The Chinese version of Posttraumatic Growth and Depreciation Inventory—Expanded version (PTGDI-X) for cancer survivors

Many patients with cancer report positive changes often referred to as posttraumatic growth (PTG). Some of these self-reported PTG may represent maladaptive illusions created by individuals to cope with the illness. A recently established Posttraumatic Growth and Depreciation Inventory – Expanded version (PTGDI-X) includes both PTG and posttraumatic depreciation (PTD) items. This inventory may provide a more balanced picture of the phenomenological world of cancer survivors. We examined the Chinese version of the PTGDI-X’s applicability to cancer patients, and how PTG and PTD were related to posttraumatic stress symptoms. Two hundred sixty-five cancer survivors in Taiwan completed the Chinese version of the PTGDI-X, along with the PTSD Checklist for the DSM-5 to measure posttraumatic stress disorder (PTSD) symptoms. Confirmatory factor analysis showed that the factor structure of the PTGDI-X established in a multi-national study fit our data from cancer patients modestly well. The PTD score had a significant and positive correlation with PTSD symptoms, whereas the PTG and PTSD showed a significant curvilinear relationship in the form of an inverted U-shape. This study’s results indicate that PTG and PTD are separated constructs with differential relationships with cancer outcomes. The Chinese version of the PTGDI-X is a viable instrument for psycho-oncological research. The PTD scores can provide useful information to guide cognitive interventions to reduce distorted cognitions. In contrast, the PTG scores can provide further information on the phenomenological world of cancer survivors. In this study, clinical implications and future studies were considered.


Introduction
Many cancer patients and survivors report positive psychological changes after their diagnosis and treatment [17,40]. This phenomenon is known as posttraumatic growth (PTG) [42,44] or stress-related growth [35,37] in the psychological literature. The PTG phenomenon affirms that psychopathology is not the only outcome of cancer diagnosis and treatment [19]. On the other hand, many researchers have proposed that self-reported PTG may be a defensive, self-enhancing, cognitive bias to temporarily cope with distress, rather than an indicator of genuine positive changes [47]. To support this hypothesis, Maercker and Zoellner [32] suggested a two-component Janus face model of PTG, comprising a constructive and an illusory component. The constructive component represents a genuine and adaptive transformation of a cognitive schema. The illusory component represents a coping mechanism to maintain psychological equilibrium after a highly stressful event, such as a cancer diagnosis [1]. In a 7-year longitudinal study of a group of breast cancer survivors in Taiwan conducted between 2009 and 2016, the latent class growth analysis revealed both illusory and constructive growth patterns [16]. Survivors in the illusory growth group tended to report a moderate level of PTG but also exhibited higher levels of maladaptive coping strategies, especially Hopelessness-Helplessness when compared to those in the constructive growth group. The researchers then conducted a follow-up study on the 10th anniversary of the project [14]. They reported that individuals in the illusory growth group, compared to those in the constructive PTG group, exhibited higher depression and anxiety levels in 2019 than they did in 2016. Supporting the above findings, a more recent study also reported a two-class model of long-term anxiety trajectories among breast cancer survivors, with a resistant group (85.2%) and a distress group (14.8%) [13].
Another issue related to PTG is that all common instruments used to measure PTG, such as the Stress-Related Growth Scale (SRGS) [36] and the Posttraumatic Growth Inventory (PTGI) [43], only contain items related to growth. Therefore, the respondents focused more on the positive outcomes of their traumatic experience, which may consequentially increase the self-enhancing biases related to illusory growth. Boals and Schuler [6] developed a revised version of the Stress-Related Growth Scale to address the above issue. First, the researchers changed the positively worded statement of the original SRGS to a neutral statement. Second, they used a bipolar response scale (− 3 = a very negative change to 0 = no change to + 3 = a very positive change) to allow for reporting of negative and positive changes. Similarly, Baker et al. [3] developed a Paired-Format Posttraumatic Growth Inventory (PTGI-42) that contains both positive and negative posttraumatic change items in a single instrument. The PTGI-42 is based on the original 21-item PTGI [43], with an addition of 21 matched but negatively worded items to represent the negative changes, referred to as depreciation (in contrast to growth). For example, "I am more likely to try to change things that need changing (a growth item)" versus "I am less likely to try to change things that need changing (a depreciation item)." The growth and depreciation items are presented in pairs to encourage the participants to consider both types of change, growth, and depreciation, at the same time [10]. It has been shown that putting the growth and depreciation items in the same domain in pairs did not create difficulties in responding for the participants. This suggests that posttraumatic growth (PTG) and posttraumatic depreciation (PTD) are separate constructs and independently influence psychological outcomes after a traumatic encounter [3,10]. Taku et al. [41] recently conducted an international study to examine the factor structure of the expanded version of the PTGI-42, known as the Posttraumatic Growth and Depreciation Inventory-Expanded version (PTGDI-X), in 10 countries (Australia, n = 2261; Germany, n = 202; Italy, n = 321; Japan, n = 130; Nepali, n = 196; Peru, n = 201; Poland, n = 287; Portugal, n = 181;Turkey, n = 169; and the USA, n = 214). The inventory was based on the PTGI-42 with four additional spiritual-existential change (SEC) items developed by Tedeschi and colleagues [45]. The resulting PTGDI-X has 25 items related to PTG and another 25 parallel items related to PTD. The PTG and PTD items of the same life domains were presented in pairs. The PTGDI-X was translated into the language of each country. A heterogeneous sample was obtained in this study, in which most countries included undergraduate students, though some countries included a community sample and emergency workers (e.g., firefighters). Most countries collected data online, whereas the paper-and-pencil survey was adopted by some (e.g., Japan and Nepali). The confirmatory factor analysis demonstrated that a standard factor structure based on the theoretical model of the original PTGI [43] fitted the data modestly well. Similar to the findings of two previous studies on PTGI-42 [3,10], higher scores were observed in PTG than in PTD in all countries [41]. Moreover, PTD had a positive correlation with PTSD symptoms in all countries, but the relationship between the PTG and PTSD symptoms was inconsistent across countries,no correlation was reported in five of the countries, while in the other five countries, a curvilinear correlation was reported [41].
The psychometric properties of the Chinese version of the PTGI were first reported in 2004 in research on cancer patients in Hong Kong [21]. Over the past 16 years, the inventory was validated among cancer survivors in Taiwan [23] as well as patients with other types of chronic illnesses in Hong Kong [15]. With the establishment of the PTGDI-X across cultures, it is necessary to examine the psychometric properties of the Chinese version of PTGDI-X to facilitate both clinical judgment and research in Chinese speaking communities. Although the revised version of the Stress-Related Growth Scale [6] is available to allow for the assessment of both positive and negative changes, this scale is less popular than the PTGI both as a cancer research and clinical instrument in Chinese speaking communities. This study aims to examine the applicability and psychometric properties of the PTGDI-X among Chinese cancer survivors. Specifically, the goodness-of-fit of the factor structure proposed by Taku et al. [41] will be examined in a sample of Chinese cancer survivors. This approach will help to facilitate comparisons of research findings across countries in the future. Furthermore, the relationships between PTG and PTD as well as the relationship of each of them to PTSD symptoms among cancer patients will be investigated.

Participants and procedures
The Research Ethical Committee of the Koo Foundation Sun Yat-Sen Cancer Center in Taiwan approved this study (reference number: 20191029A). In early 2020, 265 patients who had completed their cancer treatment at the center were recruited to participate in the study after providing informed consent. There were 58 men (21.9%) and 207 (78.1%) women. The mean age of the respondents was 51.02 years (SD = 10.33 years, range 23.0-78.0 years). The duration of onset ranged from 0 to 19 years, with a mean of 3.70 years (SD = 3.94 years). Nearly half of the participants had breast cancer (n = 122, 46%) followed by colorectal cancer (n = 40, 15.1%), lymphoma/leukemia (n = 33, 12.5%), lung and esophageal cancer (n = 24, 9.1%), head and neck cancer (n = 16, 6.0%), gynecologic cancer (n = 8, 3.0%), URO/kidney cancer (n = 6, 2.3%), and other types of cancer (each less than 2%).

Measures
The Posttraumatic Growth and Posttraumatic Depreciation Inventory-Expanded version (PTGDI-X) developed by Taku et al. [41] was adapted/translated for use in this study. The participants indicated the degree to which they did or did not experience a particular change from 0 to 5, with separate scores calculated for PTG and PTD. Moreover, higher scores indicate greater growth or depreciation. It was showed that PTG and PTD consisted of 5 domains similar to the original PTGI: relating to others (RO, 7 items), new possibilities (NP, 5 items), personal strength (PS, 4 items), spiritual and existential change (SEC, 6 items), and appreciation of life (AL, 3 items) [41].
The World Health Organization procedure for the translation and back-translation of instruments was applied to develop the Chinese version of the PTGDI-X [46]. A clinical psychologist in Taiwan with more than 10 years of clinical experience with cancer patients conducted a forward traditional Chinese translation of the English version of the PTGDI-X. PTG items were adapted from the Taiwan Chinese version of PTGI [23]. PTD items were translated with reference to the positively worded items. A research assistant with a psychology degree conducted the back-translation. Triangulation meetings were conducted with an expert panel, including the clinical psychologist in Taiwan who did the forward translation, a psychiatrist who specialized in psycho-oncology in Taiwan, and a clinical psychology professor in Hong Kong. The English and Chinese versions of the PTGDI-X are provided in the Appendix.
The Chinese version of the PTSD Checklist for DSM-5 (PCL-5) [18] was used to measure symptoms of posttraumatic stress disorder. The PCL-5 has 20 items, and the participants indicated how much they had been bothered by each symptom using a 5-point Likert scale from 0 to 4. The Cronbach's alpha of the PCL-5, according to the current sample, was.96.
Personal and medical data were retrieved from the hospital record.

Statistical analysis
The analysis of moment structures (AMOS) version 26 was used to perform the confirmatory factor analysis with a maximum likelihood optimization to examine the model established by Taku et al. [41]. We adopted a set of fit indices to evaluate the model and compared them with the results of Taku et al. [41]. χ 2 /df less than 3 was employed to determine a good fit [11,29] since the null hypothesis of χ 2 test is often rejected with p < 0.05 for large samples [33]. The comparative fit index (CFI) [4] and Tucker-Lewis index (TLI) [5,7] ranged from 0 to 1. Their values above 0.9 demonstrate a good fit. The root mean square error of approximation (RMSEA) [8] is the discrepancy between the data and model per degree of freedom. We adopted the criteria in which a value lower than 0.08 was considered a mediocre fit and less than 0.05 a good fit [26]. Moreover, the Akaike information criterion (AIC) [2,9] was also used for comparing the nonnested models, whereas Δχ 2 and Δdf were used to compare the nested models. The goodness-of-fit index (GFI) and adjusted GFI (AGFI) [27] were also reported in the current study but were not adopted as the main evaluation indices according to the guidelines suggested by some researchers [25]. Taku et al. [41] did not report these indexes as well. Multivariate analyses including regression analysis were used to examine the relationships among variables.

Confirmatory factor analysis
We conducted the CFA to examine the factor structure of the PTGDI-X tested by Taku et al. [41] in our current sample of Chinese cancer patients (Table 1).
Both PTG and PTD dimensions have five parallel dimensions similar to the original PTGI: PTG-total and PTD-total scores were calculated by summing the 25-item scores of the corresponding dimensions to represent the overall growth and depreciation, respectively. The Cronbach's reliability alphas of both PTG-total (α = 0.96) and PTD-total (α = 0.97) were high.

Correlational analysis
The Pearson product-moment correlational analyses were conducted (Table 3). Both the age and duration of onset did not relate to any of the PTG and PTD scores. PCL-5 positively correlated with PTD-total and its subscale scores; i.e., higher depreciation resulting from cancer diagnosis was related to more PTSD symptoms. PTG-personal strength was negatively related to the PCL-5 score; however, insignificant results were obtained for the other four PTG domain scores. PTG and PTD were modestly and negatively correlated with each other (r 2 = 0.017).

Curvilinear Relationship
As mentioned before, a curvilinear relationship between PTG and PTSD symptoms was obtained in some research sites of Taku et al.'s [41] study. We examined whether a curvilinear relationship would also apply to our sample of cancer survivors. Two regression analyses were performed first to examine the curvilinear relationships between PCL-5 and PTG-total as well as between PCL-5 and PTD-total, respectively. In both analyses, PCL-5 was the dependent variable. In step 1, PTG-total or PTD-total was entered, whereas in step 2, the PTG-total squared or PTD-total squared was entered. Based on the results, a significant quadratic relationship was observed between PTG-total and the PCL-5 scores: step 1, F(1,255) = 0.271, p = 0.60; step 2, F(2, 254) = 7.02, p < 0.001, ΔR 2 = 0.05, F Change = 13.75, p < 0.001. This curvilinear relationship was in the form of an inverted U-shape, as depicted in Fig. 1.
As expected, the quadratic relationship between PTDtotal and the PCL-5 scores was not significant: step 1,  Since PTD is a new concept, we examined if a curvilinear relationship could also apply to PTG and PTD in addition to the modest linear relationship mentioned above (re Table 3). A third regression analysis was conducted to examine this relationship. The result showed that a significant inverted U-shape quadratic relationship was obtained between PTG-Total and the PTD-total scores: step 1, F(1,263) = 4.04, p < 0.05; step 2, F(2, 262) = 42.48, p < 0.001, ΔR 2 = 0.25., F Change = 79.72, p < 0.001. Accordingly, increasing selfreported PTG was related to increasing self-reported PTD scores up to a certain level, and beyond that, increasing PTG scores were related to decreasing PTD scores.

Factorial equivalence of the PTGDI-X among cultures and types of traumatic events
In the current study, we compared our fit indices with the findings of Taku et al. [41] to examine whether the standard scoring method of the PTGDI-X could be applied to Chinese cancer survivors. It should be considered that Taku et al. [41] used a heterogeneous sample (mainly consisting of undergraduate students) and a variety of data collection methods (online surveys and paper-and-pencil surveys) in their study. In contrast, this study involved only cancer survivors and used face-to-face paper-and-pencil surveys in an outpatient clinic. Despite the differences, Table 2 shows that the NFI, TLI, and CFI in this study had higher values than those of Taku et al. [41], although such values in our study were below the traditional cutoff criterion of 0.9 (re Table 1). The RMSEA value (0.079) in this study was higher (hence poorer) than that of Taku et al. [41] (0.026) but still below the cutoff value of 0.08 for a moderate fit [31]. The value of χ 2 /df of this study was less than the acceptable value of 3 and indicated a good fit [11,29]. Both the GFI (0.69) and AGFI (0.66) in this study were not satisfactory but GFI and AGFI have become less popular in recent years, and some researchers suggested not to use them anymore [25,39]. Nevertheless, the reliability alphas of all the PTGDI-X scores were very good in this study. We concluded that our model fit indices were comparable to those of Taku et al.'s [41] and the same scoring method used in Taku et al.'s [41] study could be applied to the Chinese version of the PTGDI-X. Our findings suggested the PTGDI-X might show factorial equivalence among different types of trauma and data collection methods (online, paper-and-pencil, etc.). However, more studies are needed to confirm this.

PTG and PTD among cancer survivors
In corroboration with the results of other studies [3,10,41], all the PTG scores were significantly higher than the PTD scores in our sample of cancer survivors. This result could imply that all the PTG items were presented before the paired PTD items in the PTGDI-X, and this made the respondents focus more on the positive than negative changes. However, Baker et al. [3] reported that the ordering effect of the PTG and PTD items did not affect the responses of the inventory. It should, however, be noted that our participants are cancer survivors. The current results may not apply to cancer patients in early diagnosis or during active treatment. It would be interesting to examine whether the A significant positive correlation was observed between the PTD total scores and PTSD symptoms. This finding is consistent with the cognitive model that negative (and distorted) appraisals are always related to higher distress among cancer patients [34]. PTG did not show a significant linear relationship with PTSD symptoms in this study (r 2 = 0.00). Other studies on cancer patients have also reported similar results [12,17,30]. Furthermore, this study's results indicated that PTG showed a curvilinear relationship with PTSD in the form of an inverted U-shape [28]. The same curvilinear relationship occurred in a previous meta-analysis of 42 studies [38]. The above findings support the concurrent validity of the Chinese version of the PTGDI-X. Accordingly, a low level of PTG may represent a maladaptive coping mechanism exhibited by the survivors to maintain psychological equilibrium after the cancer treatment (i.e., an illusory PTG) [13]. More illusory PTG is related to more PTSD symptoms for this group of survivors. However, beyond a certain threshold, the self-reported PTG may represent a more constructive component of PTG such that more self-perceived positive change will lead to less posttraumatic stress symptoms.
Similar to the result of five countries in the study of Taku et al. [41], PTD and PTG formed a strong inverted U-shape quadratic relationship; i.e., only among cancer survivors with a high level of PTG that more self-perceived positive change will lead to less post-cancer depreciation appraisal.

Limitations
This study has several limitations that should be considered. First, the ranges of the duration of onset and age were large in our sample. Survivors with different types of cancer were included, and there were more women than men in our sample. We could not ascertain that the current results are replicable for future studies with more homogeneous samples of cancer patients. Second, two written forms of Chinese characters, often referred to as traditional Chinese and simplified Chinese, are used in the translation of instruments. The former is used in some communities such as Taiwan and Hong Kong, whereas the latter is used in other places like Mainland China. The Chinese version of the PTGDI-X used in this study is set in traditional Chinese characters. We are confident that this inventory applies to patients in Taiwan and Hong Kong. Some modifications may be necessary for the inventory to be used in places using simplified Chinese. Furthermore, the sample size of this study was not enough for both exploratory and confirmatory factor analyses.
It would be better to use an etic-emic approach [22] to develop a scoring system of the PTGDI-X with universal and culture-/trauma-specific components in future international studies.

Clinical implications
PTG and PTD are related but separate constructs that exert differential effects on the mental adjustment to cancer. Because of the robust relationships between PTD (and its five domains) and mental adjustment to cancer, the PTD scores could provide important information that is useful in cognitive therapy (CT) to reduce psychopathology [34]. For instance, high scores on some PTD items (e.g., I have a diminished feeling of self-reliance) may represent distorted cognitions related to the cancer experience. Cognitive restructuring strategies [34] could be employed to reduce the distorted cognitions. The PTG scores provide a more comprehensive picture of the phenomenological world of the patients to guide other modalities of intervention, such as supportive-expressive therapy (SET) [24]. In a supportiveexpressive group setting, clinicians may facilitate survivors to share both positive (growth) and negative (depreciation) experiences related to cancer according to the PTGDI-X results to enhance emotional support and control. Both CT and SET are evidence-based interventions in psycho-oncology [20]. The two scores together might potentially provide useful information to distinguish a cognitively biased sense of growth from constructive PTG.
It is recommended that the full PTGDI-X including both PTG and PTD items should be used in clinical research and assessment as far as possible. Nevertheless, the PTG and PTD subscales can be used separately. For example, clinicians can use the PTD subscale if they want to assess cognitive distortions of cancer patients to inform cognitive therapy.

Conclusion
In conclusion, cancer survivors report no difficulty in rating both the growth and depreciation items in pairs. The PTGDI-X is a potentially useful instrument to gauge a balanced phenomenological world of individuals with cancer. The Chinese version of the PTGDI-X, when applied to cancer survivors, has shown factorial equivalence to other language versions of the inventory. The same scoring method proposed by Taku et al. [41] could be used to facilitate the comparison of the findings. However, some goodness-of-fit issues should be investigated further. We recommended that the total PTG and PTD scores should be used for clinical assessment and research investigation as far as possible.