Prospective naturalistic research

Most of the theoretical underpinning of the contextual life-events research described above was ascertained from two separate large, prospective, longitudinal studies of non-clinical samples of working-class women living in London (Reference Brown and HarrisBrown & Harris, 1978; Reference Brown, Adler and BifulcoBrown et al, 1988). To be relevant to the work of mental health professionals, the findings of these studies must be confirmed in clinical samples of both genders studied in general practice, out-patient and in-patient psychiatric settings.

In a 6-month outcome study at a single Manchester general practice, improvement in depressive disorder was associated with relief of chronic social difficulties, especially in close interpersonal relationships, work and housing (Reference Ronalds, Creed and StoneRonalds et al, 1997). These positive life events either neutralised a previous severe life event or reduced the threat to emotional health from a chronic life difficulty. In a large Dutch study of 170 primary care patients with depressive and/or anxiety disorder followed for 3Þ years, positive life changes (fresh-start or anchoring life events, goal attainment, difficulty reduction) facilitated recovery (Reference Leenstra, Ormel and GielLeenstra et al, 1995). However, positive life events were neither necessary nor sufficient for recovery.

In female psychiatric out-patients, childhood adversity and current interpersonal difficulties predicted chronicity of depressive disorder, while good-quality emotional social support reduced the risk of chronic depression (Reference Brown, Harris and HepworthBrown et al, 1994). Similar results were found in an American out-patient sample of women and men (Reference Zlotnick, Shea and PilkonisZlotnick et al, 1996). In male and female psychiatric out-patients, life events and life difficulties in the previous 6 months were independently associated with failure to recover from first-onset and second episodes of non-psychotic depressive disorder (Reference Brugha, Bebbington and StretchBrugha et al, 1997). Better social support predicted recovery from all but first episodes of depressive disorder.

Studies of in-patients treated for depressive episodes (e.g. Paykel et al, 1996) and of out-patients with melancholic or psychotic subtypes of depressive episode (Reference Harris and BrownHarris & Brown, 1996; Reference Brugha, Bebbington and StretchBrugha et al, 1997) have generally failed to demonstrate that outcome is associated with the threat from current life events or chronic life difficulties, the creation of fresh-start events or improved social support. Post et al (1986) noted that life events often precipitate early episodes of depression, but as the number of episodes increases Òthe illness appears to evolve its own rhythmicity and spontaneity, independent of life events”. This behavioural sensitisation may explain why social adversity has little effect on outcome in patients with depression who have had out-patient (Reference Brugha, Bebbington and StretchBrugha et al, 1997) or in-patient (Reference Paykel, Cooper and RamanaPaykel et al, 1996) treatment for more than two previous episodes. On the other hand, Johnson et al (1998) report that positive life events that are internally, consistently and globally attributed by patients to their own efforts reduced hopelessness and depression in in-patients with recurrent depressive disorder.

Randomised controlled treatment trials

These explore whether interventions using contextual rating of social adversity can promote recovery from depressive episodes. Harris et al (1999a, b) showed that befriending by female volunteers to improve emotional support promoted remission compared to no treatment in a non-clinical sample of women with depressive episodes lasting more than 12 months. The offer of befriending was as important as the reality of befriending in promoting remission. These subjects had not sought help for their depression, and befriending might not be as effective in a clinical sample of patients who had already decided to do so. Fresh-start life events, the absence of additional negative life events, the ability to make supportive interpersonal life events (standard attachment) and befriending all independently promoted remission. Standard attachment enabled fresh-start events to happen.

Corney (1987) demonstrated that casework, through enhanced practical and emotional support, relieved the symptoms of depressive episodes of women with marital difficulties compared to treatment as usual by the general practitioner (GP). Marital therapy aimed at enhancing the development of a close confiding relationship significantly reduced the symptoms of depressive disorder in women attending psychiatric out-patient clinics (e.g. Waring, 1994). An intervention that successfully increased the finding of full-time work for the long-term unemployed also significantly reduced the prevalence of psychiatric disorder (Reference Proudfoot, Guest and CarsonProudfoot et al, 1997). However, in some of these interventions, it is not clear how much of the benefit of the intervention was due to improved social support, fresh-start events or reduction of threat from life difficulties, and how much was due to therapeutic ingredients uncharacteristic of the contextual approach, for example, the promotion of intimacy in marital therapy.

Psychiatric comorbidity and special groups

Depressive disorder is commonly associated with other psychiatric comorbidity. There is some evidence of specificity for the contextual meaning of life events and life difficulties in depressive and anxiety episodes. Hence, depressive episodes are preceded by loss events, and recovery or improvement in such disorders is associated with fresh-start events involving increased hope arising from a lessening of a difficulty or deprivation. Anxiety episodes are preceded by events signalling danger, while recovery or improvement in anxiety disorders is associated with anchoring events that involve increased security. Comorbid anxiety and depressive episodes are preceded by both loss and danger events, while recovery is associated with both fresh-start and anchoring events occurring together (e.g. Brown et al, 1992). Such specificity between life event and diagnosis of anxiety or depressive episodes has not been found in every study (e.g. Leenstra et al, 1995).

Childhood adversity, life events and life difficulties are widely believed to be associated with increased vulnerability to and exacerbation of substance misuse, with or without comorbid depressive disorder. Vulnerability factors such as parental separation and childhood sexual abuse lead to the risk of heavy drinking, while acute life events and chronic life difficulties are related to the onset or exacerbation of heavy drinking in large community samples. Depressive symptoms may have an independent additive effect on the risk of heavy drinking, which may be seen as a form of poor coping influenced by gender and culture (Reference NeffNeff, 1993). A similar picture has been demonstrated with respect to relapse into opiate misuse from abstinence in patients with a life-time history of misuse or dependence (Reference Kosten, Rounsaville and KleberKosten et al, 1986).

An association between life events and the onset of bulimia nervosa has been found in a retrospective community-based sample (Reference Welch, Doll and FairburnWelch et al, 1997). Clinical experience suggests that life events can precipitate the onset of comorbid depressive disorders in patients suffering from bulimia nervosa. Psychological autopsy studies suggest an association between both acute and chronic interpersonal problems and completed young suicide (e.g. Appleby et al, 1999). Interpersonal problems, social withdrawal, rootlessness, acute severe mental disorder and long-term behavioural problems such as substance misuse, repeated deliberate self-harm and personality disorder all appear to add to the risk of suicide (Reference Appleby, Cooper and AmosAppleby et al, 1999).

Finally, life events and life difficulties may play an even greater role in the onset, maintenance, recovery from and adverse consequences of depressive disorder in patients aged over 65 (Reference Harris and BrownHarris & Brown, 1996) and in non-Western settings where there is greater exposure to social adversity (Reference Broadhead and AbasBroadhead & Abas, 1998).

Conclusion

There is sufficient empirical evidence to suggest that the Brown model of social adversity outlined in Fig. 1 identifies psychosocial factors that maintain chronicity or promote recovery from depressive episodes in primary care and psychiatric out-patients with non-psychotic, non-melancholic, first or second depressive episodes. However, there is insufficient evidence that naturally occurring relieving factors or interventions to promote relieving factors alone improve outcome in these patients, or that they are necessary for remission. There is little evidence to support the use of this model in the majority of melancholic, psychotic or frequently recurring depressive episodes.

Sources of bias

Two types of bias commonly occur when making contextual ratings of life events and difficulties: search for meaning and recall bias (Reference KesslerKessler, 1997).

If patients have had a bad experience, such as becoming depressed, they may look for reasons and decide that particular social adversity was the cause of their depression. Clinicians should ask themselves why this social adversity was likely to contribute to the patient's depression. They should be non-committal about an association in the patient's mind if either the contextual meaning of the link or its timing (i.e. social adversity preceding a change in symptoms) is unclear.

Recall bias occurs when patients vary in their accuracy or willingness to recall past or current experiences. Induction of depressed mood can lead to a significant increase in reports of past adverse life events and poor-quality social support (Reference KesslerKessler, 1997). The willingness or ability to report life events as a stable personality trait appears to be inherited independently of depressive disorder (Reference Harris and BrownHarris & Brown, 1996; Reference KesslerKessler, 1997).

There are clear difficulties with relying on the contextual method of assessment of social adversity using only the patient's account. The patient's past experience, for example past abuse, or the severity of their current mood disorder may inhibit the person's ability to trust others and to communicate details of their past or current social adversity. Traditional methods of consulting an informant, exploring archival material, such as patient notes, and careful evaluation of the mental state remain helpful additions to the contextual rating of social adversity elicited directly from the patient. Unfortunately, these methods have been relatively underused in research to verify contextual ratings of social adversity and their temporal link to depressive episodes (Reference KesslerKessler, 1997).

Non-clinical samples

The contextual rating method was developed using non-clinical samples with depressive episodes. However, clinical samples may differ from non-clinical ones in several important respects. Non-clinical subjects may think that their depression is not severe enough to warrant seeking help from health professionals or that they can cope without professional help. Therefore, the severity and nature of people's depression and their resources for coping may differ between clinical and non-clinical samples. Such differences may also be reflected in different psychosocial aetiologies of depressive episodes in clinical and non-clinical subjects.

Relationhip between life events and mood

Brown's research has explored associations between life events that are independent of the subject's mood and the onset or chronicity of depressive episodes. However, life events that might be dependent on mood occur frequently before the onset of and during depressive episodes. These may play a key role in precipitating or maintaining depressive episodes, but research has not clarified their importance.

A past history of depressive episodes or current partial remission may complicate an assessment of the independence of a life event in relation to a new depressive episode. People may be less supportive of individuals showing depressed behaviour than they used to be. As a result, the depressed patient is exposed to more interpersonal loss events (Reference KesslerKessler, 1997) and does not benefit from the stress-buffering benefits of social support (Reference Brugha, Bebbington and StretchBrugha et al, 1997). Consequently, patients with a history of depressive disorder are put at increased risk of further depressive relapse, a worsening of depressive symptoms, or increased hopelessness and suicide risk.

Determining causality

Causality may be difficult to establish. Most studies of contextual rating of social adversity have been observational aggregate life-event studies (multiple life events). They interpret temporal associations between life events and depression as indicating that the former cause the latter. This interpretation assumes that life events occur randomly and independently of each other and of other causes of depression, such as genetic susceptibility, alcohol use and social support, but this assumption is unsafe (Reference KesslerKessler, 1997). Other studies assume that these confounding factors have effects additive to those of life events, but again research evidence points to more complex relationships between life events and these other causes of depression (Reference KesslerKessler, 1997). The complexity of these relationships demands experimental studies such as randomised controlled trials (Reference Harris, Brown and RobinsonHarris et al, 1999a ,Reference Harris, Brown and Robinson b ) or prospective study of the adverse effects of a single stressor (e.g. job loss, widowhood; Kessler, 1997) for their clarification. Such studies are rare.

Confounding of stress ratings with outcomes

Information about the impact of a life event on a subject's depressive episode may be used to make ratings of life-event severity. Greater weight might be placed on aspects of the environment that come to light only retrospectively, as the full impact of the depressive episode on the person's life is revealed (Reference KesslerKessler, 1997). Sometimes it is not clear whether the variability between individuals in the risk of developing depressive episodes as a reaction to life events is due to differences in their emotional reactivity to the event, qualitative or quantitative differences in their experience of the stress of the event, or their ability to cope with the event. Given that all of these factors are heavily intertwined, the contextual rating method inevitably has to make some assumptions about the meaning of the life event or difficulty for a given individual.

Nature and role of social support

The precise nature and role of effective social support remain unclear and probably depend greatly on the social context. One of the few prospective longitudinal studies to examine different types of support found that emotional support, rather than confiding alone, predicted good mental health, whereas negative aspects of close interpersonal relationships had an adverse effect on mental health (Reference Stansfeld, Fuhrer and ShipleyStansfeld et al, 1998). Effective social support is usually thought to have a role in protecting the patient from either the onset or the full consequences of depressive disorder in the face of adverse life events or life difficulty (stress buffering). Social support might be protective only in subjects who tend to have long-term feelings of lack of control over their lives pre-dating their depressive disorder (Reference Dalgard, Bjork and TambsDalgard et al, 1995). The most consistent findings linking social support to the onset or outcome of depressive episodes relate to the patient's perception that social support is present and effective. Effective social support may be obtained from one or two intense relationships or it may be the sum of many relationships. Hence, Brugha et al (1997) reported prospectively that the number of social supports predicted recovery from depressive disorder in women in the community, but that the poor quality of a few key relationships was associated with the onset of depressive disorder in women postnatally (Reference Brugha, Sharp and CooperBrugha et al, 1998). Perceptions of effective social support may differ between genders, age groups and cultures. The adverse effects of the lack of social support can also be topic-related. For example, women who developed postnatal depressive disorder regarded their husbands as generally supportive, but not supportive concerning the birth of their last baby (Reference Chapman, Hobfoll and RitterChapman et al, 1997). In some studies, ineffective or abusive relationships with others purporting to be supportive are important precipitants of depressive disorder.

Is contextual rating sufficient?

Some patients with depressive episodes, especially those with an idiosyncratic view of themselves in relation to the world, poor coping strategies or an unstable attachment style, merit more detailed assessment than the contextual rating method provides. Such assessment might include intrapsychic processes, schematic and attributional thinking, cognitive flexibility, problem-solving skills, interpersonal communication skills, personality and behaviours such as avoidance. Such assessments are particularly merited if individual or group psychotherapies are being considered.