In a survey of a random sample of the general population recent life events, collected and rated for long-term contextual threat according to the methods of Brown & Harris (1978), were also recorded where possible on an inventory of life event categories (Tennant & Andrews, 1977). Of the 82·5% of all events collected which were covered by the inventory, 12 of the 67 event categories accounted for 77% of life events with an aetiologicaly significant rating of marked or moderate long-term threat. Where practical and economic constraints oblige research workers to choose the inventory method, a brief list of event categories, such as the List of Threatening Experiences, is recommended in preference to much longer lists.

The authors review the evidence that parental death in childhood predisposes to depressive disorders in later life. The findings in general are quite inconsistent; this is due in part to the methodological limitations of most studies, principally that of inadequate control of potentially confounding variables. Where experimental and control samples were most rigorously matched, no association was found between childhood parental bereavement and depression in later life. Parental death in childhood appears to have little effect on adult depressive morbidity.

Data from a general population survey of psychiatric disorder in Camberwell were used to calculate the risk of a CATEGO-defined depressive episode before the age of 65, using a modification of Strömgren's method.

Current depression was defined as cases within the relevant categories of the CATEGO program and at threshold level or above on the Index of Definition (Wing et al. 1978). A past history of depression was elicited using key symptoms such as persistent tearfulness and depressed mood, already enquired after in the course of the PSE, to identify potential episodes, followed by questions to determine accessory symptoms, duration, and degree of social impairment. Clinical judgement was then used to decide whether the disturbance constituted a significant depressive episode.

Risk under one set of assumptions was 46 % for men and 72 % for women. Using another method based on (untenably) conservative assumptions, it became 16 % and 30 % respectively. The status and implications of these high values are discussed, particularly for genetic studies.

The reliability of Brown's contextual measure of threat is assessed and found to be highly satisfactory. Previously inexperienced raters achieve highly reliable ratings after only brief training. The implications for the concept of the threat of life events are discussed.

The effect of childhood experiences on adult psychiatric morbidity was examined in a community psychiatric survey. The Present State Examination was used to assess psychiatric morbidity. Childhood experiences assessed included childhood demographic factors and ‘loss and deprivation’ variables. The latter group comprised maternal and paternal deaths and separations and other disruptions in parental care. ‘Loss and deprivation’ in combination accounted for between 4·5 and 5·5% of the variance in adult psychiatric morbidity

This paper compares the coverage of potential life events by the inventory developed by Tennant & Andrews (1976) and the post hoc designation originated by Brown (1974). It was found that the rubric of the inventory failed to describe 5% of ‘marked’ events and 21% of ‘moderate’ events. The relative risk of minor psychiatric disorder in response to events was 3·1 for those defined by the Brown technique and 1·4 for those using the inventory. The results suggests that the post hoc designation is superior. This superiority must be weighed against its undoubted costliness.

A two-stage psychiatric survey of a random sample of adults aged 18–64 from Camberwell is described. Agency interviewers carried out the first stage (N = 800), using the shorter form of the Present State Examination (PSE). MRC interviewers, using the full PSE, saw a stratified sample of these (N = 310) in the second stage. A second interview was sought with all those of Index of Definition (ID) level 5 and above at the first interview (‘cases’) and with a random sample of those below that level. 20·9% refused or were never available for the first interview. Of the 800 subjects successfully interviewed, 10% refused a further interview and 12·4% of those finally selected for this interview were either unavailable or changed their minds. The MRC data, weighted to represent the whole sample, are used in our analyses. The prevalence of psychiatric disorder as defined in our study was calculated at 6·1% for men and 14·9% for women. Women shared a higher prevalence of disorder in the age-groups 25–34 and 45–54, but in men there was no significant association with age.

In contrast to the findings of Brown & Harris (1978), social class did not have a strong association with disorder. Single men had much higher rates than married men, while the reverse was true in women. In both sexes employment was associated with lower rates of disorder. An attempt to explain the high prevalence in women in terms of their role in marriage and child-care was only partly successful.

Assessment of the relationship of life events to minor and major affective illness is beset by several methodological problems. The reliability and validity of measures is not entirely satisfactory. When the stress concept used is that of ‘threat’ or ‘emotional distress’ it becomes of crucial importance to define clearly the criterion variable: in particular, it is necessary to distinguish normal distress responses from depressive illness. In those studies using general practice or hospital patients it is important to determine whether life events provoke depressive illness or some illness behaviour such as service utilization. Because of these problems it is still not certain that life events cause depressive illness.

Historically there has long been an interest in establishing whether children's reading levels have declined or improved in comparison with previous years. This would seem to be a valid research question and a potentially useful area of investigation. However, in reality it is a minefield. There are several reasons for the explosive nature of this topic.

First, reading holds a central position in children's education. Producing young people who are literate is the sine qua non of an adequate education system—the necessary if not sufficient demonstration that schools are doing their job. Thus, the measure of reading standards is seen as a proxy measure for educational standards more generally.

Second, the rise and fall of reading standards provides a focus for fierce debates about teaching methods, which typically crystallise around traditional versus progressive methods. Currently, the traditionalists champion methods of teaching reading that involve explicit phonics tuition. The progressive movement, sometimes referred to as “the real books approach”, emphasises the importance of purposeful reading for meaning and enjoyment. In practice it is established that eclecticism rules.

Third, there are differences of opinion concerning the most appropriate ways to assess reading levels. In order to plot reading standards over time, both valid and reliable methods of assessment are required. Preferably, too, the same test should be repeated, although parallel versions of valid tests can also be used. In practice this has frequently meant that tests involve either decontextualised word reading or sentence completion. There is agreement that such forms of assessment do not measure all the aspects of reading or literacy. There are those who argue that assessment that fails to take into account a wider range of higher-order skills, such as skimming, scanning, interpretation, and comprehension monitoring, and that uses only a narrow range of decontextualised text, will tell us little that is valuable about the level of children's reading. From this position it is difficult to measure reading standards validly over time. It is also argued that assessment should measure the areas of literacy covered by the curriculum and that as this changes, tests must also change, making comparability over time problematic.

Despite these debates, children's reading levels have been measured over time and even across countries. The question is how valid such exercises are and what conclusions we can draw from the results.

The efficacy of implementing a clinically feasible psychosocial intervention which addresses the needs of carers of the long-term mentally ill is reported. All the relatives of patients in continuous high contact with one clinical team in a local day-care facility were offered the intervention. An interactive education session at home was followed by a monthly relatives group which aimed to reduce components of expressed emotion and to alleviate burden. Patients and relatives were assessed. The controls were the patients in contact with the other teams in the day-care facility, and their relatives. The intervention was effective at reducing EE and improving family relationships. Offering this kind of support to people who are providing long-term care for the severely mentally ill can contribute significantly to the quality of life of both supporters and patients.

A psychosocial intervention is described geared to the needs of carers of the long-term mentally ill, which is feasible for a busy clinical team to implement: relatives were not selected for the group by patient diagnosis or motivation and little extra staff input was required. An interactive education session at home was followed by a monthly relatives group which aimed to reduce components of expressed emotion (EE) and to alleviate burden. The group facilitators adopted a directive but non-judgemental style, and constructive coping efforts were encouraged. The intervention was effective at reducing EE and improving family relationships. The study offers a realistic model of how to offer support to people providing long-term care for the severely mentally ill.

The relation of childhood separation experiences to adult depression, anxiety and ‘general psychological morbidity’ was assessed in a random population survey. Four causes of separation were examined in each of three age groups: childhood illness, parental illness, parental marital discord and wartime evacuation. Childhood separations occurring up to five years of age bore no relation to depression, anxiety nor to incidence of morbidity generally. From five to ten years of age, separations caused by parental illness and marital discord were related to morbidity, and in the latter instance were more likely to cause depression than anxiety. From 11 to 15 years, only separations due to parental illness were related to morbidity generally.

A longitudinal study of neurotic disorder in the community showed that half the cases identified at first interview had remitted one month later. Remission was significantly related to four variables: recency of onset and of peak of the disorder, the occurrence of recent threatening life events and the occurrence of subsequent ‘neutralizing’ life events. A neutralizing event was defined a priori as one which neutralized the impact of an earlier threatening life event or difficulty. One third of all remissions were caused by such an event. Remission of disorder was not significantly related to demographic variables, symptom severity, syndrome type, medical consultation or psychotropic drug prescription. The implications for neurotic disorder in the community are discussed, in particular its relation to life events and the favourable outcome in the absence of treatment.