Psychometric Properties of the Warwick–Edinburgh Mental Wellbeing Scale (WEMWBS) in the Iranian Older Adults

Introduction

Mental wellbeing has been referred to experience of happiness¹ in the literature and as a single construct which could embody a person’s personality structure. Having a positive approach toward life events and a balanced positive and negative emotional state² can be regarded as an important factor in older adults’ mental health.^3,4 Mental wellbeing has been reported to be associated with lower risk of premature mortality, lower risk of physical disorders, better mental health, and lower use of healthcare services.^5,6 Mental wellbeing has also been suggested to be an important factor in mental health and fruitful interpersonal relationships.⁷

Age-related changes that affect older adults’ health and functioning can negatively influence their mental wellbeing.⁸ More than 20% of people aged 55 years and over have reported having a kind of mental health problem.⁹ Different factors, such as loneliness and social isolation, social limitations, self-determination and feelings of insecurity can influence older adults’ mental wellbeing.¹⁰ While reduced mental wellbeing in older adults due to their age-related physical and mental capacities’ deterioration should be considered as natural and inevitable phenomena, it is necessary to increase the public’s attention toward the mental wellbeing and possible deviances in mental health during old age.¹¹

Impairments in mental wellbeing can reduce older adults’ ability to perform everyday tasks, reduce their independence, and affect their quality of life.¹² Therefore, early diagnosis of this impairment can potentially reduce health care costs for families and communities and help in developing strategies to cope with this inevitable challenge.¹³

Various instruments have been developed to assess mental wellbeing, such as the 20-item Positive and Negative Affect Schedule (PANAS),¹⁴ the 5-item Satisfaction with Life Scale (SWLS),¹⁵ the 54-item Ryff’s Psychological Well-Being Scales (PWB),¹⁶ the 5-item Short Depression-Happiness Scale (SDHS)¹⁷ and the Warwick–Edinburgh Mental Wellbeing Scale (WEMWBS).¹⁸

Among the invented instruments, the WEMWBS, due to its conciseness and simplicity, can be applied in research and practice settings to measure mental wellbeing and affective-emotional and cognitive aspects and psychological functioning through 14 items.

This study was conducted to translate and assess psychometric properties of the Persian translated version of the WEMWBS-P for use among Iranian and other Persian-speaking older adults.

Patients and Methods

The study participants were 300 patients admitted to the Shohada hospital affiliated to the Tabriz University of Medical Sciences, North West of Iran from May to September 2017. The sample size was selected according to the minimum sample size recommended for analysis of covariance (ANCOVA). The present study was approved by the Medical Ethics Board of Trustees (MEBoT) in the Tabriz University of Medical Sciences (approval number: 5/D/1030407-1395-12-11).

The conventional translation/back translation procedure was followed to translate the original English form of the WEMWBS into Persian by two independent proficient translators (Forward translation). The two translators reached consensus about mismatches in a face to face meeting. In the next step, the Persian translated questionnaire was translated back into English by two other translators with good knowledge of English and experience in translating texts from Persian to English (Backward translation). The back translated version was compared to the original scale at the next stage and any disagreement was resolved with consensus. The final Persian version of the WEMWBS was sent to a panel of experts including 10 psychologists and health care scholars for their opinions about the WEMWBS-P and to assess lucidity of the wordings and comprehensibility of the items. Slight modifications were made based on their feedback.

Inclusion and Exclusion Criteria

Based on the United Nations (UN) agreed cutoff point of 60 years to refer to the older population,¹⁹ those aged above 60 years and with the ability to understand and speak Persian were eligible to be included in the study. Older adults with cognitive impairments or other severely limiting complications with potential to curtail completion of the informed consent form were excluded from the study.

The Warwick–Edinburgh Mental Wellbeing Scale (WEMWBS)

The WEMWBS was developed by Tennant et al in 2007.¹⁸ It has 14 items that are rated on a 5-point Likert-type scale ranging from 1 (Never) to 5 (Always). The highest and lowest scores on the scale are 14 and 70, respectively. The higher scores indicate better mental health. The WEMWBS only assesses the positive aspect of mental health and can be used for assessment of mental health on the society, group and individual levels. It has been translated into different languages for mental health screening.^20–29 The existent research evidence has shown its applicability in different geographical and cultural groups.²³ The performed factor analysis by Tennant et al suggested a one factor solution for the WEMWBS with the Cronbach’s alpha level of 0.89 and 0.91 among the students and general public respectively.¹⁸

Data Analysis

Exploratory factor analysis (EFA) was carried out to determine the unidimensionality of the constructs in the WEMWBS-P. The EFA was performed using the SPSS software, version 22. Principal Component Analysis (PCA) as the default extraction method and the Varimax Rotation to assess the underlying structure of the data were performed and decision was made based on eigenvalues < 1 and a scree plots showing the number of factors. Varimax rotation is the common orthogonal rotation used in factor analysis to associate each original item with one factor by splitting the items as much as possible into disjoint sets of factors. Thus, every extracted factor is represented with small number of variables and interpretation of the results will be simplified.³⁰ The minimum factor loading for a variable or item was 0.4. The significance level was set at P<0.0.5 level. Ceiling and floor effects were examined based on percentage of scores at the boundaries of the scaling and considered to be a problem if more than 20% of the study attendees achieved either the best or worst possible score.³¹ The skewness and kurtosis of the scores’ distribution were also investigated to ensure that they were within acceptable limits (less than ±2) for analyses.³²

Construct Validity

Based on the EFA results which were used to examine the factor structure a CFA was conducted to confirm the factors extracted from the EFA. The Chi-Squared test, Chi-Squared to df ratio (χ2/df), Comparative Fit Index (CFI), Root Mean Square Residual (RMR), Tucker Lewis Index (TLI) and Adjusted Goodness Of Fit index (AGFI) were used to assess the output model goodness of fit.

RMSA and RMR ≤ 0.08; CFI, GFI, and TLI > 0.9; and χ²/df < 5 were considered as acceptable ranges to indicate fitness of the model to the data.²¹

The reliability of the instrument was assessed using a two-week interval test-retest reliably assessment and its internal consistency was assessed using the Cronbach’s alpha coefficient.

Those older adults who had participated in the pilot study to determine test-retest coefficients were excluded from the main study. Participation in the study was completely voluntarily, and participants were reassured about confidentiality of their personal information. A written informed consent was obtained from the study participants and all the study procedures were conducted in accordance with the Declaration of Helsinki.

Results

The mean and standard deviation of participants’ age were 70.65 and 8.71, respectively. Among the study respondents 160 (52.6%) were women, 200 (65.8%) were married, and 187 (61.5%) were illiterate. The mean and standard deviation of mental wellbeing score were 50.30 and 8.82 respectively. There was a significant relationship between the mental wellbeing score and gender, education and marital status (Table 1).

Table 1 Socio-Demographic Properties of the Study Participants and Their Association with the Total Mental Wellbeing Score

Normality and Ceiling and Floor Effects

The mental wellbeing score had a normal distribution (skewness=−0.849, kurtosis=0.272) with a skewness to the right (Figure 1).

Figure 1 Distribution of the studied Iranian older adults’ according to their WEMWBS-P scores.

The ceiling effect was below 20% for the 14 items, except for items 8 (I have been feeling good about myself), 9 (I have been feeling close to other people), 10 (I have been feeling confident), and 12 (I have been feeling loved). The floor effect was also below 20% for the 14 items, except for item 4 (I have been feeling interested in other people) and for the total mental wellbeing score. All of the items of the WEMWBS-P were positively correlated with the total mental wellbeing score (Table 2).

Table 2 Values of the Floor and Ceiling Effects, The Item-Total Correlations, and Factor Analysis Loadings in the Analysis of the Psychometric Properties of the Persian Version of Warwick–Edinburgh Mental Wellbeing Scale (WEMWBS)

Content Validity

The final, translated scale was sent to a panel of experts including 10 psychologists and health care scholars for their opinions about the WEMWBS-P. The Content Validity Index (CVI) for all the scale’s items was calculated based on a panel of experts’ ratings in terms of their relevance to the targeted construct. The Lawshe’s Content Validity Ratio (CVR) was also estimated to quantify content validity and corroborate congruence of the items with the intended construct.³⁴ The analysis results revealed good content validity (CVR = 0.84, CVI = 0.85) of the WEMWBS-P.

Reliability of the Scale

The internal consistency of the WEMWBS-P was assessed using Cronbach’s alpha coefficient which was in the acceptable range (0.93).

Factor Structure

The factor analysis using the PCA method with 152 participants, revealed a KMO of 0.94, indicating that the data were appropriate for factor analysis. The Bartlett’s test of Sphericity which is the test for null hypothesis that the correlation matrix has an identity matrix also confirmed correlations between the variables and suitability of the EFA for data analysis (P<0.0001). According to the results of factor analysis (Table 2) and the scree plot (Figure 2), one factor was extracted that explained 60% of the variance of the total score. Item number 4 was eliminated due to having a factor loading below 0.2. Finally, the WEMWBS-P with one factor and 13 items was validated among the Iranian older adults. The total variance explained by the scale’s one factor was 53% and the Scree plot supported the unidimensionality of the WEMWBS-P questionnaire (Figure 2).

Figure 2 Scree plot representing the eigenvalues of the extracted variables obtained in the principal-component analysis for psychometric validation of the WEMWBS-P.

Following the results obtained in the EFA, the CFA led to the flowing values: χ²/df = 2.10 (P<0.0001), CFI = 0.909, RMSEA = 0.084, TLI = 0.891, GFI = 0.872 and AGFI = 0.820 which are all confirmatory of the EFA outputs.

Discussion

Due to lack of a standard instrument for assessment of mental wellbeing in Iranian older adults, the present study was performed to examine validity and reliability of the Persian version of the Warwick–Edinburgh Mental Wellbeing Scale (WEMWBS-P) for use among Iranian older adults. The results indicated that the WEMWBS-P with one factor and 13 items had acceptable validity and reliability to assess the mental wellbeing of Iranian older adults.^18,25,26,35 This finding is consistent with the results of previous studies with different languages and in different populations, extracting one factor for the instrument. However, in contrary to the previous studies in which no item of the scale had been excluded, in the present study, item number 4 was eliminated due to an eigenvalue of below 0.2. This difference can be related to the cultural diversity of the population under study and the way the older adults perceived the item.

The WEMWBS-P indicated a high internal consistency. Consistent with this result, the reliability of the scale has been reported to be high in different populations.^18,24,26 The data distribution had skewness to the left. Consistent with this finding, previous studies on the psychometric properties of the WEMWBS-P in different populations, including the general population, have indicated skewness to the left.^24,26,27 However, in the Spanish and English speaking populations, the scores on the WEWBS had normal distribution. This difference between the findings can be due to differences in participants’ age and sample sizes. Therefore, it is recommended that the scale be studied in different populations using larger sample sizes.

Our findings showed that the ceiling effect for all the items, except items 6, 8, 9, 10, and 12, and the floor effect for all the items, except item 4 were below 20%. Previous studies conducted in other populations have also shown ceiling and floor effects in the items of this scale.^26,35 This discrepancy in the findings can be related to different sample sizes. In other words, sample size in the study by Dong et al, with nurses, was smaller than that in our study, and sample sizes in the studies conducted in Spanish and English populations were larger than our sample.^18,26,35

A Cronbach’s alpha coefficient of 0.93 and item-total correlations from 0.50 to 0.80 were found for the WEMWBS-P. In line with these results, previous studies have also found high Cronbach’s alphas for the WEMWBS-P.^24,26,35 These results indicate that this scale has a high internal consistency.

According to the results, there was a significant gender difference in the mental wellbeing score. This finding is consistent with that of Santos et al.²⁸ However, some previous studies have not found gender differences in mental wellbeing.^18,24 This difference in the results can be due to differences in participants’ age, health status, and disorders and different cultural beliefs regarding mental wellbeing in different populations. In addition, our population included hospitalized older adults aged 60 and over; this can also explain some differences between our findings and those of previous studies.

According to the study results, the literate older adults had better mental wellbeing than illiterate older adults. In line with this finding, Ibitoye et al showed that highly educated older adults had a better mental wellbeing profile.³⁵ A higher educational attainment appears to be an influential covariate of mental wellbeing in older adults however, in a number of other studies, significant difference was not observed in mental wellbeing between people with higher and lower educational qualifications.²⁹ These contrasting findings may imply that education level alone is not a strong determinant of older adults’ mental wellbeing, and further research is required to verify complex interplay among the education level, other predictors and mental wellbeing in old age.

Another finding was that the married older adults had better mental wellbeing than those who were single or widowed. Consistent with this result, previous studies have shown that marital status could predict mental wellbeing, with married individuals scoring higher on mental wellbeing than single individuals.^36,37 It appears that the presence of a supportive partner could have a benevolent impact on the mental wellbeing of married older adults. Old age is generally accompanied by certain stressors, such as being away from children, suffering from chronic disorders, reduced contact with peers, lower participation in the society and financial problems thence, presence of a partner could mitigate hurdles of facing these challenges.

Limitations

Interpretation of this study’s findings warrants further scrutiny due to the recruited sample attributes. Hospitalized older adults were approached in this research to collect data for psychometric appraisal of the WEMWBS-P. Therefore, probability of response bias as result of the interviewees’ health status and their subsequent imprecise responses to the study questions should not be dismissed completely. Bias due to the patients’ awkward perceptions about their overall health and illness or hospital experiences and in consequence, their tendency to inflate their responses to cast a more favorable or worse health status than the actual health condition could be another source of uncertainty in this study.^38,39

Other factors such as anxiety, depression, and resilience, socio-demographic and psychosocial variables were also shown in previous studies^40–42 to affect self-perceived burden of health conditions. Thus, reliable assessment of mental wellbeing warrants careful consideration in future studies.

Conclusion

The study findings revealed relatively robust validity and reliability properties for the WEMWBS-P to be applied in studies on assessment of Iranian and other Persian-speaking older adults’ mental wellbeing. Further cross-cultural, cross-linguistic and comparative multi-national studies are recommended to provide more cogent evidence for bridging the knowledge gap about feasibility and desirability of the WEMWBS-P application in diverse settings.