This paper sets out the rationale for the PRiSM Psychosis Study, and the research design used. Nine accompanying papers present the main results. The questions addressed by the PRiSM Psychosis Study are: can the gains of experimental studies which have demonstrated benefits arising from treatment by community mental health teams be translated to routine settings? If so, are the benefits diluted in ordinary clinical practice? What are the costs?

A prospective non-randomised controlled trial of two types of community mental health service, in two phases: case identification followed by patient interviews. For the case identification the research team conducted the complete ascertainment of all prevalent cases of psychosis in the two study catchment areas in the index year (1991–1992). From all 514 patients with psychotic disorders thus identified, 302 were randomly allocated for interview, along with a key informant clinician and a carer. Interviews were undertaken at two time points, two years apart.

This paper presents the socio-demographic, clinical and ethnic characteristics of the patients.

The people with psychosis interviewed for the PRiSM Psychosis Study are representative of the whole epidemiologically based patient population identified.

Service evaluation requires a detailed understanding of the services studied.

Community mental health services evaluated in the PRiSM Psychosis Study in south London are described. The intensive sector and standard sector services are contrasted.

The intensive sector had two teams with extended opening hours: a psychiatric acute care and emergency (PACE) team, and a psychiatric assertive continuing care (PACT) team focusing on care for people with chronic illness. In the standard sector there was a generic community team providing office-hour assessments, case management of the severely mentally ill and close liaison with in-patient services. The team made use of the local psychiatric emergency clinic and of other local resources. The intensive sector was characterised by: more admissions to fewer beds, more non-hospital residential places, extended hours, on-call rota, wider range of interventions, more medical and nursing staff, a lower nursing grade mix and higher staff turnover. The standard sector had a less highly resourced generic community psychiatric service.

Change in services has been more marked in the intensive sector.

Community care has been criticised as a hazardous policy associated with high rates of adverse events. There is little research evidence as to the truth of this claim.

Best available evidence from public records, interviews, case notes, key workers and general practitioners was assembled to establish: (a) which of the 514 subjects initially identified as having psychotic illnesses had died during an average follow-up of 4.9 years; (b) care currently received by all 286 subjects originally selected for interview; and (c) rates of major adverse events and of admission for these 286 individuals.

Twenty-eight natural and 11 unnatural deaths had occurred. Among subjects still living at the end of the follow-up, 84% were in contact with specialist mental health services and 11% only with primary care services. Rates of serious violence, imprisonment and homelessness were relatively low. Forty-one per cent had been admitted at least once during a mean follow-up of 3.2 years and 20% at least once under the Mental Health Act. After adjustment, there were no significant differences between standard and intensive care sectors.

Rates of adverse events and slipping through the net’ are relatively low among individuals receiving community-based services, whether intensive or standard care.

Community care for people with psychosis can be provided via several different service models. The study compared two models: standard community treatment with high patient:staff ratios, and intensive community treatment in which there was more of an emphasis on community involvement and lower patient:staff ratios.

All people fulfilling diagnostic criteria for psychosis and living in two defined geographical areas in south London were identified. These people were provided with either an intensive or standard service depending on their address. Their social behaviour and symptoms were measured at the beginning of the study and after two years.

There is little evidence of any effects of the two service models on levels of symptoms. However, the standard service did seem to be advantageous in reducing social disability for those with medium or low levels of disability.

The effect of intensive community services was the opposite to that predicted. Despite improvements in outcomes (e.g. hospital admission), there were no advantages for this type of service in the measures of disability or symptoms. In fact, the standard service was effective in reducing disability, whereas in the intensive service there was no change.

The costs and the effectiveness of mental health services need to be evaluated if provision is to be efficient. Service use and costs are described for two geographical areas in south London.

Service use was measured comprehensively for clients in both sectors for two six-month time periods using the Client Service Receipt Interview This information was combined with unit costs to calculate service costs. The ‘hidden’ costs of informal care and unsupported accommodation were also calculated.

At baseline significantly more intensive sector clients had in-patient stays but by the follow-up this difference had disappeared. There was significantly more use of supported accommodation in the intensive sector during both time periods. Baseline and follow-up total service costs were significantly higher for the intensive sector. Costs were spread disproportionately and a small number of services accounted for a large proportion of cost.

While the cost at Time 2 was significantly greater in the intensive sector, this was largely due to the high use of supported accommodation. There was some convergence in cost between the sectors over time.

We examined: care-giving activities in a population-based sample of carers of sufferers from psychotic disorders; putative determinants of care-giving; and changes in care-giving in an intensive community psychiatric service.

In the PRiSM interview sample 170 people had a carer; 124 were interviewed. Care-giving activities, dissatisfaction with these, and carers General Health Questionnaire (GHQ) scores were measured. Patient and illness characteristics were examined as predictors of the carer measures. The carer measures were reassessed for 62 carers after two years in two service sectors, one with an intensive community service, the other with a standard service.

Overall, 36% of carers were engaged in no, or only occasional care-giving activities. Fifty per cent expressed no dissatisfaction with their care-giving role. Patient and illness characteristics predicted care-giving poorly. Carers in the intensive community treatment sector did not experience significantly different care-giving demands or distress than those in the standard sector.

A significant proportion of carers of sufferers from psychosis do not engage in common care-giving activities, and are not dissatisfied with their role. An intensive community service did not affect the impact of the illness on carers.

Social networks are important for people with severe mental illness, and services need to assess whether they succeed in improving social contacts.

In a prospective controlled study, social network data were obtained in an epidemiologically representative sample of people with psychotic disorders both before (Time I) and two years after (Time 2) the introduction of two sectorised community mental health services in south London (one intensive service with two specialist teams, one standard service with a generic team).

There were significant baseline differences between sectors with social networks being smaller in the sector later served by the intensive service. Social network size increased within the intensive service sector, but not in the standard service sector. There was a significant sector effect for the network component of relatives (intensive > standard) and in the other (‘non friends’) component (standard > intensive) after adjusting for baseline differences.

The findings suggest that the intensive sector community mental health service enhanced peoples social networks with their relatives, relative to the standard service. The reverse is the case for other contacts.

Measurement of the impact of different types of service provision on the views of service users is important in planning mental health services.

Needs (met and unmet) and satisfaction with services, were assessed. People with psychosis (n=131) were interviewed before (Time 1) and after (Time 2) the introduction of two community mental health services in south London. One was intensive, with two specialist teams, the other standard. Outcomes were compared at Time 2, controlling for the Time 1 values.

Overall, 70% of needs were met and mean satisfaction was between ‘mixed’ and ‘mainly satisfied’. There was evidence for higher met needs in the intensive sector, but no evidence for lower unmet needs in the intensive sector or for differences in satisfaction. The additional needs met by the intensive service were associated with aspects of basic living situation. Satisfaction was negatively correlated with both unmet and met needs.

Both services were reasonably successful, with little difference from the user perspective. The intensive service provided benefits in terms of met needs, but this has to be balanced against a possible increase in unmet needs.

We report the impact on the quality of life (QOL) of people with psychosis of an intensive compared with a standard model of community care.

People with psychosis, in two sectors in south London, were interviewed with a variety of measures at baseline, and at two-year follow-up (n=138). After baseline, services within one sector were reorganised, and a more intensive model of community care was introduced. QOL was measured using the Lancashire Quality of Life Profile.

The two overall QOL measures, global QOL and the average of the domain-specific scores, were remarkably stable over time. There was weak evidence for an improvement in living situation domain in the intensive sector; this may be accounted for by a large drop in in-patient admissions. In both sectors objective QOL was poor, and there was little change in any of the objective indicators except in-patient admissions, and a suggestion of increased social activity in the intensive sector.

We failed to find an effect of intensive community care on QOL in people with psychosis. This may indicate an insensitivity to change in QOL measures, or that the intervention failed to produce the kind of changes in mental health and functioning which would be reflected in improved QOL.

The PRiSM Psychosis Study investigated the outcomes of community mental health services for epidemiologically representative cases of psychosis in London.

The results presented in the preceding nine papers are interpreted.

(a) The health and social gains reported in experimental studies of community health services are replicable in ordinary clinical settings, and are more effective than hospital-oriented services which they replace, (b) Dilution does occur - these gains are less pronounced than in experimental (efficacy) studies, (c) Both models of community services produced a range of improved outcomes, (d) Some limited extra advantages (in terms of met needs, improved quality of life, and social networks) were found in the intensive sector, (e) There is no consistent evidence that community-oriented services (which include in-patient beds) fail service users, their families or the wider public. On balance the results weigh slightly in favour of the two-team model (for acute and continuing care) in terms of clinical effectiveness, but the general model is almost as effective and is less expensive.

The evidence supports a community-oriented rather than a hospital-oriented approach and there is little difference between the community mental health team models.

There have been few epidemiological studies of the disabling and poorly understood disorder self-injurious behaviour among adults with learning disabilities.

Interviews were undertaken with the carers of adults known to the Leicestershire Learning Disabilities Register (n=2277). The Disability Assessment Schedule was used and information was also collected on demographic characteristics, developmental and physical status.

Self-injurious behaviour was present in 17.4% of the population. In 1.7% self-injurious behaviour occurred frequently and was severe. There was no gender difference between those with and without self-injurious behaviour. Both the chronological age and developmental quotient of individuals with self-injurious behaviour were lower than those of individuals without self-injurious behaviour. Autistic symptoms were more common among those with self-injurious behaviour. The association of self-injurious behaviour with a wide range of other maladaptive behaviours was highly significant. Logistic regression analysis retained age, developmental quotient, hearing status, immobility and number of autistic symptoms as explanatory variables for self-injurious behaviour.

Self-injurious behaviour is a prevalent and disabling disorder among adults with learning disabilities.

We sought to determine the extent and appropriateness of benzodiazepine use in an elderly community, by measuring prevalence and incidence of benzodiazepines and examining mental health status as a predictor of benzodiazepine use.

Data were collected from two longitudinal studies of people from the same community, sampled in 1982–1983 and again in 1989–1991.

Benzodiazepine prevalence did not decrease during the period under study, but there was a significant reduction in anxiolytic use. Prevalence of benzodiazepines in women is twice that in men, and incidence of hypnotics is slightly higher in women. Prevalence and incidence of hypnotics are strongly associated with increasing age. There were high proportions of long-term users (61 and 70%), and continued use was high (52%) among new users. A large proportion of benzodiazepine use was by those who were concurrently depressed. Similarly, anxiety predicted both current and subsequent use of hypnotics.

Many older people still use benzodiazepines, contrary to official guidelines with regard to their mental health. Our findings add to the weight of opinion that persistent and long-term use should be discouraged.