1. Indication: Maintenance treatment with antipsychotic drugs is effective (A) and indicated for all patients, except for those in whom side-effects outweigh the benefits, those with very mild episodes and unclear diagnoses (C).
In a recent review, maintenance treatment with antipsychotic drugs clearly reduced relapse rates from 64% to 27% within 11 months (Leucht et al., 2012). In naturalistic studies, approximately only 20% of patients with a first episode did not experience another episode within 5 years (Robinson et al., 1999b; Shepherd et al., 1989). As there are no valid indicators that could predict which patients will not relapse, we follow the recommendations of an early guideline that suggested continuous maintenance treatment with antipsychotic drugs for all patients, except for those in whom side-effects outweigh the benefits, those with very mild episodes and unclear diagnoses (Kissling et al., 1991).
2. Maintenance treatment should be given continuously rather than using an intermittent approach (A). The latter may be used combined with monitoring of early warning signs for those patients who do not accept continuous treatment (A-).
Intermittent treatment (tapering medication once a patient is in remission and only re-starting it when there are early warning signs such as problems in concentration or sleep, or attenuated positive symptoms) was less efficacious than continuous treatment (Carpenter et al., 1990; Herz et al., 1991; Jolley and Hirsch, 1990; Pietzcker et al., 1993; Schooler et al., 1997), even in first-episode patients (Gaebel et al., 2011; Wunderink et al., 2007). However, in Wunderink et al., 2007 gradual discontinuation was associated with higher relapse rates, but not with differences in functional outcomes. Moreover, in one of these studies, intermittent treatment started in case of early warning signs was superior to waiting for a full relapse (Pietzcker et al., 1993). Therefore, in patients who do not accept continuous treatment, intermittent treatment can be an option.
3. Maintenance treatment can be carried out with the antipsychotic drug that was effective and well tolerated in the acute phase (A). However, the higher risk of haloperidol for tardive dyskinesia and the weight gain risk of many antipsychotics should be considered (A).
A systematic review (Kishimoto et al., 2013a) found few differences between single second-generation and first-generation antipsychotics in terms of relapse prevention. As a group, second-generation antipsychotics were superior (relapse rates: 29% vs 37.5%), but the classification into first- and second generation is questionable (Leucht et al., 2009b; Lieberman et al., 2005) and almost all studies used haloperidol as a comparator. A clearer advantage includes less tardive dyskinesia, which occurred with an annual incidence of 3.0% in second-generation antipsychotics versus 7.7% with first-generation antipsychotics (again mostly haloperidol, Correll and Schenk, 2008). Many second-generation antipsychotics and low-potency first-generation antipsychotics are associated with weight gain and associated metabolic problems (Allison et al., 1999), but in contrast to tardive dyskinesia weight gain usually starts early after initiation of treatment and can therefore already be detected in the acute phase.
4. Depot formulations is an option for many patients (B).
Non-compliance is very frequent and difficult to detect in schizophrenia. Major advantages of depot drugs are that medication intake is assured and that the doctor immediately knows when patients stop their treatment. However, a large systematic review of randomized controlled trials by Kishimoto et al., 2012 found no difference between the groups of depot and oral antipsychotics for relapse prevention. According to the criteria of this guideline, depot drugs can therefore not be generally recommended for people with schizophrenia. However, the authors also highlighted the major limitation of their meta-analysis. Depot formulations are not more efficacious per se, but they could have advantages in terms of better monitored compliance. In addition, patients who consent to a randomized (and often blinded) trial may be more compliant than patients in everyday clinical care and receive more attention, reminders, etc, thereby, limiting the potential to demonstrate any superiority of depot antipsychotics (Rauch and Fleischhacker, 2013). Indeed, in epidemiological (Tiihonen et al., 2011; Tiihonen et al., 2006) and mirror image studies (Kishimoto et al., 2013b; Davis et al., 1994) in which patients were studied who received depot antipsychotics as part of clinical care, depot drugs were superior to oral antipsychotics. Moreover, depots were never less efficacious than oral forms of administration in the RCTs. Therefore, depot should at least be offered to people who accept this formulation, but should always be considered when non-compliance is a problem.
5. Dosage in relapse prevention: We recommend keeping the dose that was effective in the acute phase as long as there are no important side-effects (A-).
It is unclear whether lower doses than usually applied in the acute phase are sufficient for maintenance treatment. In an early meta-analysis, higher dosages of conventional antipsychotics than 375 mg/day chlorpromazine equivalent did not produce additional effectiveness in maintenance therapy (Bollini, 1994). In another systematic review, dosages between 50-100mg/day chlorpromazine equivalent led to more relapses than doses between 200 and 500 chlorpromazine equivalents (Barbui et al., 1996). In the most recent systematic review, standard antipsychotic doses (e.g. quetiapine ≥ 400mg/day or olanzapine ≥ 10mg/day) were consistently more effective than ‘very low’ doses (e.g. quetiapine <200mg/day or olanzapine <5mg/day), while the comparisons of relapses with ‘low’ doses (e.g. quetiapine 200-400mg/day or olanzapine 5-10mg/day) was only of borderline statistical significance (Uchida et al., 2009). The authors, however, discussed the limited database. In the study most directly addressing the question of lowering the antipsychotic dose (which was published after the meta-analysis by Uchida et al. 2009), maintaining the risperidone dose that was effective in the acute phase was significantly more effective in reducing relapse compared to 50% dose reductions after either 4 weeks or 6 months of acute-phase dosing (Wang et al., 2010). In a further RCT published after the meta-analysis by Uchida et al. 2009, low dose olanzapine depot (150mg biweekly, thought to approximately correspond to 10mg/day oral), was associated with significantly more exacerbations than 300mg/biweekly (thought to approximately correspond to oral olanzapine 20mg/day (Kane et al. 2010). The guideline group felt that, overall, the situation is not clear, as small dose reductions by for example 10% or 25% etc were not systematically assessed. We pragmatically recommended to keep the dose that was effective in the acute phase as long as it was well tolerated.
6. Excessive doses (C), antipsychotic polypharmacy (A-) and non-antipsychotic polypharmacy which may have accumulated in the acute phase should be carefully reduced (C).
Although the evidence supporting both doses considerably above the officially licensed dose ranges and polypharmacy is very limited (see above), these practices are still encountered frequently. The maintenance phase may be used to reduce excessive dosing and polypharmacy. In a recent trial, stopping one of two antipsychotics was followed by more frequent treatment discontinuation than when both antipsychotics were continued (absolute difference 26%, Essock et al., 2011). However, it was possible to successfully switch two-thirds of participants to monotherapy, and the groups did not differ with respect to necessity of hospitalization or symptoms. Moreover, monotherapy resulted in significant weight loss. These results support attempts to reduce polypharmacy and the reasonableness of prescribing guidelines encouraging trials of antipsychotic monotherapy for individuals receiving antipsychotic polypharmacy, as long as patients are allowed to restart polypharmacy if antipsychotic monotherapy does not work.
7. Any dose reduction should be performed very slowly to avoid withdrawal effects (A-) and rebound psychoses (C).
There is a theory that long-term use of antipsychotic drugs increases dopamine receptor sensitivity. If antipsychotics are withdrawn abruptly rather than gradually, this could then lead to withdrawal effects and rebound psychoses (“supersensitivity psychosis”, Chouinard and Jones, 1980; Moncrieff, 2006)). Indeed, in a meta-analysis there was significantly more dyskinesia in the withdrawal group than in the antipsychotic continuation group (Leucht et al., 2012), and various withdrawal effects, such as agitation, anxiety and insomnia, which are thought to be related to hypersensitivity of histamine or cholinergic receptors after prolonged blockade with some agents, are well known clinical phenomena (Correll, 2010). However, the same meta-analysis (Leucht et al., 2012) did not find a difference in relapse risk between the subgroups of studies that withdrew antipsychotics abruptly or gradually. In the same meta-analysis, even when excluding participants who had relapsed within the first nine months of the studies and thus only examinining “9-month survivers”, significantly more patients in the placebo groups relapsed. This finding can be used as an argument against rebound psychosis, because such effects should occur early after stopping medication and not after 9 months. Nevertheless, meta-analyses may not be sensitive enough to detect such effects, and there are animal data supporting the possibility of supersensitivity psychosis, at least in vulnerable subgroups of patients (Leucht et al., 2013b; Samaha et al., 2007; Silvestri et al., 2000). We therefore strongly recommend to reduce doses slowly, e.g. by not more than 10%-20% per month.
8. Duration of antipsychotic relapse prevention: at least 6 years for multiple episode patients and at least 1 year for first-episode patients (A-). The severity of episodes, current psychopathology, suicidal acts or aggression during acute episodes should also be considered (C).
The question is whether patients who have been symptomatically stable or even in remission for a given number of years still need maintenance treatment or whether they are “cured” and do not need it any longer.
Multiple episode patients: In a meta-regression Leucht et al., 2012 found no association between the duration participants had been stable before entering a trial and the reduction of relapse risk by antipsychotics compared to placebo. Thus, there is no evidence that the longer patients have been stable the less would they be in need of medication. In single studies withdrawing antipsychotics led to more frequent relapses than continuing them even in those patients who had been stable for at least 3-5 years (Cheung, 1981) and 6 years (Sampath et al., 1992), but both studies together comprised only 54 participants. We, therefore, tentatively recommend 6 years as a minimum duration for maintenance treatment of multiple-episode patients, because data in patients who have been stable for more than 6 years are not available. But 54 participants obviously represent a very small database and more trials are needed.
First episode patients: It should be noted that in a meta-analysis there was no difference in reduction of relapse risk by antipsychotics between first-episode and multiple-episode populations. This means that first-episode patients generally benefit as much from maintenance treatment as multiple-episode patients (Leucht et al. 2012). This is also underscored by more recent reviews by Zipursky et al., 2014 and Emsley and Fleischhacker, 2013 who reported relapse rates of up to 90 % in patients discontinuing antipsychotics. However, if the longest duration patients had been stable before entering a relapse prevention study is used as a criterion, this duration was one year (Chen et al., 2010), because there are no maintenance studies in first-episode patients who had been stable for longer than one year. We emphasize that one year is a minimum recommendation. It in part also does justice to the phenomenon that approximately 15%-20% of first-episode patients may not have a second episode, at least in the first 5 years of treatment, and will therefore receive unnecessary treatment (Robinson et al., 1999a; Shepherd et al., 1989). Unfortunately there are no valid predictors to identify these patients. Therefore, we follow previous recommendations that all patients should be treated (Kissling et al., 1991). A randomized controlled trial from the Netherlands, originally reporting higher relapse rates in patients following a guided discontinuation strategy (Wunderink et al 2007), found, in a post hoc analysis of longer-term outcomes after the controlled trial period, better recovery rates among those originally assigned to the discontinuation strategy (Wunderink et al., 2013). While this result confirms that some first-episode patients do well without long-term antipsychotics, the evidence is not equivalent to that from a randomized controlled trial designed to address the question because post-randomization selection factors very likely influenced the result.
For both subgroups we recommend that the severity of prior episodes as well as suicidal acts or aggression during acute episodes should be taken into account when considering the minimum duration of antipsychotic relapse prevention (Kissling et al., 1991). Another criterion is the current level of symptoms and functional status. If patients are just improved, but not in remission, we would be very hesitant to consider to withdraw medication.