Medication discontinuation in first episode psychosis: thinking about the offset of psychotic disorders

After recovery from a first episode of psychosis, the obvious question is ‘when can I stop my medication?’ Given the importance of this decision, there is surprisingly little evidence to guide clinician and patient. We propose that first episode psychosis (FEP) services develop a specific focus on the consequences of antipsychotic medication dose reduction and discontinuation. This would enable systematic collection of information about long-term outcomes, ideally via a registry of FEP patients that includes data about medication, symptoms and functional remission over time.

Given the lack of evidence to guide decisions about maintenance medication after recovery from FEP, it is not surprising that there is a divergence of professional opinion. Some experts argue for an early diagnosis of schizophrenia and ongoing maintenance treatment with long-acting injectable antipsychotics. Others advocate a delay in definitive diagnosis, low doses of oral medication and medication reduction/cessation where possible. There is general agreement that antipsychotic medication is essential during an episode of acute psychosis. The uncertainty is about how long medication should be continued once the acute symptoms have remitted.

A Cochrane review in 2012 assessed 65 placebo-controlled randomised controlled trials (RCTs) with 6493 participants and demonstrated that, in people with schizophrenia or schizophrenia-like psychoses, antipsychotics were significantly more effective than placebo in preventing relapse at 7–12 months (27% relapse rate with medication vs 64% relapse rate with placebo). Long-term population studies generally show benefits for those taking antipsychotic drugs, compared to people with relevant disorders not taking antipsychotics, but these studies are limited by selection biases. People who take medication reliably are likely to have more insight and engage in other positive health-related behaviours.

However, the FEP population are not the same as people with schizophrenia; some people with FEP do not progress to schizophrenia but make a full recovery without recurrence of psychosis. The diagnosis can be unclear in the early stages of psychosis, and factors such as substance abuse or psychosocial stress contribute to the clinical picture.

A recent systematic review of six studies with 209 FEP participants estimated that the weighted mean 1-year symptom recurrence rate was 77% in the medication discontinuation group, compared to 3% in the continuation group (Zipursky et al., 2014). This seems decisive, but there are important limitations. Sample sizes were small, ranging from 7 to 89 in the medication discontinuation groups. Two studies did not have a comparison arm (medication continuation group), and four studies were not blinded. The duration of these studies was limited, with three studies having follow-up for 1 year and the longest study lasting only 3 years. In contrast to Zipursky’s findings, it has been estimated in a longitudinal cohort study that up to 40% of those who achieve remission after FEP may ultimately remain well without medication (Sturup et al., 2017).

The much debated study by Wunderink et al. (2013) involved people with FEP but has some important differences from studies included in reviews such as Zipursky et al. (2014). First, Wunderink et al. followed up at least some of the original participants for 7 years. Second, the main outcome was recovery, defined by meeting both symptomatic and functional remission criteria. This is crucial, as usually severity of psychotic symptoms is the key outcome in studies of maintenance antipsychotic medication. Data on functional outcomes are often not collected or are very limited. People with psychosis tend to rate having a job, an adequate income and good social relationships as more important than persisting symptoms, indicating that from a patient perspective, functional outcome is often prioritised ahead of symptom severity or relapse.

Wunderink et al. found that relapse rates were significantly higher in the reduction/discontinuation group compared to the maintenance group in the first 3 years, but after that, the rates came on par. After 7 years, a similar proportion of participants met the symptomatic remission criteria in each arm (69.2% in the dose reduction/discontinuation group and 66.7% in the maintenance group), but there was a marked difference in functional remission rates (46.2% and 19.6%, respectively), leading to the significant difference in the rate of recovery (40.4% vs 17.6%). There are many criticisms of this study: raters were not blinded, there was a high attrition rate and most patients in both arms of the study did receive medication.

Recently, Hui et al. (2018) reported 10-year follow-up data from 178 individuals with FEP in Hong Kong. After participants had achieved full remission of positive symptoms, they were blindly randomised to either maintenance quetiapine treatment or discontinuation, with placebo, for 12 months. Relapse rates were 41% for quetiapine and 79% for placebo. After this RCT, participants received naturalistic treatments. At the 10-year follow-up, individuals who were originally in the placebo/discontinuation group had a significantly increased risk of having poor outcome compared to the maintenance group (21% vs 39%). Poor outcome was defined as persistent positive symptoms, clozapine treatment, or death by suicide. However, the secondary outcome measures, including employment, independent living and social functioning, did not differ significantly between groups. The relapse rate in the medicated group during the RCT was unexpectedly high, perhaps because some had to switch to the trial medication, quetiapine, and switching can be associated with relapse. The quetiapine dose (400 mg/day) may have been too low for some patients. Hui et al. (2018) found that relapse in the first 3 years of illness is associated with a poorer outcome – so, they propose that relapse during this critical period may have a disease-modifying effect. They therefore recommend that maintenance treatment be continued for at least 3 years after recovery from FEP. There are some differences from the Australian FEP population which may limit generalisability – people with significant drug and alcohol use in the 3 months prior to randomisation were excluded from the Hong Kong study, and the rate of comorbid substance use at follow-up was very low (1%), whereas there are much higher rates of substance use in Australian FEP patients.

Two planned RCTs should provide much needed guidance. An Australian RCT (Killackey, 2017) will investigate the effects of guided antipsychotic dose reduction in young people with FEP. The longitudinal design of the study will enable measurement of associations between antipsychotic medication and brain changes. A Danish RCT plans to compare maintenance treatment with closely monitored tapering/discontinuation of antipsychotic medication over 1 year, in people with newly diagnosed schizophrenia or persistent delusional disorder who have had 3 months remission of psychotic symptoms (Sturup et al., 2017). While these RCTs should be informative, in the meantime, we are faced with decisions about the optimal dose and duration of antipsychotic medication for people recovering from FEP.

Clinical guidelines and clinical practice

The current RANZCP clinical practice guidelines (CPG) for schizophrenia and related disorders note that the decision to reduce and discontinue antipsychotic medication requires careful discussion and evaluation of illness severity, family history, specific risk factors and other circumstances of the individual. The person should have made a full recovery and been well for at least 12 months before medication cessation is considered. The CPG recommend medical supervision and engagement of family and friends so they know to watch for changes in mental state or social functioning. It is also suggested that the individual and people important in their day-to-day life collaborate in preparing an advanced directive rescue plan.

In the current service climate where people recovering from FEP are quickly mainstreamed to primary care services, many patients have been transferred to their primary care providers well before 12 months of symptom remission. Unplanned disengagement from maintenance treatment occurs much more frequently than well-planned discontinuation of antipsychotic medication, making it difficult to determine what really worked for those who stay well after medication cessation.

Likewise, it is difficult to know what psychosocial interventions are most effective when people are coming off medication. It seems obvious that such individuals would require more intensive psychosocial support to maintain symptomatic remission and functional recovery, but the more likely clinical scenario is that biological and psychosocial treatments are often withdrawn at the same time, in the same manner (i.e. abrupt and unsupervised) when people decide to discontinue their medication through disengagement. In fact, it could be argued that the current situation, where the most common way to discontinue medication is to stop taking it in an unsupervised, non-specialist environment, is putting our patients at risk, and is not truly consistent with the patient-centred care paradigm.

Building the bridge – the need for better monitoring and data collection on medication discontinuation?

More evidence is required to inform patients about their need for maintenance medication. The risks are high; antipsychotic medication can result in metabolic syndrome and movement disorders and cause sedation which impacts negatively on engagement in social and vocational activities. But, discontinuing medication can precipitate relapse, with a potentially serious impact on the patient, their family and the community. Hui et al. (2018) extend this concern by suggesting that early relapse after recovery leads to a poorer long-term outcome.

Clinically, we need to facilitate better accumulation of knowledge and experience. Given that many patients discontinue medication in an unplanned manner, perhaps there is a need for medication reduction/discontinuation clinics which include well-organised ongoing engagement and monitoring of those who have become non-adherent against advice. These clinics would provide close supervision and psychosocial support. In such a service model, trials of patient-driven medication reduction and discontinuation would be managed under specialist clinical supervision with involvement and shared decision-making by patients and their families.

Specialist services for the onset of psychotic disorders have led to an enhanced understanding of early psychoses, and we would argue that an increased focus on the offset of psychosis would lead to an improved understanding of how to best reduce and discontinue antipsychotic medication when indicated. Collection of long-term naturalistic data over a number of sites using agreed protocols would deliver invaluable information, enabling us explore clinically relevant issues such as identifying prognostic factors and providing evidence about which are the suitable patients for successful medication reduction and discontinuation.

We need to develop appropriate down-titration regimens for different medications, when this is appropriate, and effective psychosocial strategies to assist in the process. These would include stable housing, engagement in rewarding activities (preferably work or study), good support networks and psychological treatments to improve insight, illness self-management and addressing challenges effectively. In other words, the high-quality, ongoing psychosocial care all of our patients should receive. The flip side of this focus on predictors of good outcome would be early identification those with a poor outcome much earlier, so that systematic trials of treatment can be undertaken and clozapine can be started when appropriate.