The Post-Liver Transplant Quality of Life Instrument: Translation, Reliability, and Validity of the Persian Version

Background: The Post-Liver Transplant Quality of Life (PLTQ) instrument is a specic, disease-based questionnaire to evaluate health-related quality of life (HRQOL) of liver transplantation (LTx) recipients. The aim of this study is translation, cross-cultural adaptation, and evaluation of the psychometric properties of the Persian version of PLTQ (PV-PLTQ) questionnaire. Methods: All of the stages of translation and cross-cultural adaptation of the original PLTQ instrument was carried out according to the published guideline. A total of 175 LTx recipients completed the PLTQ and short form-36 (SF-36) questionnaires. Sixty-eight randomly selected patients were asked to complete the PLTQ, 2 to 3 months later. Face validity, content validity, acceptability internal consistency, test-retest reliability, sensitivity to change, and convergent validity were evaluated. Results: The results of face validity, content validity, and missing data proportion indicate that PV-PLTQ questionnaire is acceptable and easy to understand. Cronbach’s alpha coecient for the PV-PLTQ questionnaire was 0.97 (0.82 to 0.95) for domains. The results of test-retest reliability show moderate to good intraclass correlation coecient from 0.6 to 0.86 (P< 0.05). Also, the results of the Kruskal-Wallis test shows that this questionnaire is sensitive to HRQOL changes. The correlations between PV-PLTQ domains and physical and mental components of SF-36 present a good convergent validity (P< 0.05). Conclusio: In general, The PV-PLTQ questionnaire is a valid, reliable and sensitive to change instrument to evaluate HRQOL in LTx recipients and can be applied in further researches and clinical settings in the Persian speaking population.

the absence of disease and in rmity". Moreover, the re ex of the disease and interventions on the lifestyle, mental balance, and well-being of patients, according to their own judgments and perceptions [6].
Due to the growing number of liver transplantation recipients, and moreover the importance of posttreatment quality of life among patients, there is a necessity to assess their HRQOL via a standard and speci c instrument [7]. The HRQOL assessment facilitates gathering practical data for health care professional and is an important parameter in assessing the effects of disease and evaluates the impact of therapeutic interventions on the performance of individuals' life [8].
An accurate assessment of HRQOL depends on the psychometric features of the developed instrument to measure this construct [9]. The usability of any instruments depends on the validity and reliability of each questionnaire in different populations which they will be employed and additional validation through a standardized process (translation into the local language and adaptation to the cultural context of the country) [10][11][12].
The aim of the present study was the translation, cross-cultural adaptation, and validation of the Post Liver Transplant Quality of Life (PLTQ) instrument in Persian speaking population to be used among LTx recipients.
Each item is scored by the Likert system from 1 (as persons' experience with the item all the time) to 7 (as persons' experiment with the item none of the time). Total score is expressed as the mean of all items with higher values presenting better HRQOL.
2. The Short Form Health Survey (SF-36): The SF-36 is a well-known, generic HRQOL questionnaire. This instrument has 36 questions with the different scoring systems in eight domains consists of physical function, role limitation due to physical problems, bodily pain, general health perception, vitality, social functioning, role limitation due to emotional problems, and perceived mental health. These domains form two distinct subscales; mental component summary score (MCSS) and physical component summary score (PCSS). Each domains, MCSS, and PCSS scores are calculated by mean of all domains score and can be reported from 0 (as poor QOL) to 100 (as good QOL). This survey has been translated and crossculturally validated to a Persian version. [13].
3. The Global Rating of Change (GRC): The GRC is a subjective, quick, exible and e cient instrument to track patients' conditional change [14]. In the current study, we designed a GRC questionnaire to determine any change in the participants' health condition between rst and second PLTQ instrument administrations. This question was scored on a ve-point Likert scale from − 2 (as totally deteriorated (to + 2 (as totally improved).
Translation and Cross-Cultural Adaptation: Initially, permission for translation and cross-cultural adaptation was taken from Sammy Saab. The processes of translation and cross-cultural adaptation was followed based on published guideline by Sousa and Rojjanasrirat [15]. Accordingly, two independent native Persian language speakers with good English language skills (one physiotherapist and one psychologist) translated the English version of the PLTQ questionnaire to Persian conceptually. All of the occurred discrepancies between two translated versions were discussed and resolved after comparing them with the original version by the translators and one of the authors (A.N.). A consensus version was nally created. Two professional English translators translated the forward translated version back to the English version independently. These translators were blinded to the original questionnaire. Finally, an expert panel committee including two physiotherapists, two psychologists, one psychiatrist, one post-transplant coordinator, one hepatologist and one liver transplant surgeon reviewed all versions of the questionnaires and cross-cultural adaptation processes carefully. After making a few changes to reach a better cross-cultural equivalence, the pre-nal Persian version of PLTQ (PV-PLTQ) questionnaire was provided (Online resource 1).

Face Validity and Content Validity:
In the presence of one of the authors (M.J.), the pre-nal PV-PLTQ questionnaire was read and completed by 15 LTx recipients to evaluate the level of clarity of instruction, items and response format. Also, they were asked to discuss about di culties in understanding items. The mean time to complete questionnaire was 11.6 (Standard deviation: 5.1) minutes. 87% of participants stated that all contents were totally clear and understandable. After reviewing the comments by an expert panel (with ten members) and making some corrections, the nal PV-PLTQ instrument was developed. Furthermore, we asked expert panel members to score the content equivalency of instruction, items and scales by a fourpoint Likert scale.

Participants:
A total of 180 volunteer LTx recipients were recruited consecutively from post-liver transplant hepatology clinic at Imam Khomeini Hospital Complex, Tehran, Iran over a period of 5 months. The participants were included in the study if they were LTx recipients and were older than 18 years. Patients with multiple organ transplants, re-transplant, inability to read and write in Persian, and major cognitive impairments were excluded from the study. Also, we excluded participants who omitted more than 50% of their responses. This study was approved by the Ethics Committee of Tehran University of Medical Sciences.
Informed consent was obtained from each participant.

Reliability:
The reliability of the PV-PLTQ questionnaire was tested by the test-retest statistical method. Also, Cronbach's alpha was provided for assessment of its internal consistency. To carry out test-retest reliability, 70 LTx recipients were randomly selected for longitudinal assessment and were asked to complete the PV-PLTQ questionnaire for the second time between 2 to 3 months later. Also, they were asked to answer the GRC questionnaire in the second test to measure the reproducibility and sensitivity to change of the PV-PLTQ instrument. Based on the GRC score, patients were divided into three categories: 1) patient with improvement, 2) patients with deterioration, and 3) patients with no change.

Convergent Validity:
At the time of rst assessment, participants completed the SF-36 along with the PV-PLTQ questionnaires. The PLTQ questionnaire has multiple domains which most of them are not de ned in SF-36 survey but most of the questions of PLTQ assess the patients' physical and mental health status. Therefore, a satisfactory correlation between PV-PLTQ questionnaire domains and MCSS and PCSS was expected.
Factor Analysis: The factorial structure of the PV-PLTQ questionnaire was measured by factorial analysis with varimax rotation. Factors with eigenvalues ≥ 1 were considered proper as domains of PV-PLTQ questionnaire. Also, the highest item loading of each item determined which item is associated with each domain. A Con rmatory factor analysis (CFA) with the parceling method was used in order to evaluate the factor structure of the instrument [16]. In the parceling method, based on item-scale correlation coe cients, two or three items would be aggregated and made a parcel under its subscale. Then, CFA would be run on these created parcels. Also, the parceling method can approximately deal with sample size limitation. Root mean square error of approximation (RMSEA), Con rmatory factor index (CFI), Incremental t index (IFI), Parsimony-adjusted normed t index (PNFI), Parsimony-adjusted comparative t index (PCFI), and Goodness-of-t index (GFI) were calculated to show the suitability of PV-PLTQ instrument model.

Statistical Analysis:
Sample size was calculated as 5 participants per item. After estimating about 10% possible attrition or incomplete questionnaire lling, 180 participants were determined as a suitable sample size. All data were analyzed using the Statistical Package for Social Science version 18.0 (SPSS, INC, Chicago, IL). P value < 0.05 was considered as a statistically signi cant level. Shapiro-Wilk test was established to evaluate the normality distribution of data.
The content validity index at the items (I-CVI) and at the scale (S-CVI) methods were calculated to assess the level of content validity. I-CVI > 0.78 and S-CVI > 0.9 were considered as an acceptable level for expert panel agreement.
Acceptability and simplicity to complete the PV_PLTQ questionnaire was assessed by calculation of missing data proportion for each item and domain. Missing data proportion ≤ 4% was considered satisfactory. Missing data were calculated using multiple imputation method [17]. In order to estimate internal consistency for total instrument and each subscale, Cronbach's alpha was analyzed. Cronbach's alpha ≥ 0.7 was considered satisfactory [18]. Intraclass correlation coe cient (ICC) was used to evaluate test-retest reliability in patients reported "no change" between two assessment phases. ICC < 0.5, 0.5-0.75, 0.75-0.9, and > 0.9 were considered poor, moderate, good, and excellent reliability measures respectively [19]. To evaluate sensitivity to change of PV-PLTQ instrument, Kruskal-Wallis one-way analysis of variance was used. For this purpose, the amount of change of PV-PLTQ questionnaire score was compared between patients who reported "improvement", "no change", and "deterioration" during a longitudinal period [20]. The Spearman correlation coe cient between PV-PLTQ questionnaire domains and MCSS and PCSS was established to test convergent validity.

Results:
Patients Characteristics: Out of 180 included patients, 5 did not complete more than 50% of items and therefore were excluded. From 70 randomly selected participants for longitudinal survey, 2 of them did not want to continue the study due to personal reason and one was hospitalized. Table 1 summarize the demographic characteristics of participants in cross-sectional and longitudinal assessment phases.

Content validity:
The I-CVI for the PV-PLTQ instrument ranged from 0.8 to 0.92. Also, The S-CVI was 0.95.
According to the results of principal component analysis of the parceling method, structure of the PV-PLTQ instrument could be explained by a 5-factor and 32 items solution. The path diagram of CFA with the parceling method was presented in Figure 2. Furthermore, the goodness of t indices was shown in Table 2 for CFA with the parceling method. In the parceling model, RMSEA, CFI, IFI, PNFI, PCFI showed acceptable goodness of t and GFI was nearly close to the acceptable value; i.e. 0.90. All of the regression weights between indicators and endogenous variables (healthcare, emotional, medication, physical, and cost), and endogenous variables and exogenous variable were statistically signi cant (p <0.001).
Acceptability: All of the items had less than 4% missing data except items 20 (learning to walk after surgery) and 21 (patients' concern about the ability to drive) with 6% missing data ( Table 3). The patients in the face validity phase recognized these items to be totally clear.

Reliability:
Values for Cronbach's alpha exceed the satisfactory level value more than .7 in the total instrument and each domain ( Table 3). The ICC for total PV-PLTQ questionnaire and its 5 domains were .86, ,73, .81, .79, .77, and .6 respectively (P <.001). This result shows moderate to good test-retest reliability.

Sensitivity to Change:
The PV-PLTQ questionnaire score change had no normal distribution. Therefore Kruskal-Wallis statistical analysis was utilized. The results showed that changes in the score between patients who reported "improvement", "no change" and "deterioration" was statistically signi cant different (P< 0.001)

Convergent Validity:
The result of the Spearman test presents moderate to high correlation between total and all items of PV-PLTQ questionnaire and MCSS and PCSS (P < 0.05). The highest Pearson correlation value was related to physical domain score of PV-PLTQ questionnaire score and PCSS of SF-36 (r= 0.53, P < 0.01, Table 4).

Discussion:
In the past few years, the number of LTx centers and utilization of this therapeutic intervention has been grown dramatically worldwide. Now, many countries with different cultures, religions, and languages are performing LTx for their patients. LTx is not only a cross-sectional treatment. It affects various aspects of patient's live which can be related to their HRQOL. LTx recipients have speci c health conditions that need various types of medical and healthcare management than other patients. for example, they must always take immunosuppressive drugs that may expose them to some side-effects. These speci c conditions are the key reasons for the need of a speci c HRQOL instrument.
Internationally standard HRQOL instruments allow the comparison of the patients' condition and QOL among different cultures [21]. As use of generic or speci c questionnaires is affected by different cultures, they must be translated, cross-culturally adapted, and validated based on standard guidelines [22]. PV-PLTQ questionnaire seems to be a relatively acceptable and simple to be completed based on result of pretesting face validity and missing data proportion. Items 20 and 21 had higher missing data proportions. These items are related to patients' concern about learning to walk and drive. Most patients in this center are mobilized or made to walk learn walking during hospitalization stage by a physiotherapist and they will be under follow-up of physiotherapy service after discharge. Subsequently, they can walk, and drive soon after LTx; therefore, it mays concerning about walking and driving in the condition which they are now have these functions be confusing questions.

Factor analysis:
Based on the factor analysis result, it could be postulated that the PV-PLTIQ indicated a promising and acceptable ve-factor structure in Persian speaking population. The different number of domains between original PLTQ and PV-PLTQ may be due to different conditions of LTx recipients in Persian speaking population in comparison to U.S. As an example, worry and graft rejection concern domains did not detect. Most of the PLTQ questions ask the patients about their concerns and worry. Worry and concern are two words which their equivalent translation in Persian is a similar word. Although the expert panel tried to discriminate these questions from each other, but the factor analysis could not detect questions about worry and graft rejection concern as speci c domains and most of their questions shifted to other domains. On the other hand, cost domain was totally detected in our analysis. Despite good insurance coverage for drugs and other healthcare procedures, patients have to pay some costs themselves. Furthermore, due to di cult nancial condition for Iranian people, providing this cost is mostly troublesome for patients [23]. It can be the reason why the factor analysis detected cost domain with only two questions. A similar study was reported by Xiao et al which translated and cross-cultural adapted the PLTQ instrument to Chinese language. They detected 6 domains including worry, physical function, emotional function, nancial, healthcare, and complication. Moreover, the limitation of sample size was a probable reason for failing PLTQ model with 8 domains.
Internal Consistency, Test-Retest Reliability, and Sensitivity to Change: The Cronbach's alpha of the PV_PLTQ questionnaire was 0.97 which is greater coe cient value than previously reported by Saab et al. [10], Xiao et al. [11], and Molski et al. [12]. This result presented that the PV-PLTQ instrument has satisfactory internal consistency. We selected 2 to 3 months as the period between two assessment phases. Although the results of a published systematic review indicate that the HRQOL has been increased after LTx [24], but some components of HRQOL may be attributed in the shorttime after LTx due to surgical or psychological complications or some immunosuppressive drug sideeffects. It occurs specially in the rst months after transplant and can in uence on the test-retest reliability. In order to resolve this confounder factor, GRC was used to track participants' change during 2 to 3 months. ICC was calculated in only patients who reported no change. The results of test-retest reliability showed good to moderate ICC values ranging from 0.6 to 0.86. Although, Saab et al. have not reported ICC for test-retest reliability, but our results were similar to those of the other translation and validation of the PLTQ instrument [11,12]. The changes in the PV-PLTQ questionnaire score was used to evaluate sensitivity to change power of this instrument. The results of Kruskal-Wallis showed that it seems PV-PLTQ instrument can be considered sensitive to change Persian HRQOL instrument.

Convergent Validity:
The PV-PLTQ questionnaire and the MCSS and PCSS showed a relatively good signi cant convergent validity. Although, All the result of Pearson test investigated a statistically signi cant correlation between PV-PLTQ instrument domains and MCSS and PCSS, the correlation between healthcare and cost domains were less than others. These may be due to the nature of SF36 questionnaire content. In SF-36, we don't have questions about patients' nancial condition. Moreover, there are some questions about patients' current health status but there are no question about individuals' concerns about healthcare.
Our main limitation in this study was the sample size. For performing some analysis like CFA there is a need for a larger sample size. The number of LTx procedures is relatively less than other surgical procedures and providing a large sample size is relatively di cult. The second limitation was singlecenter recruitment. Although our center has an average of 220 LTx annually from all over Iran and there are various indications for LTx in this center, but it is a fact that single-center is another main limitation.

Conclusion:
We have developed the Persian version of the Post-liver Transplant Quality of Life instrument. This tool seems to be reliable, valid, and sensitive to change. Also, the participants indicated that this instrument is acceptable and easy to understand and response. Therefore, use of PV-PLTQ questionnaire can be recommended in the researches and clinical settings to evaluate Persian linguistics/speaking LTx recipients' HRQOL.

Declarations
Author Contribution: MJ, NA, AN, MN, AJ, and BAM made substantial contributions to the conception and designed the study. MJ, NA, and MN contributed to acquisition of the data. MJ and MS analysed the data. BAM and AN was responsible for data safety monitoring during the study. MJ, and NA interpreted the data and drafted the manuscript. All authors had access to the data, revised the manuscript, and approved it.

Funding:
This research has been supported by Tehran University of Medical Sciences (grant number: 97-01-205-37867). The founder of the study had no role in study design, data collection, analysis, interpretation, or writing the manuscript. The corresponding author had full access to all the data in the study and is responsible for publishing the results.

Con ict of Interest:
The authors declare that they have no con ict of interest. DF: Degrees of freedom, GFI: Goodness-of-t index, RMSEA: Root mean square error of approximation, CFI: Con rmatory factor index (CFI), IFI: Incremental t index, PNFI: Parsimony-adjusted normed t index, PCFI: Parsimony-adjusted comparative t index. All correlations are statistically signi cant. Figure 1 Path diagram of CFA with parceling method.