Schizophrenia Research 153 (2014) 1–8

Contents lists available at ScienceDirect

Schizophrenia Research journal homepage: www.elsevier.com/locate/schres

Prospective relationship between duration of untreated psychosis and 13-year clinical outcome: A first-episode psychosis study Jennifer Yee-Man Tang a,⁎, Wing-Chung Chang a, Christy Lai-Ming Hui a, Gloria Hoi-Yan Wong a, Sherry Kit-Wa Chan a, Edwin Ho-Ming Lee a, Wai-Song Yeung b, Chi-Keung Wong b, Wai-Nang Tang b, Wah-Fat Chan b, Edwin Pui-Fai Pang c, Steve Tso d, Roger Man-Kin Ng e, Se-Fong Hung f, Eva Lai-Wah Dunn b, Pak-Chung Sham a,g, Eric Yu-Hai Chen a,g a

Department of Psychiatry, The University of Hong Kong, Hong Kong Department of Psychiatry, Pamela Youde Nethersole Eastern Hospital, Hospital Authority, Hong Kong c Department of Psychiatry, United Christian Hospital, Hospital Authority, Hong Kong d Department of Psychiatry, Castle Peak Hospital, Hospital Authority, Hong Kong e Department of Psychiatry, Kowloon Hospital, Hospital Authority, Hong Kong f Department of Psychiatry, Kwai Chung Hospital, Hospital Authority, Hong Kong g The State Key Laboratory of Brain and Cognitive Sciences, The University of Hong Kong, Hong Kong b

a r t i c l e

i n f o

Article history: Received 12 June 2013 Received in revised form 1 January 2014 Accepted 22 January 2014 Available online 11 February 2014 Keywords: Psychotic disorders Long-term outcome Prospective study Remission Duration of untreated psychosis Risk factors

a b s t r a c t Background: The adverse effects of a long duration of untreated psychosis (DUP) have been explored in numerous short-term studies. These studies support the development of early interventions that reduce treatment delay and promote recovery. However, the enduring impact of DUP is largely unknown, partly due to the paucity of prospective long-term studies. Although the DUP–outcome relationship is commonly assumed to be linear, the threshold effect has not been adequately examined. Objective: To explore the relationship between DUP and long-term symptomatic remission. Methods: This was a prospective study of a cohort of 153 first-episode psychosis patients in Hong Kong at the 13-year follow-up. The patients were categorized into short (≤ 30 days), medium (31–180 days) and long (N 180 days) DUP groups. Results: The long-term outcome was ascertained in 73% of the patients. Nearly half of the patients (47%) fulfilled the criteria for symptomatic remission. The short DUP group experienced a significantly higher remission rate over the course of the illness. The odds of long-term symptomatic remission was significantly reduced in the medium DUP (by 89%) and long DUP (by 85%) groups compared with the short DUP group. Further analysis showed that DUP had a specific impact on negative symptom remission. Conclusion: The findings support the threshold theory that DUP longer than 30 days adversely impacts the longterm outcome. The present study is one of the few studies that confirmed the enduring impact of DUP on longterm outcomes based on well-defined criteria and adequate statistical adjustment. © 2014 Elsevier B.V. All rights reserved.

1. Introduction The duration of untreated psychosis (DUP) is one of the most widely studied risk factors associated with poor outcomes such as treatment response, positive symptoms, negative symptoms and functioning (Marshall et al., 2005; Perkins et al., 2005). The effect of DUP on the outcome was supported by evidence from numerous short-term studies, but the follow-up periods in these studies were 2 years or less. In agreement with the critical period hypothesis for patients with first-episode psychosis (Birchwood, 1999), these studies supported the development of early detection programmes that aim to shorten treatment delay and ⁎ Corresponding author at: Department of Psychiatry, The University of Hong Kong, Pok Fu Lam Road, Hong Kong. Tel.: +852 2255 3064; fax: +852 2855 1345. E-mail address: [email protected] (J.Y.-M. Tang). 0920-9964/$ – see front matter © 2014 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.schres.2014.01.022

improve long-term recovery. Despite the general acceptance of early interventions for first-episode psychosis patients, the enduring impact of DUP remains unclear, which is partly due to the lack of prospective studies with longer follow-up periods. Potentially, the adverse impact of DUP may be gradually weakened by other intervening factors over the course of the illness. Alternatively, this adverse impact would not manifest during the early stage because the majority of the patients with early psychosis respond well to medications (Chang et al., 2012). Our understanding of the impact of the association between DUP and long-term outcome is very limited. Our existing knowledge about the relationship between DUP and long-term symptomatic remission has been based mainly on a few prospective studies. Many of these studies were limited by the different pre-treatments, the poorly defined pretreatment variables and the lack of key covariate adjustments, especially premorbid adjustment (Table 1). The relationship between DUP and

2

Table 1 Summary of prospective long-term studies (≥10 years) examining the relationship between DUP and long-term outcome. The pre-treatment variable

Definition

Long-term outcome as predicted by a longer DUP predicted

Remarks

Huber et al. (1980)

Treatment during their initial psychotic episodes

Uncharacteristic and characteristic residues

Results not adjusted for any covariates

Tsoi and Kua (1992)

Pre-admission duration of illness Duration of illness

“Poor outcome” (based on treatment and severity of psychotic symptoms) Poorer outcome (based on psychotic symptoms and course)

Results not adjusted for any covariates

Thara and Eaton (1996)

Harrison et al. (1996) Wiersma et al. (1998)

Duration of untreated illness prior to contact with the services Delayed mental health treatment

Wiersma et al. (2000)

Psychosis without treatment/DUP

Bottlender et al. (2003)

First psychotic symptoms to first admission

“Treated” or “non-treated” with electroconvulsive or psychopharmacological therapy during the initial psychotic episode Method to determine the time of the onset of illness was unclear Duration of illness before first contact with treatment facility; method to determine the time of the onset of illness was unclear Method to determine the time of the onset of illness was unclear The outbreak of psychosis to the initiation of mental health treatment. Subjects were categorized into “delayed” or “prompt” treatment group. Unclear cutoff value. The estimated onset of psychotic symptoms to the first contact with a mental health professional The onset of psychotic symptoms to the first psychiatric admission; assessed by clinical interview.

Kurihara et al. (2005) White et al. (2009)

DUP DUP

Hegelstad et al. (2012)

DUP

Hill et al. (2012)

DUP

Unclear The onset of first psychotic symptoms to index admission The onset of the first psychotic symptoms to the start of first adequate treatment The onset of the first psychotic symptom to the start of antipsychotic treatment

A range of poorer clinical and functional outcomes Shorter time for full remission and longer duration of first episode

Unclear reference group. Results controlled for the effect of onset type but not premorbid adjustment. Unknown magnitude of association. Results controlled for the effect of onset type but not premorbid adjustment. Results controlled for the effect of premorbid functioning and onset type.

Higher level of disability

Insignificant bivariate relationship with outcome. No multiple regression was performed.

Lower GAS and higher severity on SANS, PANSS positive subscale and PANSS general psychopathology.

Adjusted for only onset type. Premorbid adjustment had insignificant bivariate relationship with outcome, and thus was not used as a covariate. Categorisation of DUP did not account for very short duration. Results not controlled for the effect of premorbid adjustment Results controlled for the effect of premorbid adjustment

Did not predict PANSS severity and social adjustment Greater outcome symptom burden A higher proportion of patients from the early detection group had recovered relative to those from the usual-detection group More severe positive and negative symptoms and a range of functional measures

Comparison of results between subjects in the early detection group and usual-detection group. Results not controlled for the effect of premorbid adjustment Results controlled for the effect of premorbid adjustment

DUP = duration of untreated psychosis; GAS = Global Assessment Scale; PANSS = Positive and Negative Syndrome Scale; SANS = Scale for the Assessment of Negative Symptoms.

J.Y.-M. Tang et al. / Schizophrenia Research 153 (2014) 1–8

Study (year)

J.Y.-M. Tang et al. / Schizophrenia Research 153 (2014) 1–8

outcome is assumed to be linear, but this has not been adequately confirmed. Besides, an explanatory model using a log transformed DUP is not intuitively appealing. Alternatively, the threshold theory argues that a DUP beyond a critical period would substantially undermine the likelihood of a favourable outcome (Singh, 2007). These two contrasting views have different implications for identifying groups of patients who are more likely to benefit from early interventions and for estimating the resources required to obtain an improved outcome. This prospective study examined the role of DUP on the long-term outcome in a cohort of 153 patients with first-episode psychosis at the 13-year follow-up. The study aimed to identify early predictors for long-term symptomatic remission, in particular the impact of DUP. The findings should shed light on whether the DUP–outcome relationship, if any, favours the threshold effect or linear effect hypotheses. 2. Methods 2.1. Study settings A cohort of 153 patients with first-episode psychosis participated in a 3-year prospective study of neurocognitive impairment in Hong Kong between 1997 and 2000 (Chen et al., 2005; Chang et al., 2011). It was estimated that most of the first-episode patients would be incepted through the public hospitals as the public healthcare system was heavily subsidised. Participants were recruited from two public hospitals in Hong Kong, The Queen Mary Hospital and Pamela Youde Nethersole Eastern Hospital, which serve approximately 1.3 million people in the two catchment areas. Ethics approval was obtained from the local institutional review boards. 2.2. Participants Patients presenting with first-episode psychosis at the above hospitals were screened consecutively for participation in the study. The inclusion criteria were as follows: a diagnosis of schizophrenia, schizophreniform disorder, schizoaffective disorder, brief reactive psychosis not otherwise specified or affective disorder with psychotic symptoms according to the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV); had received antipsychotic treatment for less than 1 week; were aged 15 to 55 years at the study entry; and were Cantonesespeaking Han Chinese. Patients who had significant neurological disorders, severe learning disability or substance-induced psychotic disorders were excluded from the study. All participants received general psychiatric care after the initial presentation at the public hospitals. The follow-up study involved a direct interview with the patient. Collateral information from informants and medical records were considered supplementary. Written informed consent was obtained from participants for the follow-up study. 2.3. Tracing strategy The tracing followed a standardized protocol. Patients who were attending their regular psychiatric consultations were invited to participate in this follow-up study during their visit to the clinic. Patients who were no longer receiving public psychiatric care were invited to participate via telephone. Targeted patients were considered lost cases if multiple tracing attempts were unsuccessful. 2.4. Assessments Baseline demographic information including gender, age at study entry, age at illness onset, living arrangements (living alone or with others), marital status, education level in years, family history of psychotic disorders, smoker, and unemployment status were recorded. The Premorbid Adjustment Scale (PAS) was administered to assess the

3

attainment of developmental goals during childhood (up to age 11 years) and early adolescence (age, 12–16 years), specifically in relation to functioning before onset of psychosis (Cannon-Spoor et al., 1982). The four assessed areas in each developmental period were sociability, peer relations, scholastic performance, and adaptation to school. A high score (between 0 and 1) indicated greater impaired premorbid functioning (van Mastrigt and Addington, 2002). The DUP, defined as the time between the onset of positive psychotic symptoms and treatment initiation, was assessed by the Interview for the Retrospective Assessment of the Onset of Schizophrenia (IRAOS) (Häfner et al., 1992), which is an instrument incorporating information from patients, relatives and medical records. The mode of illness onset, which was also assessed by the IRAOS, refers to the time that a person develops frank psychosis from a normal condition. The mode of onset was grouped into the three categories of acute (≤ 1 week), sub-acute (1–4 weeks) and insidious (N 4 weeks). Baseline positive, negative and depressive symptoms were measured using the Positive and Negative Syndrome Scale (PANSS) positive subscale (Kay et al., 1987), the High Royds Evaluation of Negativity Scale (HEN) (Mortimer et al., 1989) and the Montgomery and Asberg Depression Rating Scale (MADRS) (Montgomery and Asberg, 1979), respectively. The diagnosis was assessed by the Chinese-bilingual Structured Clinical Interview for DSM-IV (Axis I, Patient version) (CB-SCID-I/P) (So et al., 2003). Baseline neurocognition was assessed using a battery of instruments, including logical memory immediate and delayed recall (verbal memory) (Hong Kong Psychological Society, 1989a), visual reproduction immediate and delayed recall (visual memory) (Hong Kong Psychological Society, 1989a), Modified Wisconsin Card Sorting Test (reasoning and problem solving) (Nelson, 1976) and motor soft signs (Chen et al., 1995). In addition, the Digit Span Forward and Information subtests of the Wechsler Adult Intelligence Test (WAIS-R-HK) (Hong Kong Psychological Society, 1989b) were administered to assess working memory and verbal/language skills, respectively. Symptomatic remission was defined as a symptom severity of no more than mild (≤3) based on selected PANSS items: delusions, unusual thought content, hallucinatory behaviour, conceptual disorganization, mannerisms/posturing, blunted affect, social withdrawal and lack of spontaneity (Andreasen et al., 2005). Symptomatic remission in patients was assessed at 1, 2, 3, and 13 years. Patients were divided into those having positive symptom remission and negative symptom remission based on the selected PANSS items. Psychopathological outcome at 13 years was assessed using PANSS and MADRS. The researcher (J.Y.M.T.), who conducted the follow-up interviews at 13 years, was blinded to the baseline and early outcome data. 2.5. Statistical analyses To detect any potential selection bias, baseline characteristics were compared between the completers and the non-completers of the follow-up at 13 years using chi-square tests or t-tests, as appropriate. Patients were categorized into short (0–30 days), medium (31–180 days) or long (N 180 days) DUP groups. Logistic regression using the generalised estimating equation (GEE) method was applied to explore changes in the remission status over the study period in relation to DUP length (Fig. 1). A set of logistic regression models (without GEE) were applied to identify baseline predictors for long-term symptomatic remission, positive symptom remission and negative symptom remission. We also intended to identify predictors for recovery (symptomatic remission with adequate functioning), but the number of recovered patients (n = 16) was too small to achieve sufficient statistical power. Analyses were thus not carried out. Univariate logistic regression was used to identify baseline independent variables potentially associated with the long-term outcome (p b 0.1). The selected variables were subsequently forced into the multivariate logistic regression model. DUP, the primary exposure of interest, was forced into the multivariate model even though no univariate association was found. Odds ratio

4

J.Y.-M. Tang et al. / Schizophrenia Research 153 (2014) 1–8

Percentage of remitted patients

90 80 70 60 50 40 30 20 10 0 Year 1

Year 2

Short DUP (median 14 d; IQR 17 - 30 d)

Year 3

Year 13

Medium DUP (median 120 d; IQR 75 - 180 d)

Long DUP (median 730 d; IQR 365 - 1278 d)

Fig. 1. Percentage of patients in remission over the course of the illness, grouped by short, medium and long DUP.

was used to describe the magnitude and direction of the association between the predictor and the categorical dependent variable. The area under the receiver operating characteristic (AUROC) curve, a standard measure of discrimination, was used to evaluate the ability of a model to predict the outcome. For all statistical analysis, the significance level was set at 0.05 unless otherwise specified. Statistical analyses were performed using StataIC 12 (Stata Corp, Texas Station, Texas).

3. Results 3.1. Attrition Of the 153 patients, two were excluded from the follow-up study because their diagnosis was revised to psychotic disorder due to substance abuse shortly after the baseline assessment. Among the 151 participants, 14 died during the follow-up period (4 from suicide, 3 from suspected suicide, 3 from neoplastic disease, 4 from unknown causes), 34 were lost to follow-up and seven refused assessment. Ninety-six patients participated in the follow-up study after giving written informed consent. The average length of follow-up period was 13.1 years (range, 10.7– 14.5), and the long-term outcome was ascertained in 73% (96 + 14/ 151) of the whole sample and 70% of the living subjects (96/137). To account for potential selection bias, baseline demographic characteristics were compared between the consenting patients (n = 96) and non-completers (n = 55) of this follow-up study. The two groups were highly comparable and did not differ in demographics, premorbid adjustment, illness onset, baseline symptom severity and baseline cognitive performance. We concluded that the participants who completed the assessments were representative of the whole sample.

3.2. Sample characteristics The mean age of the study population was 31.7 years (SD = 9.1) at study intake, and there were more females than males (83:70). The characteristics of the study population are shown in Table 2. The DSMIV diagnoses were schizophrenia (66%), brief psychotic disorder (9%), affective psychosis (8%), schizophreniform disorder (7%), delusional disorder (6%), schizoaffective disorder (2%) and psychotic disorder not otherwise specified (1%). More than half the patients (56%) were medication-naïve at the time of study intake, while most of the other patients had only received first generation antipsychotic drugs for less than 1 week at a median daily dose of 163 mg (IQR, 82–272) of

chlorpromazine equivalents. More than half of the patients (60%; 60/100) achieved symptomatic remission at the end of 3 years. The median DUP was 180 days (IQR, 30–730). The median values for the short, medium, and long DUP groups were 14 days (17–30), 120 days (75–180), and 730 days (365–1278), respectively. The short, medium, and long DUP groups significantly differed in terms of gender (male: 63% vs 35% vs 44%; χ2 = 6.4, df = 2, p = 0.040), mode of illness onset (insidious onset: 16% vs 41% vs 83%; χ2 = 49.7, df = 4, p b 0.001), schizophrenia diagnosis (35% vs 85% vs 75%; χ 2 = 26.1, df = 2, p b 0.001), unemployment rate (23% vs 43% vs 52%; χ2 = 8.7, df = 2, p = 0.013) and severity of negative symptoms at clinical stabilization (mean score, 4.1 vs 3.9 vs 5.8; F = 3.1, p = 0.47).

Table 2 Baseline, premorbid and onset characteristics of the study sample. Characteristics

N = 153 n (%)

Male⁎ Age in years, mean (SD) Age at illness onset in years, mean (SD) Married Education in years, mean (SD) Family history of psychotic disorder Unemployed PAS social subscale, mean (SD) PAS academic subscale, mean (SD) PSST, mean (SD) Duration of untreated psychosis, median (interquartile range)⁎ Mode of illness onset⁎ Acute Sub-acute Insidious DSM-IV schizophrenia⁎

70 (46) 31.7 (9.2) 30.2 (9.3) 38 (25) 10.5 (3.4) 35 (23) 62 (41) 0.25 (0.2) 0.23 (0.1) 10.3 (2.3) 180 (30–730)

Medication naïve at study intake Baseline antipsychotic dose of chlorpromazine equivalents in mg/day, median (interquartile range), mg/day) Baseline PANSS positive symptoms, mean (SD) Baseline HEN, sum of global items, mean (SD) Baseline MADRS, mean (SD) Clinical stabilization PANSS positive symptoms, mean (SD) Clinical stabilization HEN, sum of global items, mean (SD)⁎ Clinical stabilization MADRS, mean (SD)

39 (25) 29 (19) 80 (52) 101 (66) 85 (56) 163 (82–272) 20.0 (5.7) 6.1 (5.4) 9.7 (11.3) 9.3 (2.8) 4.8 (4.3) 3.7 (6.1)

DSM-IV = Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition; HEN = High Royds Evaluation of Negativity Scale; MADRS = Montgomery Åsberg Depression Rating Scale; PANSS = Positive and Negative Syndrome Scale; PAS = Premorbid Adjustment Scale; PSST = Premorbid Schizoid and Schizotypal Traits. ⁎ p b 0.05 by one-way analysis of variance or chi-square test.

J.Y.-M. Tang et al. / Schizophrenia Research 153 (2014) 1–8

3.3. Long-term outcome (at 13 years) A total of 96 patients (44 males and 52 females) with a mean age of 44.7 years (range, 30–67) were followed up at 13 years. The majority of the patients (59%) were single, one-third (32%) were married and the rest (8%) were separated, divorced or widowed. Most of the patients (91%) were living in domestic accommodation, and a small proportion of these patients (9%) were living alone. The other patients (9%) were living in supervised accommodation including halfway houses and private hostels. The majority of patients were engaged in some form of routine activity including full-time employment (32%), part-time employment (9%), rehabilitation (17%) and homemaking (13%), while 28% of the patients were unemployed. More than 60% of the patients were receiving social benefits. The overall severity of psychopathology was minimal or mild. Approximately half of the patients (47%) fulfilled the criteria for remission. Among the patients whose data were available, 36% of them (n = 26/73) were in remission at both 3 and 13 years (χ2 = 5.4, df = 1, p = 0.02). The DUP groups exhibited comparable numbers of patients in remission at 1 year (Fig. 1). The remission rate substantially improved at 2 years and remained stable, especially among patients with a short DUP. The generalised estimating equation revealed a significant main effect of DUP (F = 4.1, df = 2, p = 0.016) and a significant main effect of time (F = 26.9, df = 2, p b 0.001). Interaction effect (DUP × time) was not significant. Fig. 2 shows the distribution of individual DUP data with regard to the remission status at 13 years. The DUP values for the non-remitted group tended to cluster around the high end of the scale. 3.4. Predictors of remission Seven baseline variables (DUP, education, mode of illness onset, PAS social subscale, PAS academic subscale, schizophrenia diagnosis and PSST) showed a potential univariate association (p b 0.1) with symptomatic remission at 13 years. No baseline neurocognitive predictors reached a p value of b0.1. The multivariate logistic regression showed that a DUP longer than 30 days resulted in a significant reduction in the likelihood of long-term symptomatic remission. This significant association was independent of premorbid adjustment and other selected baseline variables. The high score on the PAS social subscale significantly predicted a lower likelihood of symptomatic remission. No other

predictors were identified. The AUROC of the multivariate model was satisfactory (0.83). Three baseline variables (DUP, schizophrenia diagnosis and mode of illness onset) showed a potential univariate association with positive symptom remission at 13 years. Neither DUP nor the selected variables were significant predictors in the multivariate analysis. The model predictive ability, reflected by the AUROC, was acceptable (0.67). The baseline variables (DUP, smoker, PAS social subscale, PAS academic subscale and PSST) showed a potential univariate association with negative symptom remission, similar to that of symptomatic remission. The long DUP group had a lower likelihood of negative symptom remission at 13 years than the short DUP group, even after controlling for the effects of premorbid adjustment and other selected baseline variables. The high score on the PAS social subscale significantly predicted a lower likelihood of negative symptom remission. No other predictors were identified. The AUROC of the multivariate model was also satisfactory (0.82). The Hosmer and Lemeshow tests of the above prediction models indicated the models fitted the data well. Results of the univariate analysis and multivariate analysis are shown in Tables 3 and 4, respectively. 4. Discussion This prospective follow-up study has provided us with insights into the long-standing challenges of DUP research. Previous meta-analysis reported that DUP correlated with a range of clinical outcomes including remission and the severity of positive and negative symptoms (Marshall et al., 2005; Perkins et al., 2005). The evidence supporting the relationship between DUP and long-term remission was mostly derived from the Bonn Study (Huber et al., 1980), but the variables in this study were poorly defined and statistical adjustments were lacking. Furthermore, most of the existing knowledge about DUP has been obtained from short-term studies. The present study is one of the few studies that attempted to confirm the enduring impact of DUP on long-term clinical outcomes based on well-defined criteria. Our results were consistent with the previous long-term studies, which showed a significant relationship between DUP and symptomatic outcomes (Thara and Eaton, 1996; Bottlender et al., 2003; White et al., 2009; Hill et al., 2012). DUP and premorbid adjustment provided satisfactory discrimination between subjects with and without long-term remission.

1000 800 600

DUP in days (logarithmic scale)

400 200

Not remitted at 13 years

5

Remitted at 13 years

Fig. 2. Distribution of individual DUP data by remission status at 13 years.

6

J.Y.-M. Tang et al. / Schizophrenia Research 153 (2014) 1–8

Table 3 Univariate association between baseline characteristics and the long-term outcomes.

Medium DUPa Long DUPa Gender Age Age at illness onset Education Marriedb Family history Unemployedc Schizophrenia diagnosisd Sub-acute onset of illnesse Insidious onset of illnesse Smokerf PAS social PAS academic PSST PANSS positive HEN MADRS LM immediate recall LM delayed recall VR immediate recall VR delayed recall WAIS Information WAIS Digital Span Forward MWCST Any motor soft signsg

Symptomatic remission

Positive symptom remission

Negative symptom remission

OR (95% CI)

OR (95% CI)

OR (95% CI)

0.24 (0.07, 0.81)⁎⁎ 0.15 (0.05, 0.49)⁎⁎ 0.97 (0.43, 2.22) 1.00 (0.96, 1.05) 1.02 (0.97, 1.07) 1.15 (0.98, 1.35)⁎

0.09 (0.01, 0.77)⁎⁎ 0.11 (0.01, 0.89)⁎⁎ 0.65 (0.24, 1.77) 1.01 (0.95, 1.07) 1.01 (0.95, 1.07) 1.05 (0.89, 1.24) 1.96 (0.51, 7.49) 1.19 (0.38, 3.71) 0.96 (0.35, 2.62) 0.32 (0.09, 1.19)⁎ 0.22 (0.04, 1.28)⁎ 0.21 (0.04, 1.00)⁎⁎

1.19 (0.44, 3.21) 0.74 (0.29, 1.89) 0.68 (0.29, 1.59) 0.40 (0.16, 1.02)⁎ 0.53 (0.15, 1.81) 0.29 (0.11, 0.80)⁎ 0.44 (0.14, 1.40) 0.02 (0.002, 0.39)⁎⁎ 0.005 (0.0001, 0.30)⁎⁎ 0.70 (0.54, 0.91)⁎⁎ 0.96 (0.88, 1.03) 0.96 (0.88, 1.04) 1.00 (0.96, 1.04) 0.99 (0.97, 1.01) 0.99 (0.97, 1.01) 0.95 (0.87, 1.04) 0.96 (0.89, 1.03) 1.01 (0.94, 1.10) 1.08 (0.87, 1.34) 1.02 (0.97, 1.07) 0.82 (0.34, 1.97)

0.44 (0.12, 1.59) 0.26 (0.08, 0.85)⁎⁎ 1.55 (0.66, 3.64) 0.99 (0.94, 1.04) 1.00 (0.95, 1.05) 1.08 (0.93, 1.26) 0.89 (0.32, 2.45) 0.49 (0.19, 1.26) 0.52 (0.22, 1.25) 0.48 (0.18, 1.30) 0.58 (0.16, 2.09) 0.50 (0.18, 1.44) 0.39 (0.13, 1.17)⁎

0.60 (0.18, 1.97) 1.02 (0.06, 18.85) 2.11 (0.03, 154.3) 0.94 (0.76, 1.17) 1.00 (0.91, 1.09) 0.99 (0.90, 1.10) 1.00 (0.95, 1.05) 0.98 (0.96, 1.01) 0.98 (0.96, 1.01) 0.95 (0.85, 1.07) 0.99 (0.91, 1.07) 0.96 (0.87, 1.06) 1.05 (0.82, 1.35) 0.98 (0.93, 1.03) 1.31 (0.48, 3.63)

0.01 (0.0003, 0.12)⁎⁎ 0.004 (0.0001, 0.24)⁎⁎ 0.79 (0.64, 0.98)⁎⁎ 0.98 (0.90, 1.05) 0.95 (0.87, 1.03) 1.01 (0.97, 1.05) 1.00 (0.98, 1.03) 1.00 (0.98, 1.03) 0.98 (0.90, 1.07) 0.99 (0.92, 1.06) 1.05 (0.96, 1.14) 1.00 (0.81, 1.25) 1.05 (0.99, 1.11) 0.60 (0.24, 1.53)

CI = confidence interval; HEN = High Royds Evaluation of Negativity Scale; DUP = duration of untreated psychosis; LM = Logical Memory; MADRS = Montgomery Åsberg Depression Rating Scale; MWCST = modified Wisconsin Card Sorting Test; OR = odds ratio; PANSS = Positive and Negative Syndrome Scale; PAS = Premorbid Adjustment Scale; PSST = Premorbid Schizoid and Schizotypal Traits; VR = Visual Reproduction; WAIS = Wechsler Adult Intelligence Scale. ⁎ p b 0.1. ⁎⁎ p b 0.05. a Reference: short DUP. b Reference: unmarried. c Reference: employed. d Reference: non-schizophrenia diagnosis. e Reference: acute onset of illness. f Reference: non-smoker. g Reference: no motor soft sign.

Other investigations of predictors for positive symptom and negative symptom remission (Waddington et al., 1995) corroborated our finding of the role of DUP in relation to the severity of negative symptoms at the 13-year follow-up. Our findings were not confounded by baseline neurocognition because none of the cognitive domains were associated with the long-term outcome. Our previous investigation of

the present cohort found that the association between DUP and negative symptom severity was insignificant at baseline, but the association progressed to become significant by the end of year 3 (Chang et al., 2011). The present follow-up study of the same cohort suggested the strength of the longitudinal relationship observed in the mediumterm had not diminished with time. The results were in agreement

Table 4 Predictors for long-term symptomatic remission, positive symptom remission and negative symptom remission. Symptomatic remission

Medium DUPa Long DUPa Education Schizophrenia diagnosisb Sub-acute onsetc Insidious onsetc Smokerd PAS social PAS academic PSST AUROC

Positive symptom remission

Negative symptom remission

OR (95% CI)

p

OR (95% CI)

p

OR (95% CI)

p

0.11 (0.01, 0.87) 0.15 (0.02, 0.97) 1.10 (0.91, 1.34) 1.01 (0.21, 4.82) 0.71 (0.14, 3.63) 0.68 (0.14, 3.39) – 0.01 (0.0002, 0.69) 0.06 (0.0003, 12.3) 0.79 (0.57, 1.10) 0.83

0.037 0.046 0.4 1.0 0.7 0.6

0.12 (0.01, 1.52) 0.15 (0.01, 1.71)

0.1 0.1

0.062 0.010

1.33 (0.24, 7.30) 0.33 (0.05, 2.32) 0.41 (0.07, 2.51) – – – – 0.67

0.7 0.3 0.3

0.20 (0.04, 1.08) 0.14 (0.03, 0.62) – – – – 0.71 (0.16, 3.11) 0.002 (0.00003, 0.09) 0.05 (0.0002, 17.9) 1.01 (0.75, 1.36) 0.82

0.033 0.3 0.2

0.6 0.002 0.3 1.0

AUROC = area under the receiver operating characteristic; CI = confidence interval; DUP = duration of untreated psychosis; OR = odds ratio; PAS = Premorbid Adjustment Scale; PSST = Premorbid Schizoid and Schizotypal Traits. a Reference: short DUP. b Reference: non-schizophrenia diagnosis. c Reference: acute onset of illness. d Reference: non-smoker.

J.Y.-M. Tang et al. / Schizophrenia Research 153 (2014) 1–8

with previous short-term studies that reported inconsistent findings regarding the relationship between DUP and negative symptoms (Marshall et al., 2005; Schmitz et al., 2007). However, our results were different to those of the EPPIC cohort, which found DUP was associated with the severity of positive symptoms but not with negative symptoms at 8 years (Harris et al., 2005). Considering the doubtful efficacy of antipsychotics in treating negative symptoms, future studies should evaluate the impact of DUP reduction programmes on the long-term trajectory of negative symptoms. Furthermore, the TIPS early detection programme reported that improved short-term outcomes, including negative symptoms, among patients in the early detection group were not maintained in the long-term (Hegelstad et al., 2012). Our understanding of the linear relationship between DUP and outcome is limited. There is no standardized method of how to handle the positively skewed DUP data. It has been argued that a continuous, log-transformed DUP makes better use of the data and is more sensitive to statistical tests. This approach has been adopted by many prospective cohorts such as the Hillside cohort, the South Dublin cohort (Hill et al., 2012), the EPPIC cohort (Harris et al., 2005) and the Manchester cohort (White et al., 2009). The Munich 15-year follow-up study used DUP categories with different cutoff values (≤6 months, N6 months to ≤1 year, N1 year) (Bottlender et al., 2003). However, the categories did not take into account patients with a very short DUP. Our data suggest that a threshold value for DUP as short as 31 days is sufficient to show a significant negative impact on the outcome, similar to that observed in the long DUP group (N180 days). This threshold value was much shorter than previously proposed (Singh, 2007), which indicates that the DUP– outcome relationship was unlikely to be linear. Analysis using DUP categories assumed no linearity with the outcome and that the statistical power might even be compromised. Despite the long follow-up period, a causal relationship could hardly be inferred from the observational data. To date, there have been two known cluster randomised controlled trials of GP education programmes that aimed to reduce DUP (Tait et al., 2005; Power et al., 2007), but both studies showed the programmes did not succeed. A local evaluation study of an early intervention programme with an early detection component did not find any differences in DUP between cases and historical controls (Chen et al., 2011). One possible explanation is that the intensity of the early detection and intervention was not sufficient to reduce the DUP to ≤30 days (the threshold value). An alternative explanation is that a long pre-treatment interval is typical in a group of patients who have characteristics indicative of a poor prognosis. In other words, interventions should target specific prognostic factors instead of DUP to improve the outcome. The Northwick Park Study of First Episodes (Owens et al., 2010) found that the relationship between the pre-treatment interval (duration of illness) and relapse became insignificant after adjusting for tension at admission, bizarre behaviour and unemployment status. On the contrary, our findings suggested the effect of DUP was independent. The discrepancy in the findings may be explained by the methodological differences. The primary outcome of the Northwick Park Study was relapse, which was largely determined by the severity of positive symptoms. If we take into account positive symptom remission from our data, our findings were largely consistent with the Northwick Park Study. Both studies showed that the pre-treatment interval had a significant bivariate relationship with the outcome but these effects diminished after statistical adjustment. Our data further showed that DUP appears to have a greater influence on negative symptom remission than on positive symptom remission. The relationship between DUP and positive symptoms may be less straightforward. The attrition rate of our study was largely comparable to other international studies. If the follow-up rate is calculated as the number of interviewed subjects divided by the total number of living subjects, then the follow-up rate of this study (70%) was higher than the TIPS study (56%) (Hegelstad et al., 2012) and Nottingham study (62%) (Harrison et al., 1996), comparable to the Manchester study (70%) (White et al., 2009), but was lower than the South Dublin study (80%)

7

(Hill et al., 2012). It is reasonable to assume that this follow-up cohort was representative of the whole sample since the baseline characteristics were comparable between completers and non-completers. 4.1. Limitations Missing observations reduced the statistical power of the study. The limited study population precluded further subgroup analyses, such as exploration of outcome predictors within diagnostic or gender groups. Furthermore, the sample size limited the precision of the effect size estimation. In particular, a wide confidence interval was observed in the effect of DUP on remission, indicating greater uncertainty of the effect size estimation. The outcomes used in this follow-up study were cross-sectional and thus failed to capture the long-term course of illness (specifically in terms of clinical improvement or deterioration). Other sources of measurement bias may have been present in our study. First, recall bias may have resulted from the retrospective measurement of DUP and the small number of baseline covariates. Second, different raters were responsible for the baseline and early outcome assessments, thereby potentially leading to systematic bias. This prospective observational study did not provide clear evidence for a causal effect, even though this study had an extended follow-up period and known covariates were well controlled. Although ethical issues precluded us from randomizing patients to short and long DUP groups, this study represented a good observational method to estimate the longitudinal impacts of DUP. Role of Funding Source Nil. Contributors Study conception and design: JYMT and EYHC. Data analysis: JYMT. Data interpretation: all authors. Drafting the article: JYMT. Critical revision and final approval of the article: all authors. Conflict of interest None. Acknowledgements We thank all coordinating clinicians and staff from the psychiatric units at Queen Mary Hospital, Pamela Youde Nethersole Eastern Hospital, Kwai Chung Hospital, Kowloon Hospital, United Christian Hospital, Yung Fung Shee Psychiatric Centre and Castle Peak Hospital.

References Andreasen, N.C., Carpenter, W.T., Kane, J.M., Lasser, R.A., Marder, S.R., Weinberger, D.R., 2005. Remission in schizophrenia: proposed criteria and rationale for consensus. Am. J. Psychiatry 162, 441–449. Birchwood, M., 1999. Early intervention in psychosis: the critical period. In: McGorry, P., Jackson, H. (Eds.), The Recognition and Management of Early Psychosis: A Preventive Approach. Cambridge University Press, Cambridge, pp. 226–264. Bottlender, R., Sato, T., Jäger, M., Wegener, U., Wittmann, J., Strauß, A., et al., 2003. The impact of the duration of untreated psychosis prior to first psychiatric admission on the 15-year outcome in schizophrenia. Schizophr. Res. 62, 37–44. Cannon-Spoor, H.E., Potkin, S.G., Wyatt, R.J., 1982. Measurement of premorbid adjustment in chronic schizophrenia. Schizophr. Bull. 8, 470–484. Chang, W., Hui, C.L., Tang, J.Y., Wong, G.H., Lam, M.M., Chan, S.K., et al., 2011. Persistent negative symptoms in first-episode schizophrenia: a prospective three-year followup study. Schizophr. Res. 133, 22–28. Chang, W., Hui, C., Tang, J., Wong, G., Chan, S., Lee, E., et al., 2012. Impacts of duration of untreated psychosis on cognition and negative symptoms in first-episode schizophrenia: a 3-year prospective follow-up study. Psychol. Med. 1, 1–11. Chen, E.Y., Shapleske, J., Luque, R., McKenna, P.J., Hodges, J.R., Calloway, S.P., et al., 1995. The Cambridge Neurological Inventory: a clinical instrument for assessment of soft neurological signs in psychiatric patients. Psychiatry Res. 56, 183–204. Chen, E.Y., Hui, C.L., Dunn, E.L., Miao, M.Y., Yeung, W., Wong, C., et al., 2005. A prospective 3-year longitudinal study of cognitive predictors of relapse in first-episode schizophrenic patients. Schizophr. Res. 77, 99–104. Chen, E.Y., Tang, J.Y., Hui, C.L., Chiu, C.P., Lam, M.M., Law, C.W., et al., 2011. Three‐year outcome of phase‐specific early intervention for first‐episode psychosis: a cohort study in Hong Kong. Early Interv. Psychiatry 5, 315–323. Häfner, H., Riecher-Rössler, A., Hambrecht, M., Maurer, K., Meissner, S., Schmidtke, A., et al., 1992. IRAOS: an instrument for the assessment of onset and early course of schizophrenia. Schizophr. Res. 6, 209–223.

8

J.Y.-M. Tang et al. / Schizophrenia Research 153 (2014) 1–8

Harris, M.G., Henry, L.P., Harrigan, S.M., Purcell, R., Schwartz, O.S., Farrelly, S.E., et al., 2005. The relationship between duration of untreated psychosis and outcome: an eightyear prospective study. Schizophr. Res. 79, 85–93. Harrison, G., Croudace, T., Mason, P., Glazebrook, C., Medley, I., 1996. Predicting the longterm outcome of schizophrenia. Psychol. Med. 26, 697–705. Hegelstad, W., Larsen, T.K., Auestad, B., Evensen, J., Haahr, U., Joa, I., et al., 2012. Long-term follow-up of the TIPS early detection in psychosis study: effects on 10-year outcome. Am. J. Psychiatry 169, 374–380. Hill, M., Crumlish, N., Clarke, M., Whitty, P., Owens, E., Renwick, L., et al., 2012. Prospective relationship of duration of untreated psychosis to psychopathology and functional outcome over 12 years. Schizophr. Res. 141, 215–221. Hong Kong Psychological Society, 1989a. The Wechsler Memory Scale-Revised (Cantonese Version). Hong Kong Psychological Society, Hong Kong. Hong Kong Psychological Society, 1989b. The Wechsler Adult Intelligence Scale-Revised (Cantonese Version). Hong Kong Psychological Society, Hong Kong. Huber, G., Gross, G., Schüttler, R., Linz, M., 1980. Longitudinal studies of schizophrenic patients. Schizophr. Bull. 6, 592–605. Kay, S.R., Flszbein, A., Opfer, L.A., 1987. The positive and negative syndrome scale (PANSS) for schizophrenia. Schizophr. Bull. 13, 261–276. Kurihara, T., Kato, M., Reverger, R., Tirta, I., 2005. Eleven‐year clinical outcome of schizophrenia in Bali. Acta Psychiatr. Scand. 112, 456–462. Marshall, M., Lewis, S., Lockwood, A., Drake, R., Jones, P., Croudace, T., 2005. Association between duration of untreated psychosis and outcome in cohorts of first-episode patients: a systematic review. Arch. Gen. Psychiatry 62, 975–983. Montgomery, S.A., Asberg, M., 1979. A new depression scale designed to be sensitive to change. Br. J. Psychiatry 134, 382–389. Mortimer, A.M., McKenna, P., Lund, C., Mannuzza, S., 1989. Rating of negative symptoms using the High Royds Evaluation of Negativity (HEN) Scale. Br. J. Psychiatry Suppl. 7 (155), 89–91. Nelson, H.E., 1976. A modified card sorting test sensitive to frontal lobe defects. Cortex 12, 313–324. Owens, D.C., Johnstone, E.C., Miller, P., Macmillan, J.F., Crow, T.J., 2010. Duration of untreated illness and outcome in schizophrenia: test of predictions in relation to relapse risk. Br. J. Psychiatry 196, 296–301. Perkins, D.O., Gu, H., Boteva, K., Lieberman, J.A., 2005. Relationship between duration of untreated psychosis and outcome in first-episode schizophrenia: a critical review and meta-analysis. Am. J. Psychiatry 162, 1785–1804.

Power, P., Iacoponi, E., Reynolds, N., Fisher, H., Russell, M., Garety, P.A., et al., 2007. The lambeth early onset crisis assessment team study: general practitioner education and access to an early detection team in first-episode psychosis. Br. J. Psychiatry Suppl. 191, s133–s139. Schmitz, N., Malla, A., Norman, R., Archie, S., Zipursky, R., 2007. Inconsistency in the relationship between duration of untreated psychosis (DUP) and negative symptoms: sorting out the problem of heterogeneity. Schizophr. Res. 93, 152–159. Singh, S.P., 2007. Outcome measures in early psychosis relevance of duration of untreated psychosis. Br. J. Psychiatry Suppl. 191, s58–s63. So, E., Kam, I., Leung, C., Chung, D., Liu, Z., Fong, S., 2003. The chinese-bilingual SCID-I/P project: stage 1—reliability for mood disorders and schizophrenia. Hong Kong J. Psychiatry 13, 7–18. Tait, L., Lester, H., Birchwood, M., Freemantle, N., Wilson, S., 2005. Design of the BiRmingham early detection in untREated psyChosis trial (REDIRECT): cluster randomised controlled trial of general practitioner education in detection of first episode psychosis [ISRCTN87898421]. BMC Health Serv. Res. 5, 19. Thara, R., Eaton, W.W., 1996. Outcome of schizophrenia: the madras longitudinal study. Aust. N. Z. J. Psychiatry 30, 516–522. Tsoi, W., Kua, E., 1992. Predicting the outcome of schizophrenia ten years later. Aust. N. Z. J. Psychiatry 26, 257–261. van Mastrigt, S., Addington, J., 2002. Assessment of premorbid function in first-episode schizophrenia: modifications to the Premorbid Adjustment Scale. J. Psychiatry Neurosci. 27, 92–101. Waddington, J., Youssef, H., Kinsella, A., 1995. Sequential cross-sectional and 10year prospective study of severe negative symptoms in relation to duration of initially untreated psychosis in chronic schizophrenia. Psychol. Med. 25, 849–857. White, C., Stirling, J., Hopkins, R., Morris, J., Montague, L., Tantam, D., et al., 2009. Predictors of 10-year outcome of first-episode psychosis. Psychol. Med. 39, 1447–1456. Wiersma, D., Nienhuis, F.J., Slooff, C.J., Giel, R., 1998. Natural course of schizophrenic disorders: a 15-year followup of a Dutch incidence cohort. Schizophr. Bull. 24, 75–85. Wiersma, D., Wanderling, J., Dragomirecka, E., Ganev, K., Harrison, G., An Der Heiden, W., et al., 2000. Social disability in schizophrenia: its development and prediction over 15 years in incidence cohorts in six European centres. Psychol. Med. 30, 1155–1167.