This paper examines the feasibility of evaluating innovative models of working at the interface between primary care and secondary mental health services.

Methodological problems relevant to evaluation of innovative models of working at the interface are discussed.

Although there is some evidence that neurotic disorders can be more cost-effectively treated in primary care, many general practitioners (GPs), and possibly some patients, prefer referral to community mental health teams and community psychiatric nurses, which are provided by the secondary health care services. Since the latter are provided with the intention of improving serious mental illness their involvement in the care of neurotic illness can lead to tensions between GPs, local health authorities and service providers. There is little evidence to suggest that psychiatrists working in health centres using the ‘shifted out-patient’ model have eased this problem. By contrast the ‘consultation-liaison’ (C-L) model has a number of theoretical advantages; referrals to secondary care should be limited to those most in need of this level of expertise and GP management skills should improve, so leading to better quality of care for patients who are not referred.

Studies comparing the different models of service delivery are required to address the tensions that have arisen following changes in government policy. Further work is also needed to develop the necessary research tools.

Hyperekplexia is a rare but well-delineated clinical syndrome of pathological startle response and neonatal hypertonia. Many cases result from mutations in the α1 subunit of the glycine receptor (GLRA 1).

The clinical features, management and recent genetic studies of hyperekplexia are reviewed.

Diagnosis of the disorder should not be difficult, if one is aware of the syndrome. The treatment of first choice is with the benzodiazepine drug clonazepam, which often causes a dramatic although incomplete diminution of startle. Both recessive and dominant mutations in GLRA 1 have been found in affected individuals. The study of mouse mutants with startle phenotypes suggests that the remainder of cases may well be due to mutations in the β subunit of the glycine receptor.

Hyperekplexia is the first human disease shown to result from mutations within a neurotransmitter gene. The demonstration of both dominant and recessive inheritance resulting from different mutations in the same gene is of considerable interest, as other neuropsychiatric disorders may result from mutations in ligand-gated ion channels. Mutation analysis of GLRA 1 is also likely to be useful as an aid to genetic counselling and in diagnostic evaluation of neonatal hypertonia.

A point prevalence study al schizophrenia was carried out in 1986 in the former Hampstead Health District using a key informant method to identify cases. The point prevalence al broadly defined schizophrenia was 4.7 per 1000.

A repeat census al people with schizophrenia, using the same method, was carried out in 1991 and the point prevalence calculated. The accuracy al the census method was estimated. Contact with services, social and occupational activity, and medication usage in the 1986 and 1991 samples were compared.

The point prevalence rate al broadly defined schizophrenia in 1991 was 5.1 per 1000. The results al both censuses showed we underestimated the number of individuals with DSM-III-R positive schizophrenia by about 14%. The level of contact with specialist services was greater in 1991 than 1986. Patients in 1991 were discharged on higher doses of medication, and their most recent admission was more likely to be due to non-compliance, than the 1986 group.

The point prevalence confirmed the high rate reported in 1986. The accuracy of the census was within acceptable limits for service planning. Contact with specialist services increased between 1986 and 1991.

Delusions are assumed to reflect disordered reasoning, but with little empirical support. We attempted to study this in 16 relatively intelligent deluded patients and 16 normal volunteers.

Standard tests were used which required subjects to choose between logically fallacious and valid responses, both of which were plausible. The tests were: (a) conditional statements (if… then), (b) syllogisms (e.g. no A are C, some B are C, some C are not A), and (c) judgements of probability. All three tasks incorporated neutral and emotive content.

Both normal and deluded subjects frequently made logical errors. With conditionals, deluded subjects tended to endorse fallacies more often than normal controls and this was accentuated when the content was emotive. Similarly, with syllogisms, the effect of emotional content on the endorsement of unbelievable responses was increased slightly in the deluded group. Finally, the deluded patients showed a trend to be less prone to the conjunction fallacy than normals, suggesting less reliance on existing schema.

Differences in reasoning between deluded patients and controls are surprisingly small. Patients are somewhat more prone to endorse invalid or fallacious responses, especially when emotive themes are involved.

Mental disorders have a strong association with suicide. This meta-analysis, or statistical overview, of the literature gives an estimate of the suicide risk of the common mental disorders.

We searched the medical literature to find reports on the mortality of mental disorders. English language reports were located on MEDLINE (1966–1993) with the search terms mental disorders', ‘brain injury’, ‘eating disorders’, ‘epilepsy’, ‘suicide attempt’, ‘psychosurgery’, with ‘mortality’ and ‘follow-up studies’, and from the reference lists of these reports. We abstracted 249 reports with two years or more follow-up and less than 10% loss of subjects, and compared observed numbers of suicides with those expected. A standardised mortality ratio (SMR) was calculated for each disorder.

Of 44 disorders considered, 36 have a significantly raised SMR for suicide, five have a raised SMR which fails to reach significance, one SMR is not raised and for two entries the SMR could not be calculated.

If these results can be generalised then virtually all mental disorders have an increased risk of suicide excepting mental retardation and dementia. The suicide risk is highest for functional and lowest for organic disorders with substance misuse disorders lying between. However, within these broad groupings the suicide risk varies widely.