Reliability and Validity of the Dari version of the World Health Organization Quality of Life (WHOQOL-BREF) Questionnaire in Herat, Afghanistan: A cross-sectional study


 Background

In order to predict social, mental and health care needs we must measure Quality of life of residents. In Afghanistan there was no assessment to measure quality of life of people. The aim of this study was to culturally adapt the WHOQOL-BREF into Dari language of Afghanistan and present its psychometric properties.
Methods

Cultural adaptation of the WHOQOL-BREF into Dari has consisted of two stages: translation stage, and the cognitive debriefing (pilot) stage following the translation stage. The process of translation of WHOQOL-BREF included recommended methodology, mandating forward translation, backward translation was followed. A total number of 1473 people did participate in this research. Psychometric analyses consisted of item analyses; reliability and validity analyses. Internal consistency analyses were done by Cronbach's alpha value and validity analyses consisted of construct validity (convergent validity by SF-36 and confirmatory factor analyses) and criterion validity (multiple linear regression by overall QoL item (item1)) analyses. Acceptable type 1 error was considered as 0.05 in all analyses. Analyses were done by using Lisrel v8.05 statistical package.
Results

Item domain correlations and -if item deleted- Cronbach alpha values no problematic item in Afghan version of the WHOQOL-BREF. Internal consistency of the scale was in acceptable limits for all domains (alpha values = 0.79–0.80) except for the social relations domain (alpha = 0.41). Using overall QoL item (item q1) as the dependent variable, multiple regression analysis revealed a R2 value of 57% and all of the four domains of the WHOQOL except that of Social relation domain (beta = 0.03) could explain the variance of the q1, the best domain is the environmental domain with a beta value of 0.30. Construct validity is tested by known groups validity; factorial validity and the convergent validity. All of the known categories of the age, gender, education, social class and economic status, having chronic illness were significantly sensitive to the domain scores of the WHOQOL (p < 0.001). Physical and Psychological domains of the WHOQOL-BREF was highly correlated with the related domains of the SF-36 (r = 0.60 and 0.64). Social and environmental domains showed poor correlations with the SF36. Confirmatory factor analyses showed acceptable goodness of fit results. Chi-square = 2174,748 and Degrees of freedom = 246; GFI: 0.88, CFI: 0.83; RMR: 0.066 and RMSEA: 0.073.
Conclusion

Afghan Dari version of the WHOQOL-BREF can confidently be used in clinical setting and in population level to assess the QoL of the people. The results of the social relations domain should be interpreted with caution due to its poor psychometric power. Further studies are needed to address the social aspects of quality of life in Afghan population.

Afghan Dari version of the WHOQOL-BREF can con dently be used in clinical setting and in population level to assess the QoL of the people. The results of the social relations domain should be interpreted with caution due to its poor psychometric power. Further studies are needed to address the social aspects Among the current QoL instruments available, the WHOQOL more likely to provide valid scores for comparison across community groups (4). The only problem with WHOQOL-100 is that, it is not easy to respond because of it is too long its lots of questions which makes the volunteer uncomfortable(5)(6)(7) (8) (9). The 26 item WHOQOL-BREF on the other hand, is one of the best among others and is available in more than 40 languages not only in normal populations but also recently in various diseases and conditions such as substance use disorder patients (10) and type two diabetes (11)(12) (13). It has only 26 questions as opposed to 100, though it aims to cover a broad range of quality of life facets divided into four main domains: The Physiological, Social, Physical and the Environmental. The WHOQOL-BREF has been translated and validated in many countries so far but has not been validated in Afghanistan yet (14) (15). A globally used, brief and cross culturally accepted generic quality of life tool is needed in health inequalities research and medical practice in Afghanistan. Dari is the rst o cial language of The aim of this study is to culturally adopt the WHOQOL-BREF into Dari language of Afghanistan and test the reliability and validity of the Dari version of the WHOQOL-BREF.

Methods:
The WHOQOL-BREF is a 26 item brief version of the WHOQOL questionnaire. The WHOQOL-BREF covers four individual domains such as: Physical (activities of daily living, dependence on medicinal substances and medical aids, energy and fatigue, mobility, pain and discomfort, sleep and rest, work capacity); Physiological (bodily image and appearance, negative feelings, positive feelings, self-esteem, spirituality /religion/personal beliefs, thinking, learning, memory and concentration); Social (personal relationships, social support, sexual), and Environmental ( nancial resources, freedom/physical safety and security, health and social care: accessibility and quality, home environment, opportunities for acquiring new information and skills, participation in and opportunities for recreation/ leisure activities physical environment (pollution/ noise/ tra c / climate/ issues. Higher WHOQOL-BREF scale scores indicate better quality of life.
Afghanistan is a multi-ethnic country with Pashtuns, Tajiks, Uzbeks and Hazaras making up the majority of the population although Dari is one of the languages mostly used by the people of Herat city in Afghanistan and most of them are comfortable with Dari language.
Linguistic And Cultural Adaptation: The cultural adaptation of the WHOQOL-BREF into Dari has consisted of two stages: translation stage, and the cognitive debrie ng (pilot) stage following the translation stage. The process of translation of WHOQOL-BREF included recommended methodology, mandating forward translation, backward translation was followed. Forward translations were done by four independent translators and a consensus Dari version was developed by an expert linguist. The back translation of the Dari consensus version of the WHOQOL-BREF into English was done by a bilingual person. The back translated text was then compared with the original English version of the WHOQOL-BREF and a careful comparison was done between the two texts in terms of conceptual inconsistencies. When any inconsistency is detected between the two texts, the Dari version was revised accordingly, following a panel discussion with the translators.
Cognitive debrie ng interviews were carried out on 30 healthy individuals. Following the completion of the questionnaire, each individual was asked to give their feedback about each of the items and response scales of the WHOQOL-BREF and each of the instruction sentences of the instrument. They were asked to tell: " Whether they were able to understand each of the items and each of the instruction sentences or not?" "If there is any word or phrase that they did not understand" if so, "how would they rephrase the question or the sentence?" They were also asked to do Thurston sort card exercise to con rm the correct order of response options in Dari version. It took almost 10 minutes for each individual to complete the questionnaire. The volunteers were able to understand all of the items and instructions with minimal modi cations Study Sample And Data Collection: Field trial sample is a representative sample of Herat City, Afghanistan. The 2018 census of the Herat was around 270 000 (16) The number of households is around 45 000 in the city and the research was conducted on a household based strati ed random sample of individuals aged 18 or above living in all 15 districts of Herat city. Volunteers consist of both healthy people (who stated that they are free of any kind of chronic disease) and ill participation (who stated that they have at least one or more medical condition receiving some form of medical care). Only one adult was selected from each of the households. 1473 persons were selected to the study sample (804 were males and 669 females). The gender imbalance was due to the higher willingness of men to participate to the study than women.
Of the participants, 53% were illiterate and 47% were literate. The inclusion criteria were being over age 18 and speaking the Dari language. Questionnaires were administered by an interviewer during face to face interviews in public.

Psychometric Analyses:
Psychometric analyses of the led data of the WHOQOL-BREF Dari version is consisted of distribution properties and item analyses followed by internal consistency and validity analyses. Minimal acceptable limits for Floor and Ceiling value percentages are ≤ 20% (17) and Skewness and Kurtosis limits are 2.0 (18).
Both reliability and Validity analysis were based on con rmatory approach. Internal consistency of the individual domains were tested by Cronbach's alpha (19) The average inter-item correlations for domains, and correlation of items with their intended domain were also calculated to assess scale success statistics. If Pearson's product moment coe cients between the items and their own scale score (corrected for overlap) were higher than the correlations between these items and any other domain then its regarded that WHOQOL domains represent separate entities. "If item removed alpha values" and item scale correlations were used to detect any problematic items. If any domain's alpha value increases when internal consistency analyses in repeated when a certain item removed, then that item might be a problematic item. The problematic item may be proved by looking its correlation coe cient with its own dimension score if it is less than 0.30. Convergent-divergent validity, known groups' validity and con rmatory factor analysis were employed for the assessment of construct validity of the WHOQOL-BREF Dari version. Sf-36 was used to test convergent-divergent validity of the WHOQOL-BREF and gender, education, social class and presence of any illness were used for known groups'/discriminant analyses. The discriminative ability of the instrument between a subgroups were assessed by Student's t test indicating the known groups validity of the measurement. Effect size statistics (i.e. mean differences divided by pooled sd.) were computed to determine the magnitude of the difference in mean scores (20). A Cohen's D (ES) value closer to 0.20 indicates a small effect, whereas 0.50 a medium and 0.8 and over a big effect in two groups' comparisons. A third approach that was used in this manuscript for testing the construct validity of the WHOQOL-BREF-DARI was the Con rmatory Factor Analyses (

Socio-demographic characteristics
Of the 1473 respondents, 54.6% men and 45.4% women. The 38.0% of the study sample was 18-29 years old, while 14.8% was 30-39; 24.6% was 40-49 years old and 22.6% was 50 and over, with a mean age of 37.6 ± 14.2 years (min 14, max 86). Only 47.25% (n = 696) of the respondents were graduated from any school ;41.96% (n = 618) were illiterate and 10.79% (n = 159) were just know how to read and write. 25.4% of the respondents perceived themselves as poor; 46.0% as moderate and 28.6% as good or wealthy. About one quarter of the respondents self-reported an existence of an important illness (25.2%). Only 6.5% was current smokers and a great majority of the study sample (89.1%) stated that they had no access to health services.

Psychometric Results
Item frequency analyses showed that the oor and ceiling effects range for each domains are in acceptable limits for all domains. The range of the oor effects was 0.0 to 0.3% and ceiling effects was 0.1 to 3.2% for the domains of WHOQOL-BREF. Skewness and Kurtosis values showed that all of the four domain scores were normally distributed (< 1.0) ( Table 1).   As for the know groups validity analyses, all of the known categories of the gender, level of education, social class and having chronic illness were signi cantly sensitive to all of the four domain scores of the WHOQOL (p < 0.001). We see that women were much worse than men in terms of physical, psychological and environmental dimensions, whereas social class has moderate to high level of discriminable effect on all of the four domains. The physical well-being domain score was more negatively affected by having a chronic illness compared to other domains. Environmental health domain was more sensitive to socioeconomic independent variables such as level of education and social class and BMI was only sensitive to physical domain. (Table 3). Con rmatory factor analyses showed acceptable goodness of t results for RMR (0.066) and RMSEA (0.073), but GFI (0.88) and CFI (0.83) values were lower than expected. (Table 4). Df 246 Criterion validity was tested by multiple regression analysis using overall QoL item (item q1) as the dependent variable. The regression model revealed a R2 value of 57%. All of the four domains of the WHOQOL-BREF-Dari except that of Social relation domain (beta = 0.03) could explain the variance of the Q1. The best predicting domain is environmental domain on overall QoL. Social relations domain could not predict overall QoL at al. (Table 5). Recent political developments in Afghanistan are expected to have positive in uences on the health sector of the country that covers broader approach to determinants of health and the assessment of the outcomes of the health interventions both in public health and in the clinical practice. As one of the speci c aspects of the assessment of the well-being of the population quality of life assessments need to be done in addition to objective health measures in Afghanistan. This study assessed the cultural relevance and psychometric properties of the Afghan Dari language version of the WHOQOL-Bref.
Linguistic adaptation of the original WHOQOL-Bref to Dari language followed the strict international translation procedures including multi forward and backward translation and cognitive interviews in the Herat city, the third biggest city of the country. The study sample is a representative large sample, allowing the generalizability of the results to Dari -the leading o cial language of the country-speaking community of Afghanistan. We tested the construct validity of the WHOQOL-Dari by three different methods: Convergent-divergent validity, known groups validity and the factor analyses. The generic SF-36 scale was used to assess the convergent validity of the WHOQOL-Bref-Dari version. Physical health dimension score of the WHOQOL revealed higher correlation coe cients with the physical health related dimension scores of the SF-36 than the emotional dimension scores of the SF-36. Also a moderate convergence was found between WHOQOL psychological dimension score and the Emotional Well-being dimension score of the SF-36. Social relations dimension and the Environmental dimension scores of the WHOQOL, had poorer convergence with the related dimension of the SF-36. Our results are consistent with some other papers in that, the Environmental health dimension of the WHOQOL had no signi cant correlations with any of the dimensions of the SF-36 (27). The poor convergence of the WHOQOL Environmental dimension with neither physical nor mental dimension scores may be due its content of socio-economic drive. Hence Environmental health dimension represents the socioeconomic well-being of the respondent and was not sensitive to any health related concepts which was shown in several studies including original WHOQOLgroup papers (6) (24) (28) (29) .
Lower Environmental dimension scores were also obtained from the less developed European study sites such as Romania and Turkey in the WHOQOL global study mentioned above. Our known groups validity results also con rmed the higher sensitivity of the environmental health score to the socioeconomic indicators: Quite big effect size gures were obtained for education (0.78) and social class (0.75) variables. On the other hand, several studies indicated that there is not a good convergence between WHOQOL and SF-36 (29) (30) (31).
The Known groups validity analyzes are performed to show previously proven relationships. The women, the ill and low educated people and also those people who belonged to a lower social class are previously known as disadvantaged groups in the community in regard to quality of life. Our results indicated that, all four dimensions of the WHOQOL Bref-Dari version are sensitive to gender, education, social class, objective health status, consistent with literature ndings (5) (6)(7) 29).
The con rmatory factor analyses results, as the third method of testing the construct validity of the Criterion validity of the WHOQOL-Dari version is shown by a linear regression of the Overall QoL item of the WHOQOL over the domain scores. The best Beta values were obtained for Physical health and Social relations dimensions. The only non-signi cant dimension is the environmental dimension. This is consistent with the nding that environmental dimension is the weakest related dimension with illness variable, that was shown in the known groups validity ndings of our study.
There are several limitations of this study. First, the WHOQOL-Bref has been face to face administered to the respondents due to the high portion (about 42%) of illiterate population, which might positively bias the results of this validation study. Secondly, it was not always possible to ful ll personal privacy during the interviews. And nally the men/women participation rate was higher in favor of men due to the social restrictions of the Afghan community.
To conclude we can say that, WHOQOL-Bref-Dari version can be con dently used in public health programs and clinical practice in Dari speaking communities of Afghanistan, but Social relations dimension scores should be interpreted with caution.

Conclusion
As we know, this is the rst study on Afghanistan's general population demonstrating the relation between socio-demographic variables and QoL domains. Afghan Dari version of the WHOQOL-BREF can con dently be used in clinical setting and in population level to assess the QoL of the people. The results of the social relations domain should be interpreted with caution due to its poor psychometric power.
Further studies are needed to address the social aspects of quality of life in Afghan population.