The psychometric properties test of the Malay version of the endometriosis health profile-30

A cross-sectional study, involved women with endometriosis in the Gynecology Clinic at a tertiary teaching hospital in Kuala Lumpur, Malaysia. The study was carried out between January and April 2021. The inclusion criteria were: I) Malaysian; II) an assessment carried out by a physician and a diagnosis of endometriosis based on an abdomen ultrasound; III) the subjects also had to be a pre-laparoscopy procedure; IV) be aged over 18 years old; V) have no psychosis symptoms; and VI) be able to read and give consent to participate in this study. The researcher calculated the sample size needed to obtain the minimum sample size required for confirmatory factor analysis. Based on the recommendation, the sample size needed more than 200 participants; thus, 218 respondents were selected for this study. 31 The sampling was sufficient based on the Kaiser-Meyer-Olkin (KMO) results, which measures the sampling adequacy ( p =0.90). 32 A convenience sampling method was applied to recruit respondents. The Malay version of the EHP-30 was obtained from and with the permission of Oxford University Innovation Ltd., Oxford, United Kingdom. In the validation period prior to this study, a minor adjustment was proposed in the Malay version of the EHP-30 and implemented by 6 experts in obstetrics and gynecology. The face validation carried out, which invited 30 women with endometriosis to answer the Malay version of EHP-30 and to assess the clarity and representatives of the questionnaire items. This study used the EHP-30 questionnaire to measure the validity with used psychometric properties test in Malay version. Part A referred to the participants’ sociodemographic data, including age, race, religion, education, status, occupation, income per month, and smoking habits. Part B was the revised Malay version of the EHP-30. It was composed of 30 items across 5 domains. The questionnaires were deemed valid for further studies based on their reliability, the value was between 0.87-0.96 for Cronbach’s alpha. A likert scale with 5 point used, with scores ranging from 0-100 and a scale featuring: never (0); rarely (1); sometimes (2); often (3); and always (4). The formula for scoring each dimension was based on the formula of: ‘total of the scores for all items and divided by 4 multiplied by the number of items in the dimension’. 18 This indicated that high scores meant a poor quality of life and low scores meant a good quality of life. 18 , 19 , 26 , 33 The study protocol was approved by University Malaya Medical Centre (MRECID.NO: 2020525-8672). The revised Malay version of the EHP-30 questionnaire form was answered among women who registered at the admission centre of the gynecology outpatient department at a tertiary teaching hospital. Using the ‘participant information sheet’. The researcher explained the study to the respondents who fulfilled the inclusion criteria and gave written consent before the form collection. Respondents answered the questionnaire within 10-15 minutes. The data were analyzed using The Statistical Package for the Social Sciences (SPSS), version 26.0 (IBM Corp., Armonk, NY, USA) and IBM SPSS Amos, version 24 software. To determine the domain compatibility was carried out with CFA tests for each domain. The loading factor value of ≥0.50 was the main criterion for deciding the compatibility. 32 The other criteria proposed for this fit was the fitness index ( p ≥0.05), the root mean square of error approximation ([RMSEA] values of ≤0.08), the comparative fit index (CFI), and Tucker-Lewis index (TLI) (values of ≥0.90). 34 The Chi-square/degrees of freedom (Chisq/df) (values of ≤3.0) was used. The convergent validity in this study ran for average variance extracted (AVE) and constructed validity measurements to test the validity assessment. The required AVE value for the convergent validity measurement was of ≥0.5, and the criterion for the reliability measurement value was of ≥0.7. 32 The descriptive statistics was used to analyzed sociodemographic data. The measurement of the model for Malay version of the EHP-30 for a latent construct used factor analysis for all items. This was measured based on the assessment, including the unidimensionality, validity, and reliability. Unidimensionality measures items for factor loading, whereby the value for each item should be 0.6 or higher. 32 Then, validity was measured for the latent construct types: convergent validity, construct validity, and discriminant validity. 35 This included the Cronbach’s alpha with a value of >0.7, composite reliability (CR) with a value of ≥0.6, and AVE with a value of ≥0.5. 32 , 35

A total of 218 women with endometriosis were enrolled to answer the questionnaire in this study. The sociodemographic characteristics among respondents are presented in Table 1 . The results revealed the mean age of the participants (37.31 years), the majority were Malay (73.4%, n=160), and Muslim (74.3%, n=162). Of the participants, 40.4% (n=88) had first-degree education, and 60.1% (n=131) were married. The participants generally reported working women (81.7%, n=178); the mean monthly income was USD 729.14. Only 1.8% (n=4) of women reported smoking. - Sociodemographic characteristics of the participants (N=218). Values are presented as numbers and precentages (%). SD: standard deviation, USD: United States dollars The psychometric properties for construct validity and reliability showed acceptable value. Figure 1 shows the final path diagram for the revised Malay version of the EHP-30 with 5 construct domains and 28 items after deleting 2 items (no.: 12 & 13). The following domains remained, according to the items in the Malay version of the EHP-30: domain 1: items 1-11; domain 2: items 14-17; domain 3: items 18-23; domain 4: items 24-27; and domain 5: items 28-30. The factor loading of the 28 items had an acceptable value range between 0.60-0.90, so the factor loading for all the items was highly acceptable. Domain 1 ranged from 0.79-0.90; domain 2 ranged between 0.78-0.90; the third domain ranged from 0.69-0.86; the fourth domain ranged from 0.84-0.89; and the last domain between 0.78-0.89. The model revealed that was acceptable based on the fitness of RMSEA= 0.072 (90% CI: [0.065-0.080]); CFI=0.939; TLI=0.932; and Chisq/df=2.135. - Path diagram of the revised Malay version showing the 5 domains featuring the 28 items in the EHP-30. Table 2 shows the factor loading for each item and Cronbach’s alpha, CR, and AVE for each domain in the Malay version of the EHP-30. It was revealed that Cronbach’s alpha values for each domain ranged between 0.89-0.97, indicating a highly acceptable range. The overall Cronbach’s alpha value of the Malay version of the EHP-30 was 0.963, showing good internal consistency for the 28 items. The CR values ranged between 0.88-0.96, which was highly acceptable. Meanwhile, the AVE values for the 28 items were acceptable for domain 1: 0.74, domain 2: 0.71, domain 3: 0.65, domain 4: 0.74, and domain 5: 0.72. The overall reliability results showed that the items and domains of the EHP-30 in the Malay version were acceptable to highly acceptable. - Reliability characteristics of the revised Malay version, 5 domains of the 28 items endometriosis health profile-30 (N=218). CR: composite reliability, AVE: average variance extracted

This study tested the validity and psychometric properties of the Malay version of EHP-30 questionnaire. This tool was used for the measurement of the quality of life among women with endometriosis in Malaysia. The results demonstrate that the Malay version of EHP-30 is validated and reliable in measuring the quality of life among women with endometriosis in our country. An intended latent construct used the method of psychometric properties analysis to identify reliability to produce a more accurate and advanced method. 32 Therefore, for the reliability of each item in the Malay version of the EHP-30, the authors employed internal reliability, composite reliability, and average extracted. Previous studies have reported factor analyses of the EHP-30 in table form only. 19 , 23 , 27 , 29 However, our results are shown in a path of diagram, a more advanced form than others have used. This approach also enabled accurate values of reliability to create an acceptable measurement model. 32 , 36 Identifying the reliability of the items in the Malay version of the EHP-30 would facilitate the development of novel treatments and may improve results for use in future studies. In this study, the results shown in the path diagram of the Malay version of the EHP-30 revealed the graphic latent construct and values of factor loading for the 28 items across the 5 domains retained. Items 12 and 13 were removed because the modification indices (MI) were over 15; adjustment procedures were completed. 32 It was revealed that the factor loading values for the 28 items were acceptable to highly acceptable (≥0.65 and above). The fitness values for the measurement model including the absolute fit, incremental fit, and parsimonious fit were also revealed to be acceptable based on the index. 32 Based on these results, the simple 5-domain model showed satisfactory factor loading and satisfied all the confirmatory factors for the goodness of fit criteria, namely RMSEA, CFI, TLI, and Chisq/df, for the last reduction model. The values were acceptable according to the indication for each index. 32 Therefore, the current results produced a fitness measurement model that could be used and referred to by others. Previous studies demonstrated that the results of factor loading were measured and analyzed using IBM-SPSS software for the EHP-30. 19 , 23 , 27 , 29 These studies revealed better values for factor loading. Based on the initial measurements of the EHP-30 items, the factor analysis values were of ≥0.50 and above. 19 However, using a similar analysis method, the Chinese version of the EHP-30 items had a factor loading range from 0.608-0.887, and 5 domains were maintained. 27 Meanwhile, Western studies reported that the psychometric evaluation analysis using IBM SPSS statistics software for the factor analysis of the EHP-30 items revealed factor loading ranging from 0.41-0.83 and from 0.432-0.856. 23 , 29 Based on the results, some items were low in value. However, 5 domains were still maintained in the EHP-30. Different results are presented in this study, with values of factor loading that were more accurate and produced items with higher acceptance. The internal consistency of the items in the Malay version of the EHP-30 was measured by Cronbach’s alpha, which revealed that the value of the EHP-30 for 28 items overall was acceptable and highly acceptable. These results correspond to the first EHP-30 reliability analysis, which reported that internal reliability was highly acceptable for each domain. 18 Consistent with other studies, the internal reliability of the EHP-30 was highly acceptable, even across different versions, including the English version with a range from 0.80-0.96; the Chinese version with a range between 0.89-0.97; the Portuguese version with a range between 0.876-0.981; the Swedish version with a range between 0.83-0.96; and the Spanish version ranging from 0.85-0.97. 19 , 23 , 24 , 27 , 29 This means the reliability of the EHP-30 is acceptable worldwide for use in the field. The Malay version of the EHP-30 showed that the composite reliability showed that indicated the reliability of each domain, thus achieving composite reliability for a construct model. The method can be extended to the measurement model and successfully measure an intended latent construct. 32 Thus, the Malay version of the EHP-30 was proven more concise and reliable to used. Previous studies analyzed the EHP-30 items for factor analysis using only SPSS. 26 , 29 However, one study revealed that the value was significant for convergent validity between the EHP-30 and other scales which was different compared to other studies that analyzed the EHP-30 with additional tools for construct validity and convergent validity. 29 While, a study in China revealed that the convergent validity value was good and accepted for all the EHP-30 items. 27 A study in Norway revealed that the value was good correlation for convergent validity of Norwegian version of the EHP-30 and SF-36. 26 In contrast, a study in Sweden reported that the other results concerning the reliability of the EHP-30 items involved the score distribution, floor, and ceiling effects of the items by domain. The population was limited to one city in Malaysia only. Further studies should investigate a larger cohort that includes all towns in Malaysia. Also, suggest using the Malay version of the EHP-30 when measuring the quality of life among subjects in the community or the clinical field. These findings emphasize the crucial results of the validity and reliability of EHP-30. In conclusion, the revised Malay version of the EHP-30 retained 28 items and 5 domains based on the high factor loading values. The model for psychometric properties analysis showed that the model was acceptable with the fitness of RMSEA. While internal consistency, CR, and AVE for each domain revealed highly acceptable values. This EHP-30 questionnaire showed that it was validated, reliable, and acceptable in Malay. Thus, it is valid for subsequent studies to measure and investigate the impact of life for those with endometriosis in Malaysia. This questionnaire is essential to identifying the quality of life among those women with endometriosis current and future.