Articles

Long-term effects of discontinuation from antipsychotic

maintenance following first-episode schizophrenia and
related disorders: a 10 year follow-up of a randomised,
double-blind trial
Christy L M Hui, William G Honer, Edwin H M Lee, Wing Chung Chang, Sherry K W Chan, Emily S M Chen, Edwin P F Pang, Simon S Y Lui,
Dicky W S Chung, Wai Song Yeung, Roger M K Ng, William T L Lo, Peter B Jones, Pak Sham, Eric Y H Chen

Summary
Background The long-term consequences of discontinuing antipsychotic medication after successful treatment of Lancet Psychiatry 2018
first-episode psychosis are not well studied. We assess the relation between early maintenance therapy decisions Published Online
in first-episode psychosis and the subsequent clinical outcome at 10 years. March 15, 2018
http://dx.doi.org/10.1016/
S2215-0366(18)30090-7
Methods This is a 10 year follow-up study, spanning Sept 5, 2003, to Dec 30, 2014, of a randomised, double-blind trial
Department of Psychiatry
in seven centres in Hong Kong in which 178 patients with first-episode psychosis with full positive symptom (C L M Hui PhD,
resolution after at least 1 year of antipsychotic treatment were given maintenance treatment (n=89; oral quetiapine E H M Lee MBChB,
400 mg daily) or early treatment discontinuation (n=89; placebo) for 12 months. After the trial, patients received W C Chang MBChB,
S K W Chan MBBS,
naturalistic treatment. Overall this cohort of patients will have received about 3 years of treatment before entering the
E S M Chen MPhil,
follow-up phase of the study: about 2 years of maintenance treatment before study entry and 1 year of treatment in the Prof P Sham PhD,
trial. The primary outcome of this follow-up was the proportion of patients in each group (including those for whom Prof E Y H Chen MD), State Key
direct follow-up was not available) with good or poor long-term clinical outcomes at 10 years, with poor outcome Laboratory of Brain and
Cognitive Sciences (W C Chang,
defined as a composite of persistent psychotic symptoms, a requirement for clozapine treatment, or death by suicide.
S K W Chan, Prof P Sham,
The randomised trial was registered with ClinicalTrials.gov, number NCT00334035, and the follow-up study was Prof E Y H Chen), and Centre for
registered with ClinicalTrials.gov, number NCT01926340. Genomic Sciences
(Prof P Sham), University of
Hong Kong, Hong Kong Special
Findings Poor 10 year clinical outcome occurred in 35 (39%) of 89 patients in the discontinuation group and 19 (21%) of
Administrative Region, China;
89 patients in the maintenance treatment group (risk ratio 1·84, 95% CI 1·15–2·96; p=0·012). Suicide was the only Department of Psychiatry,
serious adverse event that occurred in the follow-up phase (four [4%] patients in the early discontinuation group vs University of British Columbia,
two [2%] in the maintenance group). Canada (Prof W G Honer MD);
Department of Psychiatry,
United Christian Hospital,
Interpretation In patients with first-episode psychosis with a full initial response to treatment, medication continuation Hong Kong Special
for at least the first 3 years after starting treatment decreases the risk of relapse and poor long-term clinical outcome. Administrative Region, China
(E P F Pang MBChB);
Department of Psychiatry,
Funding Food and Health Bureau, Research Grants Council of Hong Kong, and AstraZeneca. Castle Peak Hospital,
Hong Kong Special
Introduction clozapine, the only medication with convincing efficacy Administrative Region, China
National clinical practice guidelines broadly agree that in treatment-resistant schizophrenia,9 but which is (S S Y Lui MBBS); Department of
Psychiatry, Tai Po Hospital,
antipsychotic medication is indicated for the acute associated with substantial side-effects. Hong Kong Special
treatment and initial maintenance of first-episode One report combines findings from an RCT of dose Administrative Region, China
psychosis. However, subsequent decisions to continue reduction in the early phase of illness with a follow-up (D W S Chung MBChB);
medication, reduce dose, or discontinue medication are study of long-term clinical outcome.10,11 In the follow-up, Department of Psychiatry,
Pamela Youde Nethersole
not well informed by available randomised controlled patients were reassessed 7 years after the start of the Eastern Hospital, Hong Kong
trials (RCTs). On the one hand, naturalistic studies raise original trial. A greater proportion of patients randomly Special Administrative Region,
concerns about the long-term side-effects of maintenance assigned to the early dose-reduction group had good China (W S Yeung MBBS);
medication.1,2 On the other hand, a small number of functional outcome (24 [46%] of 52 patients) than patients Department of Psychiatry,
Kowloon Hospital, Hong Kong
RCTs suggest that dose reduction or medication randomly assigned to the treatment as usual group Special Administrative Region,
discontinuation results in an increase in relapse.3 (ten [20%] of 51 patients). However, sustained medication China (R M K Ng MBChB);
However, the relation between relapse and long-term discontinuation was achieved in only 14 (22%) of Department of Psychiatry, Kwai
Chung Hospital, Hong Kong
outcome is not well studied. Some evidence suggests the 65 patients in the dose-reduction group, whereas
Special Administrative Region,
immediate consequences of relapse are limited,4 whereas 21 (32%) of 65 patients relapsed and the remainder were China (W T L Lo MBBS); and
others raise the concern that each relapse might be unable to discontinue medications, complicating the Department of Psychiatry,
associated with a poorer treatment response.5–7 Poor interpretation of the long-term outcomes.10,11 The University of Cambridge,
Cambridge, UK
clinical response is a cardinal feature of treatment- implications of early relapse were not investigated. The
(Prof P B Jones MD)
resistant psychosis8 and is the key indication for absence of adverse consequences, or perhaps even

www.thelancet.com/psychiatry Published online March 15, 2018 http://dx.doi.org/10.1016/S2215-0366(18)30090-7 1

 Articles

Correspondence to:
Dr Christy L M Hui, Department
of Psychiatry, University of
Hong Kong, Queen Mary
Hospital, Hong Kong Special
Administrative Region, China
christy@lmhui.com
Research in context
Evidence before this study
Comprehensive programmes for identifying and treating
first-episode psychosis have been implemented in many
countries to improve the health of young people and prevent
long-term disability associated with chronic illness.
Antipsychotic medication is an integral part of the treatment of
first-episode psychosis, and full remission of symptoms can be
obtained. However, the decision of whether or not to
discontinue antipsychotic medication after a sustained period
of remission is a common clinical problem. In 2012,
a systematic review of clinical guidelines available at that time
concluded that medication should continue for at least
12–24 months after remission from first-episode schizophrenia.
For clinical guidelines published between Jan 1, 2012, and
Dec 31, 2017, we searched the PubMed database using the
search terms “schizophrenia” and “guideline”, and conducted a
manual search of the reference list of relevant guidelines found.
Inclusion criteria were that the guidelines had to be developed
using a systematic and defined process, contain
recommendations about pharmacological treatment for
first-episode schizophrenia, and be written in English.
Nine relevant guidelines were found, of which five contained a
recommendation on the duration of medication maintenance.
In accordance with guidelines published before 2012, these
five guidelines also gave a recommendation of 12–24 months.
However, because of the scarcity of data concerning the
long-term consequences of early medication maintenance or
discontinuation, all such guidelines are ambivalent in their
recommendations for longer periods beyond 24 months.
The only 2 year, open-label, randomised trial of discontinuation
versus maintenance treatment in first-episode psychosis
reported outcomes at 7 years. The dose reduction group had
better psychosocial function than the maintenance treatment
group; no difference was found in terms of severity of psychotic
symptoms. However, in the 2 year initial phase of that trial,
sustained medication discontinuation was achieved in only
20% of participants, whereas approximately 30% relapsed and
the remainder were unable to discontinue medications,
complicating the interpretation of the long-term outcomes.
Added value of this study
This study is the first double-blind, randomised trial to address
the long-term consequences of the clinical decision to
discontinue antipsychotic medication treatment over a nearly
10-year follow-up period. Early discontinuation of medication
(year 2) for patients with full remission of key positive
symptoms of psychosis was associated with a higher risk of
long-term, poor clinical outcome of schizophrenia and related
psychotic illness than maintenance treatment. Mediation
analysis suggested that the consequences of early
discontinuation for long-term poor clinical outcome were
mediated in part through early relapse during the 12 month
period of the randomised trial.
Implications of all available evidence
Discontinuation of medication within the 3 years after starting
treatment in patients with first-episode psychosis (ie, from when
patients started maintenance treatment before study entry
[22 months on average] to their completion of the 12 month trial
period) should be carefully considered with awareness of the
potential costs in terms of outcome. Patients who opt for
discontinuation should be monitored with adequate support.
In patients with first-episode psychosis with a full initial response
to treatment, medication continuation for at least the first 3 years
after starting treatment is effective in preventing relapse and
decreasing the risk for a poor long-term clinical outcome.

advantages of early dose-reduction, have important
implications for patient care.
Many patients with first-episode psychosis have an
encouraging initial treatment response.12 A full
resolution of positive symptoms of psychosis presents a
compelling case for medication discontinuation after an
initial maintenance period of at least 1 year, as
recommended by many clinical practice guidelines. We
previously reported the short-term relapse outcomes of
a 1 year, randomised, double-blind, placebo-controlled
trial13 of early discontinuation compared with
maintenance treatment with quetiapine 400 mg daily in
such a group of patients. Patients were required to be
free of positive symptoms of psychosis and have already
had at least 1 year of antipsychotic treatment before
randomisation. The Kaplan–Meier estimate of the
proportion relapsed was 79% (95% CI 68–90) in the early
discontinuation group, and 41% (29–53) in the main­
tenance treatment group (p<0·0001).13 After the study,
patients received naturalistic treatment. We now report a
10 year follow-up study of these patients to determine
the effect of early medication discontinuation on long-
term clinical outcomes. Maintenance treatment during
the RCT was hypothesised to be less likely to be
associated with poor long-term outcome. We also
investigate the contribution of early relapse to the long-
term outcome.
Methods
Study design and participants
This 10 year follow-up study, spanning Sept 5, 2003, to
Dec 30, 2014, of an RCT13 investigating the effects of early
discontinuation of antipsychotic medication on risk for
relapse of positive symptoms of psychosis, was designed to
examine the long-term consequences of early medication
discontinuation for clinical outcome. The original RCT
was a multisite study involving seven psychiatric hospitals
in Hong Kong and was conducted under the Early
Assessment Service for Young People with Psychosis. In
the original RCT,13 a higher proportion of patients in the

2 www.thelancet.com/psychiatry Published online March 15, 2018 http://dx.doi.org/10.1016/S2215-0366(18)30090-7


placebo group had relapse at 12 months than in the
maintenance group.13 The study protocol is available in the
appendix.
The inclusion criteria for the RCT were a diagnosis of
schizophrenia or non-affective psychosis (schizophreniform
disorder, schizoaffective disorder, brief psychotic disorder,
or psychosis not otherwise specified) by use of criteria from
DSM-IV.14 Diagnostic assessments used the Chinese
version of the structured clinical interview for DSM-IV.
Eligible patients were aged 18–65 years, had received
antipsychotic medication continuously for at least
12 months, were free of positive symptoms of psychosis,
and had no history of relapse (no increase of positive
symptoms of psychosis requiring admission to hospital or
adjustment of medication).
Exclusion criteria were treatment with clozapine, depot
antipsychotic medication, or mood stabilising therapy
(lithium, valproate, or carbamazepine); poor adherence
to treatment (missing >50% of drug, >50% missed clinic
visits, or a history of medication discontinuation); or a
risk of suicide or violence. All patients from the original
study were traced and invited to participate in the present
study. Participants provided written informed consent.
Randomisation and masking
In the original RCT, patients were randomly assigned
to early discontinuation (placebo) or maintenance
treatment. Details of the randomisation and masking can
be found in the original study.13
The longitudinal review used case-note and centralised
records of longitudinal, monthly data extracted by
research assistants masked to the randomised trial
assignment. The first research assistant checked for
relevant information in the case-notes and redacted
these, before passing to the second research assistant for
rating.
Procedures
In the original RCT, patients with no residual psychotic
symptoms who already had received at least 12 months of
antipsychotic treatments were randomly assigned to early
discontinuation (placebo) or maintenance treatment (oral
quetiapine 400 mg daily; mean daily dose 411·9 mg
[SD 63·5]) for 12 months. Before starting the RCT, patients
had been administered antipsychotic medication for a
mean of 21·9 months (SD 8·2; table 1). Participation in
the RCT ended after 1 year or if the patient relapsed or
withdrew for other reasons, whichever came first. Overall
this cohort of patients with first-episode psychosis will
have received about 3 years of treatment in total before
entering the follow-up phase of the study: about 2 years of
maintenance treatment before study entry and 1 year of
treatment in the trial. After the trial, patients received
naturalistic treatment. Specialised teams in Hong Kong
provided the first 3 years of care for patients after their
first episode of psychosis; subsequent care was provided
by general services. We attempted to trace the 178 patients
www.thelancet.com/psychiatry Published online March 15, 2018 http://dx.doi.org/10.1016/S2215-0366(18)30090-7 3
Articles
randomly assigned to a group in the original RCT, giving
them a 10 year follow-up through face-to-face interview
and obtaining chart reviews. Direct personal assessments See Online for appendix
were done at the follow-up interview by a Master’s level
research assistant to obtain the primary outcome
component of persistent positive symptoms of psychosis.
We also did longitudinal chart reviews for all patients to
identify death by suicide or clozapine treatment as primary
outcome components. For any remaining patients for
whom death by suicide or clozapine treatment were not
identified as primary outcome components and for whom
face-to-face interviews were not possible for obtaining
data on persistent positive symptoms, we used positive
symptom assessment information from the RCT to
predict outcomes.
Medication adherence was assessed with clinical
interviews and pill counting during the RCT. Adherence
was assessed with a standardised scale during the
follow-up.15
We adapted the existing dose recommendation tables
to convert the antipsychotic medications to chlorpro-​
mazine equivalents.16–18
Outcomes
The primary outcome of this follow-up study was the
proportion of patients in each group with good or poor
long-term clinical outcomes at the 10 year follow-up. A poor
long-term clinical outcome was defined by persistent
positive symptoms of psychosis, requirement for clozapine
treatment (treatment-resistant schizophrenia), or death by
suicide. Positive symptoms of psychosis in the previous
1 month of follow-up were assessed by face-to-face interview
using the Positive and Negative Syndrome Scale.19 The
same thresholds were used as in the RCT to define presence
of positive symptoms (delusions ≥3, conceptual
disorganisation ≥4, hallucinations ≥3, suspiciousness ≥5,
or unusual thought content ≥4; appendix).13 The last
positive symptom assessment (using the Positive and
Negative Syndrome Scale) from the randomised phase was
used in our analyses for patients who could not be followed
up. The other two indicators of poor long-term outcome,
initiation of clozapine for treatment resistance at any time,
and death by suicide at any time, were assessed through the
medical record review. Absence of all three of these was
deemed to indicate a good long-term clinical outcome.
After the initial submission of the data analysis protocol,
but before data collection, the primary outcome measure
was modified on Dec 12, 2013, from “recovery”, an
aggregate measure composed of clinical outcome
(symptoms), functioning, and hospital admission defined
using the Strauss & Carpenter Scale to a composite clinical
outcome (persistent positive symptoms, use of clozapine
for treatment resistance, or death by suicide), which was
defined as the primary outcome to align the present study
with the primary outcome of symptom remission in the
original RCT,13 as well as with previously available and
subsequently published data supporting the concept that
 Articles

long-term outcome consists of several independent
dimensions.20,21
Secondary outcome assessments from direct ass­
essment were occupational status and scores on the
Social and Occupational Functioning Assessment Scale22
and the Role Functioning Scale.23 Health-related quality
of life was assessed as an additional outcome (36-item
short-form health survey).24
Serious adverse events assessed throughout the 10-year
course were death, hospital admission, or persistent

Maintenance treatment Early discontinuation All patients

(n=89) (n=89) (n=178)
Type of disorder at final review
Schizophrenia 67 (75%) 67 (75%) 134 (75%)
Schizophreniform disorder 10 (11%) 12 (13%) 22 (12%)
Schizoaffective disorder 3 (3%) 2 (2%) 5 (3%)
Brief psychotic disorder 7 (8%) 2 (2%) 9 (5%)
Delusional disorder 1 (1%) 5 (6%) 6 (3%)
Psychosis not otherwise specified 1 (1%) 1 (1%) 2 (1%)
Substance misuse at end of naturalistic open treatment
Alcohol abuse* 2/74 (3%) 0/64 (0%) 2/138 (1%)
Other drug abuse* 0/74 (0%) 0/63 (0%) 0/137 (0%)
At start of randomised trial
Age (years) 23·5 (5·2) 24·9 (7·3) 24·2 (6·4)
Men 39 (44%) 41 (46%) 80 (45%)
Education (years) 12·0 (2·6) 11·7 (3·1) 11·8 (2·8)
Employed 62 (70%) 65 (73%) 127 (71%)
PANSS score†
Total 33·0 (4·5) 34·0 (6·3) 33·5 (5·5)
Positive 7·2 (0·6) 7·3 (0·7) 7·2 (0·6)
Negative 8·7 (2·9) 9·1 (3·6) 8·9 (3·3)
Previous open-label antipsychotic duration (months) 22·4 (8·6) 21·3 (7·8) 21·9 (8·2)
Previous antipsychotic dose at baseline (mg/day)‡ 177·4 (165·9) 125·9 (84·1) 151·6 (133·6)
Duration of untreated psychosis (days)§ 88·0 (38·0–155·0) 120·0 (40·0–378·0) 91·0 (39·8–230·8)
Randomised trial
Duration (months) 6·2 (4·8) 4·9 (4·1) 5·6 (4·5)
Longitudinal antipsychotic dose (mg/day)‡ 558·9 (31·1) 40·2 (66·1) 299·5 (265·1)
Naturalistic, open treatment
Duration (months) 105·7 (18·5) 107·0 (17·7) 106·4 (18·1)
Longitudinal antipsychotic dose (mg/day)‡ 209·0 (178·8) 245·5 (195·8) 227·3 (187·8)
End of open-treatment assessment¶
Type of medication
None 17/72 (24%) 11/66 (17%) 28/138 (20%)
Typical antipsychotic only 11/72 (15%) 10/66 (15%) 21/138 (15%)
Atypical antipsychotic only 41/72 (57%) 42/66 (64%) 83/138 (60%)
Both typical and atypical antipsychotic 3/72 (4%) 3/66 (5%) 6/138 (4%)
Antipsychotic dose at end of follow-up (mg/day)‡|| 353·8 (256·7) 356·1 (313·9) 355·0 (285·4)
Medication adherence behaviour** 3·5 (0·5) 3·6 (0·5) 3·5 (0·5)
Medication adherence attitude** 3·2 (0·6) 3·4 (0·7) 3·3 (0·6)

Data are n (%), mean (SD), or median (IQR), unless stated otherwise. PANSS=Positive and Negative Syndrome Scale. Differences between the maintenance and discontinuation
groups were not significant. *Data were available in patients who had an end of open-treatment assessment. †PANSS measures the severity of symptoms in schizophrenia with a
range of 30 to 210, with higher scores indicating more severe symptoms. ‡The mean daily dose of each antipsychotic was converted to an equivalent dose of chlorpromazine.
§Duration of untreated psychosis was defined as the interval between onset of psychotic symptoms and first receipt of antipsychotic treatment. Duration of untreated psychosis
was assessed using the Interview for the Retrospective Assessment of the Onset of Schizophrenia, which is a standardised, semi-structured interview conducted on patients and
their close relatives within 4 months following the first episode. A review of the case notes was also done as collateral information to minimise recall bias. ¶Data are missing for
four of 142 patients with an end of open treatment assessment: 72 in the maintenance group and 66 in the early discontinuation group. ||For patients taking medication.
**The modified Medication Adherence Rating Scale was used to capture behaviours and attitude towards medication in the previous month. The first five items capture patients’
compliance behaviours in a 4-point scale, ranging from 1 (never) to 4 (always). The remaining nine items refer to patients’ attitude towards taking medication, ranging from 1
(strongly disagree) to 5 (strongly agree). Higher scores indicate better adherence behaviour and a more positive attitude. Data were available in patients who had completed the
end of open-treatment assessment: 60 in the maintenance treatment group and 54 in the early discontinuation group.

Table 1: Clinical and treatment characteristics of the patients

4 www.thelancet.com/psychiatry Published online March 15, 2018 http://dx.doi.org/10.1016/S2215-0366(18)30090-7

 Articles

incapacity. Each event was assessed for a relation to
treatment during the randomised trial. Side-effects were
assessed directly at the follow-up study visit, including
movement disorders assessed with three scales.25–27
Because antipsychotic drugs can cause metabolic
dysregulation, weight, height, fasting glucose, total
cholesterol, and HDL cholesterol were measured.
General side-effects were assessed using a standardised
scale.28
Statistical analysis
For the primary outcome analysis, we used risk ratios [RR]
and 95% CIs to compare the frequency of long-term poor
outcomes between all patients in the intervention groups
of the RCT, irrespective of whether direct follow-up was
available. We did sensitivity analyses to assess the use of
positive symptom assessment information from the RCT
to predict outcomes in patients for whom chart reviews
could not be obtained, reclassifying these remaining
patients as either having all good outcomes or all poor
outcomes. We also assessed the validity of our definitions
for good and poor clinical outcomes with a series of
comparisons of symptomatic and functional measures
the early discontinuation group), we used positive
symptom assessment information from the RCT to predict
outcomes. The inter-rater reliability for positive symptom
ratings (intraclass correlation), between three raters on ten
independent cases, was 0·99 for delusions, 1·00 for
conceptual disorganisation, 0·88 for hallucinations,
0·99 for suspiciousness, and 0·95 for unusual thought
content. At 10 year follow-up, there were no significant
differences across all variables at baseline between the
intervention groups (table 1).
After completion of the RCT, patients received
naturalistic treatment, with antipsychotic drug treatment
1606 patients assessed for eligibility
1428 excluded
1134 did not meet inclusion criteria
384 different diagnosis
332 relapsed
284 active psychosis
96 non-adherent
38 suicidal or other risk
294 refused to participate

between the outcome groups, irrespective of assignment

during the randomised trial. We compared secondary 178 randomly assigned to a group
outcome measures using RR or independent t tests.
To determine if relapse accounted for in whole or in
part the effects of early treatment discontinuation on
long-term outcome, we did a mediation analysis using 89 assigned to maintenance 89 assigned to early
treatment and received discontinuation and received
logistic regression models and a Sobel test. We also quetiapine placebo
calculated the percentage of the total effect being
mediated by the indirect effect of relapse.29 For serious
28 discontinued trial 18 discontinued trial
adverse events, we calculated standardised mortality 15 adverse events 6 adverse events
ratios based on age–gender population mortality rate and 1 serious adverse 1 serious adverse
age–gender death by suicide rate. All statistical analyses event event
6 withdrew consent 6 withdrew consent
were done using IBM SPSS, version 24.0. 6 other 5 other
The randomised trial was registered with ClinicalTrials.
gov, number NCT00334035, and the follow-up study was
61 completed randomised phase 71 completed randomised phase
registered with ClinicalTrials.gov, number NCT01926340.

Role of funding source 89 entered open-treatment and 89 entered open-treatment and

The funders of the study had no role in study design, given longitudinal chart review given longitudinal chart review
86 followed up 84 followed up
data collection, data analysis, data interpretation, or 3 lost to follow-up 5 lost to follow-up
writing of the report. CLMH and EYHC had full access to
all the data in the study, and the corresponding author
had final responsibility for the decision to submit for 15 did not complete 21 did not complete
endpoint interview endpoint interview
publication. 12 declined interview 16 declined interview
assessment assessment
2 died by suicide 4 died by suicide
Results 1 unable to contact 1 unable to contact
We interviewed and obtained chart reviews for 142 (80%) of
178 patients who participated in the original RCT13
(figure 1). For the 36 remaining patients, we also did 74 completed endpoint interview 68 completed endpoint interview

longitudinal chart reviews, identifying death by suicide or

clozapine treatment as primary outcome components in 89 included in intention-to-treat 89 included in intention-to-treat
eight patients (two from the maintenance group and six analysis at 10 years analysis at 10 years
from the early discontinuation group). For the remaining
28 patients (13 from the maintenance group and 15 from Figure 1: Trial profile

www.thelancet.com/psychiatry Published online March 15, 2018 http://dx.doi.org/10.1016/S2215-0366(18)30090-7 5

 Articles

Early maintenance treatment Early discontinuation p value Effect size*

(n=89) (n=89)
Composite good primary outcome† 70 (79%) 54 (61%) 0·012 1·84 (1·15–2·96)
Composite poor primary outcome† 19 (21%) 35 (39%) ·· ··
Required clozapine 3 (3%) 8 (9%) 0·13 2·76 (0·73–10·39)
Suicide 2 (2%) 4 (4%) 0·42 2·00 (0·37–10·93)
Persistent psychosis 16 (18%) 24 (27%) 0·11 1·57 (0·90–2·74)
Delusions 8 (9%) 18 (20%) 0·030 2·36 (1·08–5·13)
Conceptual disorganisation 2 (2%) 2 (2%) 1·00 1·05 (0·15–7·27)
Hallucinations 11 (12%) 16 (18%) 0·24 1·52 (0·75–3·09)
Suspiciousness 5 (6%) 6 (7%) 0·70 1·26 (0·40–3·97)
Unusual thought content 2 (2%) 4 (4%) 0·39 2·10 (0·39–11·14)
Secondary outcomes‡
Employed 51/74 (69%) 48/68 (71%) 0·65 0·02
SOFAS§ 61·9 (9·6) 64·0 (8·9) 0·19 –0·23
RFS total¶ 21·3 (3·4) 21·5 (2·8) 0·72 –0·06
RFS work 5·4 (1·1) 5·5 (1·1) 0·73 –0·09
RFS independent living 5·7 (1·0) 5·8 (0·9) 0·72 –0·11
RFS immediate social 5·6 (1·0) 5·5 (0·9) 0·88 0·11
RFS extended social 4·7 (1·2) 4·8 (1·0) 0·64 –0·09
Other outcomes
PANSS||
Total 34·8 (5·9) 35·4 (6·6) 0·55 –0·10
Positive 7·7 (2·3) 8·4 (3·1) 0·16 –0·30
Negative 8·1 (2·1) 7·9 (1·9) 0·62 0·10
Global clinical impression**
Severity of illness 1·5 (1·1) 1·7 (1·1) 0·45 –0·13
Improvement 3·9 (0·5) 3·9 (0·7) 0·96 0
Calgary Depression Scale†† 1·1 (2·6) 1·0 (2·4) 0·87 0·04
SUMD‡‡ 1·3 (0·6) 1·3 (0·6) 0·59 0·09
SF-36:MCS§§ 50·2 (9·1) 51·0 (8·4) 0·61 –0·09
SF-36:PCS§§ 56·6 (7·6) 56·9 (6·6) 0·80 –0·04

Data are n (%) or mean (SD), unless stated otherwise. Risk ratio was used for categorical variables; hazard ratio (Cox regression) was used for time to event variables
(requiring clozapine and suicide); independent t test was used for continuous variables. SOFAS=Social and Occupational Functioning Assessment Scale. RFS=Role
Functioning Scale. PANSS=Positive and Negative Syndrome Scale. SUMD=Scale to Assess Unawareness of Mental Disorder. SF-36=36-item short-form health survey.
MCS=mental component summary. PCS=physical component summary. *Risk ratio (95% CI) was used for the primary clinical outcome. For other measures, we used
Cramer’s V for categorical variables, Cohen’s d (normally distributed) and Rosenthal’s r (non-normally distributed) for continuous variables. Effect sizes were categorised as
small (0·2–0·4), medium (0·5–0·7), or large (≥0·8). †The primary long-term outcome measure was defined categorically, with poor outcome including any of the following:
persistent positive symptoms of psychosis, requirement for clozapine, or death by suicide; absence of all three of these was deemed to indicate a good long-term clinical
outcome. ‡Secondary outcomes from end of open treatment assessments were available for 74 patients in the early maintenance treatment group and 68 patients in the
early discontinuation group. §SOFAS assesses social and occupational functioning in patients with psychiatric illness using a range of scores from 1 to 100, with lower
scores representing impaired functioning. ¶RFS measures the functioning level in patients with psychiatric disorders, focusing on four domains: work, independent living
and self-care, immediate social network relationships, and extended social network relationships. Each domain is rated on a 7-point scale, with lower scores representing
lower functioning. ||PANSS measures the severity of symptoms in schizophrenia with a range of 30 to 210, with higher scores indicating more severe symptoms. PANSS has
a total of 30 items, with seven items on positive symptoms, seven on negative symptoms, and 16 on general psychopathology. Data were obtained from 74 patients in the
early maintenance treatment group and 68 patients in the early discontinuation group. **Global clinical impression assesses the global symptoms in patients with
psychotic disorders and is consisted of two individual scores—namely, the severity of illness (ranging from 1 [not mentally ill] to 7 [extremely ill]) and improvement
(ranging from 1 [very much improved] to 7 [very much worsened]). Data were obtained from 75 patients in the early maintenance treatment group and 68 patients in the
early discontinuation group. ††The Calgary Depression Scale for schizophrenia assesses the depressive level in schizophrenia, exclusive of other dimensions of
psychopathology. The scale consists of nine items rated on a 4-point scale from absent (0), mild (1), moderate (2), to severe (4). Data were obtained from 74 patients in the
early maintenance treatment group and 67 patients in the early discontinuation group. ‡‡The abridged SUMD measures insight in schizophrenia. It consists of three items:
global awareness of mental disorders, awareness of the consequences of mental disorder, and awareness of the effects of medication; with a rating of 1 indicating
awareness, 2 indicating somewhat aware or unaware, and 3 indicating severely unaware. A global mean score was computed by summing up the three items, with higher
scores indicating poor awareness. Data were obtained from 75 patients in the early maintenance treatment group and 66 patients in the early discontinuation group.
§§SF-36 consists of 36 items on functional health and wellbeing. The scores can be classified into the MCS (scores ranging from 14 to 70; data obtained from 70 patients in
the early maintenance treatment group and 58 patients in the early discontinuation group) and the PCS (scores ranging from 21 to 74; data were obtained from 71 patients
in the early maintenance treatment group and 58 patients in the early discontinuation group). Higher scores indicate better functional health and wellbeing.

Table 2: Primary-composite, secondary, and other long-term outcome measures

6 www.thelancet.com/psychiatry Published online March 15, 2018 http://dx.doi.org/10.1016/S2215-0366(18)30090-7


received for a mean duration of 106·4 months (SD 18·1;
table 1). 110 (80%) of 138 patients with complete
medication data at follow-up continued to take
antipsychotic medication at the follow-up assessment,
with a mean total dose of 355·0 mg (SD 285·4) of
chlorpromazine equivalents per day. Self-reported
adherence was high. Aside from the difference in
exposure during the RCT, no differences in duration or
doses of antipsychotic drug treatment between the early
discontinuation and maintenance groups over the
10 years of naturalistic treatment were reported. Similarly,
clinical care received did not differ between the two
intervention groups (appendix).
Poor long-term clinical outcome occurred in 35 (39%) of
89 patients randomly assigned to the discontinuation
group versus in 19 (21%) of 89 patients assigned to the
maintenance group (RR 1·84, 95% CI 1·15–2·96;
p=0·012; table 2). Sensitivity analyses indicated similar
results if analyses were limited to patients available for
direct follow-up assessment (150 of 178 patients; RR 2·05,
1·15–3·68; p=0·016) or if the remaining 28 patients were
classified as either poor outcome (RR 1·58, 1·06–2·34;
p=0·023) or good outcome (RR 2·00, 1·10–3·63;
p=0·023).
The initially positive, symptom-free patients recruited
in the present study represent a relevant group to answer
the clinical question of maintenance discontinuation. To
provide an indication of the long-term outcome of a
broader range of patients with first-episode psychosis,
we did a post-hoc analysis to compare patients with first-
episode psychosis whose initial treatment response was
less favourable than would have been included in this
study (ie, patients with at least 1 month of mild positive
symptoms or who had already relapsed) with the
maintenance and discontinuation groups (appendix).16
The early discontinuation group in our study had a
similar high prevalence of poor long-term clinical
outcomes (35 [39%] of 89 patients) as this initially less
favourable cohort (33 [49%] of 68 patients) after 10 years
(RR 0·80, 95% CI 0·56–1·14; p=0·21), whereas the
maintenance group had fewer long-term poor clinical
outcomes (19 [21%] of 89; RR 0·43, 0·27–0·69;
p<0·0001).
We also did a post-hoc analysis to investigate the
relation between poor long-term clinical outcome and
measures of symptomatic and functional outcome
(appendix). Our direct assessment of psychosis at the
outcome timepoint was a one-time face-to-face
assessment. The presence of psychosis (persistent
positive symptoms above threshold) was the largest
contributor to poor clinical outcome. Patients with poor
long-term clinical outcomes also had more severe overall
symptom scores than patients with good long-term
clinical outcomes (p<0·0001), more months with
psychosis present during the entire follow-up period
(p=0·0030) indicated by chart review, a lower score on
independent living (p=0·034), and lower mental health
www.thelancet.com/psychiatry Published online March 15, 2018 http://dx.doi.org/10.1016/S2215-0366(18)30090-7 7
Articles
A Model 1
β 0·87 (SE 0·34);
Placebo or maintenance
p=0·010
Long-term clinical
treatment outcome
B Model 2
β 1·36 (SE 0·32); p<0·0001 β 1·56 (SE 0·38); p<0·0001
Relapse of positive
symptoms of psychosis
Placebo or maintenance Long-term clinical
treatment outcome
β 0·43 (SE 0·37);
p=0·24
Figure 2: Mediation models
Model 1 shows the direct effect of treatment during the randomised trial on
long-term clinical outcome using a logistic regression analysis. Model 2 shows
the effects of testing for relapse during the randomised trial as a mediator of the
effects of the randomised treatment on long-term clinical outcome. The results
of a Sobel test indicate significant mediation was present (p=0·0030).
component scores on the 36-item short-form health
survey measure of health-related quality of life
(p=0·0030).
We did a binary logistic regression analysis to assess
whether diagnosis (schizophrenia vs non-schizophrenia)
would affect the significant relation between treatment
and outcome. Treatment group was a significant factor
related to outcome (odds ratio [OR] 0·41, 95% CI
0·21–0·80; p=0·0090): schizophrenia diagnosis was not
(OR 1·85, 0·95–3·60; p=0·070). There were no significant
differences between long-term clinical outcome and the
baseline measures of age (t=0·953; p=0·342), gender
(χ²=1·966; p=0·161), years of education (t=–0·202;
p=0·840), employment (χ²=0·070; p=0·792), medication
dose taken (t=–0·814; p=0·417), or schizophrenia
diagnosis (χ²=0·018; p=0·893).
There were no significant differences in the secondary
outcomes between the early discontinuation and
maintenance treatment groups (table 2).
As previously reported, the early discontinuation group
had more relapses during the RCT.13 However, the clinical
characteristics of relapses, including severity, duration,
and resolution of positive symptoms, did not differ
between the intervention groups (appendix), and these
relapse characteristics did not differentiate between the
good and poor long-term outcome groups (appendix).
We did a post-hoc mediation analysis of the effects of
relapse and of treatment group during the RCT on risk for
long-term poor clinical outcome (figure 2; appendix). The
total effect between treatment group and long-term
clinical outcome was significant (β 0·87 [SE 0·34];
p=0·010). Treatment group was a significant predictor of
relapse during the RCT (β 1·36 [SE 0·32]; p<0·0001), and
that relapse was a significant predictor of long-term
outcome (β 1·56, [SE 0·38]; p<0·0001). The direct effect
between treatment and outcome was no longer significant
after controlling for the mediator of relapse (β 0·43
[SE 0·37]; p=0·24). The estimated indirect effect of relapse
 Articles

on treatment group and outcome was significant,
suggesting that the increased risk for poor long-term
outcome associated with early discontinuation was
mediated through the increased risk of relapse during the
RCT (Sobel test, t=2·97 [SE 0·71]; p=0·0030). The
proportion of the total effect of the treatment group on
outcome that was mediated by the indirect effect of relapse
was 58% (0·1362 [indirect effect]/0·2337 [total effect]).
Death by suicide was the only serious adverse event and
did not differ between treatment intervention groups
(table 2). During the 1661 person-years of observation,
six (3%) of 178 participants died, all by suicide
(four patients in the discontinuation group and two in the
maintenance group; all during the follow-up period). The
crude mortality rate was 3·6 deaths per 1000 person-
years. Compared with age-matched and gender-matched
Hong Kong population data, the standardised mortality
ratio was 5·64 (95% CI 2·29–11·73; p<0·0001). The
suicide-specific standardised mortality ratio was
31·82 (12·90–179·13; p<0·0001) in the whole population,
40·4 (12·84–97·46; p<0·0001) in the early discontinuation
group, and 21·0 (3·60–70·89; p<0·0001) in the early
maintenance group.
For participants assessed directly at the 10 year
follow-up, no significant differences in movement
disorders, metabolic measures, or other side-effects were
reported (table 3).
Discussion
Early discontinuation of antipsychotics after 1 year of
maintenance was associated with a higher risk of poor
long-term clinical outcome at 10 years. A post-hoc
analysis suggested that the adverse consequences of early
discontinuation were mediated in part through early
relapse during the 1 year period following medication
discontinuation.
This study addresses a pressing question (medication
discontinuation) in an important population (patients
with first-episode psychosis who responded well to
treatment with no residual psychotic symptoms), in a
context that goes beyond available guidelines and data
(broadly suggesting at least 1 year of treatment).

Early maintenance treatment Early discontinuation (n=89) p value

(n=89)
n Measure n Measure
Serious adverse event
Death 89 2 (2%) 89 4 (4%) 0·42
Number of admissions to hospital following randomisation 89 1·0 (0·0–2·0) 89 0·0 (0·0–2·0) 0·78
Duration of hospital admission following randomisation (days)† 89 55·0 (23·5–106·0) 89 71·0 (23·5–193·0) 0·38
Side-effects of medication
Simpson-Angus Scale‡ 67 0 (0·0) 63 0·0 (0·1) 0·32
Abnormal Involuntary Movement Scale§ 66 0·0 (0·1) 63 0·3 (1·5) 0·13
Barnes Akathisia Rating Scale¶ 66 0·3 (1·0) 63 0·3 (1·2) 0·82
Weight (kg) 71 71·7 (16·9) 60 71·6 (14·2) 0·99
Body-mass index|| 71 26·4 (5·6) 60 26·1 (5·0) 0·73
Fasting glucose (mg/dL)** 65 97·3 (28·8) 59 100·9 (37·8) 0·57
Cholesterol total (mg/dL)†† 64 189·5 (34·8) 57 189·5 (36·7) 0·88
HDL cholesterol (mg/dL) 60 46·4 (11·6) 55 50·3 (11·6) 0·28
UKU Side-Effects Rating Scale‡‡ (≥mild)
Sleepiness or sedation 59 22 (37%) 60 25 (42%) 0·40
Failing memory 59 15 (25%) 60 11 (18%) 0·78
Weight gain 59 20 (34%) 60 15 (25%) 0·17
UKU Global Side-Effects (≥mild)
Assessed by patient 59 21 (36%) 60 15 (25%) 0·20
Assessed by physician 59 12 (20%) 60 13 (22%) 0·65

Data are n (%), mean (SD), or median (IQR), unless stated otherwise. UKU=Udvalg for Kliniske Undersøgelser. *Other adverse events and side-effects during the randomised trial
phase were reported in detail previously. Risk ratio was used for categorical variables; independent t tests were used for continuous variables. †Reported for 89 patients who were
admitted to hospital only: 45 in the early maintenance treatment group and 44 in the early discontinuation group. ‡The Simpson-Angus Scale ranges from 0 (normal) to
4 (severe), with a higher score indicating more severe extrapyramidal side-effects. §The Abnormal Involuntary Movement Scale ranges from 0 to 4, with higher scores indicating
more severe tardive dyskinesia. ¶The Barnes Akathisia Rating Scale ranges from 0 to 5, with higher scores indicating more severe akathisia. ||The body-mass index is the weight
in kilograms divided by the square of the height in metres. **To convert values for blood glucose to millimoles per litre, multiply by 0·05551. ††To convert values for total
cholesterol, HDL cholesterol, and LDL cholesterol to millimoles per litre, multiply by 0·02586. ‡‡The numbers and percentages of patients who had any of 42 side-effects scored
as 1 (mild) or higher according to the UKU, a side-effects rating scale. According to this scale, each item is rated from 0 to 3, with higher scores indicating more severe side-effects.
Movement-disorder side-effects are not shown because these were assessed according to more detailed scales above. Only side-effects that occurred in at least 20% of the
patients in a treatment group are reported. For all side-effects, there were no significant differences between the groups (all p values >0·05).

Table 3: Serious adverse events, movement disorders, metabolic measures, and general side-effects*

8 www.thelancet.com/psychiatry Published online March 15, 2018 http://dx.doi.org/10.1016/S2215-0366(18)30090-7


Notably, the clinical outcome started to diverge in the
cohorts included in this study from the very first episode
of psychosis. Some patients did not achieve resolution of
psychotic symptoms30 and some had relapsed within
months.5 In the real-world setting, medication
discontinuation is generally not recommended for these
groups.22,31 For patients who attained symptom resolution,
present guidelines suggest maintenance treatment for at
least 1 year,22,31 but do not offer more specific guidance for
after this point because of the paucity of long-term
outcome studies.
In an open-label study,10,11 Wunderink and colleagues
randomly assigned a broad range of patients with early
psychosis (including those with persistent psychotic
symptoms corresponding to a score of moderate
[rating of 4] using the Positive and Negative Syndrome
Scale), and found a dose reduction or discontinuation
strategy might be beneficial to the long-term functional
outcome. However, the issue of medication
discontinuation was not addressed directly because
discontinuation was implemented in only half of the
patients. In that study, only 45% of patients had a
diagnosis of schizophrenia, and 35% had comorbid
substance dependence or abuse. The patients differed
from the present study in having only 6 months of
maintenance treatment at entry.
In our double-blind RCT,13 we focused on patients
with good initial outcome (no residual psychotic
symptoms) and investigated the effect of discontinuation
because this is generally consistent with patients’
preferences. Our patients had no comorbid substance
abuse, most fulfilled the diagnosis of schizophrenia
(75%), and all had completed at least 1 year of
maintenance treatment. The present study addresses
the 10 year outcome of this cohort. Our primary
outcome was clinical (persistent positive symptoms,
need for clozapine, and death by suicide), aligned with
previous recommendations,7 suggesting that persistent
positive symptoms were the largest contributor to poor
long-term outcome.32,33 Notably, our study excluded
patients who were non-responsive to treatment from
the outset,34,35 and some of the included patients
developed a poor response later, after initially good
responses. This late poor response necessitated
clozapine treatment in some patients, with its
associated side-effects and risks.
Mediation analysis supported early relapse as a mediator
of the long-term effect of treatment discontinuation. This
finding indicates that relapse might itself alter the course
of a psychotic illness. In this regard, it is important to
recognise that 80% of patients in both intervention groups
eventually had a relapse during the 10 year period.
However, relapse early in the second to third year during
the course of illness, following treatment discontinuation,
seemed to carry more adverse implications for long-term
outcome, raising the suggestion that there might be a
time window or critical period during which a relapse
www.thelancet.com/psychiatry Published online March 15, 2018 http://dx.doi.org/10.1016/S2215-0366(18)30090-7 9
Articles
might be course-modifying. In the group of patients
studied in this Article, we found no evidence that any
specific clinical features of the relapse such as severity,
need for hospital admission, or time to resolution
contributed to long-term outcome.
The main limitation of our study is the extent to which
conclusions can be generalised. A low proportion of
the patients with first-episode psychosis studied in
Hong Kong had comorbid substance abuse and
consequently the cohort would be likely to have more
homogeneity than an equivalent cohort studied
elsewhere.
We focused on patients with complete symptom
resolution after the first episode because in practice this
is the group in whom an understandable expectation
about medication discontinuation arose. To put the
present findings in a wider first-episode psychosis
context, we compared data from the present RCT cohort
with a cohort of patients with less satisfactory initial
outcome (thus excluded from the RCT). Patients with
early symptom resolution randomly assigned to
maintenance in our study fared best. Patients with
early symptom resolution randomly assigned to
discontinuation had less favourable long-term outcome,
similar to patients without early symptom resolution.
A high proportion of our patients were employed
(71% during the RCT and 70% at follow-up), which is
different from cohorts in other countries. The high
employment might be related to the fact that we
specifically targeted a highly responsive cohort of patients
with first-episode psychosis, in addition to the cultural
expectation among Chinese patients to participate in the
workforce like other people in the community.
Non-adherence ratings based on pill counting during
the RCT and patients’ self-reports at follow-up might
overestimate true adherence. Owing to the more
collectivistic culture, with more family involvement in
facilitating medication adherence, non-adherence in
Hong Kong was lower than in other countries.
Additionally, the original RCT excluded patients who
required depot medication to ensure adherence because
the design involved randomisation to an oral medication
or placebo. Future studies designed to use depot
medication, and depot placebo, would help address
these important considerations concerning adherence.
A possible consequence of using a standardised
maintenance treatment (quetiapine) in the RCT could be
some early relapses in the maintenance group related to
switching medication.
Finally, although our study is the largest yet reported
of an RCT testing the effects of early medication
discontinuation, we cannot be certain of the possible
effects of schizophrenia versus other diagnoses of
psychotic illness on the effects of early relapse on long
term outcome, investigation of which will require larger
studies with careful documentation of diagnoses over
time.
 Articles
A 2017 review and commentary highlight the increasing
interest in the long-term consequences of antipsychotic
drug treatment.2,36 One consequence of long-term
antipsychotic medications in animals, and possibly in
some patients, is the development over time of dopamine
receptor hypersensitivity, or a so-called supersensitivity
psychosis. Characteristics of this syndrome are a severe
rebound of symptom severity after switching or
discontinuing medication, and the development of
tardive dyskinesia. The prevalence of supersensitivity
psychosis is uncertain, and although individual patients
might exhibit this pattern of symptomatology, substantial
evidence suggests that the more usual pattern of
response or changes in response to maintenance
antipsychotic medications over time is not consistent
with the supersensitivity phenomena.2,36 Our study was
not designed to investigate the possibility of
supersensitivity psychosis in detail.
In summary, our data are relevant to patients who have
symptomatic resolution following 1 year’s treatment of
first-episode psychosis (predominantly schizophrenia)
and wish to consider medication discontinuation after a
period of maintenance. The data suggest that an
awareness of the increased risk for poor long-term
outcome, in addition to the risk for relapse in the short-
term, must be taken into consideration in this decision.
In patients with first-episode psychosis with a full initial
response to treatment, medication continuation for at
least the first 3 years after starting treatment (about
2 years of maintenance treatment before study entry and
1 year of treatment in the trial) is effective in preventing
relapse and decreasing the risk for a poor long-term
outcome.
Contributors
CLMH, WGH, and EYHC designed the study, analysed the data, and
wrote and revised the Article. CLMH, EHML, WCC, SKWC, and EYHC
obtained funding and supervised the study. ESMC collected and
analysed the data and revised the Article. CLMH, EHML, WCC, SKWC,
and EYHC collected the data and revised the Article. EPFP, SSYL,
DWSC, WSY, RMKN, WTLL, and PBJ revised the Article. PS analysed
the data and revised the Article. CLMH and EYHC are guarantors.
Declaration of interests
WGH has received consultation fees from Otsuka, Lundbeck,
AphaSights, and Eli Lilly. WTLL has participated as a paid consultant for
Jansen. EYHC has received speaker honoraria from Otsuka and DSK
BioPharma, received research funding from Otsuka, participated in paid
advisory boards for Jansen and DSK BioPharma, and received funding to
attend conferences from Otsuka and DSK BioPharma; EYHC reports no
other relationships or activities that could appear to have influenced the
submitted work. All other authors declare no competing interests.
Acknowledgments
The randomised-treatment phase of the study was supported by the
Research Grants Council of Hong Kong (7655/05M) and AstraZeneca
(investigator-initiated study award). AstraZeneca prepared the quetiapine
and the placebo, and packaged the study medications according to the
randomisation schedule. The follow-up study was supported by the Food
and Health Bureau of Hong Kong (10111101). WGH was supported by
the Jack Bell Chair in Schizophrenia.
References
1 Correll C. Acute and long-term adverse effects of antipsychotics.
CNS Spectrums 2007; 12: 10–14.
10 www.thelancet.com/psychiatry Published online March 15, 2018 http://dx.doi.org/10.1016/S2215-0366(18)30090-7
2 Goff DC, Falkai P, Fleischhacker WW, et al. The long-term effects of
antipsychotic medication on clinical course in schizophrenia.
Am J Psychiatry 2017; 174: 840–49.
3 Zipursky RB, Menezes NM, Streiner DL. Risk of symptom
recurrence with medication discontinuation in first-episode
psychosis: a systematic review. Schizophrenia Res 2014;
152: 408–14.
4 Gitlin M, Nuechterlein K, Subotnik KL, et al. Clinical outcome
following neuroleptic discontinuation in patients with remitted
recent-onset schizophrenia. Am J Psychiatry 2001; 158: 1835–42.
5 Emsley R, Chiliza B, Asmal L, Harvey BH. The nature of relapse in
schizophrenia. BMC Psychiatry 2013; 13: 50.
6 Wiersma D, Nienhuis FJ, Slooff CJ, Giel R. Natural course of
schizophrenic disorders: a 15-year followup of a Dutch incidence
cohort. Schizophrenia Bull 1998; 24: 75–85.
7 Lieberman JA, Alvir JM, Koreen A, et al. Psychobiologic correlates
of treatment response in schizophrenia.
Neuropsychopharmacol 1996; 14 (suppl 3): 13S–21S.
8 Howes OD, McCutcheon R, Agid O, et al. Treatment-resistant
schizophrenia: treatment response and resistance in psychosis
(TRRIP) working group consensus guidelines on diagnosis and
terminology. Am J Psychiatry 2017; 174: 216–29.
9 Kane JM. Treatment of schizophrenia. Schizophrenia Bull 1987;
13: 133–56.
10 Wunderink L, Nienhuis FJ, Sytema S, Slooff CJ, Knegtering R,
Wiersma D. Guided discontinuation versus maintenance treatment
in remitted first-episode psychosis: relapse rates and functional
outcome. J Clin Psychiatry 2007; 68: 654–61.
11 Wunderink L, Nieboer RM, Wiersma D, Sytema S, Nienhuis FJ.
Recovery in remitted first-episode psychosis at 7 years of follow-up
of an early dose reduction/discontinuation or maintenance
treatment strategy: long-term follow-up of a 2-year randomized
clinical trial. JAMA Psychiatry 2013; 70: 913–20.
12 Zhu Y, Li C, Huhn M, et al. How well do patients with a first
episode of schizophrenia respond to antipsychotics: a systematic
review and meta-analysis. Eur Neuropsychopharmacol 2017;
27: 835–44.
13 Chen EYH, Hui CLM, Lam MML, et al. 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.
14 American Psychiatric Association. Diagnostic and statistical manual
of mental disorders: DSM-IV. Washington: American Psychiatric
Association, 1994.
15 Thompson K, Kulkarni J, Sergejew AA. Reliability and validity of a
new Medication Adherence Rating Scale (MARS) for the psychoses.
Schizophrenia Res 2000; 42: 241–47.
16 American Psychiatric Association. Practice guideline for the
treatment of patients with schizophrenia, 2nd edn. Washington, DC:
American Psychiatric Association, 2004.
17 Woods SW. Chlorpromazine equivalent doses for the newer atypical
antipsychotics. J Clin Psychiatry 2003; 64: 663–67.
18 Taylor D, Paton C, Kerwin R. The Maudsley prescribing guidelines,
10th edn. UK: CRC press, 2009.
19 Kay SR, Fiszbein A, Opler LA. The positive and negative syndrome
scale (PANSS) for schizophrenia. Schizophrenia Bull 1987; 13: 261–76.
20 Chan SKW, So HC, Hui CLM, et al. 10-year outcome study of an
early intervention program for psychosis compared with standard
care service. Psychol Med 2015; 45: 1181–93.
21 Revier J, Reininghaus U, Dutta R, et al. Ten-year outcomes of
first-episode psychoses in the MRC AESOP-10 study.
J Nerv Ment Dis 2015; 203: 379–86.
22 Goldman HH, Skodol AE, Lave TR. Revising axis V for DSM-IV:
a review of measures of social functioning. Am J Psychiatry 1992;
149: 1148–56.
23 Goodman SH, Sewell DR, Cooley EL, Leavitt N. Assessing levels of
adaptive functioning: the Role Functioning Scale.
Community Ment Health J 1993; 29: 119–31.
24 Ware JE, Kosinski M, Keller SD. SF-36 physical and mental health
summary scales: a user’s manual. Boston, MA: The Health Institute,
New England Medical Center, 1994.
25 Simpson GM, Angus JW. A rating scale for extrapyramidal side
effects. Acta Psychiatr Scand Suppl 1970; 212: 11–19.

26 Guy W. Abnormal involuntary movements scale (AIMS).
In: Guy W, ed. The ECDEU assessment manual for
psychopharmacology, revised edn. Washington, DC: Department of
Health, Education and Welfare, National Institute of Mental Health,
1976: 534–37.
27 Barnes TR. A rating scale for drug-induced akathisia.
Br J Psychiatry 1989; 154: 672–76.
28 Lingjaerde O, Ahlfors UG, Bech P, Dencker SJ, Elgen K. The UKU
side effect rating scale. A new comprehensive rating scale for
psychotropic drugs and a cross-sectional study of side effects in
neuroleptic-treated patients. Acta Psychiatr Scand Suppl 1987;
334: 1–100.
29 MacKinnon DP, Dwyer JH. Estimating mediated effects in
prevention studies. Eval Rev 1993; 17: 144–58.
30 Emsley R, Rabinowitz J, Medori R et al. Remission in early
psychosis: rates, predictors, and clinical and functional outcome
correlates. Schizophrenia Res 2007; 89: 129–39.
31 Royal Australian and New Zealand College of Psychiatrists Clinical
Practice Guidelines Team for the Treatment of Schizophrenia and
Related Disorders. Royal Australian and New Zealand College of
Psychiatrists clinical practice guidelines for the treatment of
schizophrenia and related disorders. Aust NZJ Psychiatry 2005;
39: 1–30.
www.thelancet.com/psychiatry Published online March 15, 2018 http://dx.doi.org/10.1016/S2215-0366(18)30090-7 11
Articles
32 Jordan G, Lutgens D, Joober R, Lepage M, Iyer SN, Malla A.
The relative contribution of cognition and symptomatic remission
to functional outcome following treatment of a first episode of
psychosis. J Clin Psychiatry 2014; 75: e566–72.
33 Haro JM, Altamura C, Corral R, et al. Understanding the impact of
persistent symptoms in schizophrenia: Cross-sectional findings
from the Pattern study. Schizophrenia Res 2015; 169: 234–40.
34 Meltzer H. Treatment-resistant schizophrenia—the role of clozapine.
Curr Med Res Opin 1997; 14: 1–20.
35 Lally J, Ajnakina O, Di Forti M, et al. Two distinct patterns of
treatment resistance: clinical predictors of treatment resistance in
first-episode schizophrenia spectrum psychoses. Psychol Med 2016;
46: 3231–40.
36 Leucht S, Davis JM. Do antipsychotic drugs lose their efficacy for
relapse prevention over time? Br J Psychiatry 2017; 211: 127–29.