Differences in clinical characteristics between
patients assessed for NHS specialist
psychotherapy and primary care counselling
Marco Chiesa1*, Peter Fonagy2and Anthony W. Bateman3
1West London Mental Health Trust & University College London, UK
2University College London & The Anna Freud Centre, UK
3Barnet Enfield and Haringey Mental Health Trust & University College London, UK
care counselling settings and NHS (National Health Service) specialist psychotherapy
settings, there is a paucity of studies specifically comparing differences in clinical
characteristics between the two groups of patients. The aim of this study is to ascertain if
specialist psychotherapy referrals represent a more challenging client group than primary
care counselling patients.
Although several studies have described patient populations in primary
the symptomatic, interpersonal problems and global adjustment dimensions of a sample
of patients (N ¼ 384) assessed by a primary care counselling service located in North
London and a sample of patients (N ¼ 853) assessed in eight NHS psychotherapy
centres located within urban settings in England.
We compare the socio-demographic features and severity of presentation in
Interpersonal Problems and Clinical Outcomes in Routine Evaluation Outcome
Both the groups completed the Brief Symptom Inventory, the Inventory of
dysfunctional than those referred for primary care counselling. The linear function
constructed to discriminate the groups showed that a combination of more psychotic
symptoms, social inhibitions and higher risk of self-harm effectively identified those
referred to psychotherapy services, while patients exhibiting greater levels of somatic
and anxiety symptoms and non-assertiveness were more likely to be seen in primary
care settings. However, similarities between the two samples were also marked, as
shown by the overlap in the distribution of clinical outcomes in routine evaluation
clinical scores in the two samples.
Patients referred for specialist psychotherapy services were more
service delivery of these two types of psychological therapy services.
The findings are discussed in terms of their implications for policy and
*Correspondence should be addressed to Dr Marco Chiesa, Consultant Psychiatrist in Psychotherapy, West London Mental
Health Trust & Hon. Senior Lecturer, UCL, The Cassel Hospital, Richmond TW10 7JF, UK (e-mail: firstname.lastname@example.org).
Psychology and Psychotherapy: Theory, Research and Practice (2007), 80, 591–603
q 2007 The British Psychological Society
Past reviews have emphasized the relative paucity of socio-demographic and clinical
information available concerning the patient populations which are referred for
psychological treatments within NHS psychotherapy and counselling services in the
UK (Parry & Richardson, 1996; Roth & Fonagy, 2005). A handful of district-based
reports have been published describing samples seen at NHS psychotherapy clinics,
comparing them to secondary care outpatients (Amies, 1996) or to a non-NHS
psychotherapy population (Mitchell & Freeman, 1999). These surveys show that the
levels of psychiatric morbidity, symptom distress, and personality disorder diagnoses,
are comparably high in both the NHS psychotherapy samples and general
psychiatric outpatients. Studies of populations treated within primary care
counselling services have focussed on the comparative effectiveness of various
interventions for specific disorders such as depression (Ward et al., 2000), chronic
fatigue syndrome (Chalder, Godfrey, Ridsdale, King, & Wessely, 2003), or other
emotional difficulties (Friedli, King, Lloyd, & Horder, 1997). Mellor-Clark and
colleagues used the clinical outcomes in routine evaluation (CORE) system to obtain
clients’ profiles in large populations of primary care setting users and, as a way to
improve quality of psychological service delivery, advocated the concept of practice-
based evidence as a bridge between evidence-based and routine clinical practice
(Mellor-Clark, Barkham, Mothersole, McInnes, & Evans, 2006; Mellor-Clark, Jenkins,
Evans, Mothersole, & McInnes, 2006). However, we found only one report
specifically comparing the possible clinical differences, including severity of
presentation, between patients who are referred to specialist psychotherapy services
and general practice counsellors (Barkham, Gilbert, Connell, Marshall, & Twigg,
The justification for specialist psychotherapy services is mostly constructed in
terms of the greater clinical complexity of the cases referred to these services.
While this is frequently assumed by both commissioners and providers of services,
this has not been demonstrated. In fact, community psychiatric surveys frequently
fail to show substantial differences in severity or dysfunction between those who
have been referred to and are receiving psychiatric services, and those who are
effectively maintained in the community (Department of Health, 1996; Singleton,
Meltzer, Gatward, Coid, & Deasy, 1998). To justify continued support for specialist
services, information is required to ascertain whether pathways of referrals to
different treatment settings, within which psychological therapy is provided, do
indeed reflect differences in population with regard to severity of psychiatric
morbidity. In this study, we have prospectively collected systematic demographic
and clinical information for a representative sample of 853 patients referred to eight
urban-based psychotherapy services in England over a period of 2 years, and have
compared them with data obtained from a sample of 384 patients consecutively
referred to a primary care counselling service by their general practitioners.
The two samples compared are derived from larger samples obtained in two recent
studies. The first is from a multi-centre study of 15 NHS psychotherapy services
(Chiesa & Fonagy, 1999), while the second sample is from patients recruited for a
prospective study of outcome of counselling in a primary care setting.
592 Marco Chiesa et al.
Primary care sample
Patients were recruited from an inner city primary care counselling service. All GPs
within Haringey, North London, refer patients aged 18–65 to counsellors according to
their clinical assessment of need and following agreed protocols. A maximum of 10
sessions is then offered by counsellors trained in generic counselling skills. All
patients assessed and accepted for the service during 2001–2003 (N ¼ 448) were
approached for informed consent and, on agreement, were included in the study
(N ¼ 384, 85.7% of the original sample). The primary care counselling service
provided within GP practices, is offered to a population of around 230,000 people,
and operates during and outside normal working hours to ensure easy access for
Psychotherapy services selected for this study were identified from a larger national
survey of psychotherapy departments. Out of 15 services participating in a larger
study (Chiesa & Fonagy, 1999), eight psychotherapy services (Royal Free Hospital, St
Ann’s Hospital, Ealing Hospital, Specialist Psychotherapy Services at Sheffield Care
Trust, Willesden Centre for Psychological Treatment, West Middlesex University
Hospital, Thorpe Coombe Hospital, Gordon Hospital, Charing Cross Hospital) located
in large urban areas in London and Sheffield were selected for this study in order to
match the primary care sample in terms of location of referrals. All patients
(N ¼ 1;254) aged 18 or over who attended an initial assessment interview with a
senior clinician at each of the participating psychotherapy services were approached
with a view to obtain informed consent for study participation. If consent was
granted, patients and senior assessing clinicians completed a set of self-rated and
rater-based instruments. Recruitment started in 2001 and ended in 2003. Taking into
account no consent given (N ¼ 112), failure to obtain consent (N ¼ 94) and failure
to complete and return questionnaires (N ¼ 195), data for 853 patients (68.0% of the
assessed sample) were obtained.
The few (psychotherapy sample, N ¼ 23; primary care sample, N ¼ 9) patients who
were assessed in both the primary care and the psychotherapy samples with active
psychotic disorders, learning disability, and drug and/or alcohol dependence, were
excluded from the study.
Socio-demographic (age, gender, marital status, ethnicity, employment status) data were
collected using the out-patient version of the Cassel Baseline Questionnaire (Chiesa &
The Brief Symptom Inventory (BSI) (Derogatis, 1993) was used to evaluate on a
5-point scale the patients’ symptomatic profiles: Somatization, obsessive-compulsive,
interpersonal sensitivity, depression, anxiety, hostility, phobic anxiety, paranoid ideation
and psychoticism. In addition, this measure yields one general measure of symptomatic
distress: The global severity index (GSI). Psychiatric symptom profiles in the two
samples were obtained using the BSI subscales. The BSI has shown good internal
consistency (a ranging from .71 to .85) and test–retest reliability (range .68–.90). We
Clinical characteristics in NHS psychotherapy patients593
compared the psychotherapy and primary care samples with normative data obtained
from a community British sample (Francis, Rajan, & Turner, 1990). Based on the
formulas provided by Derogatis (1993) and Schmitz et al. (2000), we calculated cut-off
points for clinical caseness and thresholds of increasing severity derived from British
The CORE outcome measure (CORE-OM) (Evans et al., 2002) is a standardized
measure of psychological distress; it is easy to apply and interpret, and is designed for
use in evidence-based practice. This instrument covers areas of subjective well-being,
symptoms/problems, functioning and risk to self or others. A clinical score is derived by
multiplying the mean score for all items by 10. Satisfactory internal consistency (a for all
domains ..90) and test–retest stability (r ¼ :81) are shown. Normative data, including a
clinical cut-off point and severity benchmarking (healthy, mild, moderate, moderate-to-
severe, severe) derived from large primary care samples (Evans, Connell, Barkham,
Marshall, & Mellor, 2003), general population samples (Connell et al., 2007) and a
psychiatric secondary care population (Barkham et al., 2001) are available. The clinical
score and the risk score have been entered into the analysis.
The Inventory of Interpersonal Problems—circumplex version (IIP-C) (Horowitz,
Alden, Wiggins, & Pincus, 2000) consists of 64 items rated on a 5-point scale
concerning difficulties with relating to others. Eight subscales (overly nurturing,
intrusive, domineering, distrusting, non-revealing, socially avoidant, non-assertive and
exploitable) and a total distress index are then derived. The items were shown to
have satisfactory internal consistency (median a ¼ :81, range ¼ .76–.94) and high
test–retest correlation coefficients (median r ¼.73, range ¼ .56–.83). Low correlation
coefficients amongst the six subscales show that they are relatively independent of
Procedure and data analysis
After obtaining informed consent from participants, the raw data collected from the
application of the intake battery of measures were sent to the central site (Cassel
Hospital) for entry and analysis. Descriptive, comparative and correlational analyses
were carried out in SPSS for Windows v. 12.
Significance of difference in the mean values of total scores on BSI, IIP and CORE
between the two samples was tested using one-way ANOVA. Chi-squared analysis was
used to test the significance of the difference in caseness and severity benchmarking on
the same measures. Owing to the large sample sizes, a significance level of .01 was used
to assess significance in the difference between dependent variables. The Browne-
Forsythe statistical test was also used when the assumption of equality of variance did
not hold, while the Bonferroni correction test was used for multiple comparisons. We
assessed the magnitude of the difference between the two samples on the three
measures’ total scores through the calculation of the respective effect sizes (Cohen’s d)
Twoseparate multivariate analyses of covariance (MANCOVA) were carried out, with
the BSI and IIP subscales as dependent variables, the assessment status (psychotherapy
or primary care) as fixed factor, and with severity of psychiatric morbidity (GSI) as
covariate, in order to evaluate the significance of the differences found in the
dimensions of functioning between the two samples. We controlled for symptom
severity in order to identify differences in profiles between the samples that went
594Marco Chiesa et al.
beyond the greater number of symptoms endorsed by one or other group. Odds ratios
were computed for analysis of caseness and severity using a binary logistic regression,
with severity banding as dependent variable and status (belonging to the psychotherapy
or primary care sample) as covariate.
In order to identify the most salient characteristics of patients referred to
psychotherapy or primary care counsellors in terms of symptomatic and interpersonal
dimensions, and in terms of CORE risk dimension, three separate stepwise discriminant
function analyses were carried out. Assessment status was entered in the analyses as the
dependent binary variable, with the individual dimensions as independent variables.
The aim of the discriminant analysis was to find the linear combination of the predictor
variables that optimally discriminated between the two groups. This linear combination
of the predictor variables is referred to as the canonical variable.
Table 1 outlines the socio-demographic profile of the samples.
No significant difference between the psychotherapy sample and the primary care
sample was found in any of these demographic variables (age: F ¼ 1:02, df ¼ 1,
p ¼ :31; gender: x2¼ 0:03, df ¼ 1, p ¼ :87; marital status: x2¼ 0:70, df ¼ 1, p ¼ :40;
employment status: x2¼ 2:01, df ¼ 1, p ¼ :16).
Patients in both the psychotherapy and the primary care samples were on average
relatively young (mean age ¼ 37.46, SD ¼ 10:8). Of the patients, 65% were of female
gender and just over half of the sample was single (54.4%), while the remaining half was
either married (29.3%) or separated/widowed or divorced (16.3%). Higher rates of
unemployment comparing to the national average (,10%) were found in both samples
(psychotherapy ¼ 35.4%, primary care ¼ 30.7%).
The BSI GSI mean score of 1.70 (SD ¼ 0:80) in the psychotherapy sample was
significantly higher than the mean score of 1.46 (SD ¼ 0:80) found in the primary care
Table 1. Demographic and clinical variables
Married or equivalent
Clinical characteristics in NHS psychotherapy patients595
sample (F ¼ 23:78, df ¼ 1;732, p , :0001) (Table 2). However, the difference
between the two samples expressed as effect size was modest (Cohen’s d ¼ 0.30,
95% CI: 0.26–0.34).
Controlling for severity, the MANCOVA for BSI individual symptom profiles showed a
significant overall difference between the psychotherapy sample and the primary care
sample (Wilks’ l ¼ 0:95, F ¼ 6:96, df ¼ 9;1226, p , :001). The psychotherapy sample
was found to be significantly more severely disturbed than the primary care sample in
the symptomatic dimensions of paranoia (F ¼ 16:95, df ¼ 1, p , :0001) and
psychoticism (F ¼ 17:63, df ¼ 1, p , :0001), with a marked difference in interpersonal
sensitivity (F ¼ 4:53, df ¼ 1, p ¼ :03). By contrast, patients in the primary care sample
were significantly more symptomatic in somatization (F ¼ 30:99, df ¼ 1, p , :0001),
anxiety (F ¼ 14:07, df ¼ 1, p , :0001), and phobia (F ¼ 10:98, df ¼ 1, p , :0001)
Table 2. Comparison of severity of symptoms (BSI), interpersonal functioning (IIP) and general
functioning (CORE) in the psychotherapy and primary care counselling samples
General severity index
Total mean score
BSI, Brief Symptom Inventory; IIP, Inventory of Interpersonal Problems; CORE, Clinical Outcomes in
Routine Evaluation Outcome Measure.
aAdjusted for level of symptom severity (BSI GSI ¼ 1.62). The MANCOVA was applied only to BSI and
IIP;bp , .01;cp , .05.
596Marco Chiesa et al.
The percentage of patients in the psychotherapy sample who scored above the
threshold for caseness (.1.04) was significantly higher than the percentage of patients
in the primary care sample (psychotherapy: N ¼ 649, 76%; primary care: N ¼ 241, 63%,
x2¼ 23:29, df ¼ 1, p , :0001). The likelihood of scoring above the threshold for
caseness for patients in the psychotherapy sample was almost double that of the
primary care sample (OR ¼ 1.86, CI 1.46–2.45).
The overall difference between the two samples in the percentage of patients in
each severity band (mild, moderate and severe) was also significant (x2¼ 25:10,
df ¼ 2, p , :0001). In particular, significantly fewer patients seen in specialist
psychotherapy services were within the mild severity range (psychotherapy:
N ¼ 204, 24%; primary care N ¼ 143, 37%, x2¼ 23:29, df ¼ 1, p , :001), while a
significantly greater percentage of the psychotherapy sample scored in the severe
range (psychotherapy: N ¼ 215, 25%; primary care: N ¼ 68, 18%, x2¼ 8:44, df ¼ 1,
p , :005). Odds ratio analysis revealed that psychotherapy patients were half as likely
as primary care cases to be within the mild range of severity (OR ¼ 0.53, CI 0.41–
0.69), and 1.5 times more likely to score within the severe band (OR ¼ 1.57, CI
1.16–2.12) (Table 3).
The discriminant analysis revealed that the canonical variable significantly
differentiated the two groups (Wilks’ l ¼ 0:941, F ¼ 25:88, df ¼ 3;1233, p , :0001).
Three variables loaded highly on the canonical variable discriminating the group,
namely psychoticism (standardized canonical discriminant function coefficient
(SCDFC) ¼ 0.71), paranoia (0.59), and somatization (20.71). The canonical variable
successfully assigned 61% to the psychotherapy sample and 63% to the primary care
sample. The overall jack-knifed correct classification was 61.0%. The result of this
analysis suggests that patients with high psychoticism and paranoia scores and low
somatization scores are more likely to be referred to specialist psychotherapy services.
Figure 1. Comparison of Brief Symptom Inventory dimensions between the primary care counselling
and NHS psychotherapy samples.
Clinical characteristics in NHS psychotherapy patients 597
The difference found in these three variables seems to account for the overall difference
in severity between the two groups.
The psychotherapy sample showed a higher level of interpersonal difficulties (IIP
mean ¼ 66.18, SD ¼ 11:23) than the primary care sample (IIP mean ¼ 62.44,
SD ¼ 11:07). The difference is significant (F ¼ 30:58, df ¼ 1;742, p , :001). A modest
difference in the magnitude of the difference between the two samples was found
(Cohen’s d ¼ 0.34, 95% CI: 20.29 to 0.96).
The difference in the rate of subjects who were beyond the threshold for caseness
(.69.99) in the psychotherapy sample (N ¼ 330, 39%) vs. the primary care sample
(N ¼ 100, 26%) was found to be significant (x2¼ 18:67, df ¼ 1, p , :0001). Odds ratio
analysis revealed that patients in the psychotherapy sample were almost twice as likely
to be above the caseness threshold as counselling primary care patients (OR ¼ 1.79, CI
1.37–2.34) (Table 3).
Controlling for severity of psychiatric morbidity (GSI), the MANCOVA revealed a
between-group significant difference in IIP subdimensions (Wilks’ l ¼ 0:98, F ¼ 2:47,
df ¼ 9;1226, p , :01). The psychotherapy sample was found to be significantly
more distressed on the socially inhibited dimension of interpersonal functioning
(F ¼ 9:35, df ¼ 1, p , :002) than the primary care sample, while marked differences
were found in the overly accommodating (F ¼ 4:36, df ¼ 1, p , :04) and self-
sacrificing (F ¼ 4:10, df ¼ 1, p , :05) subdimensions, relative to the scores of the
primary care sample (Table 2).
Table 3. Differences in caseness and severity benchmarking between the psychotherapy and primary
care samples for total means scores for BSI, CORE, and IIP
Test of significance
CORE clinical score
IIP total score
649 (76%)241 (63%) 23.29a
751 (88.0) 327 (85.2%)1.971.28 (0.91–1.82)
224 (26.3)62 (16.1)15.24a
330 (38.7)100 (26.0)18.67a
1Odds ratio indicating the relative likelihood of psychotherapy patients to score above the clinical
caseness and severity benchmarking thresholds.
ap , :001;bp , :005; andcp , :02.
598Marco Chiesa et al.
The discriminant analysis in IIP subdimensions revealed that the canonical variable
was significant in discriminating between the two samples (Wilks’ l ¼ 0.97, F ¼ 12:99,
df ¼ 5;1233, p , :0001). Socially inhibited, overly accommodating and non-assertive
were the three variables with the highest loading on the canonical variable
(SCDFC ¼ 1.00, 0.80 and 20.88, respectively). Patients who present with high levels
of social inhibition and accommodation and with low non-assertiveness tend to be
referred to specialist psychotherapy services. In the discriminant analysis, 62% and 66%
of patients were successfully assigned to the psychotherapy sample and primary care
sample, respectively. Overall, 62.7% of cross-validated group cases were correctly
classified in the analysis.
Clinical outcomes in routine evaluation
The difference between CORE clinical scores in the psychotherapy sample
(mean ¼ 19.01, SD ¼ 7:30) and in the primary care sample (mean ¼ 17.20,
SD ¼ 6:98) was found to be significant (F ¼ 16:67, df ¼ 1, p , :001). As in the
previous two standardized measures, the magnitude of the difference between the two
samples was modest (Cohen’s d ¼ 0.25, 95% CI 20.15 to 0.65). CORE risk dimension
was also significantly higher in the psychotherapy sample than in the primary care
sample (F ¼ 17:27, df ¼ 1, p , :001) (Table 2).
Although we did not find a significant difference in the percentage of patients who
scored on and above the clinical cut-off point of 10.00 (psychotherapy: N ¼ 751, 88%;
primary care: N ¼ 327, 85%, x2¼ 1:97, df ¼ 1, p ¼ :16), odds ratio analysis showed
that psychotherapy patients had a higher likelihood of belonging to the caseness status
than primary care patients (OR ¼ 1.28, CI 0.91–1.82). However, we found that
significantly more patients in the primary care sample scored in the mild severity band
than did those in the psychotherapy sample (psychotherapy: N ¼ 138, 16%; primary
care: N ¼ 85, 22%, x2¼ 6:36, df ¼ 1, p , :02), while conversely the majority of
patients referred for specialist psychotherapy were found to belong to the severe band,
with fewer patients referred to primary care counselling (psychotherapy: N ¼ 198, 23%;
primary care: N ¼ 61, 16%, x2¼ 8:59, df ¼ 1, p , :005). Odds ratio analysis indicated
that psychotherapy patients were less likely than primary care patients to belong to the
mild severity band (OR ¼ 0.68, CI 0.50–0.92), and more likely to score above the
severity threshold (OR ¼ 1.60, CI 1.17–2.20). Concerning risk, we found that
significantly more patients in the psychotherapy sample scored on or above the clinical
cut-off point (psychotherapy: N ¼ 224, 26%; primary care: N ¼ 62, 16%, x2¼ 15:24,
df ¼ 1, p , :0001). Psychotherapy patients were almost twice as likely as primary care
counselling patients to belong to the risk caseness band (OR ¼ 1.85, CI 1.35–2.53)
Despite the significant differences found, the distribution of CORE clinical scores in
the two samples showed a marked overlap (Figure 2).
When comparing our sample with normative and severity profile data derived from
large UK secondary and primary care populations in the North of England (Barkham
et al., 2005), no substantial differences between the two studies were found in the
likelihood of patients in secondary care to score above the clinical cut-off point
(OR ¼ 1.28 and 0.88, respectively) or to belong to the severe range relative to the
primary care sample (OR ¼ 1.60 and 1.34, respectively).
In CORE risk dimension, the discriminant analysis showed a significance in the
canonical variable (Wilks’ l ¼ 0:99, F ¼ 17:26, df ¼ 1;1235, p , :001). Patients with
Clinical characteristics in NHS psychotherapy patients 599
higher risk were more likely to be referred to specialist psychotherapy services, while
patients with lower risk tended to be seen, at least initially, in primary care counselling
services. However, classification just on the basis of risk was fairly poor, with 68% of the
sample correctly assigned to psychotherapy, and 60% to primary care, on the basis of
risk score alone.
The results of this comparative study show that patients referred to NHS secondary care
psychotherapy services are on average significantly more likely to have a diagnosable
psychiatric disorder and more self-reported symptoms, as well as self-reported
interpersonal problems, than patients seen in primary care counselling services. As
shown by the analysis of the BSI, the overall difference in symptom severity is chiefly
accounted for by paranoid and psychotic symptoms in the psychotherapy sample.
Although both groups presented with considerable psychiatric morbidity, we were
able to discriminate the diagnostic/symptom profiles in the two samples. Thus, patients
with higher levels of psychotic symptoms and social inhibitions were more likely to be
seen in psychotherapy services. By contrast, once severity of symptoms was controlled,
higher degrees of somatic and anxiety symptoms and of non-assertiveness predicted a
higher likelihood of being seen in primary care settings.
The use of the CORE measure in this study enables us to compare our dataset with
other large datasets available in the UK for secondary care samples (Barkham et al.,
2001). Data from this instrument suggest that both the specialist service and the
community population sample for this study were comparable in severity to other
psychological therapy samples assessed using CORE. Similar to previous studies
comparing primary and secondary care using the CORE measure (Barkham et al., 2005),
the distribution of severity band scores differed significantly at extreme ends. Both
studies found that the CORE risk scale differentiated the two samples. This suggests that
patients referred to specialist services are on average more likely to indicate suicidal and
Figure 2. Distribution of CORE clinical scores in the psychotherapy and primary care samples.
600 Marco Chiesa et al.
other self-harm risks and relationship and social problems. However, we need to keep in
mind when considering this result that these were only two of the dimensions that
differentiated the groups and only explained a small percentage of the variability.
We also found evidence of higher levels of self-reported interpersonal problems in
the specialist services group. On the whole, people with greater interpersonal
difficulties present with more personality problems, and hence are more difficult to
treat and require longer therapies with experienced clinicians (Parry & Richardson,
1996). These patients, who are likely to have longer standing and more chronic
disturbance, are sensibly referred to specialist services for psychological treatment, as
one might expect. In contrast, people with anxiety disorders and somatic conditions are
either preferentially referred to primary care counsellors, or are as often treated in
primary care as in secondary care.
Despite the differences found in severity and in profiles, the similarities between the
two samples were more marked than the differences. The effect sizes for the difference
between the means in the three measures used are small, and the distribution shows a
very marked degree ofoverlap between the twosamples. While there appear to be some
differences between patients seen in primary care and specialist settings, the extent of
the observed differences do not indicate that the key reason for referral for secondary
care is either severityor risk. Our data do not support the suggestion that chronicity may
account for the choice of referral, and does not indicate what proportion of those in
secondary care may have exhausted primary care treatment options. The step-care
model proposed in a recent document (Care Services Improvement Partnership (CSIP),
2006) concerning the desirability of offering increasing doses of complexity of
psychological intervention, assumes that persistence of problems, rather than severity,
should be the main determinant of the care pathway. This approach may pave the way
for broadening the availability of psychological treatment in primary care, as was
recently proposed (Layard, 2006). Indeed, our study suggests the need to rationalize and
improve access to treatment and the desirability for greater collaboration between
primary and secondary care as a way to improve outcome (Barkham et al., 2005; von
Korff & Goldberg, 2001).
This study also highlights the usefulness of utilizing relatively easy-to-apply measures
in routine evaluations of treatment settings to effectively assess symptoms and other
dimensions of psychiatric morbidity. The adoption by psychological services of a shared
core evaluative battery of measures in routine clinical practice would provide a national
database for service profiling and much required pooled outcome data. In-turn, both
would improve the quality of service delivery across the country.
There are several limitations to the study. These include a potential bias in sampling
selection, as a nationally drawn sample was compared with only one local sample.
Although we tried to minimize bias by selecting only urban-based psychotherapy
services, results may have been different had a nationally drawn primary care data been
available. Generalizability is challenged by extrapolating from one primary care sample
and contrasting this with eight specialist service samples; while the primary care sample
was representative of most UK urban primary care referred samples in terms of the
means and distributional properties, there may be specific factors related to referral
patterns associated with each metropolitan location that might create difficulties for
Other limitations include the lack of formal diagnostic data for the two samples,
which would have given additional comparative strength, particularly in terms of
co-morbidity, in matching the two samples. In addition, sole reliance on self-rated
Clinical characteristics in NHS psychotherapy patients601
measures for evaluation of severity of psychopathological presentation limits the
generalizabilityof data and makes it hard to compare with investigationswhere clinically
based assessments were used.
Notwithstanding these limitations, we feel that this study has given a contribution to
the current debate concerning the need to rationalize psychological therapies provision
and underscores the desirability for a more integrated and better coordinated tier of
services in primary and secondary care. While current findings suggest that genuine
differences between primary and specialist care exists, the size of the overlap between
the two populations indicates a need for a more rational approach to the provision of
The Educational Trust Fund of the Association for Psychoanalytic Psychotherapy in the NHS
provided a grant towards the study. Miriam Wright BSc, Melissa Harrison BSc and Susanna Waern
BSc, helped with data collection.
Amies, P. (1996). Psychotherapy patients: Are they ‘the worried well’? Psychiatric Bulletin, 20,
Barkham, M., Gilbert, N., Connell, J., Marshall, C., & Twigg, E. (2005). Suitability and utility of the
CORE-OM and CORE-A for assessing severity of presenting problems in psychological therapy
services based in primary and secondary care settings. British Journal of Psychiatry, 186,
Barkham, M., Margison, F., Leach, C., Lucock, M., Mellor, C. J., Evans, C., et al. (2001). Service
profiling and outcomes benchmarking using the CORE-OM: Toward practice-based evidence
in the psychological therapies. Journal of Consulting and Clinical Psychology, 69(2),
Care Services Improvement Partnership (CSIP). (2006). Primary care services for depression:
A guide to best practice. London: Department of Health.
Chalder, T., Godfrey, E., Ridsdale, L., King, M., & Wessely, S. (2003). Predictors of outcome in a
fatigued population in primary care following a randomized controlled trial. Psychological
Medicine, 33, 283.
Chiesa, M., & Fonagy, P. (1999). From the efficacy to the effectiveness model in psychotherapy
research. Psychoanalytic Psychotherapy, 13, 259–272.
Chiesa, M., & Fonagy, P. (2000). Cassel personality disorder study: Methodology and treatment
effects. British Journal of Psychiatry, 176, 485–491.
Cohen, J. (1992). Quantitative methods in psychology: A power primer. Psychological Bulletin,
Connell, J., Barkham, M., Stiles, W. B., Twigg, E., Singleton, N., Evans, O., et al. (2007). Distribution
of CORE-OM scores in a general population, clinical cut-off points and comparison with the
CIS-R. British Journal of Psychiatry, 190, 69–74.
Department of Health (1996). Psychiatric morbidity report 4 (OPCS surveys of psychiatric
morbidity in Great Britain). London: The Stationary Office.
Derogatis, L. R. (1993). Brief symptom inventory: Administration, scoring & procedures
manual. Minneapolis: National Computer Systems.
Evans, C., Connell, J., Barkham, M., Margison, F., McGrath, G., Mellor-Clark, J., et al. (2002).
Towards a standardised brief outcome measure: Psychometric properties and utility of the
CORE-OM. British Journal of Psychiatry, 180, 51–60.
Evans, C., Connell, J., Barkham, M., Marshall, C., & Mellor, C. J. (2003). Practice-based evidence:
Benchmarking primary care counselling services at national and local levels. Clinical
Psychology and Psychotherapy, 10, 374–388.
602Marco Chiesa et al.
Francis, V., Rajan, P., & Turner, N. (1990). British community norms for the Brief Symptom
Inventory. British Journal of Clinical Psychology, 29, 115–116.
Friedli, K., King, M. B., Lloyd, M., & Horder, J. (1997). Randomised controlled assessment of non-
directive psychotherapy versus routine general-practitioner care. Lancet, 350, 1662.
Horowitz, L., Alden, L., Wiggins, J., & Pincus, A. (2000). Inventory of interpersonal problems
manual. Odessa, FL: Psychological Corporation.
Layard, R. (2006). The case for psychological treatment centres. British Medical Journal, 332,
Mellor-Clark, J., Barkham, M., Mothersole, G., McInnes, B., & Evans, R. (2006). Reflections on
benchmarking NHS primary care psychological therapies and counselling. Counselling and
Psychotherapy Research, 6, 81.
Mellor-Clark, J., Jenkins, A. C., Evans, R., Mothersole, G., & McInnes, B. (2006). Resourcing a CORE
Network to develop a National Research Database to help enhance psychological therapy and
counselling service provision. Counselling and Psychotherapy Research, 6(1), 16.
Mitchell, J. R., & Freeman, C. P. (1999). Psychiatric morbidity in patients referred for individual
psychotherapy within and without the NHS. Psychiatric Bulletin, 23, 146–149.
Parry, G., & Richardson, A. (1996). NHS psychotherapy services in England: Review of strategic
policy. London: Department of Health.
Roth, A., & Fonagy, P. (2005). What works for whom: A critical review of psychotherapy research
(2nd ed.). New York: Guilford Press.
Schmitz, N., Hartkamp, N., & Franke, G. H. (2000). Assessing clinically significant change:
Application to the SCL-90-R. Psychological Reports, 86, 263–274.
Singleton, N., Meltzer, H., Gatward, R., Coid, J., & Deasy, D. (1998). Psychiatric morbidity among
adults in England and Wales. London: The Stationary Office.
von Korff, M., & Goldberg, D. (2001). Improving outcomes in depression. British Medical
Journal, 323, 948–949.
Ward, E., King, M., Lloyd, M., Bower, P., Sibbald, B., Farrelly, S., et al. (2000). Randomised
controlled trial of non-directive counselling, cognitive- behaviour therapy, and usual general
practitioner care for patients with depression I: Clinical effectiveness. British Medical
Journal, 321, 1383.
Received 15 October 2006; revised version received 10 March 2007
Clinical characteristics in NHS psychotherapy patients 603