patients. They can reinforce these dimensions, and use them as coping strategies for patients. Considering the culture of different societies, a different factor structure may be defined for PTGI; however, the results of CFA in the present study confirmed a five-dimensional factor structure, which emphasizes the importance of CFA in assessing theory-driven scales.
Conflicts of interest
The authors have no funding or conflicts of interest to disclose.
Acknowledgement
This study was supported and approved by the ethics committee of Ardabil University of Medical Sciences. The authors would like to thank all the patients, their families, the personnel of oncology units and clinics and authorities of Tabriz hospital, the vice-presidency of research, Ardabil University of Medical Sciences, and all who helped them in doing this study.
References
Schroevers M.J.
Helgeson V.S.
Sanderman R.
Ranchor A.V.
Type of social support matters for prediction of posttraumatic growth among cancer survivors.
in: Weiss T. Berger R. Posttraumatic Growth and Culturally Competent Practice: Lessons Learned From Around the Globe. 978-0-470-35802-3 Wiley,
New York2010
Principles of good practice for the translation and cultural adaptation process for patient-reported outcomes (PRO) measures: report of the ISPOR task force for translation and cultural adaptation.
Confirmatory factor analysis of the posttraumatic growth inventory with a sample of soldiers previously deployed in support of the Iraq and Afghanistan wars.
30077-9&id=gr1.jpg){kind=link}
30077-9&id=gr2.jpg){kind=link}