Transcultural observations offer an opportunity to study attitude to mental illness in different societies and family structures. The disparity between industrialised and lower-income societies reflects greater tolerance due to the ability of extended families to compensate for the patient's limitations.

This paper is something of a patchwork, incorporating many issues that have intrigued me during 34 years of research. I have included the importance of maintaining a solid base in clinical work, alongside research activities, and being alert to the possibility of a somatic condition contributing to psychiatric symptoms. I stress the value of careful observation of patients, their response to treatments and reasons for dropping out. In addition, I have included 14 more lessons, learned from my experience of research, which I hope will be of use to those readers who aspire to become researchers.

Long-stay patients who were discharged from Friern and Claybury hospitals during the 6-year period between September 1985 and August 1991 were followed up for at least 1 year and the details of all readmissions were recorded. Of the 615 hospital leavers, some 178 (29%) were readmitted at some time during the follow-up period. A proportional hazards regression analysis revealed a number of independent baseline factors which affect the risk of readmission at any time during the follow-up period. Younger patients were more at risk of readmission than were older patients and ‘new’ long-stay patients had a greater risk than baseline patients. Risk was also greater in patients who had a larger number of pre-discharge admissions,a higher abnormal behaviour score (PSE BSO) or a diagnosis of manic psychosis. An ordinal logistic regression analysis revealed that these same five factors tended to increase the actual number of readmission episodes. The proportion of post-discharge time spent in hospital among the readmitted patients was found to be greater for males, new long-stay patients and for patients with a higher number of positive social behaviour symptoms (SBSPOS score).

A carefully controlled comparison of communication defects and deviances in parents of schizophrenics and neurotics was undertaken using the 41-category Rorschach manual of Singer and Wynne. Subjects were tested and scored blindly and scoring was shown to be reliable. Although a statistically significant difference was found between the two groups, the marked overlap of the distribution of scores found in this study is not compatible with theories advanced by other authors concerning communication abnormality in parents of schizophrenics. Evidence is presented which provides a more parsimonious explanation of the difference between groups and leads to a hypothesis which would explain the failure of this study to confirm the striking differences between groups reported by previous workers.

A group of 72 long-term patients, who were retained in Friern Hospital until the final stage of its closure programme, underwent a comprehensive psychiatric and social assessment shortly before moving to other hospitals. These patients were then compared with the rest of the long-stay patients who were considered suitable for community placement.

The residual group consisted of younger patients with shorter duration of stay, compared with the rest of the hospital population. Although not excessively disabled functionally or physically, the residual patients were slightly more disturbed in mental state in comparison with the rest, with notably higher levels of subjective anxiety.

A range of 13 problem behaviours were designated by staff as a direct impediment to placing a patient in a community setting. Most common problem areas were aggressiveness, non-compliance with medication and inappropriate sexual behaviour. Fourteen per cent of the residual group had none of these special problems. Some other patients have remained in hospital solely because they refused to leave.

It is argued that a distinction should be made between the residual patients who were actually transferred to other hospitals, and a core group of extremely disturbed patients for whom further hospitalization is likely to be the only alternative.

Data from two studies, one naturalistic and the other a controlled trial, were analysed to clarify the relationships between independent life events, Expressed Emotion of a key relative, maintenance neuroleptics and the relapse of schizophrenia. It was found that patients in the community who are unprotected by medication are vulnerable either to acute stress in the form of life events or to chronic stress in the form of living with a high Expressed Emotion relative. Patients on regular medication are protected against one or other stress, but are very likely to relapse if the two forms of stress occur together. A model of schizophrenic susceptibility to environmental stress is constructed to incorporate these observations.