Background: Although continuous treatment with antipsychotics is still recommended as the gold standard treatment paradigm for all patients with schizophrenia, some clinicians question whether continuous antipsychotic treatment is necessary, or even justified, for every patient with schizophrenia who has been stabilized on antipsychotics.

Objective: The primary objectives of this systematic review and meta-analysis were (i) to compare relapse/hospitalization risks of stabilized patients with schizophrenia under active versus intermittent or placebo treatment conditions; (ii) to examine the role of several study characteristics, possibly intervening in the relationship between relapse risk and treatment condition; and (iii) to examine whether time to relapse is associated with antipsychotic treatment duration.

Methods: A systematic literature search, using the MEDLINE database (1950 until November 2014), was conducted for English-language published randomized controlled trials, covering a follow-up time period of at least 6 months, and investigating relapse/rehospitalization and/or time-to-relapse rates with placebo or intermittent treatment strategies versus continuous treatment with oral and long-acting injectable first- or second-generation antipsychotics (FGAs/SGAs) in stabilized patients with schizophrenia. Additional studies were identified through searches of reference lists of other identified systematic reviews and Cochrane reports. Two meta-analyses (placebo versus continuous and intermittent versus continuous treatment) were performed to obtain an optimal estimation of the relapse/hospitalization risks of stabilized patients with schizophrenia under these treatment conditions and to assess the role of study characteristics. For time-to-relapse data, a descriptive analysis was performed.

Results: Forty-eight reports were selected as potentially eligible for our meta-analysis. Of these, 21 met the inclusion criteria. Twenty-five records, identified through Cochrane and other systematic reviews and fulfilling the inclusion criteria, were added, resulting in a total of 46 records. Stabilized patients with schizophrenia who have been exposed for at least 6 months to intermittent or placebo strategies, respectively, have a 3 (odds ratio [OR] 3.36; 95% CI 2.36-5.45; p < 0.0001) to 6 (OR 5.64; 95% CI 4.47-7.11; p < 0.0001) times increased risk of relapse, compared with patients on continuous treatment. The availability of rescue medication (p = 0.0102) was the only study characteristic explaining systematic differences in the OR for relapse between placebo versus continuous treatment across studies. Studies reporting time-to-relapse data show that the time to (impending) relapse is always significantly delayed with continuous treatment, compared with placebo or intermittent treatment strategies. Although the interval between treatment discontinuation and symptom recurrence can be highly variable, mean time-to-relapse data seem to indicate a failure of clinical stability before 7-14 months with intermittent and before 5 months with placebo treatment strategies. For all reports included in this systematic review, median time-to-relapse rates in the continuous treatment group were not estimable as <50% of the patients in this treatment condition relapsed before the end of the study.

Conclusions: With continuous treatment, patients have a lower risk of relapse and remain relapse free for a longer period of time compared with placebo and intermittent treatment strategies. Moreover, 'success rates' in the intermittent treatment conditions are expected to be an overestimate of actual outcome rates. Therefore, continuous treatment remains the 'gold standard' for good clinical practice, particularly as, until now, only a few and rather general valid predictors for relapse in schizophrenia are known and subsequent relapses may contribute to functional deterioration as well as treatment resistance in patients with schizophrenia.