Child and adolescent mental health services (CAMHS) are inadequately developed over much of the country (Health Advisory Service, 1995), so effective commissioning will be vital if CAMHS are to thrive in the new NHS. The latest health service reorganisation (Department of Health, 1997) has made primary care groups or trusts — which include general practitioners (GPs) — the sole commissioners for local services. Previous studies on the relationship between GPs and CAMHS have examined GP referrals and found them broadly appropriate (Reference Garralda and BaileyGarralda & Bailey, 1988), but some studies suggest GPs' understanding of CAMHS may be too limited for a commissioning role (Reference Bailey and GarraldaBailey & Garralda, 1989; Reference Evans and BrownEvans & Brown, 1993). Although the knowledge that a GP should possess in order to commission services has never been defined (Reference SimpsonSimpson, 2000), it seems uncontentious to assert that commissioning GPs should have good awareness of CAMHS. This paper describes an attempt to define this ‘awareness’ operationally and then evaluate it in a survey of GP principals in North Staffordshire.
The study
Awareness was defined as the expression of attitudes conforming to current knowledge about CAMHS in primary care. GPs were asked (a) to rate how valuable further CAMHS training was for GPs; (b) whether they usually referred cases they identified to CAMHS; and (c) how commonly they thought they saw CAMHS cases. Each question was dichotomised and had a preferred direction assigned, on the basis of current knowledge. The preferred directions were (a) thinking further CAMHS training useful or more; (b) usually doing other than referring cases to CAMHS (Reference Schowalter and SolnitSchowalter & Solnit, 1998); and (c) seeing CAMHS cases commonly (Reference GoodmanGoodman, 1999). GP sex, experience and qualifications were recorded and years of practice after registration was recorded in preference to chronological age. Postgraduate training experience for each discipline was recorded on a four-point scale: up to 1 month; up to 3 months; up to 6 months; and more than 6 months. The disciplines were classified into those with a significant child or psychiatric component (child and adolescent psychiatry, community paediatrics, paediatrics and psychiatry) and those without. The item scores were summed for each of these two groups of variables to create two continuous variables for each GP, estimating CAMHS-related expertise and other expertise respectively. The questionnaire was discussed with a study group of local GPs with an academic interest, to ensure face validity and acceptability before posting. No problems were identified. The questionnaire was posted to all 238 GP principals in North Staffordshire. A three-pass design was intended, but resource issues allowed only two.
Analysis
No returned questionnaire needed to be discarded and no question had more than 7% missing data. Each ‘awareness’ question was used as a dependent variable in separate logistic regression analyses. Independent variables in all three logistic regression models included the two ‘experience’ variables, time spent as a GP, GP gender, undergraduate training, the form training took and possession of Membership of the Royal College of General Practitioners by examination or election. Data analysis was performed using SPSS for Windows 7.5.1, supplemented by StatXact 3 for exact tests.
Findings
The characteristics of the sample, their training and perceived training opportunities in child and adolescent psychiatry and their responses to the awareness questions are given in Table 1.
General sample characteristics | |
General practitioner principals returning questionnaire | n=176 (74% of 238) |
Time since registration | 21.1 Years (SD=8.8) |
Overseas graduates | n=59 (35%) |
Women | n=31 (18%) |
Qualified from organised training scheme | n=60 (35%) |
Organised own training scheme | n=71 (41%) |
Trained without scheme | n=42 (24%) |
MRCGP by examination or election | n=41 (24%) |
Experience and perceived access to training in child and adolescent psychiatry | |
No undergraduate child psychiatry teaching | n=81 (47%) |
More than 1 month's postgraduate experience in child and adolescent psychiatry | n=12 (7%) |
Child and adolescent psychiatry available on an organised rotation | n=2/60 (3%) |
Considered it easy to get further training in child and adolescent psychiatry | n=12 (7%) |
Response to awareness questions | |
Considered further child psychiatry training not valuable | n=48 (27%) |
Usually referred CAMHS cases | n=112 (64%) |
Thought they saw CAMHS cases frequently | n=48 (27%) |
We were unable to trace any information on four non-respondents. Returners differed from non-returners solely on Membership of the Royal College of General Practitioners (MRCGP) (9% of 58: 24% of 176, P<0.001). Only one of the two GPs who reported child and adolescent psychiatry to be available on their training rotations acquired between 1 and 3 months' experience in CAMHS. This was the sole UK graduate with more than 1 month's experience.
Experience in child and adolescent psychiatry alone was not associated with answers to any awareness question (Fisher exact test: all probabilities above 0.3). The results of the three logistic regression analyses on these questions are given in Table 2. This shows that the analysis of the question ‘would further training in child and adolescent psychiatry be useful?’ was not significant overall, so CAMHS related experience and GP gender alone were included in an additional logistic regression, which was significant (P=0.005). Child-related expertise was individually significant at P=0.01 (odds ratio 1.26, 95% Cl 1.05-1.51). GP sex was not significant at P=0.1 (odds ratio 2.4, 95% Cl 0.85-6.79).
Would further training in child and adolescent psychiatry be useful? (overall P < 0.08) | Do you usually refer child and adolescent psychiatry cases? (overall P < 0.0001) | Do you see child and adolescent psychiatry cases frequently? (overall P < 0.003) | |||||||
---|---|---|---|---|---|---|---|---|---|
Odds ratio | 95% CI | P value | Odds ratio | 95% CI | P value | Odds ratio | 95% CI | P value | |
Characteristics and experience | |||||||||
Years since registration | 0.97 | 0.92-1.03 | 0.34 | 1.01 | 0.94-1.07 | 0.82 | 1.11 | 1.04-1.18 | <0.001 |
CAMHS related experience | 1.23 | 1.01-1.51 | 0.044 | 0.77 | 0.63-0.94 | 0.009 | 0.98 | 0.81-1.19 | 0.85 |
General experience | 1.07 | 0.97-1.18 | 0.17 | 1.19 | 1.05-1.34 | 0.005 | 1.07 | 0.96-1.09 | 0.24 |
Not MRCGP | 0.78 | 0.31-1.97 | 0.59 | 4.97 | 1.95-12.72 | <0.001 | 2.03 | 0.64-6.47 | 0.23 |
Female general practitioner (GP) | 2.9 | 0.92-9.11 | 0.068 | 0.58 | 0.22-1.57 | 0.29 | 3.14 | 1.17-8.44 | 0.024 |
Undergraduate CAMHS training v. none | |||||||||
Undergraduate lectures | 0.75 | 0.31-1.8 | 0.52 | 2.24 | 0.81-6.18 | 0.12 | 1.87 | 0.75-4.68 | 0.18 |
Undergraduate experience | 0.33 | 0.05-2.0 | 0.23 | 0.62 | 0.09-4.31 | 0.63 | 1.6 | 0.13-19.59 | 0.71 |
Undergraduate lectures and experience | 0.46 | 0.15-1.37 | 0.16 | 0.66 | 0.22-1.99 | 0.46 | 1.15 | 0.34-3.87 | 0.82 |
Other GP training v. organised rotation | |||||||||
Planned own GP training rotation | 1.30 | 0.51-3.32 | 0.59 | 0.98 | 0.36-2.63 | 0.97 | 0.92 | 0.33-2.51 | 0.86 |
No rotational GP training | 1.16 | 0.33-4.07 | 0.82 | 0.31 | 0.08-1.18 | 0.08 | 0.63 | 0.16-2.44 | 0.51 |
Although child-related experience did not predict seeing CAMHS cases frequently, those who did think they saw such cases frequently were more likely to have a great deal of relevant experience (Moses test of extreme reactions (trimmed) P=0.02). Cross-tabulating the three awareness questions showed that, among those who considered further training was useful, those who thought they saw CAMHS cases rarely were more likely to say they usually referred such cases on (Fisher's exact test P=0.04).
Discussion
GPs were less likely to have postgraduate training in CAMHS than any other speciality covered in the survey. Most thought they saw child mental health problems rarely, despite more than 20 years of research, and referred those they did detect to others. Personal qualities (for example, years in practice) rivalled professional training in determining their awareness. It follows that, as commissioners and managers for CAMHS to their local communities, most UK GPs are likely to underestimate the size of the problem, may prefer solutions that engage people other than themselves and show wide variation in their preferences based on personal characteristics and training experience. The current situation, where an uneven but inadequate spread of secondary and tertiary services tries to provide a service without much support from primary care, looks set to continue under GP commissioning.
There is some good news. The logistic regression analyses suggest that relevant training is associated with improved awareness and most GPs considered further training in CAMHS to be at least valuable. The Royal College of General Practitioners (Anonymous, 1992) may also be having a beneficial influence. Unfortunately, those who did not think that further training was useful were likely to be those with the least relevant training, and most GPs thought access to such training was poor. Undergraduate training seems ineffective.
If the individual questions of the ‘awareness’ construct are valid, increasing child-related expertise should predict, for each question, an answer in the preferred direction. The logistic regression analyses showed this for attitudes to training and referral. For attitude to frequency, predictive validity was supported by the Moses test only. The less convincing validation here probably reflects wide variation in interpretation of the term ‘rarely’, causing conservative error, but previous studies also suggest that GPs lack sensitivity to child psychiatric disorders (Reference Evans and BrownEvans & Brown, 1993). The questions were interrelated: those less confident in their knowledge, and who thought they saw CAMHS cases more rarely, were more likely to refer on. This is consistent with measuring awareness. The failure of specific training in CAMHS to predict any awareness question probably reflects the small number of GPs who had such training, and so does not invalidate the major validity findings. Further development of the method would involve the assessment of larger numbers of questions and the comparison of questionnaire responses with the respondents' assessment of standardised cases or vignettes.
The survey sample represents a sub-sample of the GP principals across all of Staffordshire surveyed by Chambers and Campbell (Reference Chambers and Campbell1996), which is considered nationally representative of GP principals. This differed only in having slightly more single-handed practices than they did (18.9% of 176: 12.3% of 609 GPs, P<0.05), which probably reflects the inner-city bias of this simple, as North Staffordshire is dominated by the conurbation of Stoke-on-Trent. As child and adolescent psychiatric disorders concentrate in cities (Reference Offord, Fleming and LewisOfford & Fleming, 1996), the bias is towards GPs who are most likely to see psychiatrically disturbed children, and increases the relevance of the study sample. It is likely that those GPs who did not return their questionnaires were less interested in CAMHS than those who did. Therefore, the preponderance of GPs who were not Members of their Royal College in the non-responding group is likely to have resulted in conservative bias in relation to that variable.
The major conclusion to be drawn from this survey is that child and adolescent psychiatrists need to increase their involvement with GPs, both as collaborators in the commissioning process (Reference SimpsonSimpson, 2000) and as trainers, possibly in collaboration with other paediatric or psychiatric courses for GPs. The alternative may be to see ‘The New NHS’ pass CAMHS by.
Acknowledgements
A. Wilson assisted in the design of the GP questionnaire and undertook the data collection and computer entry. A. Wolton undertook a comparative audit of GP referral letters, which was used to ensure appropriateness of the awareness questions.
eLetters
No eLetters have been published for this article.