Skip navigation
Link to Back issues listing | Back Issue Listing with content Index | Subject Index

Treatment adherence in psychosis

Systematic review

Over a quarter of a century ago, David Sackett and Brian Haynes defined compliance (concordance) as the extent to which a person's behaviour coincides with the medical advice given. Non-adherence can just be about taking the pills, or prematurely ending treatment, or not entering treatment in the first place. Systematic reviews are infrequent, and one about failure of adherence to treatment programmes in people with mental health problems tells us that one in four fails to adhere to treatment [1].

Systematic review

Searching involved two electronic databases, including a specialist mental health database, plus reference lists. Searching was from 1980, but the date of the last search was not given. Any report was included if it included patients with schizophrenia, psychoses, or severe mental disorders, if adherence was the primary outcome and if patients were recruited in a psychiatric setting. Compulsory treatment studies, and those relating to initial appointments were not included.

The definition of non-adherence was either not taking drugs as prescribed, or not keeping appointments as scheduled. Any method of determining these outcomes was allowed.


There were 103 studies found, with data for analysis on just under 24,000 patients in 86 of them. The diagnosis was equally split between patients with schizophrenia, psychoses, or severe mental disorder. Most patients were already on treatment (84%) or were first contacts (14%), with few studies on patients with a history of low adherence. Most studies (85%) were in outpatients or in patients after discharge from hospital. Determination of non-adherence was predominantly made by use of case notes, physician interview or rating scales.

Study sizes ranged from 20 to 2,300 patients, and half the studies had fewer than 100 patients. The bulk of patients (84%) were in studies with more than 151 patients. Duration was as short as two weeks and as long as four years. The median duration was eight months, and 60% lasted longer than six months.

The overall weighted mean rate of non-adherence was 26%, where weighting was by size of study. The unweighted average was 38%. Trial size was a major (and significant) determinant of non-adherence rate (Figure 1), with much lower rates of non-adherence in the bulk of patients in studies with more than 151 patients, whether or not weighting was used. Length of follow up made no difference to non-adherence rates.

Figure 1: Adherence rates and study size for treatments of psychoses

Non-adherence was similar whether the definition was not keeping appointments as scheduled (24%) or not taking drugs as prescribed (30%). Similar rates were seen for inpatients, outpatients, and patients after discharge, and for the different diagnoses of schizophrenia, psychosis and severe mental disorder.


Knowledge is good, methodological insight is better. This study gives us both.

We know from this study that about one in four patients with severe mental illness will fail to adhere to treatment programmes. Neither setting nor diagnosis seems to make much difference.

What we also have is an insight into methodological issues for this type of study, and one that might usefully be applied when looking at similar studies in different settings. This insight is the double one of trial size, and the importance of weighting. There was a clear relationship, with small studies giving much higher non-adherence rates. Weighting by trial size to some extent might obviate the problem, but the implication is perhaps that small studies could have other problems that make them less reliable.

What might then have been interesting would have been a sensitivity analysis of different factors only in larger studies.


  1. 1 M Nosé et al. How often do patients with psychosis fail to adhere to treatment programmes? A systematic review. Psychological Medicine 2003 33: 1149-1160.

previous or next story