General practitioner and psychotherapist referrals to a specialist psychotherapy centre

the design of specialist psychotherapy services as they are responsible for purchasing and commis sioning to an increasing extent. This is particu larly true for GPs who are fundholders, but the views of primary care physicians are important also because of their overall clinical responsibility for the majority of patients with psychiatric illness. It would be useful to know: (a)whether GPs have opinions about the clinical indications for referral; (b) whether there is any difference between those GPs who have referred to a specialist service recently and those who have not; (c)whether the GPs' views


General
practitioners (GPs) were surveyed by questionnaire to determine their views on the indications for specialist psychotherapy and their estimates about the number of patients they thought should be referred.GPs had good awareness of the types of treatments available.
There was close agreement between referring and non-referring GPs and with specialists on the indications for treatment.Estimates about demand for services showed that those who had not referred in the previous year estimated higher rates than those who had referred.GPs and specialists show good overall agreement on level of service and the indications for its use.
Little is known of the reasons why general practitioners (GPs) refer to specialist psychother apy departments or whether their reasons bear any relation to the expectations of the therapists themselves.However, GPs' views are crucial in the design of specialist psychotherapy services as they are responsible for purchasing and commis sioning to an increasing extent.This is particu larly true for GPs who are fundholders, but the views of primary care physicians are important also because of their overall clinical responsibility for the majority of patients with psychiatric illness.
It would be useful to know: (a)whether GPs have opinions about the clinical indications for referral; (b) whether there is any difference between those GPs who have referred to a specialist service recently and those who have not; (c)whether the GPs' views correspond to those of the service providers; and (d) whether GPs tend to agree about the overall need for such specialist services.Shepherd et cd (1966) estimated the 'conspic uous psychiatric morbidity' at 140/1000 at risk/ year in a general practice setting.The manage ment of most of this morbidity is at primary care level and the progression to specialised care settings (such as a specialist psychotherapy service) can be conceptualised as a series of 'filters' (Goldberg & Huxley, 1980).
The specific filters operating to determine which patients are thought to be amenable to psychotherapeutic help have not been delineated.Neither is it known whether the specialist psychotherapists agree in principle with these filters.It seems likely that the extent of service provision tends to be based on historical factors.
Previous work on assessment and allocation to treatment has largely focused on the processes after a referral is received.Alexander & Eagles (1990) described the pattern of referrals to a psychotherapy service, based on the Aberdeen case register, where 49% were offered therapy.Hence about half of the patients referred were assessed but did not receive therapy.Diagnosis, duration of symptoms, which team received the referral, distance and marital status were the factors shown to influence allocation to treatment type.Diagnosis was significantly associated with treatment type for certain illnesses; for example, behaviour therapy being offered preferentially for phobic states.They found no particular associa tion with age or gender and no confirmation of the cultural and social class biases found elsewhere (Siassi & Messer, 1976;Shen & Murray, 1981).Reder & Tyson (1980) have studied the alloca tion process to a particular type of therapy (psychodynamic therapy) in a specialist service.Single patients were more often offered psychodynamic therapy although married patients were more likely to stay in therapy once offered.
The assessment process itself has been de scribed in detail, but, typically, authors have examined the method of assessing suitability for a particular type of therapy rather than the choices between therapies.The attitudes of the clinician carrying out the assessment influence treatment allocation (Calne & Small, 1967) as does the practicality of treatment availability (Tyrer et al, 1987).Little information is, however, available about influences prior to the referral.

The Central Manchester psychotherapy service
The specialist service in Central Manchester accepts direct referrals from GPs as well as tertiary referrals from other psychiatrists.The department offers therapies of cognitive, psychodynamic and systemic orientation, although its main training function is in brief and long-term psychodynamic therapy.The service takes about 500 referrals yearly, about half of which are Manchester residents.There is no formal catch ment area and so the service tends to be demand led via the referrers.For the purpose of this study we did not define to which psychotherapy department the specialist referral might be made so that a fairer comparison of recent referrers and non-referrers could be made.
Psychotherapy in an inner city area such as Manchester does not fit the media stereotype of a predominantly middle class 'life choice' activity.Fifteen per cent of referrals are unemployed, and 10% come from non-UK ethnic backgrounds.
There is a high level of psychiatric morbidity: 71.4% have a DSM-III-R (American Psychiatric Association, 1987) Axis 1 (episode) psychiatric disorder, and 47.8% have an Axis 2 personality disorder diagnosis; 5% have no formal disorder (mainly work and family-related problems).Hence, 23.6% are 'co-morbid' for both Axis 1 and 2 disorders.
The levels of symptoms and disability reported on the Central Manchester psychiatric case register show that the patients seen are of comparable overall severity to an out-patient psychiatric population drawn from the same catchment area.Twenty-six per cent of female referred patients have a history of child sexual abuse and over 20% have family backgrounds rated as emotionally deprived or very deprived.

The study
The questionnaire elicited information in four areas: the clinicians' experience (years' experi ence in psychiatry and in psychotherapy) and knowledge about the psychotherapy service (by asking which of six types of therapy actually available and one 'dummy item' they thought were offered by the department); the clinicians' beliefs about need and appropriate referral rates; the factors from the history most likely to prompt them to refer to a specialist psychotherapy department; and problems thought most likely to benefit from such a referral.The last two were rated on an ordinal 0-5 scale where 0=ideally suitable and 5=not at all suitable.

Sample of general practitioners
The study included GPs in Central and South Manchester.The 51 GPs who had referred patients within the last 12 months were con tacted, of whom 27 (53%) replied.A comparison group of the 48 GPs from the same area who had not referred a patient in the previous 12 months were also contacted.Of these, 24 replied includ ing 21 (44%) with sufficient data to be analysed.There were no significant differences in training in psychiatry between the two groups (22% of referrers and 24% of non-referrers).There was a non-significant trend towards greater experience in the non-referrer group (15% and 24% respec tively).
The questionnaire about factors in the history and about appropriate problems to be referred was also given to all the senior staff of the department (n=7) to act as a comparison group of trained therapists.

Analysis
Responses to the ordinal scales were analysed using the Mann-Whitney U test with a relatively conservative test of significance (P<0.01) because of the multiple measures.
Kendall's coefficient of concordance W was used as the measure of agreement, a nonparametric test relying on rank order with three or more sets of judgements.1 Estimates of rates were elicited by asking GPs to state how many referrals in a notional 2000 patients practice they would refer to a specialist service (and also how many they thought would be likely to benefit with no resource constraint).In view of the asymmetric distribution a test of goodness of fit confirmed a positive skew.There fore median values were used in calculating estimated referral rates.

Findings
How well were referrers aware of services available?Knowledge about the services varied, referrers showing greater awareness of each of the specified aspects of the service, but with both groups showing good awareness of the nature of the service.The GPs were asked whether parti cular types of service were available.Of six items tested the mean awareness was 68% for referrers and 40.3% for non-referrers.
A dummy item ('assertiveness group' which was not held within the department but in the adjacent day hospital) was used to test for response bias.This showed the lowest rate of response at 23% for the whole samples suggest ing that most respondents were replying on the basis of a reasonable knowledge of the depart ment's work.
1. Correlations can lie between 0 and 1, significance Is expressed by calculating x2 and correlations between any two sets of raters as -1 to +1, expressed as rsav.

Estimating need
Non-referring GPs estimated a substantially greater level of need for referral despite their actual behaviour and so the estimates of need have been calculated using the median of the more conservative rates given by the referring GPs.
The median level of need identified by the GPs who did refer was 20 patients per 2000 practice (1%) of whom eleven (0.55%) would actually be referred on.Hence for a population of 200 000 in a typical district the GPs thought that about 1100 patients a year would be referred.This is higher than their actual rates, perhaps reflecting the difficulty of estimating rates as an abstract figure.

Relevant factors in the history
There was excellent overall agreement between referreers, non-referrers and specialist staff about the factors in the history which would incline towards or against referral (Kendall's W= 0.917,x2=35.75,d.f.=134. P<0.001. rsav=0.875).
Clinicians agreed that early loss, early trauma, positive requests for psychotherapy all inclined toward referral whereas violent behaviour, chao tic substance abuse, failure of other therapies and desperateness were not good indicators for referral.
There were few significant differences between referrers and non-referrers or between the referrer group and the specialists, which suggests an encouraging overlap of views between the service purchasers and service providers.

Patient problems and referral
The next analysis concerned the ratings about the nature of the patient's current problem.This showed a good convergence in beliefs again between referrers, non-referrers, and specialists (Kendall's W=0.903,x2=37.93,d.f.=14. P<0.001,rsav=0.896).
Positive indications (score < 2.5) included poor self-esteem, relationship and couple problems, problems regarding sexuality, phobic and obses sional symptoms, mild to moderate depression, anxiety, and psychosomatic symptoms, whereas psychosis (particularly schizophrenia), severe depression, and loneliness were seen as relatively poor indicators (>3.5).

Comment
There is inadequate information linking services to population-based health needs assessment.As an initial step this study shows that there is at least a consensus between GPs (particularly those who refer) and specialist staff about the problems which might be addressed in a specia list service, and those factors in the history favouring referral.Hence, the primary care filter is at least intending to operate rationally to the extent that there is a consensus between those referring and those providing the service.This study does not show how the GPs actually behaved but merely asks about their intentions, but nevertheless the findings are encouraging.
The referring GPs showed a good awareness of the facilities available although there were some gaps In the knowledge of available services for non-referring GPs.
Wide variations in the estimated levels of need and numbers to be referred suggest that there are no established norms in the speciality which GPs can use to calibrate the suitability of referrals.The high rates of predicted referral (i.e.11 per      v. specialistâ€¢***â€¢ â€¢¿ Significant difference, PssO.Ol, Mann-Whitney L/test.Figuresare means based on an ordinal 0-5 scale where 0=ideally suitable and 5=not at all suitable. practice of 2000 which is equivalent to 1100 per district of 200 000) may reflect the high level of awareness of the referring GPs and possibly reflects a bias of interest in those returning the questionnaires.
These results need to be treated with caution as the sample of GPs may be atypical, the response rate was moderate and the estimates show a wide range.Moreover, there is no certainty that these practitioners would actually behave in the way they stated.
Further work is currently in progress to investigate the agreement between referrers and the patients themselves about the perceived problem; on the decision path to different treatment allocations; and to link assessment with outcome.The present study does, however, provide one discrete piece of a jigsaw through the investigation of referrers and potential referrers.Further work is needed to examine the effect of the additional niter provided by the general psychiatrist for tertiary referrals, and also on the actual behaviour of the referrers and how this reflects their intentions.
The referral process is governed by a rational decision-making process by which service design can be seen as an aggregation of multiple individual decisions by referrers.Knowledge of the local service suggests that the picture is much more complex and that referrals may be partly a result of satisfaction or otherwise with the way in which previous particular clinical problems have been treated, i.e. a much more personal 'custo mer/supplier' relationship.Nevertheless, there is good agreement between GPs and specialist psychotherapists about why patients need refer ring and the factors in the history which alert them to this possibility.

Table 1 .
Factors in history and suitability for referral to a specialist psychotherapy service Significant difference.PssO.01,Mann-Whitney I/test.Figures are means based on a 0-5 ordinal scale where 0=ideally suitable and 5=not at all suitable.

Table 2 .
Type of problem and suitability for referral to a specialist psychotherapy service