SWEMWBS

SWEMWBS Metric Scores and Mental Wellbeing Utilities

SWEMWBS keeps seven of the fourteen WEMWBS items, the set that Stewart-Brown and colleagues found in 2009 to fit a Rasch model, and reports a metric score between 7 and 35 obtained by converting the raw item total through a published table. For a health economist its main interest is a route to preference-based values, through a UK value set for mental wellbeing adjusted life years (MWALYs) and a mapping to the ReQoL utility index. Seven items also take less respondent time than fourteen, which matters in population surveys and in programme evaluations that collect several outcomes. Reading a SWEMWBS result correctly depends on knowing whether a reported figure is the raw total or the metric score, because the conversion between them is not linear.

Seven items selected by Rasch analysis

The short form came out of a test of the full WEMWBS against the Rasch measurement model, which asks whether items behave as a single scale with equal meaning across groups. Stewart-Brown and colleagues applied the model to 779 respondents aged 16 to 74 (average 41.9 years) in Wave 12 of the Scottish Health Education Population Survey, collected in autumn 2006.

The fourteen items fitted the model poorly. Three items, about feeling good about oneself, being interested in new things and feeling cheerful, showed significant misfit and were deleted. Further deletions left a strict unidimensional seven-item scale, which the authors named the Short Warwick Edinburgh Mental Well-Being Scale.

The seven-item scale was largely free of bias. Many of the removed items had shown considerable bias by gender, and two retained items also showed gender bias, but at scale level the two cancelled out. The authors noted that the short form presents a more restricted view of mental well-being than the fourteen-item scale, with most items representing psychological and eudemonic well-being and few covering hedonic well-being or affect. They still judged that its measurement properties and brevity made it preferable to the full scale, at that time, for monitoring mental well-being in populations.

Raw scores, metric scores and the conversion table

Each of the seven items is answered on the same five-point scale as the full WEMWBS, so the raw total runs from 7 to 35. The raw total is an ordinal count. Stewart-Brown and colleagues described SWEMWBS as providing a raw score to interval scale transformation for use in parametric procedures, published as a conversion table that turns each raw total into a metric score on an interval scale.

The University of Warwick's guidance says SWEMWBS is more complicated to score than the full scale and must be scored correctly for comparisons to be made across studies, because the raw scores need to be transformed. The conversion keeps the endpoints, so a raw 7 is a metric 7.00 and a raw 35 is a metric 35.00, and the published rows for raw totals 7 to 13 show it stretching the distances at the floor.

$$M_i = T(R_i), \qquad R_i = \sum_{k=1}^{7} x_{ik}$$

where $M_i$ is the metric score of respondent $i$, $T$ is the published conversion table, $R_i$ is the raw total and $x_{ik}$ is the response of respondent $i$ to item $k$, scored 1 to 5.

The first rows show the non-linearity. One raw point at the floor is worth 2.51 metric points, and one raw point between 12 and 13 is worth 0.67.

Illustrative worked example of change near the floor of the scale

All figures here are illustrative, and the conversion values are those in the table above. Two participants in a wellbeing programme each improve by one raw point. Participant A starts at a raw 7 and participant B at a raw 12.

1. Raw change. Each participant gains 1 raw point, so the mean raw change is 1.

2. Metric change. Participant A gains 9.51 − 7.00 = 2.51 metric points. Participant B gains 14.75 − 14.08 = 0.67 metric points.

3. Mean metric change.

$$\bar{\Delta}_M = \frac{2.51 + 0.67}{2} = \frac{3.18}{2} = 1.59$$

where $\bar{\Delta}_M$ is the mean change in metric score across the two participants.

4. Cost per point. If the programme costs GBP 100 per participant, the cost per raw point is 100 / 1 = 100, or GBP 100, while the cost per metric point is 100 divided by 1.59, about GBP 62.89.

The same programme looks more effective on the metric scale because one participant started at the floor, where the Rasch conversion widens the gaps between raw scores. In a real evaluation the choice between raw and metric scores changes effect sizes and cost per point whenever baseline scores differ between groups, so a reported result has to state which score was analysed. Neither figure can be compared with a cost per QALY, because a metric point has no utility value.

National norms from the Health Survey for England

Population norms let an evaluator judge where a trial sample starts relative to the general public. Ng Fat and colleagues reported norms in 2017 from the Health Survey for England 2010 to 2013, covering 27,169 adults aged 16 and over.

They produced age-specific and sex-specific norms from means and percentiles. They tested criterion validity against the GHQ-12, a happiness index and the EQ-5D visual analogue scale, and compared the short form with the full fourteen-item WEMWBS in the same survey.

Responsiveness in a clinical sample

Responsiveness decides whether the scale can show the effect of an intervention within a trial's follow-up. Shah and colleagues reported in 2018 on 172 clients receiving up to four sessions of cognitive hypnotherapy.

Mean SWEMWBS scores rose from 19.28 at baseline to 23.32 by the fourth session. Effect sizes ranged from 0.20 to 1.41 and standardised response means from 0.30 to 0.88, with responsiveness increasing with the number of sessions. At individual level, a change of between 1 and 3 points met thresholds for statistically important change, depending on the standard used. The authors noted that anchor-based studies are needed to confirm that such change is minimally important from the participants' perspective.

Preference-based values for SWEMWBS

A summed or metric SWEMWBS score is not a utility, but two lines of work now attach preference-based values to SWEMWBS responses. One values SWEMWBS states directly, and the other predicts utilities on another instrument.

A UK value set for mental wellbeing adjusted life years

Yiu and colleagues published a UK preference-based value set for SWEMWBS in 2023. They present the MWALY as an alternative outcome measure to the QALY in economic evaluations of interventions aimed at improving mental well-being. Between December 2020 and August 2021, 225 participants each completed 10 composite time trade-off tasks and 10 discrete choice experiment tasks, administered by an interviewer.

The time trade-off responses were modelled with a heteroskedastic Tobit model and the choice responses with a conditional logit model. The choice results were rescaled to a scale comparable with the time trade-off, and an inverse variance weighting hybrid model combined the two sets of coefficients into weighted-average values. Because the authors present the MWALY as an alternative to the QALY, a cost per MWALY is a separate quantity from a cost per QALY and cannot be read against a cost-effectiveness threshold set for QALYs.

Mapping SWEMWBS to the ReQoL utility index

Keetharuth and colleagues built a mapping function in 2024 for predicting ReQoL-UI utilities from SWEMWBS responses. Service users with mental health difficulties were recruited in two cross-sectional UK studies, and 2,573 respondents had complete data on both instruments, age and sex.

Direct models (ordinary least squares, Tobit and generalised linear models) performed similarly. An indirect, response-based approach predicted the answer to each ReQoL-UI item and then applied the UK tariff, and simulations that allowed for uncertainty favoured it. The authors recommend the indirect algorithm.

SWEMWBS alongside the EQ-5D-5L

Comparisons with the EQ-5D-5L show how far a wellbeing measure and a health-related quality of life measure overlap. Hong and colleagues compared the two instruments in 500 online respondents with one of four chronic conditions in China. Apart from anxiety or depression, the EQ-5D-5L dimensions correlated only weakly with the SWEMWBS items, and in exploratory factor analysis the seven SWEMWBS items loaded on one factor and four EQ-5D-5L dimensions on another. Both instruments showed good convergent and known-group validity, with the EQ-5D-5L the more sensitive even for mental conditions, and the authors concluded that they measure different constructs and can be seen as complementary.

How SWEMWBS differs from WEMWBS and other wellbeing measures

The short and long forms share their origin, items and response scale but not their scoring or their economic uses. The differences matter most when a result has to feed a utility-based analysis.

Outside the Warwick-Edinburgh family, the EQ-HWB was developed to measure health and wellbeing for patients, social care users and carers in economic evaluation, and ICECAP-A measures capability wellbeing in adults. ReQoL-10 belongs to the Recovering Quality of Life measures used in mental health. SWEMWBS differs from these in measuring positive mental wellbeing only, with no items on symptoms, physical health or capability.

Limitations and common misreadings

The short form is easy to administer and easy to misanalyse. Several recurring errors change effect estimates or the meaning of an economic result.

• Reporting raw scores as SWEMWBS scores. A raw total and a metric score share the range 7 to 35, so a table that does not say which was used cannot be compared with another study or with national norms.
• Pooling raw and metric scores. Combining studies that analysed different scores mixes two non-linear versions of the same responses.
• Ignoring the restricted content. With few affect items, an intervention whose main effect is on mood may move SWEMWBS less than the full scale.
• Treating a MWALY as a QALY. The value set rests on 225 interviews valuing SWEMWBS states, and its values describe mental wellbeing, not health-related quality of life.
• Reading distribution-based thresholds as patient-important change. The 1 to 3 point thresholds from the clinical sample describe change beyond measurement error, and the authors called for anchor-based work.

Sources

• Hong Y, Jiang X, Zhang T, Luo N, Yang Z. Examining the relationship between the Short Warwick-Edinburgh Mental Well-being Scale (SWEMWBS) and EQ-5D-5L and comparing their psychometric properties. Health and Quality of Life Outcomes. 2023;21:25. doi:10.1186/s12955-023-02108-y
• Keetharuth AD, Gray LA, McGrane E, Worboys H, Orozco-Leal G. Mapping Short Warwick and Edinburgh Mental Wellbeing Scale (SWEMWBS) to Recovering Quality of Life (ReQoL) to estimate health utilities. Health and Quality of Life Outcomes. 2024;22:7. doi:10.1186/s12955-023-02220-z
• Ng Fat L, Scholes S, Boniface S, Mindell J, Stewart-Brown S. Evaluating and establishing national norms for mental wellbeing using the short Warwick-Edinburgh Mental Well-being Scale (SWEMWBS): findings from the Health Survey for England. Quality of Life Research. 2017;26(5):1129-1144. https://pmc.ncbi.nlm.nih.gov/articles/PMC5376387/
• Shah N, Cader M, Andrews WP, Wijesekera D, Stewart-Brown SL. Responsiveness of the Short Warwick Edinburgh Mental Well-Being Scale (SWEMWBS): evaluation a clinical sample. Health and Quality of Life Outcomes. 2018;16:239. doi:10.1186/s12955-018-1060-2
• Stewart-Brown S, Tennant A, Tennant R, Platt S, Parkinson J, Weich S. Internal construct validity of the Warwick-Edinburgh Mental Well-being Scale (WEMWBS): a Rasch analysis using data from the Scottish Health Education Population Survey. Health and Quality of Life Outcomes. 2009;7:15. https://pmc.ncbi.nlm.nih.gov/articles/PMC2669062/
• University of Warwick. WEMWBS, how to use and score the scales. https://warwick.ac.uk/services/innovations/wemwbs/how/ (accessed 17 October 2026)
• Yiu HHE, Buckell J, Petrou S, Stewart-Brown S, Madan JJ. Derivation of a UK preference-based value set for the Short Warwick-Edinburgh Mental Well-being Scale (SWEMWBS) to allow estimation of Mental Well-being Adjusted Life Years (MWALYs). Social Science and Medicine. 2023;327:115928. https://pubmed.ncbi.nlm.nih.gov/37201343/