Welcome to the Schizophrenia Resource Centre

Welcome, this website is intended for healthcare professionals in EMEA with an interest in the treatment of schizophrenia. By clicking the link below you are declaring and confirming that you are a healthcare professional

You are here

Risk of symptom recurrence with medication discontinuation in first-episode psychosis: A systematic review

Schizophrenia Research, 2-3, 152, pages 408 - 414


The large majority of individuals with a first episode of schizophrenia will experience a remission of symptoms within their first year of treatment. It is not clear how long treatment with antipsychotic medications should be continued in this situation. The possibility that a percentage of patients may not require ongoing treatment and may be unnecessarily exposed to the long-term risks of antipsychotic medications has led to the development of a number of studies to address this question. We carried out a systematic review to determine the risk of experiencing a recurrence of psychotic symptoms in individuals who have discontinued antipsychotic medications after achieving symptomatic remission from a first episode of non-affective psychosis (FEP). Six studies were identified that met our criteria and these reported a weighted mean one-year recurrence rate of 77% following discontinuation of antipsychotic medication. By two years, the risk of recurrence had increased to over 90%. By comparison, we estimated the one-year recurrence rate for patients who continued antipsychotic medication to be 3%. These findings suggest that in the absence of uncertainty about the diagnosis or concerns about the contribution of medication side effects to problems with health or functioning, a trial off of antipsychotic medications is associated with a very high risk of symptom recurrence and should thus not be recommended.

Keywords: First episode psychosis, Schizophrenia, Relapse, Recurrence, Antipsychotic, Withdrawal, Discontinuation.

1. Introduction

Specialized programs have been developed at many centers around the world for individuals experiencing a first episode of schizophrenia and related non-affective psychoses (FEP) with the goal of optimizing clinical outcomes and reducing morbidity (Jackson and McGorry, 2009 and Zipursky and Schulz, 2001). Over most of the past century, schizophrenia has been thought of as a progressive deteriorating illness that inevitably leads to poor outcomes ( Jablensky et al., 1993 ). It is now appreciated, however, that approximately 80% of people will experience a remission of symptoms following a FEP ( Lieberman et al., 1993 ) and that with intensive efforts, both sustained remission and functional recovery may be achieved (Girgis et al, 2011, Lambert et al, 2008, Menezes et al, 2006, and Zipursky et al, 2013). The observation of high rates of symptomatic remission has raised the question of whether all such patients require long-term prophylaxis with antipsychotic medication. Given the long-term medical risks of remaining on antipsychotic medication, in particular tardive dyskinesia ( Correll et al., 2004 ), obesity and metabolic problems ( Allison et al., 2009 ), it has been important to determine whether there are some patients who may not require ongoing maintenance treatment.

It is now known that the large majority of remitted FEP patients will experience a relapse following their remission, but less clear how this risk is related to medication discontinuation. Robinson et al. (1999) reported that 82% of remitted FEP patients had a relapse in the first five years, with those who discontinued their medication having a five times greater risk of relapse than those who remained on medications. Published treatment guidelines and algorithms have been unclear about how long antipsychotic medications should be continued in remitted FEP patients. Takeuchi et al. (2012) reported that of the 11 published treatment guidelines and algorithms that address the issue of medication discontinuation in first-episode schizophrenia, six do not argue against discontinuing antipsychotics after one to two years of treatment. For example, the Canadian clinical practice guidelines for the treatment of schizophrenia (1998) suggest that it is pragmatic to recommend that “patients who have made a functional recovery and have been in remission on medication for at least one to two years may be considered candidates for a trial of no medication”. The Practice Guideline for the Treatment of Patients with Schizophrenia published by the American Psychiatric Association ( Lehman et al., 2004 ) recommends indefinite antipsychotic maintenance medication for patients who have had multiple prior episodes or two episodes within five years. In the case of remitted FEP patients, however, it is suggested that clinicians discuss two possible options with their patients, either indefinite antipsychotic maintenance medication or discontinuation following at least one year of symptom remission.

Discussing a trial of medication discontinuation requires a clear articulation of the risks and costs of relapse. A second episode of schizophrenia may be experienced as a major setback for patients who have made a good recovery from their FEP; it can lead to losses in hard-won social and vocational gains, and may increase the risk of violence and suicide (Amore et al, 2008, Hor and Taylor, 2010, Llorca, 2008, and Masand et al, 2009). Tragically, some patients are not able to achieve a remission following their second episode and develop chronic treatment resistant symptoms (Lieberman, 1993 and Wiersma et al, 1998).

Studies to date have yielded a wide range of recurrence and relapse rates after a remitted FEP when patients discontinue their medications. The risk of relapse in schizophrenia with and without maintenance treatment has been the subject of a recent meta-analysis by Leucht et al. (2012) who reported that one-year relapse rates averaged 65% in patients who had discontinued and 27% in those who had continued medication. These rates were not found to differ when the analysis was limited to studies of patients with a FEP. However, methodological issues in many of the included studies limit their relevance to addressing the clinical dilemma that clinicians face in determining whether to recommend a trial off antipsychotic medications for patients who have remitted from a FEP. In this study, we carried out a systematic review to determine the risk of psychotic symptom recurrence after medication discontinuation in patients who have remitted from a FEP.

2. Materials and methods

2.1. Search strategy

A systematic computerized search was performed to find relevant English language articles published before Oct. 10, 2012, using the following keywords: “[schizophrenia or psychotic disorders]” and “[first episode or early episode]”, and “antipsychotic agents”, and “[recurrence or relapse or withdrawal or discontinuation or taper]”. The following databases were searched for title, abstract and index terms of reference: Medline, PsycINFO, Embase, and Cochrane. Furthermore, the references of retrieved articles were manually searched for additional references. Abstracts were perused for relevance to the systematic review. If relevant studies did not provide the data being sought, individual authors were contacted personally to obtain missing data.

2.2. Selection criteria

Studies were included in the systematic review if: (1) they included a first episode non-affective psychosis population; (2) the population had responded to treatment or experienced a remission of symptoms (e.g. was out of the acute phase of illness and in the maintenance phase) for a minimum of six months prior to medication discontinuation; and (3) the study reported some measure of symptom recurrence, worsening, or relapse as an outcome of interest a minimum of six months after medication discontinuation.

While the primary objective of this review was to estimate the rate of symptom recurrence off of medications, we were also interested in estimating the rate of symptom recurrence with continued treatment. From those studies that met our inclusion criteria above, we also extracted the rate of symptom recurrence if: 1) subjects were randomized to medication continuation for the duration of the follow-up period, and 2) medication continuation did not involve switching to a different medication.

2.3. Data extraction

Two authors (RBZ and NMM) independently reviewed all abstracts and studies, applied the selection criteria, and extracted information about study design, treatment, and outcome. Any discrepancies were resolved by consensus.

2.4. Statistical analysis

As studies varied in their criteria for relapse and symptom recurrence, it was necessary to select criteria for each study that could be used to compare results. When a study used multiple measures of relapse and recurrence, we elected to use the lowest threshold for clinical deterioration as our aim was to estimate the risk of symptom recurrence off of medications rather than the risk of a full relapse. This was also necessary as some studies reinstituted antipsychotic treatment at these lower thresholds in order to prevent relapses and hospitalizations. As a result, many patients who experienced an exacerbation of symptoms off of medications were prevented from deteriorating to a point where they would meet the criteria for the higher threshold of relapse. A weighted mean one-year recurrence rate was calculated by multiplying the mean one-year recurrence rate reported for each study by the number of subjects in the group, summing these, and dividing by the total number of subjects in all studies combined.

3. Results

Two hundred and eighty-six abstracts were reviewed through database searching. Twenty-seven full-text articles, representing 19 unique cohorts, were assessed for eligibility. Six studies met selection criteria and were included in the systematic review (Boonstra et al, 2011, Chen et al, 2010, Emsley et al, 2012, Gaebel et al, 2011, Gitlin et al, 2001, and McCreadie et al, 1989). Details of the six studies included are presented in Table 1 . Studies were excluded if they did not include a distinct group whose medications were discontinued (Crespo-Facorro et al, 2011, Emsley et al, 2008, Gaebel et al, 2007, Gleeson et al, 2009, Harrow et al, 2012, Malla et al, 2008, and Wunderink et al, 2007), if they involved patients who had not been stable on treatment for at least 6 months prior to discontinuing medications (Crow et al, 1986, Gaebel et al, 2002, Hogarty and Ulrich, 1998, Jolley et al, 1990, Kane et al, 1982, Rabiner et al, 1986, Rifkin et al, 1977, and Robinson et al, 1999), or if they included patients who had more than one episode of psychosis (Jolley et al, 1990 and Herz et al, 1991). Three of the six studies provided estimates of the rate of symptom recurrence with medication continuation that met our criteria for inclusion (Boonstra et al, 2011, Gaebel et al, 2011, and McCreadie et al, 1989). Two studies did not have a medication continuation group (Emsley et al, 2012 and Gitlin et al, 2001) and one study was excluded because subjects in the continuation arm had been switched to a different medication ( Chen et al., 2010 ).

Table 1 Characteristics of the included studies and observed relapse rates.

Study Diagnoses Study design Remission duration Intervention group Control group Other treatments Sample size of discontinuation group Sample size of


Method of discontinuation Follow-up duration after discontinuation Definition of relapse/recurrence Relapse rate in medication discontinuation group Relapse rate in medication continuation group
Boonstra et al. (2011) S, SCP, SCA (SCID-IV) RCT (open) 1 year in remission Medication discontinuation Medication continuation for a minimum of 6 months; 4/9 discontinued medications before the 9-month point Permitted, not controlled (case management, psychoeducation, psychosocial 11 9 Gradual over 6–12 weeks 2 years > 4 any PANSS core item and 20% increase in total PANSS, or hospitalization Relapse (6 months): 73%

Relapse (9 months): 82%

Relapse (12 months): 91%

Hospitalization: (9 months): 36%
Relapse (6 months): 0%

Relapse (9 months):12%

Relapse (12 months): 12%

Hospitalization (9 months): 0%
Chen et al. (2010) S, SCP, SCA, brief psychotic disorder, psychosis not otherwise specified (clinical, SCID-IV) RCT, doubleblind 8 weeks psychosis-free

1 year relapse-free on treatment
Placebo 400 mg quetiapine   89 89 Cross-titration to study drug or placebo over 4–6 weeks 1 year Reappearance of psychotic symptoms (beyond threshold on PANSS) and CGI-Severity ≥ 3 Relapse: 63%

Relapse (Kaplan–Meier): 79%

Hospitalization: 16%
Relapse: 30%

Relapse (Kaplan–Meier): 41%

Hospitalization: 6%
Emsley et al. (2012) S, SCP, SCA (SCID-IV) Open, nonrandomized intervention 2 years clinically stable Medication discontinuation and subsequent intermittent treatment None Benzodiazapines and antidepressants permitted 33   IM stopped or tapered over 6 weeks for doses > 25 mg biweekly 3 years CGI-Change ≥ 6, 25% increase in PANSS, hospitalization, SI, HI, self-injury, or violence Relapse (12 months): 79%

Relapse (24 months): 94%

Relapse (36 months): 97%
Gaebel et al. (2011) S, SCP (clinical, ICD-10 F20.x) RCT, open 1 year relapse-free on treatment   R or H continuation (2–8 mg) CBT and psychoed adjunct in random design (5 centers) EI with lorazepam in randomized

19 23 Open step-wise

guided withdrawal

over a maximum of 3

1 year PANSS positive change > 10,

CGI-Change ≥ 6, and GAF

decrease > 20
Relapse: 19%

Marked clinical deterioration: 42.9%

Deterioration (Csernansky

criteria): 57.1%

Hospitalization: 21%
Relapse: 0%

Marked clinical deterioration: 0%

Deterioration (Csernansky criteria):


Hospitalization: 0%
Gitlin et al. (2001) S, SCA (clinical, RDC) Open, nonrandomized

1 year on IM

3 months psychosis-free and stable during 24 week crossover trial
Medication discontinuation None   50   Stopped IM 1.5 years ≥ 6 or 7 on any BPRS psychosis item Relapse (12 months): 78%

Relapse (24 months): 96%

Relapse (36 months): 98%
McCreadie et al. (1989) S (clinical), also

assessed with

PSE, Feigner, RDC
RCT, doubleblind 1 year relapse-free on treatment Placebo Pimozide or flupenthixol IM   7 8 Not specified 1 year Hospitalization Hospitalization: 57% Hospitalization: 0%

Abbreviations: S, schizophrenia; SCP, schizophreniform; SCA, schizoaffective; SCID-IV, Structured Clinical Interview for DSM-IV; ICD, International Classification of Diseases; RDC, Research Diagnostic Criteria; PSE, Present State Examination; PANSS, Positive and Negative Syndrome Scale; CGI, Clinical Gobal Impression; GAF, Global.

Assessment of functioning; BPRS, Brief Psychiatric Rating Scale; EI, early intervention; AP, antipsychotic; BZ, benzodiazepine; R, risperidone; H, haloperidol; IM, intramuscular depot; RCT, randomized controlled trial; CBT, Cognitive Behavioral Therapy; SI, suicidal ideation, HI, homicidal ideation.

3.1. Methodological aspects of the studies reviewed

3.1.1. Study design

Of the six studies included, four were randomized-controlled trials, of which two involved placebo groups (Chen et al, 2010 and McCreadie et al, 1989) and two involved open medication discontinuation (Boonstra et al, 2011 and Gaebel et al, 2011). The other two studies were non-randomized interventions involving medication discontinuation (Emsley et al, 2012 and Gitlin et al, 2001). Two of the six studies evaluated intermittent treatment or early targeted intervention (Emsley et al, 2012 and Gaebel et al, 2011), with the endpoint of relevance to this review being relapse or symptom recurrence requiring medication resumption.

3.1.2. Sample size

Sample sizes ranged from 7 to 89 patients in the included studies for a total of 209 patients in the medication discontinuation groups.

3.1.3. Diagnostic criteria

All studies included first episode non-affective psychosis patients with schizophrenia-related disorders (schizophrenia, schizoaffective disorder, or schizophreniform disorder). One study also included patients with diagnoses of brief psychotic disorder and psychosis not otherwise specified ( Chen et al., 2010 ). Diagnosis was determined clinically for three studies (Gaebel et al, 2011, Gitlin et al, 2001, and McCreadie et al, 1989) and via structured clinical interview (SCID) for three studies (Boonstra et al, 2011, Chen et al, 2010, and Emsley et al, 2012). Two of the studies reconfirmed the diagnosis at a later time (Chen et al, 2010 and Gaebel et al, 2011).

3.1.4. Treatment

In two studies, medication discontinuation involved being randomized to a placebo arm (Chen et al, 2010 and McCreadie et al, 1989). The two studies of medication discontinuation each involved discontinuing use of an intramuscular depot antipsychotic (Emsley et al, 2012 and Gitlin et al, 2001). In those studies that included a maintenance treatment group, three specified the antipsychotics used (Chen et al, 2010, Gaebel et al, 2011, and McCreadie et al, 1989), while one study did not control what medication was used ( Boonstra et al., 2011 ). Two studies specified that adjunctive benzodiazepine or antidepressant use was permitted (Emsley et al, 2012 and Gaebel et al, 2011), and two studies specified that adjunctive non-pharmacological treatments (e.g. psychoeducation, CBT, psychosocial support, etc.) were permitted (Boonstra et al, 2011 and Gaebel et al, 2011).

3.1.5. Length of antipsychotic taper and method of medication discontinuation

Three studies gradually discontinued medications over a maximum of three months (Boonstra et al, 2011, Chen et al, 2010, and Gaebel et al, 2011), two studies stopped a long-acting depot (Emsley et al, 2012 and Gitlin et al, 2001) (intrinsically involving a gradual titration), and one study did not specify the method of discontinuation ( McCreadie et al., 1989 ) although a switch to placebo could be assumed. One randomized study cross-titrated patients from their maintenance medication to either placebo or a new medication ( Chen et al., 2010 ). Finally, one study involved a double-blind placebo cross-over phase prior to open withdrawal being carried out in those subjects who did not relapse in the crossover phase, ensuring a certain clinical stability of subjects whose treatment was withdrawn ( Gitlin et al., 2001 ).

3.1.6. Follow-up

Patients were stabilized on medications for an average of 15 months (range 12–24 months) prior to medication discontinuation. Patients were followed for an average of 16.5 months after medication discontinuation, often until they relapsed. The frequency of monitoring varied from every two weeks to every three months at different points in time. Three studies specified that adherence was monitored through either pill count and/or clinical or structured interviewing (Boonstra et al, 2011, Chen et al, 2010, and Gaebel et al, 2011).

3.1.7. Definition of primary outcome of interest

All studies used some measure of relapse or symptom recurrence as the primary outcome of interest. Three studies used change in the Positive and Negative Syndrome Scale (PANSS) ( Kay et al., 1987 ) total score (10-point or 20–25% increase) (Boonstra et al, 2011, Emsley et al, 2012, and Gaebel et al, 2011), three utilized a minimum threshold score on PANSS or Brief Psychiatric Rating Scale (BPRS) ( Overall and Gorham, 1962 ) total score (Boonstra et al, 2011, Chen et al, 2010, and Gitlin et al, 2001), one used clinical evaluation of relapse ( McCreadie et al., 1989 ), and three used additional signs of relapse such as admission to hospital, change in Global Assessment of Functioning scale (GAF) ( Endicott et al., 1976 ) or Clinical Global Impression (CGI) ( Guy, 1976 ) scale (Chen et al, 2010, Gaebel et al, 2011, and McCreadie et al, 1989). When more than one measure of relapse, symptom recurrence, or clinical deterioration was specified, the measure that reflected the lowest threshold was used as described above.

3.1.8. Risk of recurrence following medication discontinuation

The one-year recurrence rates with medication discontinuation ranged from 57% to 91% and resulted in a weighted mean one-year recurrence rate of 77% in the six studies included ( Table 1 ).

3.1.9. Risk of recurrence with medication continuation

Three of the six studies provided estimates of the recurrence rate associated with continuing antipsychotic medication. None of the nine patients in the study by Boonstra et al. (2011) relapsed during a six-month period of follow-up. Similarly, none of eight patients receiving maintenance medication relapsed at one-year in the study by McCreadie et al. (1989) . Gaebel et al. (2011) reported no relapses in the continuation group at one-year follow-up but 1/23 treated patients had experienced clinical deterioration. Taking the two studies with one-year follow-up together yields a weighted mean recurrence rate of 3%.

4. Discussion

Our systematic review of the rate of recurrence following a remission from a FEP found that the risk of symptom recurrence in remitted patients following discontinuation of antipsychotic medication is extremely high. We estimated the risk of recurrence within the first year following medication discontinuation to be 77%. In contrast, the one-year risk of recurrence with maintenance medications was found to be very low and was estimated at 3%.

These estimates differ substantially from the figures of 64% and 27% reported in a recent meta-analysis for one-year relapse rates with medication discontinuation and continuation, respectively ( Leucht et al., 2012 ). That meta-analysis was not limited to studies involving patients who had been treated for a FEP. However, their one-year relapse rate estimated from a subgroup of eight studies that involved only FEP patients did not differ from their overall results; the one-year relapse rates were 61% for those who discontinued medication and 26% in those who remained on medications. That our estimates differ greatly from those reported by Leucht et al. likely reflects the methodological differences in the studies included in each of the reviews. The meta-analysis by Leucht et al. involved studies dating as far back as 1959 when very different diagnostic systems and clinical management practices were in place. Thirty-one of the 65 randomized controlled trials included in their analysis involved only hospital inpatients, a feature which may limit the generalizability of the findings to current practice. Their analysis was also restricted to randomized controlled trials, which resulted in the exclusion of a number of carefully conducted studies designed to define the risk of relapse following medication discontinuation.

The weighted mean one-year risk of recurrence of 77% is very likely an underestimate and cannot be generalized to those who abruptly discontinue oral medication. A number of studies included in this systematic review involved patients on long-acting depot antipsychotics. Studies by Gitlin et al. (2001) and Emsley et al. (2012) included only patients who had been stabilized on long-acting formulations of fluphenazine and risperidone, respectively. As brain levels of antipsychotics decrease very slowly in patients receiving these medications in depot forms (Gitlin et al, 2001 and Wistedt et al, 1982), it is likely that the percentage of patients relapsing in the first year off medications is smaller than would be seen in patients treated with oral antipsychotics. Other studies utilized gradual discontinuation protocols (Boonstra et al, 2011 and Gaebel et al, 2011), which would also be expected to lead to lower risk of recurrence during the one-year study period than would be associated with a more abrupt discontinuation ( Viguera et al., 1997 ).

The risk of symptom recurrence persists beyond the first year following medication discontinuation. The two non-randomized intervention studies involving medication discontinuation (Emsley et al, 2012 and Gitlin et al, 2001) also reported cumulative recurrence rates after two and three years of follow-up. Gitlin et al. (2001) reported that 96% and 98% of patients relapsed while Emsley et al. (2012) reported that 94% and 97% relapsed within two and three year periods of follow-up, respectively. The subjects studied by Gitlin et al. all met RDC criteria for schizophrenia or schizoaffective disorder while those studied by Emsley et al. all met DSM-IV criteria for schizophrenia or schizophreniform disorder.

The estimates of recurrence risk while continuing on medications in this report are limited by the inclusion of only three studies all with very modest samples sizes (Boonstra et al, 2011, Gaebel et al, 2011, and McCreadie et al, 1989). The study by Boonstra et al. (2011) provided a six-month rate of relapse with medication continuation; we did not include their one-year rate of relapse in our weighted average as 4/9 patients assigned to medication continuation had their medication discontinued after 6 months. No relapses occurred during the 6-month period of medication continuation but two patients were reported to have relapsed at one-year follow-up. We also excluded the results from the randomized controlled trial by Chen et al. (2010) from our estimate of the recurrence rate associated with medication continuation. They found that 27/89 (30%) patients randomized to maintenance treatment with quetiapine relapsed in the first year. However, only 14/89 patients assigned to maintenance treatment with quetiapine had been treated with quetiapine prior to randomization. That is, for 75/89 subjects, maintenance treatment involved switching to a different antipsychotic medication, a process which itself has been reported to be associated with a one-year relapse rate of 39% ( Gardos, 1974 ). Crow et al. (1986) carried out a randomized placebo-controlled study of prophylactic antipsychotic treatment in 120 patients with a FEP and reported that 46% of patients assigned to continued treatment relapsed in the first year compared to 62% assigned to placebo. This study was excluded from our systematic review as patients may have been stable for as little as two months prior to discontinuing medications, and inclusion criteria did not specify the degree of treatment response required but only that they be discharged from inpatient or day hospital care. The one-year relapse rates with medication continuation from the studies by Boonstra et al., Chen et al. and Crow et al. were all included in the meta-analysis by Leucht and contributed to their finding a much higher average rate than reported in our review.

A number of other studies comparing treated to untreated patients that were not included in our analysis reported no relapses at one-year in FEP patients receiving maintenance medication. Kane et al. found that 0/11 patients with a FEP who were randomized to medication continuation relapsed at one-year compared to 7/17 (41%) patients who discontinued medication ( Kane et al., 1982 ). This study was not included in our review as patients were required to have been in remission for only 4 weeks prior to study entry and some patients were determined to have had diagnoses outside of the schizophrenia spectrum when diagnoses were reassigned using Research Diagnostic Criteria (RDC) ( Spitzer et al., 1978 ). Of patients meeting RDC criteria for schizophrenia in that study, 0/6 assigned to maintenance treatment with fluphenazine had relapsed at one-year follow-up ( Kane et al., 1982 ). Similarly, Rabiner et al. (1986) , using a naturalistic design that involved the systematic follow-up of 64 patients with a FEP, found that 0/10 patients who met RDC criteria for schizophrenia and had remained on antipsychotic medication had relapsed after one-year compared to 8/18 (44%) who had discontinued their medications. Gaebel et al. investigated the rate of relapse in 151 patients with a first episode of schizophrenia who were treated for one year with risperidone or haloperidol after having a period of acute treatment which lasted eight weeks in most cases. No relapses were observed in either group but clinical deterioration was observed in 34 patients (23%). As neither the degree of remission achieved at the onset of the one-year follow-up period nor the degree of medication adherence achieved during this time were defined, it is unclear how these results relate to the question being addressed in the current review, i.e. the risk of recurrence with continued medications in patients who had been in remission for at least 6 months. The very low rate of relapse with maintenance treatment that has been reported in a number of studies is likely biased by the inclusion of more cooperative and adherent patients, the exclusion of patients with substance use disorders from some studies (Gaebel et al, 2011, Gitlin et al, 2001, and Kane et al, 1982), and the ongoing treatment these subjects received in research-intensive comprehensive specialized FEP services. However, if these factors lead to an underestimation of the risk of recurrence with medication continuation, one would assume that the weighted mean recurrence rate of 77% that we report with medication discontinuation also likely represents an underestimate.

Our systematic review has led to a higher estimate of one-year recurrence rates off medications and a lower estimate of one-year recurrence rates with medication continuation than has been reported in previous reviews (Gilbert et al, 1995 and Leucht et al, 2012). This likely reflects the different inclusion criteria used for this analysis, which were selected to more accurately reflect the situations that clinicians are now likely to encounter. To date, the broad range of one-year recurrence rates that have been reported in studies with both medication continuation and discontinuation is likely impacted by a number of design features that differ between published studies. These include the diagnostic groups enrolled, the choice of maintenance medication, the rate of medication discontinuation, the degree and duration of clinical stability prior to medication discontinuation, the severity and duration of the recurrence, the duration of follow-up, the frequency of monitoring appointments, the fidelity of treatment assignment and the level of adherence to the assigned treatment. While it would be prudent to suggest that future studies standardize these design features so that results will be more comparable, the question of whether further studies investigating the risk of recurrence with medication withdrawal are warranted should be very carefully considered. Given the very high risk of recurrence associated with medication discontinuation and the strikingly low rate of recurrence associated with maintenance treatment, there are likely to be very few situations in which future clinical trials of this question would be clinically and ethically justified.

This systematic review does have a number of limitations. These include the small number of studies included and the very modest number of subjects in some of these studies. We did not include naturalistic studies that investigated the rate of recurrence with medication continuation as such studies are likely to include patients whose degree of adherence to medication varies greatly, which would erroneously inflate the estimates of recurrence rates on medication. This being said, it is not known how closely relapse rates reflect rates of adherence. It is likely that the low rates of relapse with maintenance medication reported in this analysis reflect very high degrees of adherence in these clinical research studies. It would be important for future studies to document adherence as well as measures of antipsychotic plasma levels. This review was also limited to quantifying the rates of symptom recurrence and did not evaluate the clinical impact of these recurrences. The impact of a recurrence of symptoms may be minor for some individuals whose symptoms are identified early and effectively treated. It would be valuable for future research to document the impact of such relapses in both research and clinical samples of patients.

It is often stated in the literature that 20% of patients who have experienced a first episode of schizophrenia will not require long-term antipsychotic treatment (Emsley et al, 2012 and Leucht et al, 2012). The results of our review do call this assumption into question. Robinson et al. (1999) reported that approximately 18% of patients with a first episode of schizophrenia did not relapse over a period of five years; their study design did not yield an absolute risk of relapse associated with medication continuation but did find that those who stopped medication were almost five times more likely to relapse. Prospective studies of patients with a first episode of schizophrenia have suggested that as many as one third of first admission patients are relapse-free at two years ( Ram et al., 1992 ). However, this finding also begs the issue of how many of these individuals remained well because they were on medications. Prospective studies of first onset patients are also likely to include a percentage of patients whose initial diagnosis of schizophrenia is not confirmed at long-term follow-up. Bromet et al. (2011) reported that 11% of 126 patients initially diagnosed with schizophrenia did not have the diagnosis confirmed at 10-year follow-up. It would not be surprising if the samples recruited for the discontinuation studies reviewed in this paper were less likely to include individuals who have been misdiagnosed as the inclusion criteria required that they be treated and remitted for a minimum of 6 months prior to medication discontinuation. Ascertainment for these studies occurred on average more than a year after starting treatment which might be predicted to result in samples with very high percentages of patients who do truly have schizophrenia. This might in part explain why so few patients remained in remission off medications in the studies included in our review.

In this paper, we have estimated the risk of recurrence following a remission from a first episode of non-affective psychosis. Our review suggests that approximately three quarters of all patients who have experienced a remission from a FEP will have a recurrence of symptoms without medications within a year of discontinuation and that 95–100% are likely to experience clinical deterioration with additional time. The one-year risk of recurrence with medication continuation appears to be very low and in the range of 0–5%. These numbers do not in themselves argue that such patients should not be given a trial off of medications. There are situations where discontinuation of medication requires consideration. In some cases, there may be ongoing uncertainty about the diagnosis that warrants a trial off of medication. There may be other patients for whom it is unclear whether their symptoms or disability may be better accounted for by the effects of medication rather than the effects of illness; a brief trial off medication may resolve such questions. More commonly, patients may decide that they want a trial off of medication. In such cases, physicians should first discuss the high risk of recurrence expected when medications are discontinued and then monitor their patients carefully for early signs that the illness is returning in order to prevent a full relapse. Our findings do call into question the wisdom of those treatment guidelines that suggest that patients who have been in remission from a first episode of schizophrenia for one to two years may be considered for a trial of medication withdrawal ( Takeuchi et al., 2012 ). In the absence of uncertainty about the diagnosis or concerns about the contribution of medication side effects to problems with health or functioning, it would not seem reasonable for psychiatrists to recommend a trial off of antipsychotic medications.

Role of funding source

There was no funding source for this research.


Robert Zipursky and Natasja Menezes designed the study, reviewed the literature and wrote the first draft of the manuscript. David Streiner provided advice on the study design and analysis of the data. All authors contributed to and approved the final manuscript.

Conflict of interest

Robert Zipursky has received research contract funding from Roche and served as a consultant to Roche. He has served on Advisory Boards for Amgen and Sunovion Pharmaceutics.


The authors wish to gratefully acknowledge Lois Cottrell for her assistance with literature searches and Darshika Mistry for her assistance with the preparation of Table 1 .


  • Allison et al., 2009 D.B. Allison, J.W. Newcomer, A.L. Dunn, J.A. Blumenthal, A.N. Fabricatore, G.L. Daumit, M.B. Cope, W.T. Riley, B. Vreeland, J.R. Hibbeln, J.E. Alpert. Obesity among those with mental disorders: a National Institute of Mental Health meeting report. Am. J. Prev. Med.. 2009;36(4):341-350 Crossref
  • Amore et al., 2008 M. Amore, M. Menchetti, C. Tonti, F. Scarlatti, E. Lundgren, W. Esposito, D. Berardi. Predictors of violent behavior among acute psychiatric patients: clinical study. Psychiatry Clin. Neurosci.. 2008;62(3):247-255 Crossref
  • Boonstra et al., 2011 G. Boonstra, H. Burger, D.E. Grobbee, R.S. Kahn. Antipsychotic prophylaxis is needed after remission from a first psychotic episode in schizophrenia patients: results from an aborted randomised trial. Int. J. Psychiatry Clin. Pract.. 2011;15(2):128-134 Crossref
  • Bromet et al., 2011 E.J. Bromet, R. Kotov, L.J. Fochtmann, G.A. Carlson, M. Tanenberg-Karant, C. Ruggero, S.W. Chang. Diagnostic shifts during the decade following first admission for psychosis. Am. J. Psychiatry. 2011;168(11):1186-1194
  • Canadian Psychiatric Association, 1998 Canadian Psychiatric Association. Canadian clinical practice guidelines for the treatment of schizophrenia. Can. J. Psychiatry. 1998;43(Suppl. 2):25S-40S
  • Chen et al., 2010 E.Y. Chen, C.L. Hui, M.M. Lam, C.P. Chiu, C.W. Law, D.W. Chung, S. Tso, E.P. Pang, K.T. Chan, Y.C. Wong, F.Y. Mo, K.P. Chan, T.J. Yao, S.F. Hung, W.G. Honer. Maintenance treatment with quetiapine versus discontinuation after one year of treatment in patients with remitted first episode psychosis: randomised controlled trial. BMJ. 2010;341:c4024 Crossref
  • Correll et al., 2004 C.U. Correll, S. Leucht, J.M. Kane. Lower risk for tardive dyskinesia associated with second-generation antipsychotics: a systematic review of 1-year studies. Am. J. Psychiatry. 2004;161(3):414-425 Crossref
  • Crespo-Facorro et al., 2011 B. Crespo-Facorro, R. Perez-Iglesias, I. Mata, O. Caseiro, O. Martinez-Garcia, G. Pardo, M. Ramirez-Bonilla, J.M. Pelayo-Teran, J.L. Vazquez-Barquero. Relapse prevention and remission attainment in first-episode non-affective psychosis. A randomized, controlled 1-year follow-up comparison of haloperidol, risperidone and olanzapine. J. Psychiatr. Res.. 2011;45(6):763-769 Crossref
  • Crow et al., 1986 T.J. Crow, J.F. MacMillan, A.L. Johnson, E.C. Johnstone. A randomised controlled trial of prophylactic neuroleptic treatment. Br. J. Psychiatry. 1986;148:120-127 Crossref
  • Emsley et al., 2008 R. Emsley, P. Oosthuizen, L. Koen, D.J. Niehaus, R. Medori, J. Rabinowitz. Oral versus injectable antipsychotic treatment in early psychosis: post hoc comparison of two studies. Clin. Ther.. 2008;30(12):2378-2386 Crossref
  • Emsley et al., 2012 R. Emsley, P.P. Oosthuizen, L. Koen, D.J. Niehaus, G. Martinez. Symptom recurrence following intermittent treatment in first-episode schizophrenia successfully treated for 2 years: a 3-year open-label clinical study. J. Clin. Psychiatry. 2012;73(4):e541-e547 Crossref
  • Endicott et al., 1976 J. Endicott, R.L. Spitzer, J.L. Fleiss, J. Cohen. The global assessment scale. A procedure for measuring overall severity of psychiatric disturbance. Arch. Gen. Psychiatry. 1976;33(6):766-771 Crossref
  • Gaebel et al., 2002 W. Gaebel, M. Janner, N. Frommann, A. Pietzcker, W. Kopcke, M. Linden, P. Muller, F. Muller-Spahn, J. Tegeler. First vs multiple episode schizophrenia: two-year outcome of intermittent and maintenance medication strategies. Schizophr. Res.. 2002;53(1–2):145-159 Crossref
  • Gaebel et al., 2007 W. Gaebel, M. Riesbeck, W. Wolwer, A. Klimke, M. Eickhoff, M. von Wilmsdorff, M.C. Jockers-Scherubl, K.U. Kuhn, M. Lemke, A. Bechdolf, S. Bender, D. Degner, R. Schlosser, L.G. Schmidt, A. Schmitt, M. Jager, G. Buchkremer, P. Falkai, S. Klingberg, W. Kopcke, W. Maier, H. Hafner, C. Ohmann, H.J. Salize, F. Schneider, H.J. Moller. Maintenance treatment with risperidone or low-dose haloperidol in first-episode schizophrenia: 1-year results of a randomized controlled trial within the German Research Network on Schizophrenia. J. Clin. Psychiatry. 2007;68(11):1763-1774 Crossref
  • Gaebel et al., 2011 W. Gaebel, M. Riesbeck, W. Wolwer, A. Klimke, M. Eickhoff, M. von Wilmsdorff, M. Lemke, I. Heuser, W. Maier, W. Huff, A. Schmitt, H. Sauer, M. Riedel, S. Klingberg, W. Kopcke, C. Ohmann, H.J. Moller. Relapse prevention in first-episode schizophrenia—maintenance vs intermittent drug treatment with prodrome-based early intervention: results of a randomized controlled trial within the German Research Network on Schizophrenia. J. Clin. Psychiatry. 2011;72(2):205-218 Crossref
  • Gardos, 1974 G. Gardos. Are antipsychotic drugs interchangeable?. J. Nerv. Ment. Dis.. 1974;159(5):343-348 Crossref
  • Gilbert et al., 1995 P.L. Gilbert, M.J. Harris, L.A. McAdams, D.V. Jeste. Neuroleptic withdrawal in schizophrenic patients. A review of the literature. Arch. Gen. Psychiatry. 1995;52(3):173-188 Crossref
  • Girgis et al., 2011 R.R. Girgis, M.R. Phillips, X. Li, K. Li, H. Jiang, C. Wu, N. Duan, Y. Niu, J.A. Lieberman. Clozapine v. chlorpromazine in treatment-naive, first-episode schizophrenia: 9-year outcomes of a randomised clinical trial. Br. J. Psychiatry. 2011;199(4):281-288 Crossref
  • Gitlin et al., 2001 M. Gitlin, K. Nuechterlein, K.L. Subotnik, J. Ventura, J. Mintz, D.L. Fogelson, G. Bartzokis, M. Aravagiri. Clinical outcome following neuroleptic discontinuation in patients with remitted recent-onset schizophrenia. Am. J. Psychiatry. 2001;158(11):1835-1842 Crossref
  • Gleeson et al., 2009 J.F. Gleeson, S.M. Cotton, M. Alvarez-Jimenez, D. Wade, D. Gee, K. Crisp, T. Pearce, B. Newman, D. Spiliotacopoulos, D. Castle, P.D. McGorry. A randomized controlled trial of relapse prevention therapy for first-episode psychosis patients. J. Clin. Psychiatry. 2009;70(4):477-486 Crossref
  • Guy, 1976 W. Guy. Clinical global impressions, ECDEU assessment manual for psychopharmacology. (, 1976) 218-220
  • Harrow et al., 2012 M. Harrow, T.H. Jobe, R.N. Faull. Do all schizophrenia patients need antipsychotic treatment continuously throughout their lifetime? A 20-year longitudinal study. Psychol. Med.. 2012;42(10):2145-2155
  • Herz et al., 1991 M.I. Herz, W.M. Glazer, M.A. Mostert, M.A. Sheard, H.V. Szymanski, H. Hafez, M. Mirza, J. Vana. Intermittent vs maintenance medication in schizophrenia. Two-year results. Arch. Gen. Psychiatry. 1991;48((4):333-339 Crossref
  • Hogarty and Ulrich, 1998 G.E. Hogarty, R.F. Ulrich. The limitations of antipsychotic medication on schizophrenia relapse and adjustment and the contributions of psychosocial treatment. J. Psychiatr. Res.. 1998;32(3–4):243-250 Crossref
  • Hor and Taylor, 2010 K. Hor, M. Taylor. Suicide and schizophrenia: a systematic review of rates and risk factors. J. Psychopharmacol.. 2010;24(4):81-90 (Suppl.) Crossref
  • Jablensky et al., 1993 A. Jablensky, H. Hugler, M. Von Cranach, K. Kalinov. Kraepelin revisited: a reassessment and statistical analysis of dementia praecox and manic-depressive insanity in 1908. Psychol. Med.. 1993;23(4):843-858 Crossref
  • Jackson and McGorry, 2009 H.J. Jackson, P.D. McGorry. The Recognition and Management of Early Psychosis: A Preventive Approach. 2nd ed. (Cambridge University Press, Cambridge, UK; New York, 2009)
  • Jolley et al., 1990 A.G. Jolley, S.R. Hirsch, E. Morrison, A. McRink, L. Wilson. Trial of brief intermittent neuroleptic prophylaxis for selected schizophrenic outpatients: clinical and social outcome at two years. BMJ. 1990;301(6756):837-842 Crossref
  • Kane et al., 1982 J.M. Kane, A. Rifkin, F. Quitkin, D. Nayak, J. Ramos-Lorenzi. Fluphenazine vs placebo in patients with remitted, acute first-episode schizophrenia. Arch. Gen. Psychiatry. 1982;39(1):70-73 Crossref
  • Kay et al., 1987 S.R. Kay, A. Fiszbein, L.A. Opler. The positive and negative syndrome scale (PANSS) for schizophrenia. Schizophr. Bull.. 1987;13(2):261-276 Crossref
  • Lambert et al., 2008 M. Lambert, D. Naber, A. Schacht, T. Wagner, H.P. Hundemer, A. Karow, C.G. Huber, D. Suarez, J.M. Haro, D. Novick, R.W. Dittmann, B.G. Schimmelmann. Rates and predictors of remission and recovery during 3 years in 392 never-treated patients with schizophrenia. Acta Psychiatr. Scand.. 2008;118(3):220-229 Crossref
  • Lehman et al., 2004 A.F. Lehman, J.A. Lieberman, L.B. Dixon, T.H. McGlashan, A.L. Miller, D.O. Perkins, J. Kreyenbuhl. Practice guideline for the treatment of patients with schizophrenia, second edition. Am. J. Psychiatry. 2004;161(2):1-56 (Suppl.)
  • Leucht et al., 2012 S. Leucht, M. Tardy, K. Komossa, S. Heres, W. Kissling, G. Salanti, J.M. Davis. Antipsychotic drugs versus placebo for relapse prevention in schizophrenia: a systematic review and meta-analysis. Lancet. 2012;379(9831):2063-2071 Crossref
  • Lieberman, 1993 J.A. Lieberman. Prediction of outcome in first-episode schizophrenia. J. Clin. Psychiatry. 1993;54:13-17 (Suppl.)
  • Lieberman et al., 1993 J. Lieberman, D. Jody, S. Geisler, J. Alvir, A. Loebel, S. Szymanski, M. Woerner, M. Borenstein. Time course and biologic correlates of treatment response in first-episode schizophrenia. Arch. Gen. Psychiatry. 1993;50(5):369-376 Crossref
  • Llorca, 2008 P.M. Llorca. Partial compliance in schizophrenia and the impact on patient outcomes. Psychiatry Res.. 2008;161(2):235-247 Crossref
  • Malla et al., 2008 A. Malla, R. Norman, L. Bechard-Evans, N. Schmitz, R. Manchanda, C. Cassidy. Factors influencing relapse during a 2-year follow-up of first-episode psychosis in a specialized early intervention service. Psychol. Med.. 2008;38(11):1585-1593 Crossref
  • Masand et al., 2009 P.S. Masand, M. Roca, M.S. Turner, J.M. Kane. Partial adherence to antipsychotic medication impacts the course of illness in patients with schizophrenia: a review. Prim. Care Companion J. Clin. Psychiatry. 2009;11(4):147-154 Crossref
  • McCreadie et al., 1989 R.G. McCreadie, D. Wiles, S. Grant, G.T. Crockett, Z. Mahmood, M.G. Livingston, J.A. Watt, J.G. Greene, P.W. Kershaw, N.A. Todd, et al. The Scottish first episode schizophrenia study. VII. Two-year follow-up. Scottish Schizophrenia Research Group. Acta Psychiatr. Scand.. 1989;80(6):597-602
  • Menezes et al., 2006 N.M. Menezes, T. Arenovich, R.B. Zipursky. A systematic review of longitudinal outcome studies of first-episode psychosis. Psychol. Med.. 2006;36(10):1349-1362 Crossref
  • Overall and Gorham, 1962 J. Overall, D. Gorham. The Brief Psychiatric Rating Scale. Psychol. Rep.. 1962;10(3):799-812
  • Rabiner et al., 1986 C.J. Rabiner, J.T. Wegner, J.M. Kane. Outcome study of first-episode psychosis. I: Relapse rates after 1 year. Am. J. Psychiatry. 1986;143(9):1155-1158
  • Ram et al., 1992 R. Ram, E.J. Bromet, W.W. Eaton, C. Pato, J.E. Schwartz. The natural course of schizophrenia: a review of first-admission studies. Schizophr. Bull.. 1992;18(2):185-207 Crossref
  • Rifkin et al., 1977 A. Rifkin, F. Quitkin, D.F. Klein. Fluphenazine decanoate, oral fluphenazine, and placebo in treatment of remitted schizophrenics. II. Rating scale data. Arch. Gen. Psychiatry. 1977;34(10):1215-1219 Crossref
  • Robinson et al., 1999 D. Robinson, M.G. Woerner, J.M. Alvir, R. Bilder, R. Goldman, S. Geisler, A. Koreen, B. Sheitman, M. Chakos, D. Mayerhoff, J.A. Lieberman. Predictors of relapse following response from a first episode of schizophrenia or schizoaffective disorder. Arch. Gen. Psychiatry. 1999;56(3):241-247 Crossref
  • Spitzer et al., 1978 R.L. Spitzer, J. Endicott, E. Robins. Research diagnostic criteria: rationale and reliability. Arch. Gen. Psychiatry. 1978;35(6):773-782 Crossref
  • Takeuchi et al., 2012 H. Takeuchi, T. Suzuki, H. Uchida, K. Watanabe, M. Mimura. Antipsychotic treatment for schizophrenia in the maintenance phase: a systematic review of the guidelines and algorithms. Schizophr. Res.. 2012;134(2–3):219-225 Crossref
  • Viguera et al., 1997 A.C. Viguera, R.J. Baldessarini, J.D. Hegarty, D.P. van Kammen, M. Tohen. Clinical risk following abrupt and gradual withdrawal of maintenance neuroleptic treatment. Arch. Gen. Psychiatry. 1997;54(1):49-55 Crossref
  • Wiersma et al., 1998 D. Wiersma, F.J. Nienhuis, C.J. Slooff, R. Giel. Natural course of schizophrenic disorders: a 15-year followup of a Dutch incidence cohort. Schizophr. Bull.. 1998;24(1):75-85 Crossref
  • Wistedt et al., 1982 B. Wistedt, A. Jorgensen, D. Wiles. A depot neuroleptic withdrawal study. Plasma concentration of fluphenazine and flupenthixol and relapse frequency. Psychopharmacology (Berl). 1982;78(4):301-304 Crossref
  • Wunderink et al., 2007 L. Wunderink, F.J. Nienhuis, S. Sytema, C.J. Slooff, R. Knegtering, D. Wiersma. Guided discontinuation versus maintenance treatment in remitted first-episode psychosis: relapse rates and functional outcome. J. Clin. Psychiatry. 2007;68(5):654-661 Crossref
  • Zipursky and Schulz, 2001 R.B. Zipursky, S.C. Schulz. The Early Stages of Schizophrenia. 1st ed. (American Psychiatric Pub, Washington, DC, 2001)
  • Zipursky et al., 2013 R.B. Zipursky, T.J. Reilly, R.M. Murray. The myth of schizophrenia as a progressive brain disease. Schizophr. Bull.. 2013; (Epub ahead of print, Dec 7, in press)


a Department of Psychiatry and Behavioural Neurosciences, Michael G. DeGroote School of Medicine, McMaster University, St. Joseph's Healthcare Hamilton, 100 West 5th Street, Hamilton, Ontario L8N 3K7, Canada

b Department of Psychiatry, Faculty of Medicine, University of Toronto, Toronto, Ontario, Canada

lowast Corresponding author. Tel.: + 1 905 522 1155x36250; fax: + 1 905 381 5633.