Article Text

Does somatic illness explain the association between common mental disorder and elevated mortality? Findings from extended follow-up of study members in the UK Health and Lifestyle Survey
  1. G David Batty1,2,3,
  2. Mark Hamer3,
  3. Geoff Der2,3
  1. 1Department of Epidemiology and Public Health, University College London, London, UK
  2. 2MRC Centre for Cognitive Ageing & Cognitive Epidemiology, University of Edinburgh, Edinburgh, UK
  3. 3Medical Research Council Social & Public Health Sciences Unit, Glasgow, UK
  1. Correspondence to David Batty, Department of Epidemiology and Public Health, University College London, 1-19 Torrington Place, London WC1E 6BT, UK; david.batty{at}


Background Common mental disorder (psychological distress) is associated with an increased risk of disease-specific mortality. Given that physical illness is related to both exposure and outcome, it may explain this relation through confounding or mediation.

Methods The authors used a 20-year follow-up of the UK Health and Lifestyle Survey (6127 men and women) in which common mental disorder was ascertained at baseline using the 30 item General Health Questionnaire and physical illness using a range of enquiries. Study members were an average of 45.2 years (SD 17.0) at study induction.

Results In age-adjusted analyses, a 1 SD increase in common mental disorder score was associated with an elevated risk of mortality outcomes coronary heart disease (CHD) in men (HR 1.11, 95% CI 0.96 to 1.27), CHD in women (1.33, 1.16 to 1.51); plus, in men and women combined, stroke (1.13, 0.96 to 1.30), respiratory disease (1.31, 1.15 to 1.48), lung cancer (1.11, 0.92 to 1.33), ‘other’ cancer (1.14, 1.03 to 1.26) and all causes (1.18, 1.12 to 1.23). Controlling for prior physical illness effectively eliminated the common mental disorder–mortality relation in all analyses with the exception of CHD in women.

Conclusion That physical illness largely explains the link between common mental disorder and mortality in the present cohort is compatible with either a confounding or mediation explanation.

Statistics from


The first investigation of the impact of mental illness on life expectancy was probably undertaken over seven decades ago when institutionalised patients in New York City were found to have a higher death rate than the general population.1 Counter to expectations, this mortality differential was not solely ascribed to violent causes—unintentional injury and suicides—rather, elevated rates of death from chronic illness, such cardiovascular disease, were also observed.

Beyond the sphere of serious psychiatric conditions, investigators have recently explored the influence on disease-specific mortality of common mental disorder, which represents a less severe array of non-specific psychological symptoms that have greater population prevalence. A series of studies have shown that mental disorder is associated with cardiovascular disease2 and all-cause mortality,3 while fewer have examined links with cancer.4

Importantly, it has been postulated that the apparent common mental disorder–mortality gradient is explained by the close relation of both with physical comorbidities. As such, it may be physical illness rather than psychiatric illness that is generating a link with mortality (confounding) or physical illness lies on the pathway between psychiatric illness and mortality (mediation).5 This hypothesis can be easily tested by simply adding a physical illness variable(s) to a multivariable model containing mental disorder and mortality experience or, alternatively, excluding from analyses people with baseline illness. In the few studies conducted, the results are discordant: in some, the elevated risk of mortality in people reporting psychological disorder essentially holds,6 whereas in others, it is diminished3 7 or eliminated.8

We examine this issue using the UK Health and Lifestyle Survey, a prospective cohort study with 20 years of mortality surveillance in which common mental disorder was assessed at baseline using the 30-item General Health Questionnaire (GHQ30).9


Data were taken from the UK Health and Lifestyle Survey, a random sample of community-dwelling adults aged 18 years and over who were surveyed in 1984/1985.10 Of the 9003 men and women interviewed in the first phase of the study, 7414 took part in a nurse interview and 6572 mailed back a self-complete questionnaire, which included the GHQ30. Study participants were ‘flagged’ electronically at the NHS central registry and copies of death certificates are provided (updated until June 2005). Full data for the present analyses were available for 6127 respondents (3446 women). The BMA Ethical Committee approved the study.

The GHQ30, a self-administered inventory, comprises 30 questions covering a range of neurotic symptoms with an emphasis on those typical of anxiety and depression with a deliberate avoidance of those that might also reflect physical illness.9 The responses are made on a 4-point ordinal scale. The response for each item is scored from 0 to 3 and then summed over the 30 items (range 0–90).

Physical health status was extensively assessed at baseline via self-reported conditions, symptoms and self-rated health. Current medical conditions were ascertained using a standard question on longstanding illness (“Do you have any longstanding illness, disability or infirmity?”). Positive responders then reported their illness verbatim (eg, heart disease, stroke, emphysema, diabetes, cancer, etc). Respondents were also asked whether they had ever had any of 18 specified conditions (eg, asthma, bronchitis, other chest trouble, etc). The occurrence of common symptoms in the last month was also recorded, with ‘palpitations or breathlessness’ and ‘persistent cough’ used herein. Finally, relative to a counterpart of their own age, study members were asked to classify their own health as: ‘excellent’, ‘good’, ‘fair’ or ‘poor’. Given that smoking is related to both mortality outcomes featured herein11 and anxiety/depression,12 we also adjusted for this behaviour so allowing comparison with the impact of controlling for physical illness. Respondents smoking status was recorded (never smoked, ex-smoker, occasional or light smokers) as well as the number of cigarettes smoked and whether or not the smoke was inhaled.


The mean (SD) age of the sample at baseline was 45.2 years (17.0) (men: 45.8 years (17.2); women: 44.7 years (16.8)). In table 1, we show the relation between a 1 SD increase (worse) (10.6) in GHQ30 score and the risk of mortality from a variety of causes. Twenty years of follow-up gave rise to 804 deaths from all causes in men and 720 in women. In preliminary analyses, there was some evidence that the relationship between common mental disorder was more strongly related to coronary heart disease (CHD) mortality risk in women than in men (p value for interaction: 0.021), hence we present HRs for this outcome only according to gender—for all other outcomes, there was no evidence of such effect modification.

Table 1

Relation of a 1 SD increase in common mental disorder score with mortality experience: the UK Health and Lifestyle Survey (N=6127)

In age- and sex-adjusted analyses, common mental disorder was associated with an elevated risk of all six mortality outcomes, although statistical significance was not apparent on all occasions. In model 2, we added smoking as a covariate and computed the per cent attenuation in HRs relative to this comparator model (model 1). As previously described,14 15 key in interpreting these results is the change in HRs, if any, rather than loss of statistical significance. On controlling for smoking habit (model 2), the greatest attenuation in the common mental disorder–mortality association was, unsurprisingly, apparent for lung cancer. Effect estimates were attenuated most when physical illness was added to the multivariable model (model 3). The addition of socioeconomic status did not impact upon the pattern of the afore described relationships. Throughout these analyses, the common mental disorder–CHD gradient in women was most robust to statistical adjustment.


The main finding of these analyses was that while common mental disorder was associated with mortality from a range of causes, these associations were most heavily attenuated after taking into account prior physical illness. Some previous studies have reported similar results such that taking into account for physical comorbidity essentially abolishes the common mental disorder–mortality gradient,8 while in others, the magnitude of the association was more modestly affected.3 7

The observation that the only common mental disorder–mortality gradient to survive adjustment for somatic morbidity was in the analyses of CHD in women contrasts with results from one of the few other studies to have explored gender-specific associations where the reverse was apparent.16 Given that there is some evidence to suggest that women tend to have non-specific prodromal symptoms as opposed to chest pain, and these symptoms are therefore not identified as being cardiac in origin,17 may have led to an under-reporting of the presence of this condition in women in the present study. This may perhaps explain why controlling for baseline CHD did not eliminate the common mental disorder–mortality gradient. It is also equally plausible that women are, for reasons that are not clear, biologically more vulnerable to the impact on CHD mortality of anxiety and depression.

This study has a number of strengths, including its moderately large sample size, which, allied to extended follow-up, yields greater statistical power than previous studies. It is also unusually well characterised for baseline morbidities, although biological material was not collected. Furthermore, a comparison of the present sample with the 1981 UK Census concluded that the present study population provides a representation of the UK population, suggesting our results have a high degree of generalisability.10 Our study is not of course without its shortcomings. The 30-item GHQ is a screening rather than a diagnostic instrument and was designed to be followed by a standardised interview with a psychiatrist during which clinical psychiatric cases could be confirmed. Without such a second-stage procedure in the present study, it is possible that there is misclassification of cases. Additionally, the present analyses are based on a single-point assessment of common mental disorder and covariates. There would have been some variation in these levels during follow-up that were not taken into account herein. This is a shortcoming germane to most cohort studies.

In conclusion, physical illness largely explains the link between common mental disorder and mortality in the present cohort. This observation is compatible with either a confounding or mediation explanation; it was not possible to identify which in the present analyses.

What is already known on this subject

  • In people with common mental disorder, there appears to be an elevated rate of later mortality risk.

  • The role of baseline physical comorbidity, a potential explanatory factor in this relationship, is unclear.

What this study adds

  • Our findings suggest that, for the large part, taking into account physical comorbidity eliminated the common mental disorder–mortality gradient.

  • This observation is compatible with either a confounding or mediation explanation.


View Abstract


  • Funding The UK Health and Lifestyle Survey was funded by the Health Promotion Research Trust. GDB is a Wellcome Trust Career Development Fellow. MH is supported by the British Heart Foundation. The Medical Research Council (MRC) Social and Public Health Sciences Unit receives funding from the UK MRC and the Chief Scientist Office at the Scottish government Health Directorates. The Centre for Cognitive Ageing and Cognitive Epidemiology is supported by the Biotechnology and Biological Sciences Research Council, the Engineering and Physical Sciences Research Council, the Economic and Social Research Council, the Medical Research Council and the University of Edinburgh as part of the cross-council Lifelong Health and Well-being initiative.

  • Competing interests None.

  • Ethics approval Various LRECs UK-wide.

  • Provenance and peer review Not commissioned; externally peer reviewed.

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.