Article Text

Download PDFPDF
P114 How are health behaviours associated with mental wellbeing using the short WaRwick Edinburgh Mental Wellbeing Scale (SWEMWBS)? An evaluation of SWEMWBS using the Health Survey for England
  1. L Ng Fat1,
  2. S Scholes1,
  3. S Boniface2,
  4. J Mindell1,
  5. S Stewart-Brown3
  1. 1Epidemiology and Public Health, University College London, London, UK
  2. 2Addictions, Kings College London, London, UK
  3. 3Statistics and Epidemiology, University of Warwick, Coventry, UK


Background The Warwick-Edinburgh Mental Well-being scale (WEMWBS), a set of 14 positive worded statements, has been validated as an instrument to measure mental wellbeing on a population level. A shorter, seven-item version (SWEMWBS) was developed after applying the Rasch measurement model, however little is known on how SWEMWBS is distributed in the population, and how well it performs as an instrument to measure wellbeing. Using data from nationally-representative samples, we computed age- and sex-specific norms for the SWEMWBS for use in interpreting individual- and group-scores, and examined relative- and criterion-validity.

Methods Based on adults aged 16+ from The Health Survey for England 2010–13 (n = 27,169), four nationally representative surveys of the population living in private households, age- and sex-specific norms were estimated using means and percentiles. Criterion-validity was examined using: (1) Spearman correlations (ρ) for SWEMWBS versus instruments expected to be correlated with wellbeing (e.g. General Health Questionnaire (GHQ-12), Happiness Index, EQ-5D), and (2) a multinomial logit model with SWEMWBS (grouped into low, medium, and high wellbeing) as the outcome variable and the following explanatory variables: sex, age, general health, marital status, ethnicity, education, income, body mass index, fruit and vegetable intake, alcohol consumption on the heaviest drinking day, and smoking status. Relative validity was examined by comparing SWEMWBS with WEMWBS using: (1) Spearman correlations (continuous data), and (2) the weighted Kappa statistic (categorical), within population sub-groups.

Results Mean (median) SWEMWBS was 23.7 (23.2) for men and 23.2 (23.2) for women (p < 0.01). Spearman correlations were moderately sized for the Happiness Index (ρ = 0.53, P < 0.001), GHQ-12 (ρ = −0.52, p < 0.001) and EQ-5D (ρ = 0.40, p < 0.001). Weaker correlations were found for self-rated health (ρ = −0.33, p < 0.001) and longstanding illness (ρ = −0.21, p < 0.001). Results from the multinomial logit model with the medium category as reference showed that participants with bad/very bad vs very-goodself-rated health were significantly more likely to have low vs medium wellbeing (Odds ratio (OR) 10.47, 95% Confidence Interval (CI) 8.90–12.31). Participants consuming <1 portion of fruit and vegetables a day vs ≥5 (1.43, 1.22–1.66) and current smokers vs non-smokers (1.28, 1.15–1.41)) were also more likely to have low vs medium wellbeing. Participants who binge drank versus non-drinkers were less likely to have high versus medium wellbeing (0.81 (0.71–0.92)). Spearman correlations between SWEMWBS and WEMWBS were above 0.95; weighted Kappa statistics showed almost perfect agreement (0.79 to 0.85).

Conclusion SWEMWBS distinguishes wellbeing between groups, similarly to WEMWBS. SWEMWBS can be used to measure wellbeing where there are time and resource constraints.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.