Background. Fatigue is a common, non-specific, subjective symptom associated with several medical and psychiatric illnesses. The purpose of this investigation was to explore further the epidemiology of unexplained fatigue in the general population and the relationship between fatigue and depression.

Methods. The design was a prospective population-based study. Subjects included community-dwelling adults who were participants of the Baltimore sample of the Epidemiologic Catchment Area Program in 1981 and who were reinterviewed 13 years later. Lay interviewers using the Diagnostic Interview Schedule interviewed subjects.

Results. Number of somatization symptoms and history of a dysphoric episode at baseline were the two strongest predictors of both new onset of fatigue as well as recurrent/chronic fatigue over the 13-year follow-up interval. In addition, individuals who reported a history of unexplained fatigue at baseline as well as during the follow-up, were at markedly increased risk for new onset major depression as compared to those who never reported such fatigue, (RR = 28·4, 95% CI) (11·7, 68·0). Similarly, respondents who developed new fatigue or had remitted fatigue after 1981 were also at increased risk for developing major depression.

Conclusions. Somatization was the strongest predictor of both new and chronic fatigue with unknown cause. In addition, fatigue was both predictive and a consequence of the depression syndrome.

Background. Concordance for major depressive disorder (MDD) between parents could happen for different reasons. Regardless of the origin and the frequency of the concordance, the effect on offspring of having two parents affected with MDD may be serious. The sex of the affected parent and offspring may also be a important risk factor for MDD in offspring.

Methods. We examined the increased risk of psychopathology among offspring of the four parental mating groups: both parents affected with MDD (N = 53); only mother affected (N = 31); only father affected (N = 65); and, neither parents affected (N = 33). Parents and offspring were assessed by direct interview, conducted blind and independently of each other.

Results. Among the four parental mating groups, offspring of both parents affected had the highest risk of MDD, anxiety disorder and alcohol dependence, and the earliest age of onset for MDD. There were two exceptions: the highest risk of conduct disorder and of drug dependence was in the groups where only the father was affected and where only the mother was affected, respectively. Mother's MDD was a stronger predictor of MDD in male compared to female offspring. Father's MDD was a stronger predictor of MDD in female compared to male offspring.

Conclusion. Having two parents with MDD increases the risk of psychiatric disorders in offspring. A clear dose–response relationship between the number of affected parents and psychiatric disorders in offspring was observed. The sex of the affected parent and of the offspring is important in determining the risk to offspring. For an examination of the risk to psychopathology in offspring, diagnosis status of both parents should be considered.

Background. Brain 5-hydroxytryptamine (5-HT) function is implicated in the pathophysiology of schizophrenia and the action of new generation antipsychotic drugs. By the method of acute tryptophan depletion (ATD) 5-HT can be selectively manipulated. The aim of this study was to examine the effects of ATD on symptoms, mood and cognition in schizophrenic patients.

Methods. Twenty-eight schizophrenic patients participated in a within subject, double-blind, placebo-controlled counterbalanced cross-over study. Patients with a concurrent DSM-IV axis I diagnosis were excluded. Symptoms, mood and cognitive function were evaluated following ATD or ingestion of a control drink.

Results. The depleting drink significantly reduced plasma total and free tryptophan. Tryptophan/LNAA ratios did not alter with the administration of the control drink, but differed significantly with ATD; however there was no significant change in tyrosine/LNAA ratio. ATD led to impairment in executive function that was dependent upon the order of administration. Tests of sustained attention, speed of processing, and everyday memory were not affected. No effects were observed on subjective mood ratings, movement disorders or PANSS scores.

Conclusions. Acute tryptophan depletion selectively alters cognition in schizophrenia, but has no effect on symptoms, mood ratings or movement disorders.

Background. Few studies have examined the extent to which populations of suicides and attempted suicides are similar, or different. This paper compares suicides and serious suicide attempts in terms of known risk factors for suicidal behaviour.

Methods. Using case–control methodology, risk factors for suicidal behaviour were examined in 202 individuals who died by suicide, 275 individuals who made medically serious suicide attempts and 984 randomly selected control subjects. Based on data from significant others, measures used spanned sociodemographic factors, childhood experiences, psychiatric morbidity and psychiatric history, exposure to recent stressful life events and social interaction.

Results. Multiple logistic regression identified the following risk factors that were common to suicide and serious suicide attempts: current mood disorder; previous suicide attempts; prior out-patient psychiatric treatment; admission to psychiatric hospital within the previous year; low income; a lack of formal educational qualifications; exposure to recent stressful interpersonal, legal and work-related life events. Suicides and suicide attempts were distinguished in the following ways: suicides were more likely to be male (OR = 1·9, 95% CI 1·1, 3·2); older (OR = 1·03, 95% CI 1·02, 1·04); and to have a current diagnosis of non-affective psychosis (OR = 8·5, 95% CI 2·0, 35·9). Suicide attempts were more likely than suicides to have a current diagnosis of anxiety disorder (OR = 3·5, 95% CI 1·6, 7·8) and to be socially isolated (OR = 2·0, 95% CI 1·2, 3·5). These findings were confirmed by discriminant function analysis, which identified two functions that described the three subject groups: the first function discriminated the two suicide groups from control subjects on a dimension corresponding to risk factors for suicide; the second function discriminated suicide from suicide attempt subjects on a series of factors including gender, non-affective psychosis and anxiety disorder.

Conclusions. Suicides and medically serious suicide attempts are two overlapping populations that share common psychiatric diagnostic and history features, but are distinguished by gender and patterning of psychiatric disorder.

Background. Treatment resistant depression (TRD) continues to present a formidable challenge to clinicians, accounts for over half of the annual costs associated with treatment for depression and causes great frustration to patients. Although there have been studies attempting to define TRD, little information is available as to the cause of TRD. One suggestion is that patients with TRD have a greater frequency of co-morbid psychiatric disorders, which explains their resistance to standard antidepressant treatments. The objective of this study was to compare the co-morbidity of Axis I disorders between a sample of TRD patients and a sample of non-TRD patients.

Methods. TRD and non-TRD patients, recruited from two separate antidepressant treatment studies, were assessed for Axis I co-morbidity using the SCID-P for the DSM-III-R. Patients for the two samples were then matched for baseline HAM-D-17 total score and gender.

Results. Results reveal that non-TRD patients had a higher rate of both lifetime and current generalized anxiety disorder co-morbidity than did the TRD patients. No other statistically significant differences in Axis I co-morbidity were found.

Conclusions. These findings do not support the idea that current or lifetime Axis I co-morbidity is more common in TRD than non-TRD patients. In fact, the only statistical difference showed non-TRD patients with higher co-morbidity rates.

Background. Previous neuroimaging studies of children with attention deficit hyperactivity disorder (ADHD) have demonstrated anatomic and functional abnormalities predominantly in frontal and striatal grey matter. Here we report the use of novel image analysis methods, which do not require prior selection of regions of interest, to characterize distributed morphological deficits of both grey and white matter associated with ADHD.

Methods. Eighteen children with a refined phenotype of ADHD, who also met ICD-10 criteria for hyperkinetic disorder (mean age 10·4 years), and 16 normal children (mean age 10·3 years) were compared using magnetic resonance imaging. The groups were matched for handedness, sex, height, weight and head circumference. Morphological differences between groups were estimated by fitting a linear model at each voxel in standard space, applying a threshold to the resulting voxel statistic maps to generate clusters of spatially contiguous suprathreshold voxels, and testing cluster ‘mass’, or the sum of suprathreshold voxel statistics in each 2D cluster, by repeated random resampling of the data.

Results. The hyperkinetic children had significant grey matter deficits in right superior frontal gyrus (Brodmann area (BA) 8/9), right posterior cingulate gyrus (BA 30) and the basal ganglia bilaterally (especially right globus pallidus and putamen). They also demonstrated significant central white matter deficits in the left hemisphere anterior to the pyramidal tracts and superior to the basal ganglia.

Conclusions. This pattern of spatially distributed grey matter deficit in the right hemisphere is compatible with the hypothesis that ADHD is associated with disruption of a large scale neurocognitive network for attention. The left hemispheric white matter deficits may be due to dysmyelination.

Background. There is now evidence that the availability of plasma tryptophan is decreased during pregnancy and the puerperium and also in patients with major depression and inflammation. The aims of the present study were to examine: (i) the effects of pregnancy and delivery on plasma tryptophan and the amino acids known to compete for the same cerebral uptake mechanism (CAAs), valine, leucine, tyrosine, phenylalanine and isoleucine; (ii) the relationships between the availability of plasma tryptophan and postpartum depression or anxiety; and (iii) the relationships between the availability of plasma tryptophan to the brain and inflammatory markers, such as serum interleukin-6 (IL-6), interleukin-1 receptor-antagonist (IL-1RA) and the leukaemia inhibitory factor receptor (LIF-R).

Methods. The above variables were measured in 13 healthy non-pregnant and in 98 pregnant women 3 to 6 days before delivery and 1 and 3 days after delivery. On each occasion the parturient women completed the state version of Spielberger State–Trait Anxiety Inventory (STAI) and the Zung Depression Rating Scale (ZDS).

Results. Plasma tryptophan and the tryptophan/CAA ratio were significantly lower at the end of term and after delivery than in the plasma of non-pregnant, healthy women. The tryptophan/CAA ratio was significantly lower in the early puerperium than at the end of term. There were no significant relationships between the availability of plasma tryptophan and either post-partum depression or changes in the STAI or ZDS scores in the early puerperium. The changes in the tryptophan/CAA ratio from the end of term to the early puerperium were significantly and inversely related to serum IL-6, IL-1RA and LIF-R.

Conclusions. The results show that the reduction in the availability of plasma tryptophan from the end of term to the early puerperium is related to immune activation; and that the lowered availability of plasma tryptophan is not related either to depressive or anxiety symptoms in the early puerperium or to post-partum depression ensuing some months later.

Background. Bulimic, impulsive and depressive syndromes have all been associated with abnormalities in brain serotonin (5-hydroxytryptamine; 5-HT) mechanisms.

Methods. We had 26 bulimic women and 22 normal-eater women report impulsive, affective, self-destructive and bulimic symptoms, and then provide serial blood samples for measurement of: [3H]-paroxetine binding in platelets; and, prolactin (PRL) responses following oral meta-chlorophenyl-piperazine (m-CPP).

Results. Bulimic status was associated with markedly reduced density of paroxetine-binding sites, modest blunting of m-CPP stimulated PRL response, and greater nausea following m-CPP. Biological variables did not co-vary with most psychopathological or eating-symptom indices. However, there were inverse associations (in bulimic women only) between scores indicating impulsivity (largely ‘unreflectiveness’) and density of platelet 5-HT uptake sites.

Conclusions. Our observations link bulimia nervosa to altered 5-HT functioning, and suggest that there may be a relatively symptom-specific association between impulsivity and reduced 5-HT reuptake.

Background. It has recently been shown that 3,4-methylenedioxymethamphetamine (MDMA) or ‘ecstasy’ causes long-lasting alterations to brain structure and function in animals, and there is mounting evidence that recreational users of the drug show impairments in some aspects of cognitive functioning including memory for verbal information. The present study investigates possible effects on other cognitive functions and explores the temporal course of development and resolution of these impairments by comparing novice, regular and abstaining users with a matched group of non-users.

Methods. Eighty participants categorized as non-users, novice users, regular users or currently abstinent users of MDMA were assessed on tests of verbal IQ, reversed digit span, immediate and delayed recall of a prose passage and of a complex geometric figure and verbal fluency.

Results. The four groups were well-matched for verbal IQ and on demographic variables. They differed in frequency of cannabis use over the last month, but this did not correlate with any cognitive test scores. All three groups of MDMA users showed significantly poorer verbal fluency and immediate and delayed prose recall than non-users. Days since last use and total lifetime consumption of MDMA made separate contributions to the variance in recall scores, accounting jointly for almost half of the variance in delayed recall. By contrast, the groups did not differ on either visual recall or reversed digit span.

Conclusions. The observed deficits provide further evidence of impairments of verbal but not visual memory in MDMA users, and indicate that the deficits are not attributable either to differences in general reasoning ability or to impairment of working memory. The data further suggest that the observed impairments may be attributable to a combination of reversible acute effects of MDMA resolving over a period of 2–3 weeks and more long-term changes associated with extent of lifetime consumption.

Background. Many, but not all, studies have reported that job strain is related to cardiovascular morbidity and mortality. To date, this relationship has not been tested on an English full population sample. This study examines whether the demand–control model of job strain contributes to our understanding of the determinants of coronary heart disease.

Methods. The analysis uses data from 4350 working men aged 20–64 in the 1993 Health Survey for England. Job demand and control characteristics were determined by questionnaire. Several health outcomes were examined: self-rated health; psychiatric health; angina and possible myocardial infarction, measured by the Rose questionnaire; doctor-diagnosed heart disease; any heart disease. The relationship between job strain and the health outcomes was determined by logistic regression analyses after controlling for known confounders.

Results. Those in high strain jobs consistently reported poorer health on all measures than men with lower strain. Similarly, men reporting low job strain were least likely to report poor health in 5/6 health outcomes. Those with intermediate levels of strain tended to have intermediate prevalence rates for poor health. The pattern of association between job strain and the CHD was independent of coronary risk factors.

Conclusions. The analyses broadly support Karasek's demand–control model of job strain. Health selection into low strain jobs may account for the lack of an association between job strain and doctor diagnosed heart disease while independent associations between job strain and all CHD measures considered together indicate that job strain may have aetiological significance for heart disease.

Background. Chronic fatigue syndrome (CFS) and depression share a number of common symptoms and the majority of CFS patients meet lifetime criteria for depression. While cognitive factors seem key to the maintenance of CFS and depression, little is known about how the cognitive characteristics differ in the two conditions.

Methods. Fifty-three CFS patients were compared with 20 depressed patients and 38 healthy controls on perceptions of their health, illness attributions, self-esteem, cognitive distortions of general and somatic events, symptoms of distress and coping. A 6 month follow-up was also conducted to determine the stability of these factors and to investigate whether CFS-related cognitions predict ongoing disability and fatigue in this disorder.

Results. Between-group analyses confirmed that the depressed group was distinguished by low self-esteem, the propensity to make cognitive distortions across all situations, and to attribute their illness to psychological factors. In contrast, the CFS patients were characterized by low ratings of their current health status, a strong illness identity, external attributions for their illness, and distortions in thinking that were specific to somatic experiences. They were also more likely than depressed patients to cope with their illness by limiting stress and activity levels. These CFS-related cognitions and behaviours were associated with disability and fatigue 6 months later.

Conclusions. CFS and depression can be distinguished by unique cognitive styles characteristic of each condition. The documented cognitive profile of the CFS patients provides support for the current cognitive behavioural models of the illness.

Background. We investigated whether there is regional variation in the incidence of schizophrenia and if so, whether it is caused by urban–rural differences, larger spatial clustering, or both. To control for the effect of migration, we examined regional variation in the incidence according to place of birth.

Methods. Finnish birth cohorts born from 1950 to 1969 were followed in the National Hospital Discharge Register from 1969 until 1991, and all cases of schizophrenia (ICD-8 or ICD-9 295) were identified (N = 14828). Forty-eight of the 559 municipalities were classified as urban and 25% of the Finnish population lived in these municipalities in 1960. For the analysis of spatial clustering, municipalities were grouped into 57 functional small-areas. We used Poisson regression model with the number of births of individuals who later developed schizophrenia as a response variable, and place of birth (urban/rural), birth cohort (1950–54, 1955–9, 1960–64, and 1965–9), functional small-area units, and sex as response variables.

Results. The incidence was slightly higher among the rural-born in the oldest birth cohort. In the other cohorts, it was higher among the urban-born, and the difference between urban and rural born increased in the youngest cohorts. Significant spatial clustering of schizophrenia was observed in eastern Finland.

Conclusions. Urban birth is a risk factor for schizophrenia in Finland in cohorts born since 1955. However, genuine spatial clustering of schizophrenia in eastern Finland was also observed, possibly caused by genetic isolation.

Background. The importance of stressful life events and long-term difficulties in the onset of episodes of unipolar depression is well established for young and middle-aged persons, but less so for older people.

Method. A prospective case–control study was nested in a large community survey of older people. We recruited 83 onset cases during a 2-year period starting 2½ years after the survey, via screening (N = 59) and GP monitoring (N = 24), and 83 controls, a random sample from the same survey population. We assessed depression with the PSE-10 and life stress exposure with the LEDS.

Results. Risk of onset was increased 22-fold by severe events and three-fold by ongoing difficulties of at least moderate severity. Severe events accounted for 21% of all episodes but ongoing difficulties for 45%. The association of onset with life stress, often health-related such as death, major disability and hospitalization of subject or someone close, was most pronounced in the cases identified by screening. While a clear risk threshold for events was found between threat 2 and 3 (on a scale of 1–4), the risk associated with difficulties increased more gradually with severity of difficulty. Compared with controls, severe events involved a larger risk for cases without a prior history of depression (OR = 39·48) than for cases with (OR = 8·86). The opposite was found for mild events (OR = 2·94 in recurrent episodes; OR = 1·09 in first episodes). The impact of ongoing difficulties was independent of severity of episode and history of depression.

Conclusion. Although the nature of life stress in later life, in particular health-related disability and loss of (close) social contacts, is rather different from that in younger persons, it is a potent risk factor for onset of a depressive episode in old age. Severe events show the largest relative risk, but ongoing difficulties account for most episodes. The association of severe events with onset tends to be stronger in first than in recurrent episodes. Mild events can trigger a recurrent episode but not a first one.

Background. Patients with chronic fatigue syndrome (CFS) often make somatic attributions for their illness which has been associated with poor outcome. A tendency to make somatic attributions in general may be a vulnerability factor for the development of CFS.

Methods. This cross-sectional study based on self-report questionnaire data aimed to investigate the type of attributions for symptoms made by patients with CFS and to compare this to attributions made by their partners. It was hypothesized that patients with CFS would make more somatic attributions for their own symptoms than control subjects and that partners of patients with CFS would make more somatic attributions for their ill relative's symptoms but would be similar to controls regarding their own symptoms. Fifty patients with CFS were compared to 50 controls from a fracture clinic in the same hospital and 46 relatives living with the patients with CFS. A modified Symptom Interpretation Questionnaire was used to assess causal attributions.

Results. CFS patients were more likely to make somatic attributions for their symptoms. The relatives of patients with CFS made significantly more somatic attributions for symptoms in their ill relative. However, they were like the fracture clinic controls in terms of making predominantly normalizing attributions for their own symptoms.

Conclusions. The data support modification of existing cognitive behavioural treatments for CFS to investigate whether addressing partners' attributions for patients' symptoms improves recovery in the patient. Furthermore, a tendency to make somatic attributions for symptoms may be a vulnerability factor for the development of CFS.

Background. Neuroimaging studies of tuberous sclerosis complex (TSC) have previously focused mainly on tubers or subependymal nodules. Subtle pathological changes in the structure of the brain have not been studied in detail. Computationally intensive techniques for reliable morphometry of brain structure are useful in disorders like TSC, where there is little prior data to guide selection of regions of interest.

Methods. Dual-echo, fast spin-echo MRI data were acquired from 10 TSC patients of normal intelligence and eight age-matched controls. Between-group differences in grey matter, white matter and cerebrospinal fluid were estimated at each intracerebral voxel after registration of these images in standard space; a permutation test based on spatial statistics was used for inference. CSF-attenuated FLAIR images were acquired for neuroradiological rating of tuber number.

Results. Significant deficits were found in patients, relative to comparison subjects, of grey matter volume bilaterally in the medial temporal lobes, posterior cingulate gyrus, thalamus and basal ganglia, and unilaterally in right fronto-parietal cortex (patients −20%). We also found significant and approximately symmetrical deficits of central white matter involving the longitudinal fasciculi and other major intrahemispheric tracts (patients −21%); and a bilateral cerebellar region of relative white matter excess (patients +28%). Within the patient group, grey matter volume in limbic and subcortical regions of deficit was negatively correlated with tuber count.

Conclusions. Neuropathological changes associated with TSC may be more extensive than hitherto suspected, involving radiologically normal parenchymal structures as well as tubers, although these two aspects of the disorder may be correlated.

Objective. This study investigates the relationship of specific anxiety and affective disorders to obsessive–compulsive disorder (OCD) in a blind, controlled family study.

Method. Eighty case and 73 control probands, as well as 343 case and 300 control first-degree relatives of these probands, participated in the study. Subjects were examined by psychologists or psychiatrists using the Schedule for Affective Disorder and Schizophrenia-Lifetime Anxiety version (SADS-LA). Two experienced psychiatrists independently reviewed all clinical materials, and final diagnoses were made according to DSM-IV criteria, by consensus procedure.

Results. Except for bipolar disorder, all anxiety and affective disorders investigated were more frequent in case than control probands. Substance dependence disorders were not more frequent. Generalized anxiety disorder (GAD), panic disorder, agoraphobia, separation anxiety disorder (SAD) and recurrent major depression were more common in case than control relatives. These disorders occurred more frequently if the relative was diagnosed with OCD. Only GAD and agoraphobia were more frequent in case relatives independent of OCD.

Conclusion. GAD and agoraphobia share a common familial aetiology with OCD. The other anxiety and affective disorders, when comorbid with OCD, may emerge as a consequence of the OCD or as a more complex syndrome.

Background. The present paper aimed to: (a) provide Australian estimates of the population-level association between psychotic ‘caseness’ and substance use; (b) examine liability to problematical substance use according to ‘caseness’ via the conditional prevalence (prevalence among users); and (c) examine associations between problematical substance use and the number of psychotic symptoms using ordinal logistic regression.

Method. Data were from the National Survey of Mental Health and Well-Being (NSMHWB), a stratified multi-stage probability sample of Australian adults, using a subset of persons under the age of 50 years (N = 6722). A screener assessed the presence of characteristic psychotic symptoms. Associations between ‘case’ status and DSM-IV alcohol, cannabis and other drug use disorders were examined. Ordinal logistic regressions predicting psychosis scores were carried out, including demographic, mental health and drug use variables.

Results. Ninety-nine persons (1·2 %) screened positively for psychosis. Regular tobacco, alcohol and cannabis use were much more common among persons screening positively, as were alcohol, cannabis and other drug use disorders. Among alcohol and cannabis users, psychosis ‘cases’ were much more likely to be dependent. Ordinal logistic regressions revealed that regular tobacco use, cannabis and alcohol dependence, and opiate abuse were predictors of psychosis scores.

Conclusions. The mental health risks of problematical substance use need to be disseminated to persons at risk of, or suffering from, psychotic illness, and to heavy substance users. Work is needed to develop effective treatment approaches for problematical substance use among persons with psychosis.

Background. We pursue an observation that age may influence the clinical features of melancholia and, in particular, psychomotor disturbance.

Methods. Two large clinical databases were amalgamated allowing the clinical features of 124 depressed subjects meeting DSM-III-R and clinical criteria for melancholia to be contrasted with 218 subjects diagnosed as having a non-melancholic depression by both criteria sets. Psychomotor disturbance was assessed by the CORE measure and by seven classical endogeneity symptoms of melancholia which, when summed, created a ENDOG score.

Results. There was no impact of age on ENDOG scores in either the melancholics or non-melancholics. In the melancholics, increasing age was associated with increasing CORE scores and with agitation scale scores in particular. In a set of discriminant function analyses seeking to identify the comparative utility of a set of predictors of melancholic (versus non-melancholic) groups, age was significant, and while CORE and ENDOG scores were individual predictors, their combined entry established that the CORE score alone made the ENDOG score redundant, and that the addition of age then made little impact.

Conclusions. Melancholia appears to have a later age of onset than non-melancholic depression, while its phenotypic expression appears to change with age, with psychomotor disturbance being more distinct in older subjects. Such an effect may have a number of clinical implications, including possible differential effects of varying antidepressant treatments.

Background. Marital status is a strong correlate of psychiatric morbidity in adulthood, but debate continues on how far this association reflects causal influences or selection effects based on prior psychological characteristics.

Method. Prospective data from the National Child Development Study were used to examine effects of adolescent emotional and behavioural problems on transitions into and out of first partnerships, and their implications for psychiatric morbidity at age 33.

Results. Emotional and behavioural problems in adolescence showed systematic links with early partnership transitions (age at partnership formation, type of first partnership, and risks of first partnership breakdown). More detailed tests suggested that these effects only accounted for a modest proportion of the associations between partnership status and psychiatric morbidity at age 33.

Conclusions. In a non-referred community sample selection effects associated with adolescent emotional and behavioural problems appear to play only a modest role in links between partnership status and adult mental health.

Background. It is not known why the most common mental disorders (CMD), anxiety and depression, are more prevalent among women. This gradient has not been explained by differences in the number or type of social roles occupied by men and women. Given the dearth of longitudinal studies, these negative findings could reflect reverse causality, if men with CMD relinquish social roles more readily than women.

Methods. Cohort study using data from the first seven (annual) waves of the British Household Panel Survey. The prevalence of CMD was assessed using the General Health Questionnaire (GHQ), 12 months after ascertaining occupancy of five social roles. Of 12379 participants aged 16–70, 9947 completed the GHQ on at least two consecutive occasions, resulting in 44139 paired observations. Random effects models adjusted for the correlation of repeated measures and for baseline GHQ score.

Results. The odds ratio for the gender difference in the future prevalence of CMD (adjusted for baseline GHQ score) was 1·92 (95% CI 1·75–2·10). Neither the number or type of social roles occupied, nor socio-economic status explained the gender difference in these conditions (adjusted OR 1·82, 95% CI 1·66–1·99). While CMD at baseline was associated with a subsequent reduction in social role occupancy, this did not vary between men and women.

Conclusions. Gender differences in CMD were not explained by differences in the number or type of social roles occupied by men and women, or by reverse causality. Future studies should consider characteristics of social roles, such as demand, control and reward.