Medication refusal among patients treated in a community mental health rehabilitation service

were associated, identifying a consistent core of patients at continuing risk of refusal. However of refusers, 84% were persuaded within one month to take treatment, mostly by their community keyworker. Treatment refusal was associated with diagnosis of affective disorder, higher Brief Psychiatric Rating Scale score and failure to usea dosett dispenser,but not with insight, cognitive function, satisfaction with treatment, knowledge of treatment, a range of illness and demographic factors or any particular treatment type. Clinicalimplications Drugrefusalwas evidentlymostly managed by community keyworkers, and without recourse to the Mental Health Act. Simple techniaues such as the use of dosett dispenser may be valuable. Problemsof communicating with the severely mentally ill about their long-term treatment were discussed.

Relapse prevention is at the core of modern psychiatric practice. Immeasurable human suf fering and enormous cost to society result from relapse (Kissling, 1994). Following a psychotic or affective episode, continuing drug treatment will roughly half the risk of relapse over the next year. Poor compliance has been cited as a prominent reason for treatment resistance in schizophrenia (Barnes & Pantelis, 1992), up to 80% of psychotic patients failing to comply with their treatment (Corrigan et al, 1990). Putative reasons for noncompliance including side-effects of treatment, the complexity of the drug regime, lack of social supervision, more severe psychopathology and the doctor's attitude, were identified in an early paper by Blackwell (1972). Kemp et al (1996) have argued that psychotic patients often fail to accept treatment that may help them "largely because of their illness", and have presented promising results from new therapeutic approaches, target ing compliance problems, with techniques such as motivational interviewing.
Methodological problems are inevitable in compliance research, which generally involves the inference of non-compliance with treatment from indirect measures of patient behaviour, and in which the patient's report implicitly cannot be relied on. It is possible to distinguish treatment refusal as a sub-group of non-compliance pro blems, most of which are probably passive and 'partial', meaning a failure to take some but not all the medication. Studying treatment refusal has theoretical advantages, being more reliably based on the professional's account and argu ably representing the type of problematic patient attitude more likely to lead to a requirement for compulsory treatment.
The present study aimed to examine rates of drug refusal and compliance problems in a representative group of patients with severe mental illness of varying aetiology, being treated largely in community settings. The study aimed also to assess the relationship between compli ance problems and relevant clinical factors.

The study
This study was carried out as part of a larger examination of patient satisfaction (Macpherson et al, 1998). One hundred and seventy-four of 199 patients on the Gloucester rehabilitation register were surveyed. The work of the rehabil itation service studied is focused on challenging behaviour in severe mental health, and is inevitably biased towards problems such as treatment refusal. Two research assistants

744
Psychiatric Bulletin (1998). 22, 744-748 assessed all patients in their home settings, with the following instruments: (a) The Insight Scale (Birchwood et al 1993), a seven-item interview scale deriving total insight scores ranging zero to 14. (b) The abbreviated Roth Hopkins Scale (Blessed & Thompson, 1987) for cognitive function. (c) Patient satisfaction was measured as the total of six satisfaction domain scores, on a scale designed by R. M. (Macpherson et al 1997). (d) An abbreviated form of the Understanding of Medication Questionnaire (Macpherson et al, 1996) was used for knowledge of treatment, referring to the patient's main (in his/her view) psychiatric medication. (e) The Brief Psychiatric Rating Scale (BPRS; Overall & Gorham, 1964) was used to rate psychopathology.

Compliance with medication
The compliance sub-scale of the Insight Scale (Birchwood et al 1993), range zero to four, was rated by research assistants. Keyworkers identified whether their patients had refused treatment (defined as explicitly and actively declining treatment in communication with the keyworker) over the past month. Where refusal had occurred, its outcome was recorded by keyworkers on a seven-point scale, giving a spectrum of outcomes from being persuaded immediately to take the treatment, to continued refusal for the full month of treatment. Keyworkers also rated the patient's predominant attitude to treatment over the previous month as 'actively pursued', 'passively accepted', 'actively refused', following a modified form of Van Putten's (1974) method. Keyworkers also rated the patient's acceptance of treatment over the past month as: 'always', 'usually', 'not usually' or 'never'.
In addition to current compliance difficulties, keyworkers recorded the number of compliancerelated admissions over the course of the patient's illness.
Data were entered in to the SPSS computer package for statistical analysis.

Findings
Full results were obtained for 174 of 199 patients (87% response rate). Patient diagnostic, demo graphic and illness related details are listed in Table 1.
Predominant attitude to medication was sig nificantly correlated with acceptance of medi cation and with the compliance sub-scale of the Insight Scale (Kendall's tau, r^0.27, P<0.001; r^â€"0.19, P=0.004). Acceptance of medication was significantly correlated with the compliance sub-scale (Kendall's tau, n=-0.29, P<0.001).
Associations oj treatment refusal Drug refusal was significantly related to more negative attitude to treatment (Kendall's tau, r=0.24, P=0.001), see Table 2. There was no association between treatment refusal and gen der, race, current Section status, accommoda tion type, mode of supervision of medication, or any particular current medication including oral/depot antipsychotic, atypical antipsychotic, antidepressant. lithium or carbamazepine (/2 tests). Treatment refusal distributed significantly differently by diagnosis, being more likely with a diagnosis of affective disorder, and less likely with 'other' diagnosis (mainly primary substance misuse and various neurotic conditions) (/2=18.5. P=0.002). Treatment refusal was less common in the group given a medidose to help take their medication (x2=5.5,P=0.02). See Table  3 for distribution of refusal by diagnosis, use of medidose and accommodation type.
Drug refusal was associated with significantly greater compliance-related and Section admis sions, and higher BPRS score (Mann-Whitney Litests, z=-2.8, P=0.004; z=2.5, P=0.01 respec tively), but not with age, total years since illness started, Insight Score, Roth Hopkins Score or modified Understanding of Medication Question naire score. Total satisfaction score and the satisfaction with treatment sub-scale scores were similar in the treatment refusal and accepting groups (Mann-Whitney U-tests in all cases). The results of this study should be treated with some caution, as the population may be biased in favour of drug refusers; patients in the care of a rehabilitation service may have more severe, resistant illnesses, and drug refusal may be more common than in general psychiatric populations.
The service studied has an assertive out reach policy; periods of drug refusal are anti cipated and do not cause discharge from the service. The response rate was high, suggesting the results are at least representative in this type of population, although as non-participa tion resulted mainly from non-cooperation, the rate of treatment refusal may be slightly underestimated. Around a third of the patients had refused their treatment in the month prior to assess ment. This rate appears high, but is in keeping with a previous finding (Macpherson et al, 1997) of 24% over a two-week period. Marder et al (1984) found that 15 of 31 psychiatric inpatients said they would, if permitted, refuse treatment. The demonstration of higher rates of refusal in hospital is not surprising, particu larly as patients in hospital may be detained. While high rates of refusal appear alarming, the keyworkers' reports of the consequences of refusal are illuminating: half the refusers were persuaded immediately, and 84% within a month, to take their treatment. Although multi-disciplinary reviews contributed to the persuasion, the keyworkers were mostly re sponsible. Evidently, there is a spectrum of patient attitudes, ranging from intermittent refusal responding to reassurance and encour agement, to a more persistent and trenchant refusal expressed by a small proportion of patients (roughly 6% in the present sample). It is also encouraging, and will be of no surprise to clinicians, that the great majority of patients were persuaded to take their treatment by explanation, education and encouragement, and without recourse to compulsory treatment. The crucial role of the community keyworker, with a good relationship with the patient, is emphasised.
Treatment refusal was not associated with any particular psychiatric treatment. Neither was it associated with duration of treatment, cognitive function or satisfaction with or knowledge of treatment. More severe psychopathology on the BPRS and diagnosis of affective disorder were associated with refusal, both findings previously demonstrated (Marder et al 1984;Zito et al 1985). Although the association between refusal and affective diagnosis should be viewed with some caution in this population, which may for example include abnormally severe and difficult to treat depressives, it seems likely that even in chronic conditions, episodic mood disorder may account for treatment refusal. The link between current and historical indicators of refusal suggests that there is a sub-group of patients who persistently and predictably maintain treat ment refusal over time, presumably the familiar 'revolving door' patient who improves with compulsory treatment in hospital and then relapses after discharge due to non-compliance. These findings could be construed to argue for a community treatment order (as proposed by the Royal College of Psychiatrists, 1987), in that this group is easily identifiable and could in theory be maintained well and relapse free, with compul sory community treatment. However, these re sults show that most refusing patients can be persuaded, and the effect of more formalised compliance-promoting therapies shows great potential (Kemp et al 1996). The feasibility of forcing treatment on highly resistive patients in the community is questionable. Logically,as half our patients are in reality partially complying with their treatment, we must dispense with the myth of full compliance in favour of practice which aims for sufficient compliance to prevent relapse.
Finding that depot antipsychotic was not more (or less) associated with drug refusal was interesting, and reinforces previously expressed concern (Babiker, 1986) about the use of depot treatment in situations where compliance is problematic. It is as easy to say no to tablets or injections. More positive was the finding that the use of a dosett dispenser to administer medica tion was associated with lower treatment refusal. Although a causal relationship cannot be in ferred from these cross-sectional findings, it is implied that aids to memory and compliance training may have a secondary effect on improv ing attitudes and minimising refusal. It is surprising that a device so simple and cheap, and apparently effective, should not have passed in to more routine use.
We were surprised that there was no relation ship between satisfaction with drug treatment and drug refusal. Finding a meaningful way to communicate with patients about these,issues is a difficult problem, and the validity of satis faction ratings is inevitably questionable in populations where insight is expected to be impaired.
The results of this study suggest a need for regular discussion with patients about their attitudes to treatment, its perceived value and the risks of non-compliance. Our experience is that patients with severe mental illness are surprisingly open to this approach, and the potential for longer-term intervention packages targeted at improving understanding and attitudes to drug treatment is increasingly clear.
Medication refusal in a community mental health rehabilitation service