LAUSR

Abstract Background Patients are often treated with high doses or combinations of antipsychotics. High doses are associated with more severe side effects and reduction of motivation and drive, which may hamper recovery. Nevertheless, dose-reduction (DR) or discontinuation of antipsychotic medication in chronic patients, carries the risk of psychotic relapse. In order to identify risk factors of psychotic relapse after DR or discontinuation, we performed a meta-analysis, aimed (i) to determine the rate of relapse after DR or discontinuation in patients with chronic schizophrenia, and (ii) to assess risk factors for psychotic relapse. Methods We searched PubMed, EMBASE and PsycINFO for studies on dose-reduction of antipsychotics from January 1950 through June 2019. We extracted data and calculated event rates (ER=relapse rate) per person-years including 95% confidence intervals (95%CI). The following data were extracted: (1) patient characteristics (age, percentage of male subjects, setting, duration of illness), (2) dose-reduction/discontinuation characteristics (start-dose before dose-reduction, end-dose after dose-reduction, dose-reduction in milligrams, dose-reduction as percentage of start-dose, time period of dose-reduction), (3) follow-up characteristics (time after dose-reduction), and (4) study characteristics (blinding, year of publication and relapse definition). To account for sample variation we pooled the results following the Dersimonian and Laird random effects method (CMA; Borenstein et al 2009). Between-study heterogeneity was assessed with Cochran’s I2-statistic. We examined the risk of bias in the included studies based on five aspects that could affect the association between exposure and outcome from the Newcastle-Ottawa scale (NOS). Results 46 unique cohorts, presenting 1677 patients in which doses were reduced/discontinued were included in meta-analysis. Most included patients were man at middle age, and with a mean duration of illness of 15 years. There was a considerable risk of bias in studies (48% of studies with NOS≤3). We found an overall event rate (ER) per person-years on psychotic relapse of 0.55 (CI95% 0.46–0.65; p<0.0001; I2 =79). We present various variables that influence event rates. Highest rates were found for inpatients with a short duration of illness. Most robust event rates for psychotic relapse were seen for discontinuing antipsychotics, and if not discontinuing, dose-reduction till under 5mg haloperidol equivalents daily (HE). Abrupt reduction yielded higher rates than gradual reduction. During short follow-up time more relapses occurred than in studies with long follow-up time. Older studies and studies in which relapse was defined as a clinical decision (without applying a psychometric scale) also yielded high event rates, explained by the fact that older studies mostly reduced antipsychotics abruptly till zero or at least doses under 5mgHE, while more recent studies did not, and used rating scales for relapse. Discussion In patients with chronic schizophrenia discontinuing, and to a lesser extent DR till end-dose <5mgHE, patients who reduce doses abrupt, inpatients, and patients with a short duration of illness carry highest relapse risk. Most relapses occur during the first half year after DR.