Mental health, mental illness and mental well-being

Mental illnesses are among the main causes of disease burden globally.^{1–Reference Whiteford, Degenhardt, Rehm, Baxter, Ferrari and Erskine3} Mental health, by contrast, is described by the World Health Organization (WHO) as ‘a state of well-being’⁴ and as ‘a positive phenomenon that is more than the absence of mental illness’ by Westerhof & Keyes.^{Reference Westerhof and Keyes5} The concept of mental well-being was first introduced by the WHO in 1948 and is said to occur when an individual can cope with the normal stresses of life, work productively and fulfilling, feels happy and satisfied with their life, and is able to make a contribution to his or her community.⁶ Although mental illness and mental well-being were once seen as two ends of a single continuum (from flourishing to languishing),^{Reference Huppert, Baylis and Keverne7} empirical evidence now supports the view that these are two correlated dimensions.^{Reference Westerhof and Keyes5,Reference Weich, Brugha, King, McManus, Bebbington and Jenkins8} This is consistent with the view that it is possible to have moderate or high levels of mental well-being (including enjoying life and/or having a sense of purpose and fulfilment) despite experiencing mental illness.

Although governments and others are committed to promoting mental well-being as a means of enhancing prosperity,^{Reference Aked, Marks, Cordon and Thompson9–13} it remains unclear whether, and to what degree, promoting mental well-being can reduce the burden of mental illness and improve outcomes for people with established mental illness. An observational study by Meyer compared 30 patients over 6 weeks with different types of mental illness.^{Reference Meyer14} Considering coping as a proxy for functioning, the association between mental well-being (measured using the Psychological Well-Being Questionnaire^{Reference Ryff and Keyes15}) and functioning was statistically significant in the schizophrenia group. Current clinical guidance for managing serious mental illnesses (such as bipolar disorder and psychosis and schizophrenia) focuses on achieving symptom resolution.^{16–Reference Slade19} It is not yet known whether and to what extent mental well-being may be associated with, or contribute to, functional improvements in people with mental illness, and whether this is independent of symptoms. Such an association would support the importance of developing and delivering interventions to enhance mental well-being in people experiencing episodes of mental illness.

Aims

The aim of this study was to test whether mental well-being was associated with differences in functioning in people with a self-reported history of mental illness, after adjusting for differences in symptoms of mental distress.

Mental illness groups

Mental illness groups were defined according to self-reported history of mental illness. Respondents were asked whether they had experienced or were diagnosed with any mental health conditions using a list of 17 conditions. If the response was ‘yes’ to any of these, the interviewer proceeded to ask if a diagnosis had ever been given or confirmed by a doctor, psychiatrist or other professional. Serious mental illness was defined as a reported diagnosis (i.e. one that had been confirmed by a professional) of bipolar disorder, eating disorder, nervous breakdown, personality disorder, psychosis or schizophrenia. History of mental illness was classified as follows: (a) no mental illness; (b) mental illness experienced, but not diagnosed by a professional; (c) diagnosed mental illness other than serious mental illness; (d) diagnosis of serious mental illness.

Mental well-being

The Warwick-Edinburgh Mental Well-Being Scale (WEMWBS)^{Reference Tennant, Hiller, Fishwick, Platt, Joseph and Weich21} was administered by self-completed questionnaire. WEMWBS comprises 14 items covering psychological functioning, and cognitive-evaluative and affective-emotional aspects of well-being.^{Reference Tennant, Hiller, Fishwick, Platt, Joseph and Weich21} All 14 items are worded positively and responses are recorded using a Likert scale.²² An example is: ‘I've been feeling optimistic about the future’, with responses coded from one for ‘none of the time’ to five for ‘all of the time’. Although WEMWBS includes hedonic and eudemonic items, principal component analysis only extracted one component and hence total WEMWBS score was used, derived as the sum of item responses (minimum possible score 14, maximum 70).

Functioning

As there was no questionnaire set within HSE 2014 dedicated to measuring functioning, this was derived from responses to EQ-5D items.²³ EQ-5D is a self-completed questionnaire that asks about general health and functioning using items covering mobility, self-care, usual activities (i.e. work, study, housework, family and leisure activities that the respondent would normally undertake), pain or discomfort, and anxiety and depression.^{Reference Gusi, Olivares and Rajendram24} We summed scores for the mobility, self-care and usual activities items. A response was recorded for each item using a Likert scale, with 1 representing the least problems or impediment and 3 representing extreme inability. We included only the mobility, self-care and usual activities items; the other two EQ-5D items were excluded as ‘anxiety’ is closely related to mental health and ‘pain’ is not a measure of functioning.

Principal components analysis using data from the three included EQ-5D items extracted one component, with similar weightings for all three items. We conducted further exploratory principal components analysis to include employment status.^{Reference McKnight and Kashdan25} The factor loading for employment status was modest, and hence this variable was not included in the derivation of functioning scores. Consequently, functioning scores were derived by summing the mobility, self-care and usual activities item scores.

Symptoms of mental distress

Symptoms of mental distress were ascertained using the General Health Questionnaire (GHQ-12).^{Reference Goldberg and Williams26} GHQ-12 is a self-report questionnaire with excellent psychometric properties, commonly used to screen for psychiatric disorders in non-clinical settings.^{Reference Goldberg and Williams26} It comprises 12 items, 6 worded positively and 6 worded negatively. Responses for each item are recorded using a Likert scale. The direction of scores for each item is such that a higher score indicates more severe symptoms. An example item is ‘Have you felt constantly under strain?’, with responses coded as 1 for ‘better than usual’ to 4 for ‘much less than usual’. Previous psychometric research has shown that GHQ-12 scores are best described by two factors, namely symptoms of mental distress and positive mental health.^{Reference Hu, Stewart-Brown, Twigg and Weich27} Six items were found to load onto the former factor. To avoid conflating our measures of mental well-being and mental distress, only the six negatively worded items were used to derive the mental distress score in this study, which was taken as the sum of scores on these items. The items were: ‘Have you recently (1) lost much sleep over worry; (2) felt constantly under strain; (3) felt you couldn't overcome your difficulties; (4) been feeling unhappy or depressed; (5) been losing confidence in yourself; and (6) been thinking of yourself as a worthless person?’.

Potential confounders

Other potential confounders of the association between mental well-being and functioning were age,^{Reference Blanchflower and Oswald28–Reference Keyes30} gender,^{Reference Pinquart and Sörensen31} ethnicity,^{Reference Heim, Hunter and Jones32–36} long-standing illness^{Reference Mossabir, Morris, Kennedy, Blickem and Rogers37–Reference Shields and Price39} and income. The HSE survey asks about the presence of long-standing illnesses, up to a maximum of six conditions. A summary physical illness variable was derived by summing the number of long-standing illness categories that respondents identified, excluding ‘mental disorders’. Equivalised household income was derived using McClements scores to account for household composition.

Statistical analysis

Descriptive statistics regarding respondent characteristics and other variables of interest were reported by mental illness group. Preliminary one-way analysis of variance (ANOVA) was conducted to test for significant differences in WEMWBS score between mental illness groups. The median WEMWBS score of the group with no mental illness was calculated and used to compare the proportion of people within the serious mental illness group above this level.

Moderated regression analyses were used to explore whether, and to what extent, the association between mental well-being and functioning scores varied between mental illness groups.^{Reference Dawson40,Reference Keith41} The moderator was represented as an interaction term for mental illness group × mental well-being (referred to as group × well-being) in the analyses. The size and statistical significance of these interaction terms were thus estimated.

Three models were developed in which functioning was the outcome: (a) model excluding interaction terms; (b) model including the interaction term for group × well-being; (c) model including group × well-being interaction and covariates. Where interactions in model 2 were not statistically significant, this was omitted from the final model.

Predicted functioning scores were derived from the regression coefficients (B) for each exposure group. The gradient for the slope (m) of each group was derived through algebraic manipulation by arranging the equation into the form of y = mx + C, where y is the dependent variable, functioning score; and x is the independent variable, WEMWBS score. All regression models used the group with self-reported history of ‘no mental illness’ as the reference.

To aid interpretation, results of models 2 and 3 were represented visually by means of a plot of functioning score against WEMWBS score. For both graphs, ‘low’ and ‘high’ WEMWBS scores were calculated as 1 s.d. below and above the mean for the study sample, respectively. As model 3 included covariates that are continuous variables, the graph for this model was plotted using centred variables to ensure that the value ‘zero’ is meaningful for these variables.^{Reference Dawson40} Centring was done by subtracting the variable's sample mean from its original value.

The effect size for the interaction term (group × well-being) was calculated using Cohen's f ² statistic, which calculates the ratio of variance explained by the interaction term alone to the unexplained variance in the final model. Cohen's f ² is expressed mathematically as,

$$f^2 = \displaystyle{{R_2^2 -R_1^2 } \over {1-R_2^2 }}$$

where R ₁² is the variance explained by the model excluding the term of interest, and R ₂² is the variance explained by the model including the term of interest. In general, a larger f ² denotes a larger effect size and vice versa.^{Reference Dawson40} It has been suggested that the values of 0.10, 0.25, and 0.40 represent small, medium, and large effect sizes, respectively.^{Reference Chen, Chen and Salkind42}

Analyses were performed using SPSS Version 24 for Macintosh. A published SPSS syntax for an ANCOVA version of two-way interaction with a categorical moderator was used in constructing and executing the regression models.^{Reference Dawson40}

Ethical considerations

This study involved the use of secondary data from the HSE 2014. Ethics clearance was sought and received from the Ethics Review Committee of the School of Health and Related Research at the University of Sheffield. The data-set was supplied by the UK Data Service who granted permission to use the data in its anonymised form for the purposes of this study.