Schizophrenia Bulletin

Schizophrenia Bulletin Advance Access originally published online on February 16, 2005
Schizophrenia Bulletin 2005 31(3):735-750; doi:10.1093/schbul/sbi017

This Article Abstract FREE Full Text (PDF) All Versions of this Article: 31/3/735    most recent sbi017v1 Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Add to My Personal Archive Download to citation manager Request Permissions Google Scholar Articles by Rosen, K. Articles by Garety, P. Search for Related Content PubMed PubMed Citation Articles by Rosen, K. Articles by Garety, P. Social Bookmarking          What's this?

© The Author 2005. Published by Oxford University Press on behalf of the Maryland Psychiatric Research Center. All rights reserved. For permissions, please email: journals.permissions@oupjournals.org.

Predicting Recovery From Schizophrenia: A Retrospective Comparison of Characteristics at Onset of People With Single and Multiple Episodes

Kate Rosen, M.Phil.
Clinical Psychologist, Department of Psychology, Institute of Psychiatry, King's College London, London, UK

Philippa Garety, Ph.D.
Professor of Clinical Psychology, Department of Psychology, Institute of Psychiatry, King's College London, London, UK

	Abstract

Top Abstract Heterogeneity of Outcome in... Single Episodes Predictors of Outcome The Present Study Method Results Predictors of Outcome Discussion References

 
A retrospective study was designed to determine whether socio-demographic and clinical factors at onset, previously shown to relate to outcome, differentiated those with a single episode with no persistent symptoms of schizophrenia from other outcome groups. In a geographically determined sample of 436 people with schizophrenia spectrum disorder and a minimum followup period of at least 6 years, 68 people (15.6%) had a single episode with complete remission. The single episode group differed significantly from the rest of the sample at onset on nine variables. On a logistic regression, employment status independently predicted single episodes. Although those with single episodes differed from the rest of the sample on a number of variables, they did not differ significantly at onset from the other better outcome group (repeated episodes without persistent symptoms) on any variables with the exception of insight. Two possibilities are considered: (1) the two better outcome groups differed very little at onset but subsequent treatment or experiences accounted for the differences in outcome; or (2) important differences, not routinely assessed at onset, influenced outcome. The implications of these findings for research into the prevention of relapse in psychosis are considered.

Keywords: Schizophrenia / psychosis / course / outcome / single episode / predictors

---

The course and outcome of schizophrenia have received considerable attention since the condition was first described as dementia praecox by Kraepelin in 1896. While a century's research has given rise to an extensive literature regarding its natural history, some of the original questions regarding the prognosis and outcome of schizophrenia remain just as pertinent today. Kraepelin originally considered dementia praecox to have a deteriorating course, altering this view when he found some of his severe hospital-based patients showed spontaneous remission. Long-term followup studies have investigated this question of full recovery without subsequent relapse. In all studies, a proportion of patients with a diagnosis of schizophrenia are found to have only a single episode of illness. The estimates of this proportion range from 12 percent to 22 percent. Investigations of the characteristics of this group should aid our understanding of recovery and relapse in schizophrenia. This article describes a retrospective case note study of a geographically determined sample of cases of schizophrenia in which a single episode sample is identified and studied.

	Heterogeneity of Outcome in Schizophrenia

Top Abstract Heterogeneity of Outcome in... Single Episodes Predictors of Outcome The Present Study Method Results Predictors of Outcome Discussion References

 
While there had been a tendency to dichotomize outcome into either a chronic, deteriorating course or a full remission, research has increasingly viewed the outcome of schizophrenia spectrum psychosis (hereafter, "psychosis") as being on a continuum (Vaillant 1963, 1964, 1978; Strauss and Carpenter 1972) and has emphasized the heterogeneity and complexity of outcome (e.g., Strauss 1969; Strauss and Carpenter 1977; Carpenter and Strauss 1991; Harding et al. 1992). Taking mode of illness onset, number of episodes, and level of remission into account, Ciompi (1980a, 1980b) described 8 main patterns of outcome, Huber et al. (1980) described 12 possible courses, and Shepherd et al. (1989) referred to 4. Thus, while it is now generally accepted that the course of illness is heterogeneous, there is little consensus regarding the complexity of this heterogeneity.

Longitudinal studies are valuable for providing a comprehensive evaluation of the outcome of psychosis. One important collection of studies is the World Health Organization Collaborative Study on the Determinants of Outcome of Severe Mental Disorders, known as the International Study of Schizophrenia (ISoS; Jablensky et al. 1992), and its precursor, the International Pilot Study of Schizophrenia (IPSS; Sartorius et al. 1977, 1987; Leff et al. 1992). The IPSS found that outcome of schizophrenia was better in developing countries than in developed countries, even after 5 years of followup, with outcome improving at 5 years for some individuals. The ISoS again showed outcome to differ according to country, with developing countries showing a better course of illness than developed countries. At 2-year followup, the majority of patients in the study had a remitting course of illness, with 50.3 percent having a single episode of psychosis, 31.1 percent having repeated episodes, and 15.7 percent showing an unremitting course.

In a followup study of the Washington cohort of the IPSS, Carpenter and Strauss (1991) found that patients doing well at 2 years were likely to be doing well at 11 years and that patients doing poorly at 2 years often changed status at 11 years. They argue that progressive deterioration is uncommon and is probably seen less frequently than late-course improvement and conclude that many patients with schizophrenia reach a plateau early in the course of psychosis. Harrison et al. (2001) found that course of illness at 2 years predicted 15-year outcome but also that it was possible for early unremitting cases to achieve late-phase recovery.

A followup of the English cohort 13 years after the ISoS study was conducted by Mason et al. (1995, 1996, 1997). They followed up 94 percent of their original sample of 67 and found that 60.9 percent of their sample had an episodic course, 32.8 percent had a continuous course, and 6.3 percent had neither an episodic nor a continuous course. Twenty-five percent of their study's participants were never readmitted, while 18 percent had no further episodes. This is comparable to Shepherd et al. (1989), who reported in a 5-year followup study that 22 percent of their sample of 49 first episode subjects had not relapsed. Outcome after 13 years of followup did not suggest either progressive deterioration or amelioration, and Mason et al. (1996) conclude that the course of schizophrenia "is most stormy at onset and early in its manifest course, plateauing thereafter" (p. 585).

The idea of schizophrenia patients reaching a plateau after the early course was suggested by some earlier long-term followup studies (e.g., Watts 1985; McGlashan 1986) and is further supported by, among others, Carpenter and Strauss (1991), Leff et al. (1992), and Harrison et al. (1996). Ram et al. (1992), in a review of the literature, conclude that patterns of illness course are variable but that some of this variability relates to different lengths of followup, as course of illness changes over time and outcome assessed after 1 year may not represent outcome assessed after 10 years. While studies including such different lengths of followup may therefore be skewed (Bartko et al. 1988), it is generally accepted that the early course of psychosis is most changeable and that there is greater stability from 5 years after onset.

	Single Episodes

Top Abstract Heterogeneity of Outcome in... Single Episodes Predictors of Outcome The Present Study Method Results Predictors of Outcome Discussion References

 
In the literature documenting different courses, one group is described as having the best outcome, the single episode patients who make a good recovery from first onset and do not subsequently relapse. There are a range of estimates of the proportion of presentations with this picture. Jablensky et al. (1992) report that 50.3 percent of their sample had a single episode, in a short followup of only 2 years from onset; Shepherd et al. (1989) report 22 percent at 5 years; Mason et al. (1995, 1996, 1997) report 18 percent at 13 years; and Wiersma et al. (1998) report 12.2 percent at 15 years. Evidently, the proportion of single episodes decreases with length of followup. By what date can we be reasonably confident that a person will not relapse subsequently? Wiersma et al. (1998), who had the longest followup, conclude that a large majority of the sample who would relapse had relapsed by 5 years (70% of the total sample) and that the risk of relapse seemed to disappear after 9 years. This is further supported by Jackson and Birchwood (1996), in a review of the literature of first episode followup studies, who conclude that deterioration occurs in the prepsychotic and early phases of illness but that it usually stabilizes by 5 years.

	Predictors of Outcome

Top Abstract Heterogeneity of Outcome in... Single Episodes Predictors of Outcome The Present Study Method Results Predictors of Outcome Discussion References

 
Early reviews of the literature (Vaillant 1964, 1978; Stephens 1970) found a number of factors to be associated with course of illness: good prognosis was suggested by having an acute onset of illness, exhibiting prominent affective features, showing good premorbid functioning, being married, having a precipitating event, being confused, being preoccupied with death (later found not to be significant), and having a history of manic-depressive illness in the family. Poor prognosis was suggested by having an insidious onset, exhibiting an asocial premorbid personality, being withdrawn, having never been married, and having a family history of schizophrenia. However, in a retrospective review of courses of illness in 51 patients who had shown full remission of their schizophrenic symptoms, Vaillant (1978) found that only 31 patients maintained remission during their followup of 4 to 16 years, while 20 patients subsequently became "chronic." Those who remitted had almost the same presentation as those who developed a chronic course, suggesting that course of psychosis is not reliably predicted at onset.

Since the early studies, predictors of course have been found in numerous areas of functioning and can be grouped into demographic factors and socioeconomic situation, premorbid situation, biological factors, and factors relating to the illness itself, for example length of illness prior to treatment and age at onset of illness.

Better outcome has been found to relate to being married (e.g., Shepherd et al. 1989; Jablensky et al. 1992; Leff et al. 1992; Harrison et al. 1996; Wiersma et al. 1998), attaining higher levels of education (Geddes et al. 1994), being female (Watt et al. 1983; Shepherd et al. 1989; Jablensky et al. 1992; Leff et al. 1992; Geddes et al. 1994; Harrison et al. 1996; Murray and van Os 1998; Moriarty et al. 2001), showing better premorbid functioning (Shepherd et al. 1989; Carpenter and Strauss 1991; Jablensky et al. 1992; Loebel et al. 1992), having no family history of schizophrenia or affective illness (Vaillant 1964; Murray and van Os 1998), having no organic impairment in the brain (e.g., Murray and van Os 1998), having a shorter length of untreated psychosis (Loebel et al. 1992; Harrison et al. 1996; Wiersma et al. 1998; Drake et al. 2000), having an older age at onset (Shepherd et al. 1989; Jablensky et al. 1992; Loebel et al. 1992), and having a trigger in the 3-month period prior to the onset of psychosis (Vaillant 1964; Leff et al. 1992; Murray and van Os 1998). Social factors such as contact with family and friends have been found to predict outcome, with Jablensky et al. (1992) and Carpenter and Strauss (1991) finding that increased social contact with friends and family predicted better course, but the nature of this contact is also important, as findings suggest that relapse is predicted by the presence of a critical atmosphere or high expressed emotion in these social interactions (Butzlaff and Hooley 1998).

Presence of depression at first onset of psychosis has been found to relate both to better outcome (e.g., Vaillant 1964, 1978; Stephens 1970) and to worse outcome (e.g., Geddes et al. 1994; Sands and Harrow 1999). Furthermore, some of the findings that are strongly predictive of outcome in some studies have not been supported by others. An example of this is later age at onset of psychosis, which as outlined above has been associated with better prognosis. Wiersma et al. (1998) found no relationship between age at onset and outcome, while Harrison et al. (1996) found that older age at onset was associated with worse outcome.

While some literature exists on the effect of certain psychological factors on the course of illness—for example, "recovery style" (McGlashan et al. 1975)—this research is much more limited. Jackson et al. (1998) found, in an early psychosis sample, that those people who showed a style of adaptation concerned with "integrating" rather than "sealing over" the experiences had subsequent higher levels of depression. Furthermore, we noted above that the duration of untreated psychosis (DUP) has been found in a number of studies to be a predictor of course and outcome. In a recent study, insight was linked to DUP (Drake et al. 2000). In a sample of first admissions for schizophrenia, long DUP not only predicted poorer short-term treatment outcome at 12 weeks but was itself predicted by poorer insight, avolition, and poor social integration.

While many of the findings have focused on predicting good outcome, it has been difficult to predict who will do best—that is, who will have a single episode, recover well, and not relapse. In a study examining the predictive power of various prognostic scales for schizophrenia, Fenton and McGlashan (1987) found that a poor prognostic score seemed to almost guarantee poor outcome, while a good prognostic score did not ensure good outcome: some of the patients with the best prognosis remained at risk of persistent disability. They conclude that "it is far easier to predict which patients will do poorly, than it is to predict, with much power, those who will do well" (p. 283).

There are, therefore, important questions to be answered about the course of psychosis and factors associated with better outcome. One way of addressing these questions could be to explore those with best outcome in more detail. Studies with at least a 5-year followup have found that between 12 percent and 22 percent of all new incidences will have only a single episode of psychosis, but very little is known about this group. To our knowledge, no study has focused on identifying which factors predict single episodes. As a relatively small group, the identification of which can be reasonably reliably confirmed only after at least 5 years following first onset, it is not an easy group to study. A study that identifies a sample of people with a single episode may offer new insights about the factors associated with recovery and relapse prevention.

Outcome studies employ either a retrospective or a prospective design, the relative merits of which are outlined by McGlashan et al. (1988). While prospective studies enable full assessment information and certainty about course and outcome, they suffer from attrition, are costly, and usually require drawing a sample from a very large pool of people. Retrospective studies allow large sample sizes and long followup intervals but are often subject to missing information and reliability problems. McGlashan et al. (1988) argue that the study design should fit the study hypotheses and questions, and that while both retrospective and prospective designs have inherent difficulties and benefits, careful consideration of the aims of the study would suggest the most appropriate study design.

	The Present Study

Top Abstract Heterogeneity of Outcome in... Single Episodes Predictors of Outcome The Present Study Method Results Predictors of Outcome Discussion References

 
The aims of the present study were to investigate a cohort of people with a single episode of schizophrenia spectrum psychosis, assessing factors at onset previously shown to relate to course, and to determine whether there are differences in these factors between those with the best outcome (single episode) and the rest of the groups, and between groups with different patterns of course and outcome.

Questions were posed regarding the pattern of course of psychosis over time, the survival time to first relapse, and whether any demographic, psychological, or illness factors recorded at first contact predict full subsequent recovery. As the purpose of the study was to investigate factors associated with a single episode of psychosis, it was necessary to have an adequately large sample of people with a single episode. This necessitated there being both a long followup period, to reduce the chances of falsely identifying people as having had a single episode, and a large pool of people. It was therefore decided that a retrospective case note review best achieved these aims.

	Method

Top Abstract Heterogeneity of Outcome in... Single Episodes Predictors of Outcome The Present Study Method Results Predictors of Outcome Discussion References

 
A sample of subjects was identified using the Oxford Record Linkage System (ORLS). ORLS was established in 1963 as a collection of summaries of all hospitalizations in Oxford, U.K., for an individual together with demographic, social, and clinical information, and records of births and deaths. ORLS then links all successive records for that individual, providing a summary of the person's medical care. The system was extended to cover all of Oxfordshire, an English county with a population of approximately 550,000, in 1966 and to include data since 1974 on all psychiatric contacts, including outpatient contacts, day patient care, and domiciliary visits. In view of the findings discussed above, that a minimum of 5 years' followup is required to identify single episodes, a list of people with a first contact with a psychotic disorder in ORLS between 1980 and 1991 was obtained at the end of 1997. The sample therefore consisted of people whose first onset of psychosis dated back at least 6 years, and because data collection continued through 1998 and 1999, provided followup data to a maximum of 19 years (for those with onset in 1980). This list includes information on all outpatient, domiciliary, and day patient contacts, and thus the sample is not restricted to admitted patients. It was expected that ORLS would generate an overinclusive sample to be screened for eligibility (including, e.g., people with major affective disorders, such as bipolar disorder) and that there would be a number of false-positives. An original cohort of people with a psychotic diagnosis was identified for screening, and attempts were made to examine all psychiatric case notes for the sample.

Ethical approval and consent were sought from the appropriate authorities prior to obtaining the list and examining the psychiatric case notes, and all subjects' information was anonymized for confidentiality. The majority of the case notes were located in medical record centers in the area, and the remainder were held in the community mental health teams responsible for the patients' care.

The inclusion criteria were as follows:

1. First contact with a psychotic disorder between 1980 and 1991 (i.e., at least 6 years since onset of psychosis)
   
2. Age 16 to 45 at first contact with psychotic disorder
   
3. A primary diagnosis of a functional psychosis of a nonaffective type according to ICD–9 (295–schizophrenic psychoses, 297–persistent paranoid state, 298.4–psychogenic paranoid psychosis; WHO 1978) or
   
4. Symptoms consistent with above in the absence of an actual diagnosis
   
5. Oxfordshire resident at first episode
   

The exclusion criteria were as follows:

1. Evidence of organic cerebral disorder, drug abuse, or alcohol abuse as the primary problem
   
2. IQ below 70
   
3. Departure to reside abroad immediately after first psychotic illness
   
4. Evaluation for legal assessment only
   
5. Anonymous client (employee or relation of employee of the service who has contact with a psychiatric disorder and whose confidentiality is maintained)
   
6. A diagnosis that does not match the information recorded or absence of documentation of symptoms consistent with the diagnosis
   

As can be seen from the inclusion criteria, a decision was made not to rely on the front-sheet diagnosis alone but to check that the diagnosis was consistent with the information available in the case notes regarding presenting symptoms. This was in order not to exclude people with changing diagnoses or situations where an early diagnosis of schizophrenia spectrum disorder was avoided by the psychiatrist, and to achieve a reasonably homogeneous group of individuals according to presenting symptoms. We also restricted the sample to an upper age limit of 45 years at first presentation. This was in order to reduce the possibility of false-positives in an older age group who had had an earlier psychotic episode about which information had been lost or not documented in existing case notes.

The first author, a qualified clinical psychologist, conducted the case note review of the 1,801 patients initially identified by ORLS as meeting criteria for diagnosis and date of onset. A consensus approach was used for inclusion, so that any uncertainty regarding possible inclusion was resolved by detailed discussion with the second author, a clinical psychologist with specialist expertise and training in psychosis and diagnostic assessment. For those meeting the inclusion criteria, information was gathered from the case notes at first contact with mental health services with a psychotic disorder, where possible, on age, gender, date of first contact with psychiatric services, date of first psychotic symptoms, length of illness episode, dates of further relapses or admissions, symptoms expressed at index episode, insight at onset as recorded in the psychiatric evaluation, personal background, duration of prodrome and untreated psychosis (calculated from data given in the reports of psychiatric assessment at first contact), duration of first episode (defined as the length of time from first treatment until resolution of symptoms to a recovered or stable state), medication prescribed, alcohol or drug use, self-harm, family history of mental illness, and circumstances at first onset as reported in the psychiatric assessment, including whether there seemed to be an identifiable stressor or trigger. Duration of prodrome refers to the time from appearance of nonpsychotic psychiatric symptoms to appearance of psychotic symptoms, and DUP refers to the time from the appearance of positive psychotic symptoms to initiation of treatment with medication (Drake et al. 2000). Descriptions were recorded as they appeared in the notes, and where appropriate, were grouped and coded for analysis. Psychotic symptoms recorded were divided into Liddle's (1987) categories of positive, negative, and disorganization symptoms.

Case notes do not invariably provide the same information entered clearly and unambiguously. This is an acknowledged problem with studies of this type (McGlashan et al. 1988). To minimize the risk of invalid data, two approaches were taken. First, where information was ambiguous or unclear, a detailed discussion between the authors was used to reach a consensus; this approach (of reaching agreement rather than trying to achieve reliable independent ratings) has been proposed when retrospectively rating information (Tennant et al. 1979). Second, a high threshold was set for the inclusion of data. Where a clear consensus could not be achieved following detailed discussion, "missing" data were recorded. Some information was so patchily available, it was not included as a variable; one example was ethnicity, which, especially in the earlier sets of notes, was frequently unrecorded.

Outcome was classified for the course of illness for each individual. It is known that whereas some show full recovery from the first episode (and subsequent episodes), others show only partial recovery (Tohen et al. 1992; Lieberman et al. 1993). Being considered a "full" recovery required a case record of complete symptom remission with no persisting symptoms, while "partial" recovery required evidence of significant clinical improvement, with less severe ("residual") symptoms persisting (Craig et al., 2004). Accordingly, different patterns of illness course were identified, using categorizations similar to those of Shepherd et al. (1989): single episode with no residual difficulties, single episode with residual difficulties (positive, negative, and disorganization symptoms), repeated episodes without residual difficulties, and repeated episodes with residual difficulties. Outcome was recorded both for the whole followup period available with variable lengths of followup and at a cutoff point of 6 years after first episode. Because a variable length of followup is prone to biases, and in practice there were few differences in the results, all analyses reported are for the 6-year cutoff.

Care was taken with the definition of relapse, given the criticisms that have been made of the looseness of some studies (Falloon and Talbot 1981; Gleeson 2001). Falloon and Talbot recommend that psychotic symptoms be used as relapse indicators and be treated separately from readmission or deterioration in social or affective states. Furthermore, a minimum duration of symptom change is recommended to differentiate relapse from a brief "flurry" of symptoms; a week is used as a criterion by a number of researchers (e.g., Linszen et al. 1996). While the required duration of the change does vary in the literature, we set a minimum of 7 days' duration of change; given our reliance on case note data, a longer period would have increased the probability of recording false-positives (i.e., recording as "single episode" those who had in fact relapsed). Relapse was therefore defined in terms of a clearly recorded worsening of psychotic symptoms requiring a change of response such as an increase in medication. It could occur as change from either (1) the absence of any psychotic symptoms (fully recovered) to the reemergence of psychotic symptoms, or (2) a previously stable state of persistent symptoms (partially recovered) to a clear exacerbation of psychotic symptoms (Leff and Vaughn 1985).

Additionally, to further reduce the risk of falsely identifying people as having a single episode, only those people who ended treatment, cooperated with followup, and were discharged without symptoms were identified as having had a single episode of psychosis. The period of an absence of symptoms recorded had to be at least 7 days, although in practice this was normally recorded over a much longer period. If the person had dropped out of contact without a clear decision to discharge as recovered, or the suggestion was raised in the notes that the person had moved away from the area, outcome was recorded as not known.

Descriptive data are first presented. Differences between the groups on each of the variables outlined above were explored using parametric and nonparametric analyses, as appropriate to the nature of the variable. With the factors that significantly differentiated the outcome groups established, regression analyses and discriminant analyses were conducted to determine which of the significant factors independently predicted outcome. To check that results were not being affected by potential sample biases arising from differential levels of missing data, regression analyses were repeated, with the successive removal of factors with the most missing data. In addition, categories within variables were collapsed, where appropriate to do so, to avoid having too many cells with no cases. Furthermore, in view of the large standard deviations (SDs) of the variables prodrome, DUP, combined length of prodrome and DUP, and duration of first episode, analyses were conducted using logarithmic transformations, thereby making the variation consistent.

	Results

Top Abstract Heterogeneity of Outcome in... Single Episodes Predictors of Outcome The Present Study Method Results Predictors of Outcome Discussion References

 
Case Identification and Classification in Groups
Of the original list of 1,801 patients, 1,238 (68.7%) did not, on further examination, meet the inclusion criteria. The reasons are detailed in table 1. It was not possible to trace 56 sets of notes, and a further 71 of the names on the list were duplicates of others already identified. Exactly 436 patients were included in the study.

View this table: [in this window] [in a new window]   Table 1. Summary of reasons for exclusion

 
At 6 years, 68 people (15.6% of the sample) had only a single episode of illness with complete remission (table 2). One hundred and fifty-three people (35.1%) had multiple episodes without persisting symptoms between episodes, 136 (31.2%) had multiple episodes with persisting symptoms, and 30 people (6.9%) had a single episode from which there was no recovery. Seven people died in their first episode, 5 of whom were reported to have committed suicide. These individuals were considered to be most similar to the group of people whose first episode had not ended (i.e., poorest outcome, the fourth group), and given the already small size of that group, it was decided to merge these two groups for analyses. It was not possible to determine outcome for 42 people (9.6% of the sample), as they moved or dropped out of contact with the psychiatric services before the end of their first psychotic episode or of followup. A check was made on those for whom no outcome or followup data were available. They were more likely to be students (p < 0.05) and had attained a higher level of education by the time of their first episode (p < 0.05). These findings are consistent with the large student population in Oxford, many of whom would have returned to their parental home around the time of leaving the hospital.

View this table: [in this window] [in a new window]   Table 2. Patterns of illness course at 6 years

 
Relapse
The mean time to first relapse was 2.2 years (SD 2.4), ranging from 1 week (the minimum as defined by the criteria) to 15 years. Within the first 2 years, 191 (49.4%) of the available sample (i.e., excluding those who had died in their first episode, and those for whom outcome could not be determined) had relapsed. Two hundred and eighty-eight (74.4%) of this sample had relapsed within 6 years. When considering only those who did relapse at some point (i.e., also excluding the single episode groups with and without persistent symptoms), 91.7 percent had relapsed within 6 years and 98.1 percent (all but six people) had relapsed at 9 years. The final six people relapsed between 10 and 15 years.

Death
At 6 years' followup, 15 (3.8%) people had died, 7 of whom had died in their first episode. Of these deaths, 11 were recorded as suicide, 3 were of unknown causes, and 1 appears to have been an accident. When considering the entirety of the followup information available, 23 (5.8%) people had died, 12 of whom were recorded as having committed suicide.

Sample Characteristics at First Presentation to Psychiatric Services With Psychosis
Details of the sociodemographic variables at first episode for the total group and compared for two groups—the single episode with no persistent residual symptoms group and the remainder of the sample—are displayed in table 3. Clinical variables at first episode are detailed in table 4.

View this table: [in this window] [in a new window]   Table 3. Sociodemographic variables at first presentation for single episode sample versus remainder

 

View this table: [in this window] [in a new window]   Table 4. Clinical variables at first presentation for single episode sample versus remainder

 
The total sample had an average age of 29 years at first episode, and 58 percent were male. Sixty-eight percent were single, and 41 percent were employed, while a further 12 percent were students (consistent with Oxford being a university town). Approximately two-thirds of the sample were treated at first episode by admission to the hospital, and 98 percent had clear positive symptoms. About 60 percent were recorded as having no insight, although there are high levels of missing data on this variable as well as on the variables DUP and duration of the prodrome, mean durations recorded as 130 days and 563 days, respectively, with very large SDs.

Differences Between Groups
Three main ways of grouping the patients on course of illness were undertaken. First, two groups, the single episode with no persistent residual symptoms group and the remainder of the sample, were considered for analysis (tables 3 and 4). Second, the four different patterns of illness were considered. Finally, the different courses were considered in two groups (as is common in other studies, e.g., Jablensky et al. 1992; Harrison et al. 2001), "worse outcome" (single episode with persistent symptoms and repeated episode with persistent symptoms groups collapsed together) and "better outcome" (single and repeated episode groups with no persistent symptoms collapsed together). For each of these three methods of grouping, statistical analyses were conducted in two stages, as outlined in the Method section. The first stage was to identify which variables differed between groups. Once identified, the significant variables and those with a trend toward significance were entered into regression analyses to determine which, if any, variables independently predicted outcome.

Single episode with no persistent symptoms versus remainder of sample
Sixty-eight people (15.6% of the total sample and 17.3% of the sample on whom outcome information was available) were found to have a single episode with no persistent symptoms. Unless otherwise specified, the term "single episode" will be used to refer to the best outcome group of people found to have a single episode with no persistent residual symptoms (tables 3 and 4). There were significant differences between the single episode group and the remainder of the sample in the variables of gender, marital status, employment status, premorbid adjustment, insight, reported use of drugs or alcohol at or before the first episode, duration of the prodrome, combined length of prodrome and DUP, and duration of the first episode. Trends toward significance were found in the variables disorganization symptoms and DUP.

Four patterns of illness course
A large number (13) of significant differences were identified when examining the four patterns of illness. These are shown in tables 5 and 6.

View this table: [in this window] [in a new window]   Table 5. Sociodemographic variables at first presentation for the four patterns of illness course

 

View this table: [in this window] [in a new window]   Table 6. Clinical variables at first presentation for the four patterns of illness course

 
Better versus worse outcome
Eleven significant differences were found between the better and worse outcome groups; these were very similar to those differentiating the four patterns (table 7 summarizes the significant differences between groups, in the three methods of comparisons between groups). Significant differences were found in gender (²(1, 394) = 18.721, p < 0.0001), educational attainment (²(1, n = 343) = 5.879, p = 0.015), marital status (²(1, n = 389) = 12.901, p = 0.002), employment status (²(2, n = 371) = 18.532, p < 0.0001), premorbid adjustment (²(1, n = 382) = 18.714, p < 0.0001), trigger to illness identified (²(1, n = 366) 10.541, p = 0.001), negative symptoms (²(1, n = 394) = 8.249, p = 0.004), duration of the prodrome (Mann-Whitney U (n = 304) = 7783, z = –4.849, p < 0.0001), DUP (Mann-Whitney U (n = 283) = 7,415, z = 3.219, p = 0.001), duration of the prodrome and DUP combined (Mann-Whitney U (n = 298) = 7,563, z = –4.721, p < 0.0001), and duration of the first episode (Mann-Whitney U (n = 360) = 10,338.5, z = –5.533, p < 0.0001). There was a trend toward significance for the variables age (Mann-Whitney U (n = 394) = 17,067, z = –1.756, p = 0.079) and disorganization symptoms (²(1, n = 394) = 3.763, p = 0.052).

View this table: [in this window] [in a new window]   Table 7. Summary of significant differences in three methods of group comparison

 
Examination of the data suggested that there were few, if any, differences within the better outcome group between those who had only a single episode of psychotic symptoms with no residual symptoms and those who had repeated episodes with no residual symptoms. Analyses were therefore conducted to determine whether this was the case. A significant difference was found in only the variable insight (²(2, n = 167) = 6.998, p = 0.030), with trends toward significance in the variables drug use and marital status.

Given the exploratory nature of these analyses, there are a large number of statistical comparisons; therefore, all significant differences at below the p < 0.01 level should be treated with caution.

	Predictors of Outcome

Top Abstract Heterogeneity of Outcome in... Single Episodes Predictors of Outcome The Present Study Method Results Predictors of Outcome Discussion References

 
Single episode with no persistent symptoms versus remainder of sample
The significant and almost significant variables (11 in total) were entered into logistic regression analyses. When these variables were entered, 155 participants were included in the sample. Marital status (p = 0.036, odds ratio [OR] = 3.429) and employment (p = 0.080, OR = 0.307) emerged as being significant or almost significant. When the sample size was maximized by successively removing variables with missing data, 343 cases were used in the analyses. The variables gender, marital status, employment status, premorbid adjustment, disorganization symptoms, and drug use were entered; employment status emerged as being significant overall (p = 0.025), with unemployment predicting membership in the rest of the sample and not having only a single episode (p = 0.007, OR = 0.398). When employment was excluded from the analyses, 363 cases were included in the analyses and no factors emerged as being independently significant in predicting group membership.

Four patterns of illness course
Discriminant analyses were conducted to determine whether any of the significant variables could reliably predict group membership. When all 14 of the significant or nearly significant variables were entered, 144 of the possible sample of 394 were included in the analyses. The first function was found to be significant (Wilks lambda = 0.633, ² = 61.444, p = 0.012), with 55.6 percent of the sample correctly assigned. However, only 25.0 percent of the single episode group with no persistent residual symptoms were correctly assigned. This function was best at correctly assigning those with repeated episodes and no residual positive symptoms, with 70.0 percent correctly assigned. Overall, therefore, while the function was significant, this was due to its success at assigning individuals to the repeated episode group without persistent symptoms. The function was poor at identifying factors predicting membership in the single episode, no persistent symptoms, group. Analyses were repeated successively, removing variables with the most missing data. As variables were removed, the discriminant analyses became poorer at correctly predicting group membership. For example, when 377 participants were included in the analysis, the first function was significant (Wilks lambda = 0.833, ² = 67.810, p < 0.0001), but only 49.6 percent overall were correctly classified. No participant with a single episode without persistent symptoms was correctly assigned, with the majority (67.7%) being classified as belonging to the repeated episodes with no persistent symptoms group.

Better versus worse outcome
When all the significant variables listed were entered, 170 participants were included in the logistic regression analyses. Four variables emerged as significant: gender (p = 0.016, OR = 2.694), educational attainment (p = 0.026, OR = 2.694), marital status (p = 0.026, OR = 2.734), and trigger to illness identified (p = 0.048, OR = 2.199). When the sample size was maximized by limiting the logistic regression to the variables of gender, marital status, employment status, premorbid adjustment, negative symptoms, and disorganization symptoms, 356 participants were included in the analyses. Gender (p = 0.001, OR = 2.319), employment status (p = 0.001, OR = 0.440), premorbid adjustment (p = 0.006, OR = 0.446), and negative symptoms (p = 0.032, OR = 1.863) were independently significant in discriminating better from worse outcome groups.

	Discussion

Top Abstract Heterogeneity of Outcome in... Single Episodes Predictors of Outcome The Present Study Method Results Predictors of Outcome Discussion References

 
To our knowledge, this is the first study to investigate factors associated with single episodes of psychosis. The sample generated is likely to be fairly representative of the clinical population of people with schizophrenia; and the sample's demographic characteristics, except the rather high age at onset, are comparable with those of many prospective studies (e.g., Harrison et al. 2001).

Sixteen percent of the total sample had a single episode with full recovery. This is toward the midpoint of the range of percentages discussed earlier, where it was estimated that between 12 and 22 percent fall into this group. All but 6.5 percent of the sample who would go on to relapse had done so by 6 years. These findings are consistent with prospective followup studies with long followup periods, such as that of Wiersma et al. (1998), who also identified a sample with ICD–9 schizophrenia spectrum disorders and found that by 5 years after onset, 70 percent of the sample had experienced a psychotic relapse and no new relapse occurred after 9 years, with a total followup period of 15 years.

By 6 years, approximately 4 percent of the sample had died, most of the deaths being recorded as suicide. This is similar to the rate of suicide described by Stephens et al. (1999) although less than that reported by some other studies (e.g., Caldwell and Gottesman 1990; Ram et al. 1992).

A major source of methodological limitations of the study derives from its retrospective case note design. Most important, there is the potential for some of the single episode sample to have had further episodes of illness, and either to have had contact with psychiatric services elsewhere or to have been managed outside the psychiatric system. However, strenuous attempts were made to reduce this risk of false-positives by assuming that outcome was unknown unless a clear resolution to the psychotic episode was identified and the participant cooperated with followup. The reasonably low percentage rate of single episodes (16%) in this study in comparison to some previous research (e.g., 23%, Linszen et al. 2001; 22%, Shepherd et al. 1989) would suggest that our allocation was appropriately conservative and that the risk of false-positives is low. Second, the reliability of case note data entered by different clinicians over a long time period can be questioned. Reliance on case note data also results in variable amounts of missing data; this particularly affected two variables of interest, DUP and insight, where there were up to 25 percent of cases with missing data, and thus these variables could not be entered into the regression analyses. However, our main findings are consistent with the established literature, indicating a reasonable degree of reliability in our measures. In contrast, a strength of the study is that we have a geographically determined sample of good size, permitting us to address questions about single episodes.

Our key findings were that the single episode with no persistent symptoms group differed significantly from the rest of the sample on nine variables at onset. However, membership in this single episode group could not be reliably determined by a discriminant function analysis; indeed, this group, despite being of adequate size (n = 68), was only significantly different on one variable, insight, from the group with a relapsing course, without persistent symptoms. These two groups combined (the better outcome) differed from the worse outcome groups on eleven variables, with trends on a further two variables. These findings replicate those of many of the studies referred to above. Gender, employment status, premorbid adjustment, and negative symptoms independently discriminated these two groups (better or worse outcome).

The results of this study are therefore consistent with the large body of research in finding a considerable number of sociodemographic and clinical variables to be associated with better or worse outcome in schizophrenia spectrum psychosis (e.g., Jablensky et al. 1992). The difficulty in predicting full recovery from psychosis has been documented since Vaillant (1964). More recently, Torgalsbøen (1999) failed to find factors that significantly predicted full recovery, although good premorbid adjustment and gender showed a "substantial association in the predicted direction" (p. 149). In our study, a number of factors were found to be associated with the best outcome of single episode with no persistent symptoms, including marital status, employment status, and premorbid adjustment, consistent with Torgalsbøen.

As Shepherd et al. (1989) found, employment was a predictor of better outcome; here, we found employment status at onset to be associated with, and predictive of, both better outcome and single episode of psychosis. Those who were employed at onset were more likely to have a single episode. Being employed following the acute episode has previously been shown to assist recovery (see Warner 1994 for a full discussion; Bell and Lysaker 1997), and many now emphasize its value to service users (Perkins et al. 1999; British Psychological Society 2000). However, while there are very probably direct benefits, it seems likely that employment status at onset may also be an indirect reflection of other factors related to outcome. For example, those with a gradual and lengthy premorbid deterioration and a long DUP are less likely to maintain employment before contact with services.

If there are few, if any, sociodemographic or clinical variables found to reliably distinguish the two better outcome groups, and the study is adequately powered, we can draw one of two conclusions. First, it is possible that these groups do not differ at onset but that subsequent circumstances, such as adequacy of treatment, adverse life events, employment, social support, or stress, are major factors accounting for the difference in relapse between these two groups. If this is so, it is intriguing to consider whether this suggests that there are additional clinical or social interventions, as well as medication, that would assist further in relapse prevention. Indeed, family interventions, which aim to influence the social environment to reduce stress and therefore relapses, are now well established in this regard (Pilling et al. 2002); and there is some evidence that cognitive-behavioral therapy (CBT) interventions may also help prevent relapse (Pilling et al. 2002; Gumley et al. 2003). We have noted above that work is an important contributor to better outcome (Warner 1994). Investigating the mechanisms by which these interventions prevent relapse would be informative. Given that such a high proportion of relapses occur within 2 years, and nearly all within 6 years, these efforts should not be delayed. Furthermore, more comprehensive early intervention services, providing effective pharmacological and psychological treatments and followup from onset, together with social support and vocational programs, might also be expected to reduce relapses (Edwards and McGorry 2002); publication of the results of randomized controlled trials (RCTs) of such newly established early intervention services is awaited.

A recent long-term followup study from the ISoS found that the response to treatment in the first 2 years following onset was the strongest predictor of 15-year outcome (Harrison et al. 2001). A shorter period of time spent experiencing psychotic symptoms in the first 2 years was the best predictor of all outcome measures at 15 years. The authors therefore argue for effective early intervention services, embracing social as well as pharmaceutical approaches, to counter the possible "toxic" effect of early symptom exposure. Interestingly, in the present study, we found a nonsignificant difference between the single episode with no persistent symptoms group and the rest of the sample on the duration of the first episode, with the single episode group experiencing a shorter period of time unwell. In another recent study, with a 6-year followup, de Haan et al. (2003) found that both delay in treating with medication (the traditional measure of DUP) and delay in providing intensive psychosocial treatment (DIPT) were associated with outcome; at 6 years both were associated with negative symptoms, DUP was associated with mild relapse, and DIPT was associated with rehospitalization.

A second possibility is that there are indeed some factors that differ at onset between those with single episodes and those who recover well but relapse, which are not routinely assessed and that may be important in protecting a recovered individual from subsequent relapse. We note that there was a finding of better insight in our single episode group. This emerged as significant in the comparison of single episode versus the remainder of the sample, was a trend in the examination of the four patterns of illness course, and was significant in the comparison between the two better outcome groups (although we note that none of the significance levels were high). We noted above that poorer insight has been linked to longer DUP (Drake et al. 2000). The authors suggest that poor insight is one aspect of lack of concern, which reduces the likelihood of presenting for help. Poor insight is also likely to have effects subsequent to onset, such as influencing attitudes toward treatment, whether medication or psychological interventions, such as CBT. Our data on insight and DUP/prodromes must be treated with caution, given the high rates of missing data and the questionable reliability of case note records of these variables, but they do offer some support for insight and DUP as possible factors distinguishing the single episode group from the time of onset.

The potential importance of insight is also proposed in a new cognitive model of psychosis put forward by Garety et al. (2001). In this model, set within a stress-vulnerability framework, the appraisal of psychosis (which is one aspect of insight), social support, and reasoning are all hypothesized to contribute to the recurrence or maintenance of symptoms. The significance of a person's manner of representing or appraising the threat of an illness in the outcome of disorders other than psychosis has already been established, particularly in the field of general medicine (e.g., Leventhal et al. 1984; Skelton and Croyle 1991; Weinman et al. 1996; Lobban et al. 2003). Recently, it has been argued that the illness representation approach may be fruitfully applied to extend the investigation of the role of insight in psychosis (Barrowclough et al. 2001; Jolley and Garety 2004; Lobban et al. 2003). We speculate that insight and illness representations over the early postdiagnosis phase may play an important role in influencing course and contributing to the prevention or triggering of relapse. Additionally, two other variables, drug use before or during the first episode and marital status, were (weakly) associated with single episodes with full recovery. This is consistent with other research. The importance of protective social relationships has been frequently noted before, as discussed above. Recently, there has been a growing recognition of the particular importance of drug misuse as a trigger of and as a factor influencing outcome from the first episode (see Murray et al. 2003 for a review of recent research). We conclude that working on insight, promoting close and supportive relationships, and changing habits of drug use, in the context of the effective and early provision of medication and psychological and social interventions, including promoting return to work, may improve the prevention of relapse from the first episode. A researchable goal of first episode services might therefore reasonably be not simply to reduce relapses but, more ambitiously, to increase the proportion of those who have a single episode with full recovery.

	Footnotes

 
Send reprint requests to Dr. P. Garety, Department of Psychology, Henry Wellcome Building, Institute of Psychiatry, De Crespigny Park, London SE5 8AF; e-mail: p.garety{at}iop.kcl.ac.uk.

	References

Top Abstract Heterogeneity of Outcome in... Single Episodes Predictors of Outcome The Present Study Method Results Predictors of Outcome Discussion References

 

Barrowclough, C, Lobban, F, Hatton, C, Quinn, J. An investigation of models of illness in carers of schizophrenia patients using the Illness Perception Questionnaire. British Journal of Clinical Psychology 2001; 40:371–385.[CrossRef][Web of Science][Medline]

Bartko, JJ, Carpenter, WT, McGlashan, TH. Statistical issues in long-term follow-up studies. Schizophrenia Bulletin 1988; 14:575–587.

Bell, MD and Lysaker, PH. Clinical benefits of paid work activity in schizophrenia: 1-year follow-up. Schizophrenia Bulletin 1997; 23:317–328.

British Psychological Society, Division of Clinical Psychology. Recent Advances in Understanding Mental Illness and Psychotic Experiences Leicester, UK: British Psychological Society 2000.

Butzlaff, RL and Hooley, JM. Expressed emotion and psychiatric relapse: A meta-analysis. Archives of General Psychiatry 1998; 55:547–552.[Abstract/Free Full Text]

Caldwell, CB and Gottesman, II. Schizophrenics kill themselves too: A review of risk factors for suicide. Schizophrenia Bulletin 1990; 16:571–589.

Carpenter, WT and Strauss, JS. The prediction of outcome in schizophrenia: IV. Eleven-year follow-up of the Washington IPSS cohort. Journal of Nervous and Mental Disease 1991; 179:517–525.[Web of Science][Medline]

Ciompi, L. Catamnestic long term study on the course of life and aging of schizophrenics. Schizophrenia Bulletin 1980a; 6:606–618.

Ciompi, L. The natural history of schizophrenia in the long term. British Journal of Psychiatry 1980b; 136:413–420.[Free Full Text]

Craig, TKJ, Garety, PA, Power, P, Rahaman, N, Colbert, S, Fornells-Ambrojo, M, Dunn, G. The LEO Randomised Controlled Trial: An investigation of assertive outreach for early psychosis: I. Primary outcomes—relapse and readmission. British Medical Journal 2004; 329:1067.[Abstract/Free Full Text]

De Haan, L, Linszen, DH, Lenior, ME, de Win, ED, Gorsira, R. Duration of untreated psychosis and outcome of schizophrenia: Delay in intensive psychosocial treatment versus delay in treatment with antipsychotic medication. Schizophrenia Bulletin 2003; 29:341–348.

Drake, RJ, Haley, CJ, Akhtar, S, Lewis, SW. Causes and consequences of duration of untreated psychosis in schizophrenia. British Journal of Psychiatry 2000; 177:511–515.[Abstract/Free Full Text]

Edwards, J and McGorry, P. Implementing Early Intervention in Psychosis London, U.K.: Martin Dunitz 2002.

Falloon, IRH and Talbot, RE. Persistent auditory hallucinations: Coping mechanisms and implications for management. Psychological Medicine 1981; 11:329–339.[Web of Science][Medline]

Fenton, WS and McGlashan, TH. Prognostic scale for chronic schizophrenia. Schizophrenia Bulletin 1987; 13:277–286.

Garety, PA, Kuipers, E, Fowler, D, Freeman, D, Bebbington, PE. A cognitive model of the positive symptoms of psychosis. Psychological Medicine 2001; 31:189–195.[CrossRef][Web of Science][Medline]

Geddes, J, Mercer, G, Frith, CD, Macmillan, F, Owens, DGC, Johnstone, EC. Prediction of outcome following a first episode of schizophrenia: A follow-up study of Northwick Park first episode study subjects. British Journal of Psychiatry 1994; 165:664–668.[Abstract/Free Full Text]

Gleeson, J.F. Early warning signs and risk factors for relapse in early psychosis. Unpublished Ph.D. thesis, University of Melbourne, Australia, 2001.

Gumley, A, O'Grady, M, McNay, L, Reilly, J, Power, K, Norrie, J. Early intervention for relapse in schizophrenia: Results of a 12-month randomized controlled trial of cognitive behavioural therapy. Psychological Medicine 2003; 33:419–431.[CrossRef][Web of Science][Medline]

Harding, CM, Zubin, J, Strauss, JS. Chronicity in schizophrenia: Revisited. British Journal of Psychiatry 1992; 161:27–37.

Harrison, G, Croudace, T, Mason, P, Glazebrook, C, Medley, I. Predicting the long-term outcome of schizophrenia. Psychological Medicine 1996; 26:697–705.[Web of Science][Medline]

Harrison, G, Hopper, K, Craig, T, Laska, E, Siegel, C, Wanderling, J, Dube, KC, Ganev, K, Giel, R, an-der-Heiden, W, Holmberg, SK, Janca, A, Lee, PW, Leon, CA, Malhotra, S, Marsella, AJ, Nakane, Y, Sartorius, N, Shen, Y, Skoda, C, Thara, R, Tsirkin, SJ, Varma, VK, Walsh, D, Wiersma, D. Recovery from psychotic illness: A 15-year and 25-year international follow-up study. British Journal of Psychiatry 2001; 178:506–517.[Abstract/Free Full Text]

Huber, G, Gross, G, Schuttler, R, Linz, M. Longitudinal studies of schizophrenic patients. Schizophrenia Bulletin 1980; 6:592–605.

Jablensky, A, Sartorius, N, Ernberg, G, Anker, M, Korten, A, Cooper, JE, Day, R, Bertelsen, A. Schizophrenia: Manifestations, incidence and course in different cultures. A World Health Organization ten-country study. Psychological Medicine 1992; 20:1–97.

Jackson, C and Birchwood, M. Early intervention in psychosis: Opportunities for secondary prevention. British Journal of Clinical Psychology 1996; 35:487–502.

Jackson, H, McGorry, P, Edwards, J, Hulbert, C, Henry, L, Francey, S, Maude, D, Cocks, J, Power, P, Harrigan, S, Dudgeon, P. Cognitively oriented psychotherapy for early psychosis (COPE): Preliminary results. British Journal of Psychiatry 1998; 172:93–100.[Web of Science]

Jolley, S and Garety, P. Insight and delusions. In Amador, X and David, A (Eds.). Insight and Psychosis 2nd ed. Oxford, UK: Oxford University Press 2004.

Kraepelin, E. Barclay, RM, trans. Dementia Praecox and Paraphrenia (1896) Chicago, IL: Chicago Medical Books 1919.

Leff, J, Sartorius, N, Jablensky, A, Korten, A, Ernberg, G. The International Pilot Study of Schizophrenia: 5-year follow-up findings. Psychological Medicine 1992; 22:131–145.[Web of Science][Medline]

Leff, J and Vaughn, C. Expressed Emotion in Families New York, NY: Guilford Press 1985.

Leventhal, H, Nerenz, DR, Steel, DJ. Illness representations and coping with health threats. In Baum, A, Taylor, SE, Singer, JE (Eds.). Handbook of Psychology and Health Hillsdale, NJ: Erlbaum 1984 Vol; 4: pp. 219–252.

Liddle, PF. The symptoms of chronic schizophrenia: A re-examination of the positive-negative dichotomy. British Journal of Psychiatry 1987; 151:145–151.[Abstract/Free Full Text]

Lieberman, J, Jody, D, Geisler, S, Alvir, J, Loebel, A, Szymanski, S, Woerner, M, Borenstein, M. Time course and biological correlates of treatment responses in first-episode schizophrenia. Archives of General Psychiatry 1993; 50:369–376.[Abstract/Free Full Text]

Linszen, D, Dingemans, P, Lenior, M. Early intervention and a five-year follow-up in young adults with a short duration of untreated psychosis: Ethical implications. Schizophrenia Research 2001; 51:55–61.[CrossRef][Web of Science][Medline]

Linszen, D, Dingemans, P, Van der Does, JW, Nugter, A, Scholte, P, Lenior, R, Goldstein, MJ. Treatment, expressed emotion and relapse in recent onset schizophrenic disorders. Psychological Medicine 1996; 26:333–342.[Web of Science][Medline]

Lobban, F, Barrowclough, C, Jones, S. A review of the role of illness models in severe mental illness. Clinical Psychology Review 2003; 23:171–196.[CrossRef][Web of Science][Medline]

Loebel, AD, Lieberman, JA, Alvir, JM, Mayerhoff, DI, Geisler, SH, Szymanski, SR. Duration of psychosis and outcome in first-episode schizophrenia. American Journal of Psychiatry 1992; 149:1183–1188.[Abstract/Free Full Text]

Mason, P, Harrison, G, Croudace, T, Glazebrook, C, Medley, I. The predictive validity of a diagnosis of schizophrenia. A report from the International Study of Schizophrenia (ISoS) coordinated by the World Health Organization and the Department of Psychiatry, University of Nottingham. British Journal of Psychiatry 1997; 170:321–327.[Abstract/Free Full Text]

Mason, P, Harrison, G, Glazebrook, C, Medley, I, Croudace, T. The course of schizophrenia over 13 years. A report from the International Study on Schizophrenia (ISoS) coordinated by the World Health Organization. British Journal of Psychiatry 1996; 169:580–586.[Abstract/Free Full Text]

Mason, P, Harrison, G, Glazebrook, C, Medley, I, Dalkin, T, Croudace, T. Characteristics of outcome in schizophrenia at 13 years. British Journal of Psychiatry 1995; 167:596–603.[Abstract/Free Full Text]

McGlashan, TH. The prediction of outcome in chronic schizophrenia: IV. The Chestnut Lodge follow-up study. Archives of General Psychiatry 1986; 43:167–176.[Abstract/Free Full Text]

McGlashan, TH, Carpenter, WT, Bartko, JJ Jr. Issues of design and methodology in long-term followup studies. Schizophrenia Bulletin 1988; 14:569–574.

McGlashan, TH, Levy, ST, Carpenter, WT. Integration and sealing over: Clinically distinct recovery styles from schizophrenia. Archives of General Psychiatry 1975; 32:1269–1272.[Abstract/Free Full Text]

Moriarty, PJ, Lieber, D, Bennett, A, White, L, Parrella, M, Harvey, PD, Davis, KL. Gender differences in poor outcome patients with lifelong schizophrenia. Schizophrenia Bulletin 2001; 27:103–113.

Murray, RM, Grech, A, Phillips, P, Johnson, S. What is the relationship between substance abuse and schizophrenia? In Murray, RM, Jones, PB, Susser, E, van Os, J, Cannon, M (Eds.). The Epidemiology of Schizophrenia Cambridge, UK: Cambridge University Press 2003 pp. 317–342.

Murray, RM and van Os, J. Predictors of outcome in schizophrenia. Journal of Clinical Psychopharmacology 1998; 18:2S–4S.[CrossRef][Web of Science][Medline]

Perkins, RE, Choy, D, Davidson, B. Pathfinder User Employment Programme: Progress Report London, UK: Pathfinder Mental Health Trust 1999.

Pilling, S, Bebbington, P, Kuipers, E, Garety, P, Geddes, J, Orbach, G, Morgan, C. Psychological treatments in schizophrenia: I. Meta-analysis of family intervention and cognitive behavioural therapy. Psychological Medicine 2002; 32:763–782.[CrossRef][Web of Science][Medline]

Ram, R, Bromet, EJ, Eaton, WW, Pato, C, Schwartz, JE. The natural course of schizophrenia: A review of first-admission studies. Schizophrenia Bulletin 1992; 18:185–207.

Sands, JR and Harrow, M. Depression during the longitudinal course of schizophrenia. Schizophrenia Bulletin 1999; 25:157–171.

Sartorius, N, Jablensky, A, Ernberg, G, Leff, J, Korten, A, Gulbinat, W. Course of schizophrenia in different countries: Some results of a WHO International 5-year follow-up study. In Hafner, H, Gattaz, WF, Janzarik, W (Eds.). Search for the Causes of Schizophrenia Berlin, Germany: Springer Verlag 1987 pp. 107–113.

Sartorius, N, Jablensky, A, Shapiro, R. Two-year follow-up of the patients included in the WHO International Pilot Study of Schizophrenia: Preliminary communication. Psychological Medicine 1977; 7:529–541.[Web of Science][Medline]

Shepherd, M, Watt, D, Falloon, IRH, Smeeton, N. The Natural History of Schizophrenia: A Five-Year Follow-up Study of Outcome and Prediction in a Representative Sample of Schizophrenics Cambridge, UK: Cambridge University Press 1989.

Skelton, JA and Croyle, RT. Mental Representation in Health and Illness New York, NY: Springer-Verlag 1991.

Stephens, JH. Long-term course and prognosis in schizophrenia. Seminars in Psychiatry 1970; 2:464–485.[Web of Science][Medline]

Stephens, JH, Richard, P, McHugh, PR. Suicide in patients hospitalised for schizophrenia: 1913–1940. Journal of Nervous and Mental Disease 1999; 187:10–14.[CrossRef][Web of Science][Medline]

Strauss, JS. Hallucinations and delusions as points on a continua function. Archives of General Psychiatry 1969; 21:581–586.[Abstract/Free Full Text]

Strauss, JS and Carpenter, WT. The prediction of outcome in schizophrenia: I. Characteristics of outcome. Archives of General Psychiatry 1972; 27:739–746.[Abstract/Free Full Text]

Strauss, JS and Carpenter, WT. Prediction of outcome in schizophrenia. Archives of General Psychiatry 1977; 34:159–163.[Abstract/Free Full Text]

Tennant, C, Smith, A, Bebbington, P, Hurry, J. The contextual threat of life events: The concept and its reliability. Psychological Medicine 1979; 9:525–528.[Web of Science][Medline]

Tohen, M, Stoll, AL, Strakowski, SM, Faedda, GL, Mayer, PV, Goodwin, DC, Kolbrener, ML, Madigan, AM. The McClean First-Episode Psychosis Project: Six-month recovery and recurrence outcome. Schizophrenia Bulletin 1992; 18:273–282.

Torgalsbøen, A. Full recovery from schizophrenia: The prognostic role of premorbid adjustment, symptoms at first admission, precipitating events and gender. Psychiatry Research 1999; 88:143–152.[CrossRef][Web of Science][Medline]

Vaillant, GE. The natural history of the remitting schizophrenias. American Journal of Psychiatry 1963; 120:367–375.[Abstract/Free Full Text]

Vaillant, GE. Prospective prediction of schizophrenia remission. Archives of General Psychiatry 1964; 11:509–518.

Vaillant, GE. A 10-year followup of remitting schizophrenics. Schizophrenia Bulletin 1978; 4:78–85.

Warner, R. Psychiatry and the Political Economy London, UK: Routledge 1994.

Watt, DC, Katz, K, Shepherd, M. The natural history of schizophrenia: A 5-year prospective follow-up of a representative sample of schizophrenics by means of a standardised clinical and social assessment. Psychological Medicine 1983; 13:663–670.[Web of Science][Medline]

Watts, CAH. A long-term follow-up of schizophrenic patients: 1946–1983. Journal of Clinical Psychiatry 1985; 46:210–216.[Web of Science][Medline]

Weinman, J, Petrie, KJ, Moss-Morris, R, Horne, R. The Illness Perception Questionnaire: A new method for assessing the cognitive representation of illness. Psychology & Health 1996; 11:431–445.[CrossRef][Web of Science]

Wiersma, D, Nienhuis, FJ, Slooff, CJ, Giel, R. Natural course of schizophrenic disorders: A 15-year followup of a Dutch incidence cohort. Schizophrenia Bulletin 1998; 24:75–85.

World Health Organization, R. International Classification of Diseases Geneva, Switzerland: World Health Organization 1978 9th revision.

CiteULike    Connotea    Del.icio.us    What's this?

This article has been cited by other articles:

				S. Mohamed, R. Rosenheck, J. McEvoy, M. Swartz, S. Stroup, and J. A. Lieberman Cross-sectional and Longitudinal Relationships Between Insight and Attitudes Toward Medication and Clinical Outcomes in Chronic Schizophrenia Schizophr Bull, March 1, 2009; 35(2): 336 - 346. [Abstract] [Full Text] [PDF]

				M. Bodnar, A. Malla, R. Joober, and M. Lepage Cognitive markers of short-term clinical outcome in first-episode psychosis The British Journal of Psychiatry, October 1, 2008; 193(4): 297 - 304. [Abstract] [Full Text] [PDF]

				A. S. Bellack Scientific and Consumer Models of Recovery in Schizophrenia: Concordance, Contrasts, and Implications Schizophr Bull, July 1, 2006; 32(3): 432 - 442. [Abstract] [Full Text] [PDF]

This Article Abstract FREE Full Text (PDF) All Versions of this Article: 31/3/735    most recent sbi017v1 Alert me when this article is cited Alert me if a correction is posted Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Add to My Personal Archive Download to citation manager Request Permissions Google Scholar Articles by Rosen, K. Articles by Garety, P. Search for Related Content PubMed PubMed Citation Articles by Rosen, K. Articles by Garety, P. Social Bookmarking          What's this?

Online ISSN 1745-1701 - Print ISSN 0586-7614

Copyright © 2009 Maryland Psychiatric Research Center and Oxford University Press

Oxford Journals Oxford University Press

• Site Map
• Privacy Policy
• Frequently Asked Questions

Other Oxford University Press sites: Oxford University Press Oxford Journals China Oxford Journals Japan American National Biography Booksellers' Information Service Children's Fiction and Poetry Children's Reference Corporate & Special Sales Dictionaries Dictionary of National Biography Digital Reference English Language Teaching Higher Education Textbooks Humanities International Education Unit Law Medicine Music Online Products Oxford English Dictionary Reference Rights and Permissions Science School Books Social Sciences Very Short Introductions World's Classics