Evaluation of patient held shared care records for people with mental illness
p 40 □ record used B record not used
8.1 Summary o f results
Participation in shared care did not significantly affect rates of admission to hospital, or the length of in-patient episodes. A large proportion of subjects, mostly patients with schizophrenia and related psychoses, were admitted during the study, reflecting the fact that the subjects recruited had severe mental illness and were vulnerable to relapse. Subjects in the shared care limb of the trial were no more likely to receive, or to keep outpatient appointments. Only 76 of the 90 patients recruited had outpatient appointments. Since an inclusion criteria was that follow- up by the hospital services had been organised, it appears that not all patients are followed-up in outpatients. This may reflect the increasing shift to community psychiatry, with some patients only followed up in the community.
Mental health, as assessed by the BASIS-32 and BPRS, did not differ significantly between shared and standard care at follow-up. There was a non-significant trend for patients in shared care to have worse BPRS scores than those receiving standard care. Patient satisfaction with the service was moderate, and shared care had no impact on satisfaction at follow-up.
Most patients and professionals were strongly in favour of the proposed shared care scheme at the start of the study, although this was not translated into practice.
Uptake of the shared care record scheme was low. Professionals and patients alike appeared reticent to use the records. Patients with psychotic illness appear to be particularly reluctant to engage, possibly because of their chaotic lifestyle, or paranoid ideation affecting their judgement about the purpose of the record. By the end of the study, professionals’ enthusiasm for the scheme had waned; they were significantly more pessimistic about patients carrying their records, and the utility of the scheme.
8.2 Use of BASIS-32
The BASIS-32 (Eisen et al 1994), a relatively new instrument for measuring mental health, was chosen as an outcome measure in this study over other, more established instruments. The BASIS-32 was chosen because it is a self-report instrument, and thus affords patients with psychosis a rare opportunity to complete a questionnaire about serious psychological symptoms. It also overcomes some of the difficulties inherent in observer rated scales where the rater is not blind to the condition of the subject.
There was moderate correlation between the global score o f the BASIS-32 and the global score of the BPRS (correlation coefficient o f 0.51), and the depression and psychosis domains o f the two scales. The BASIS-32 and the BPRS regression residuals demonstrated that the relationship between the two scales was linear across a range of scores. However, the confidence intervals of the regression
coefficient of the BPRS and the BASIS-32 score are wide, suggesting that an individual’s BASIS-32 score does not necessarily predict well the BPRS score.
Ideally, two scales purportedly measuring the same thing should be strongly correlated, although some care is needed when comparing different scales using correlation coefficients alone (Bland and Altman 1996). A significant correlation coefficient merely proves there is a relationship between scores, not that they are measuring the same thing. For example, there is likely to be a positive correlation between weight and height in a population, although these are quite separate entities. Another approach in comparing a new instrument with an established one is to assess criterion validity. This is the ability o f the new scale to distinguish case from non-case, compared with an established one, often by plotting receiver operating characteristic (ROC) curves using the sensitivity and specificity for various values. This was not possible in this study, because o f the lack o f case defining criterion of the BPRS.
One possible explanation for the correlation coefficient not being higher may be that patients completing the BASIS-32 who lack insight may underestimate the impact of their disease. The lack of correlation between the “self-care” and “relationships” sub-scales on the BASIS-32 and the “withdrawal retardation” sub scale on the BPRS may be due to lack of insight causing an over-estimation of ability and underscoring on questions. For example, difficulties sought in BASIS-32 questions such as “To what extent are you experiencing difficulty managing day-to- day life” or “To what extent are you experiencing difficulty in the area of leisure
time or recreational activities” may be underestimated, resulting in an artificially low score on the BASIS-32.
The BPRS thinking disturbance sub-scale measured at baseline was strongly associated with a diagnosis classed as a “psychotic illness” i.e. schizophrenia and delusional disorder. However, there was no such relationship with the BASIS-32 psychosis sub-scale. The BASIS-32 may not therefore be a good measure of psychotic symptoms. The components on the BASIS-32 that measure psychosis include the following questions:
“To what extent are you experiencing difficulty in the area o f disturbing or unreal thoughts or beliefs? ”,
"To what extent are you experiencing difficulty in the area o f hearing voices or seeing things?
Patients lacking insight into their psychotic symptoms and may not think that their psychotic thoughts are “disturbing or unreal”, or that their auditory hallucinations are causing difficulty. Manic patients rarely feel their symptoms are problematic.
Another reason for the difference between the two scales is that respondents are aware of their symptoms, but chose not to admit them. Even if insight is present, patients may not wish to reveal these on a questionnaire. Answering positively to questions like those listed above may not only prove embarrassing to the
respondent, but may also be distressing to admit to. An individual with psychosis and consequently, eroded ego boundaries and the classic defences o f denial and regression may be overwhelmed by having to confront the ego-damaging admission that they are experiencing delusions or hallucinations. I view the admission of these phenomena as different to admitting the label o f the phenomena, even if the admission is only to self. The label carries with it stigma, prejudice and a sense of self-accountability that the phenomenon does not. This is one o f the strengths of observer-rated scales of severe mental illness; the patient may manifest symptoms, but it is left to the clinician, often without the raters knowledge, to label those symptoms.
Another benefit of observer-rated scales is that behaviours are more easily recorded. For example, the BASIS-32 asks the respondent
“To what extent are you experiencing problems in the area o f impulsive, illegal or reckless behaviour? ”
“To what extent are you experiencing problems in the area o f manic, bizarre behaviour? ”
“To what extent are you experiencing problems in the area o f apathy, lack o f interest in things? ”
Rating ones own behaviours, especially in the context of an abnormal mental state, possibly without insight, is unlikely to yield a highly accurate response, especially if one of the core questions asks about illegal behaviours. Similar probes exist on the BPRS (excitement, blunted affect), which are rated mainly by the observed behaviour during the interview. This may account for the relatively low correlation between those sub-scales measuring psychosis on the BPRS and the BASIS-32 compared with, say, the depression sub-scales.
I have presented arguments that two o f the main domains germane to assessing psychosis; psychotic experiences and abnormal behaviours, are difficult to gauge by self-report. Nevertheless, there is still an argument for retaining self-report scales for severe mental illness. Patients may have experiences that they would not divulge to an interviewer, or may be missed by a superficial assessment, they do not require the presence o f an experienced clinician and the responses will not be contaminated by the assessors’ preconceptions of the patient, or their knowledge o f the subject’s randomisation status.