High levels of psychiatric illness have been reported in general neurology inpatients and outpatients. Fink et al¹ found a prevalence of 55.1% for current mental disorder in new neurological patients. Carson et al² reported a prevalence of 47% for anxiety and depressive disorders in patients referred to general neurology outpatient clinics. Bridges and Goldberg³ showed the prevalence of psychiatric illness in neurology inpatients to be 39%. Carson et al⁴ reported that 30% of patients referred to a neurologist had symptoms not explained by organic disease. A comparison of neurology outpatients with and without emotional disorders has shown that physical function, physical role function, bodily pain and social functioning are worse in those with emotional disorders.²

It has also been shown that psychiatric illnesses are commonly missed in neurological patients. Bridges and Goldberg³ showed that 72% of psychiatric illness was unrecognised by neurologists, and Fink et al¹ found that only 1.5% of the patients with a mental illness were referred to a psychiatrist.

Carson et al² found that only 11% of neurology patients with an emotional disorder expressed any perceived need for psychiatric input, whereas Bridges and Goldberg³ reported that 58% of neurology patients with a psychiatric illness, and 50% of those without, wished that an enquiry into their mood had been made.

To date, no study has investigated the prevalence rates of mental illness in neurology inpatients admitted to a regional neurosciences centre. It cannot be assumed that the rates would be the same as those reported for neurology outpatients.

The aims of this study were to:

• Estimate the prevalence of psychiatric disorder in neurology inpatients in a regional neuroscience centre

• Identify any high-risk groups

• Assess the effectiveness of a batch of screening questionnaires in the identification of mental illness.

METHODS

All new admissions to a neurology ward during a 6-month period were seen for screening and a psychiatric assessment. The only patients excluded were repeat admissions and private patients. Screening was undertaken using the following questionnaires:

• The anxiety and depression modules of the Primary Care Evaluation of Mental Disorders ⁵

• Mini-Mental State Examination

• The Frontal Assessment Battery⁶

• Alcohol Use Disorders Identification Test questionnaire⁷

• A neurologist rating scale of “organicity”⁴

The screening questionnaires were used as a “batch”. If patients scored above the threshold on one or more of the questionnaires, they were deemed to have a positive result on screening.

All patients had a clinical interview (“the gold standard”) by the same clinician who carried out the screening. If relevant, diagnosis according to the Diagnostic and statistical manual of mental disorders,4th edition (DSM-IV)were assigned. Age, sex and neurological diagnosis were also recorded, so we could calculate whether certain groups of patients were more likely to require psychiatric attention.

Treatment recommendations were made for patients found to have a mental illness.

RESULTS

Characteristics of the sample

There were 265 consecutive admissions during the study period (fig 1). In total 197 (74%) patients were seen by the researchers; 57.4% were women and 42.6% were men. The age range was 18–91 years (mean (SD) 54.5 years). In all, 56% of patients had a known neurological diagnosis, and 44% had been admitted for diagnostic investigations.

Prevalence of mental disorder in our sample

In total, 51.3% of patients fulfilled the DSM-IV criteria for diagnosis of one or more mental disorders (95% confidence interval 44% to 58%), 18.7% fulfilled the criteria for two diagnoses and 5.1% fulfilled the criteria for three or more diagnoses (table 1).

Risk factors for mental illness

There were no differences in rates of mental illness between groups in terms of age, sex, or based on whether subjects had a known neurological diagnosis.

Are the screening questionnaires correctly identifying the people with mental illness?

Presence of a DSM-IV diagnosis on the Primary Care Evaluation of Mental Disorders was taken as a positive result. A cut-off of ⩽26 was used for the Mini Mental State Examination and ⩾8 for the Alcohol Use Disorders Identification Test questionnaire. As there was a high correlation between the Mini Mental State Examination and the Frontal Assessment Battery in detection of caseness (correlation coefficient = 0.776), the Frontal Assessment Battery was not used in the final analysis of the sensitivity and specificity.

Of the 197 patients seen, 104 scored positively on screening. Of the 101 who had a diagnosis of mental illness, 82 were picked up on screening and 19 were missed. The screening questionnaires falsely recognised 22 respondents as having mental illness.

Therefore, the sensitivity of the questionnaires as a whole is 81.2% (71.9–88%) and the specificity is 77.1% (67.2–84.8%).

There was no difference in age, sex, and neurological or psychiatric diagnosis for the 19 patients who had a diagnosis of mental illness but were missed by screening compared with the rest of the sample. Of the 19 patients who were missed by screening, only 12 had completed the full batch of questionnaires, whereas six had not been able to complete any of them.

DISCUSSION

In this study, we have shown that there are high prevalence rates of mental illness in neurology inpatients. These rates are similar to those previously found in neurology outpatients (47–64.6%),^1,2 but higher than the rates reported in neurology inpatients in generic wards (39%).^1,3

We found similar rates of anxiety and depression as previous researchers, but the prevalence of certain groups of psychiatric diagnoses differed in our group compared with those previously studied. Carson et al⁴ have previously shown prevalence rates of non-organic disease to be as high as 30% in a general neurology outpatient clinic. We found the prevalence rate of somatoform disorders to be only 4.5%. This can be explained by a number of factors. Some patients were assessed before an organic cause for their symptoms had been excluded. The prevalence rates for somatoform disorders may differ between inpatients and outpatients, or there may be variation in clinical practices between centres.

We showed a low rate of substance misuse in neurology inpatients (3%). Previous prevalence studies looking at patients in a general hospital show rates of alcohol misuse or dependence to be as high as 30% in men and 8% in women.⁸

The batch of screening questionnaires used was shown to have a high sensitivity (81%) and specificity (77%). However, 19 people who had a diagnosis of mental illness were missed by screening. Six of the 19 (32%) patients missed by the screening procedure had been unable to complete any of the screening questionnaires. Only nine of the remaining 182 (5%) patients had not completed any of the questionnaires. If only those who were able to complete the full screening examination were studied, the sensitivity and specificity would be more favourable. The group of patients who were unable to complete any of the questionnaires have a high rate of neuropsychiatric illness. Six of the 15 (40%) patients who did not complete any of the screening questionnaires were found to have mental illness after psychiatric assessment. When instituting a screening protocol in clinical practice, it will be important to ensure that this group is not forgotten.

We saw a high proportion (74%) of all the patients admitted during a 6-month period. However, patients who were missed by the researchers were typically those who were admitted for shorter periods of time. It is not clear whether these patients were significantly different from the rest of the study population and whether this had an effect on our reported prevalence figures.

The fact that the same researcher carried out both the screening questionnaires and clinical assessment might have introduced bias. However, our prevalence figures were in line with previous research. This maximised the number of patients who completed both components of the assessment and minimised inconvenience for patients. Patients were seen at various times during their admission. It would be useful in future research to follow up on the outcome of the admission so that a longitudinal view could be applied.

The screening and psychiatric assessments were well received by both patients and also the neurology staff. No patients refused to be seen. Many made positive comments about the improved service. Later survey of the neurologists showed that they found this service useful and acceptable, and were keen for it to be continued.

CONCLUSIONS

There is a high prevalence of psychiatric disorder in inpatients on a dedicated neurology ward. The screening questionnaires used had a high sensitivity and specificity and could therefore be used as a simple way of identifying those with psychiatric illness. Considering the high rate of undetected psychiatric comorbidity in this population, which results in distress and burden, this screening process may offer a cost-effective solution that is acceptable to patients and clinicians.