Results

Of 364 new patients booked to attend the designated clinics during the study period 48 did not attend their appointments, leaving 316 eligible to participate. Of these, 12 refused to participate, one was too cognitively impaired to be assessable, two patients disappeared and were untraceable, and one patient found the assessment questions distressing and withdrew. This left 300 patients, a participation rate of 96% of attenders and 82% of all those referred.

One hundred and seventy four (58%) of the patients were female. The mean age was 43 (SD 16.2) years with a range from 14 to 88 years. Table1 describes the neurologists' final “diagnoses” of patients in the sample. In some cases the “diagnoses” are merely symptom descriptions.

The median HAD total score for the sample was 10, with a median HAD depression score of 3 and anxiety score of 6. Criteria for one or more DSMIV diagnoses of emotional disorder were met by 140 (47%) of patients (table 2). Sixty seven patients met criteria for more than one disorder, in 58 cases this was major depression plus another disorder.

Table 3 shows correlations between the HAD scores (emotional distress) and the SF-36 scores (disability). The HAD depression score was more closely associated with disability, particularly in physical function and role functioning physical domains, than the HAD anxiety score.

The main finding was that patients with emotional disorders were substantially more disabled on all eight subscales of the SF-36 (table4). Four of the subscales of the SF-36, general health perceptions, vitality, emotional role functioning, and mental health, measure symptoms that substantively overlap with those recorded by the PRIME-MD and the HAD scale. We will not comment further on scores from these four subscales as they add no meaningful data. (However, interested readers are invited to contact the senior author for full details). Furthermore, subdivision of the sample into those with neurological disease (210 (70%)), and those with unexplained symptoms (90 (30%); see table 5) indicated that this association with emotional disorder was independent of the presence of neurological disease.

An analysis of the associations between emotional disorder and number of somatic symptoms showed that patients with an emotional disorder reported twice as many somatic symptoms as patients without an emotional disorder. Again, independent of the presence of neurological disease (tables 3, 4, 5).

We did not examine the associations between individual emotional disorder diagnoses and disability for two reasons. Firstly, there would be small numbers in individual cells increasing the likelihood of errors. Secondly, about one third of patients with emotional disorders had more than one diagnosis.

Only 11% of the 140 patients with an emotional disorder expressed any perceived need for some form of psychological/psychiatric treatment. Both GPs and neurologists perceived a greater need for such treatment but were more likely to recommend this in cases where the patient had unexplained symptoms, irrespective of the presence of emotional disorders (table 6).

Discussion

Almost half of new attenders at neurology clinics in Edinburgh had an emotional (anxiety or depressive) disorder. The most common diagnostic category was major depression (26%). These emotional disorders were strongly associated with increased disability, more pain, and more somatic symptoms. This association was present whether analysed using a continuous variable (the HAD score) or a categorical approach (the PRIME-MD produced DSM IV diagnoses). On the HAD scale the relation was strongest for depressive symptoms. The association of disability and emotional disorder was equally strong in those with neurological disease and those with unexplained symptoms.

It is of interest to note that despite the apparent importance of emotional disorders, few of these patients expressed any interest in psychological/psychiatric treatment and rather expressed their preference for use of the traditional medical approach. There was a discrepancy between patients and their doctors in that the neurologists and GPs were more likely to consider psychiatric/psychological treatment to be appropriate. However, they were more likely to suggest this for those patients with unexplained symptoms rather than those with neurological disease and comorbid emotional disorder.

When interpreting results the representativeness of the sample must be considered. The participating clinics were representative of the overall service and referrals were taken from a common pool. However certain specialised clinics, in particular neurovascular and memory clinics, are not included and potentially render our sample unrepresentative of patients seen in less specialised services. Nonetheless, the very high participation rate (96%) indicated that this sample is representative of new attenders to the Edinburgh general neurology clinics.

The PRIME-MD is based on the structured clinical interview for DSM III-R15 and has been tested extensively in four American primary care settings,10 three of which were hospital based clinics and one was a family based clinic. The overwhelming majority of patients in the original sampling frame had a comorbid physical disorder. In this patient group it was specific for the major diagnostic categories; 92% for any mood disorder, 98% for major depression, 96% for dysthymia, 99% for panic disorder, and 97% for generalised anxiety disorder. Its weakness was sensitivity, with a detection rate of 67% for any mood disorder and 83% for any psychiatric condition. We tried to improve on this sensitivity by asking all the questions from each module in the interview schedule, as it had been originally designed, and omitted the screening questionnaire that was later developed to decrease the time spent in completing the interview. If it remained insensitive, this will have resulted in an underestimate of psychiatric morbidity. An examination of the difference in median total HAD scores, in this sample, between those identified by the PRIME-MD as having an emotional disorder (median 17) and those who do not (median 6), demonstrated a difference of 11 points, providing further reassurance of the validity of the PRIME-MD in patients with comorbid physical illness.

In accordance with the authors' instructions we converted the raw SF-36 scores into percentages. It should be noted that although making the findings easier to interpret, this conversion can give the appearance that some domains, particularly the role functioning domains, have a greater range of scores than they actually do. For example, in the physical role functioning domain, the raw scores range from 4 to 8, and converted they range from 0 to 100.

It may be asked whether the emotional disorders identified persist. Follow up studies of inpatient populations have found that such disorders often remit after discharge.16 The suggested explanation is that the emotional disorder reflects the stress of a stay in hospital. This hypothesis may be less applicable to outpatients. We are, however, following up this cohort and will report on this separately.

There have been only four previous studies of anxiety and depression in neurology outpatient clinics, one in the north east of England,4 two in the United States,5 and one in the Netherlands.6 Two4 5 were carried out using self rating scales, the general health questionnaire,17 and the symptom check list-90.18 In these studies the prevalence of patients with emotional disorder was 27% and 51% respectively. Neither scale allowed for the generation of individual diagnoses, but were limited to the use of cut offs to describe a patient as “likely” to have a emotional disorder. The Dutch study used a diagnostic interview, the present state examination (PSE),19 and found that 25% of new patients met PSE criteria for an emotional diagnosis, with depressive disorders (15%) being most common. The final study7 relied on the clinical impression of a neurology resident and reports a much lower rate of depression. None of these studies commented on disability or the perception of treatment need.

The results from our study are of particular interest because they demonstrated a strong relation between emotional disorders and disability. In the study of Wells et al,3 they also found that the combined presence of depression and a medical condition had a magnified effect on disability. Our findings replicate this in patients with neurological diseases, a group not included in the study of Wellset al. Of particular interest is the finding that a similar effect exists in patients with unexplained symptoms.

What should be done? The important questions now are firstly, do these emotional disorders persist? Secondly, can they be successfully treated? Thirdly, does successful treatment of an emotional disorder lead to reduction in the patient's disability? It should be noted however, that whereas the patients' doctors were of the view that many of the patients should receive psychiatric or psychological treatment, by and large the patients themselves were not. This has important implications for how psychiatric treatments are “sold” to these patients. If it can be shown that effective treatment can be provided then it seems likely that the first part of any treatment strategy will involve some form of symptom reattribution, possibly using some of the techniques of cognitive therapy. It may be that future neurologists will have to be conversant in the basics of these techniques.

Conclusions

Almost half of the new referrals to neurology outpatient clinics met criteria for a DSM IV mood or anxiety disorder. These patients had greater disability than those without an emotional disorder irrespective of whether they had neurological disease or unexplained symptoms. Despite this, few patients expressed enthusiasm for psychiatric treatment. Both general practitioners and neurologists were more likely to recommend psychiatric treatment when the patients' symptoms were medically unexplained.