Use of Quantitative Rating Scales to Assess Outcome in Schizophrenia Prevention Studies

Abstract

Objective

To present a summary of quantitative scales relevant to schizophrenia prevention studies.

Method

Fifteen scales were reviewed and summarised in terms of structure, domains assessed, previous use and psychometric properties. Instruments of symptom measurement, role functioning and global functioning were considered, along with multidimensional instruments and other scales of potential interest to research in schizophrenia prevention.

Results and conclusions

No scales of potential value in measuring premorbid risk for schizophrenia have been sufficiently tested for reliability and validity in the context of primary prevention of schizophrenia. The absence of a sufficiently sensitive and specific means for identifying those at high risk of schizophrenia before the onset of psychosis is a major barrier to valid measurement of the outcome of attempts at primary prevention. However, there have been advances in the development of instruments relevant to the goals of secondary and tertiary prevention. Most studies use instruments developed for patients with established psychoses and have applied them to early psychosis groups with some success, although possible ‘floor’ effects may confound measurement in the ‘prodromal’ period.

Interest in the early detection and treatment of psychosis has recently turned towards the possibility of primary prevention of schizophrenia and related psychoses. McGlashan noted that primary prevention in mental health involves reducing the occurrence of new cases (i.e. incidence), while secondary prevention aims to delay onset or prevent recurrence (i.e. reduce prevalence), and tertiary prevention aims to decrease the duration or severity of symptoms and associated disability [1]. Three strategies have been proposed to achieve the primary, secondary and tertiary prevention of schizophrenia and related psychoses [2]. These are: (i) the early detection of groups at increased risk of developing a psychotic illness; (ii) the early treatment of first episode psychosis; and (iii) targeting interventions at the early ‘critical period’ of psychosis. In terms of primary prevention, it has been proposed that if the distinguishing features of those people who subsequently develop psychosis can be identified, then it may be possible to prevent the development of frank psychotic symptoms [1,3–5]. Therefore, the main focus of primary prevention studies of schizophrenia has been to develop strategies and criteria for the identification of groups at increased risk of developing a psychosis.

Known risk factors for schizophrenia include family history (i.e. genetic risk), season of birth, migrant/minority status, perinatal complications, neurodevelopmental deficits, and behavioural or social maladjustment. However, each of these is associated with relatively small increases in risk relative to the general population. Certainly, none alone can be taken to indicate a sufficient degree of risk of schizophrenia to warrant preventive interventions at a clinical level. This handful of risk factors of small magnitude, together with the fact that schizophrenia is a disorder of low prevalence, suggest that even if multiple risk factors were taken into account there are still such gaps in our knowledge that the development of a multifactorial risk index of sufficient sensitivity and specificity to have clinical utility represents a formidable challenge. It is not sufficient, for primary prevention in the clinical (as opposed to population) domain, to be merely at ‘increased risk’, but to be at ‘high risk.’

The attempts by McGorry and colleagues to identify young people at sufficiently increased risk to justify intervention before the onset of psychotic symptoms has relied on state and trait risk variables that can be readily identified in a clinical setting [3,4,6]. This has entailed identifying individuals during what may otherwise have later been regarded, retrospectively, as the prodromal period of schizophrenia.

However, there are significant problems with this approach. First, the risk categories based on these state and trait variables [3] have not been sufficiently tested for predictive validity at this stage. Indeed, very little research has been done in following the progress of such groups to evaluate whether the purportedly ‘high risk’ individuals do go on to experience a first episode of psychosis. Although the early work in this area is very promising [6], there is a need for large scale studies to determine whether, using this or other methods of identifying ‘high risk’ groups, primary prevention is indeed feasible. Rigorous longitudinal measurement of outcomes in purportedly ‘high risk’ individuals is essential for validation of the criteria for determining risk status. Without such fundamental information, systematic prospective studies of primary prevention will lack a sound scientific basis. Second, even if a risk index of acceptable sensitivity, specificity and positive predictive value were available, then for low incidence/prevalence disorders such as schizophrenia it is by no means clear that such an instrument could, both theoretically and practically, be applied in a clinical setting sufficiently early to provide opportunity for primary prevention.

Notwithstanding the above, clinical studies of putative ‘high risk’ individuals have typically used published work from studies of established psychoses to guide the investigation of possible predictive factors and domains of interest [1,3,7,8], thereby influencing the choice of quantitative outcome measures of potential applicability in primary prevention. It is not yet clear whether these measures are suitable for this purpose.

By comparison, the situation with respect to outcome measurement in secondary and tertiary prevention is much clearer. Falloon et al. have indicated that the secondary and tertiary prevention of schizophrenia could best be achieved through the early detection and treatment of psychotic symptoms, and continued monitoring for any recurrence of such symptoms [9]. Outcome measurement in this context needs to involve the investigation of the efficacy of treatment interventions and detection systems for patients in the early stages of an established psychotic illness. Thus, the aims of outcome measurement in schizophrenia prevention are similar to those discussed by Stedman et al. in the context of general mental health [10]; that is, the reliable and valid measurement of outcomes is necessary for the evaluation of treatment effectiveness. In addition, outcome measurement will allow the identification of groups for whom intervention is most effective, and the description of service provider and patient characteristics associated with a positive clinical outcome. In combination, these factors influence decision making, both at a clinical level, and at a resource allocation level.

Given that outcome measurement is vital to the investigation of schizophrenia prevention, it is necessary to describe suitable dimensions to be measured, as well as the quantitative tools used to do so. Andrews and colleagues outlined important considerations in choosing appropriate outcome measures [11]. Measures should be applicable, clearly assessing the domains of interest. In psychometric terms, measures should be reliable and valid, and should also be sensitive to change in the domain of interest. Studies of outcome in groups with established psychosis have focused on change in positive and negative symptoms following treatment intervention, evaluation of social and occupational role functioning, and investigation of the predictive validity of diagnosis in longitudinal follow-up studies [12,13].

The present paper aims to summarise a selection of quantitative measures commonly used in the field of schizophrenia research. The measures were selected on the basis of their widespread application, their use as outcome measures in schizophrenia prevention or intervention studies, and their potential applicability to studies of the ‘prodromal’ or first-episode psychosis populations. Table 1 provides a summary of the measures detailed in this article, including their structure and the domains they attempt to measure.

Table 1

Summary of measures commonly used in schizophrenia research

Scale	Structure	Domains assessed
Symptom measurement
Brief Psychiatric Rating Scale (BPRS) [14]; Expanded Brief Psychiatric Rating Scale (BPRS-E)[15]	This is an 18- or 24-item clinician-rated scale. Scores for each item range from 1 to 7.	Areas rated include: somatic concern, anxiety, depression, guilt, hostility, suspiciousness, unusual thought content, grandiosity, hallucinations, disorientation, conceptual disorganisation, excitement, motor retardation, blunted affect, tension, mannerisms and posturing, uncooperativeness, emotional withdrawal, suicidality, self-neglect, bizarre behaviour, elevated mood, motor hyperactivity, and distractibility.
Scale for the Assessment of Negative Symptoms (SANS) [23]	The SANS is a clinician-rated scale designed to assess the experience of negative symptoms, typically in the month before assessment. Scores for each item range from 0 to 5. Global ratings are made for major areas of assessment, reflecting subject responses to several sub-items (provided in brackets at right).	Affective flattening or blunting (unchanging facial expression; decreased spontaneous movements; paucity of expressive gestures; poor eye contact; affective non-responsivity; inappropriate affect; lack of vocal intonations); alogia (poverty of speech, poverty of content of speech, blocking, increased latency of response); avolition–apathy (grooming and hygiene, impersistence at work or school, physical anergia); anhedonia– asociality (recreational interests and activities, sexual interest and activity, ability to feel intimacy and closeness, relationship with friends and peers); and attention (social inattentiveness, inattentiveness during mental state testing).
Scale for the Assessment of Positive Symptoms (SAPS) [24]	The SAPS is a clinician-rated scale designed to assess the experience of positive symptoms, typically in the month prior to assessment. Scores for each item range from 0 to 5. Global ratings are made for four major areas of assessment, reflecting subject responses to several sub-items (provided in brackets at right).	Hallucinations (auditory, voices commenting, voices conversing, somatic or tactile, olfactory, visual); delusions (persecutory, jealousy, sin or guilt, grandiose, religious, somatic, reference, being controlled, mind reading, thought broadcasting, thought insertion, thought withdrawal); bizarre behaviour (clothing and appearance, social and sexual, aggressive and agitated, repetitive or stereotyped); and positive formal thought disorder (derailment, tangentiality, incoherence, illogicality, circumstantiality, pressure of speech, distractible speech, clanging).
Positive and Negative Syndrome Scale (PANSS) [27]	Consists of a structured interview and 30 items each rated on a 7-point scale.	Consists of three subscales: (i) positive symptoms (delusions, conceptual disorganisation, hallucinatory behaviour, excitement, grandiosity, suspiciousness, hostility); (ii) negative symptoms (blunted affect, emotional withdrawal, poor rapport, passive-apathetic social withdrawal, difficulty in abstract thinking, lack of spontaneity and flow of conversation, stereotyped thinking); and (iii) general psychopathology (somatic concern, anxiety, guilt feelings, tension, mannerisms and posturing, depression, motor retardation, uncooperativeness, unusual thought content, disorientation, poor attention, lack of judgement and insight, disturbance of volition, poor impulse control, preoccupation, active social avoidance).
Positive and Negative Syndrome Scale–Child and Adolescent Version (Kiddie-PANSS) [31]	Consists of a semi-structured interview section to be completed with the primary caregiver, as well as a longer (25 min) patient interview. Scoring is based on 30 symptoms derived from the adult PANSS.	Subscales of the Kiddie-PANSS are based on those of the adult PANSS. However, the definitions of 10 of the 30 items were modified to take into account developmental differences of children in the processes of cognition, affect regulation and impulse control.
Manchester Scale (MS) [32]	Eight-item clinician-rated symptom scale, with each item rated from 0 to 4.	The items are divided into two groups. The first assesses symptoms on the basis of responses to questions (depression, anxiety, delusions, hallucinations); and the second is based on the examiner's observations of the subject (incoherent or irrelevant speech, poverty of speech, flattened or incongruous affect, psychomotor retardation).
Symptom Checklist-90-Revised (SCL-90-R) [34]	Ninety-item self-report measure, rated on a 5-point Likert scale ranging from 0 ‘not at all’ to 4 ‘extremely’.	The items are divided into 10 clinical scales of somatization, depression, anxiety, obsessive-compulsiveness, hostility, interpersonal sensitivity, phobic anxiety, paranoid ideation and psychoticism. In addition, three global symptom indices are also assessed.
Measurement of role functioning
The Quality of Life Scale (QLS) [39]	The QLS is administered as a 21-item semistructured interview. It focuses on symptoms and functioning during the 4 weeks prior to assessment. Each item is composed of three parts: (i) a brief descriptive statement; (ii) a set of suggested probe questions; and (iii) a 7-point scale with descriptive statements for every second point. High scores on each scale represent normal functioning, while low scores represent severely impaired functioning.	Four broad areas are covered: (i) interpersonal relations (household members, intimate relationships, active acquaintances, level of social activity, involved social network, social initiative, social withdrawal, sociosexual relations); (ii) instrumental role (occupational role, occupational functioning, level of accomplishment, level of satisfaction); (iii) intrapsychic foundations (sense of purpose, motivation, curiosity, anhedonia, time utilisation, empathy, emotional interaction); and (iv) common objects and activities.
Role Functioning Scale (RFS) [40]	Four items scored on 7-point rating scales with a global index based on a summed score.	Assesses working; independent living and self-care; immediate social network relationships; and extended social network relationships.
Life Skills Profile [42]	This is a 39-item questionnaire, with five subscale scores and a total score. Higher scores represent better functioning. May be completed by service providers or primary carers.	Subscales are: self-care, non-turbulence, social contact, communication and responsibility.
Global scales
Global Assessment Scale (GAS) [43]	The GAS is a clinician-rated general functioning scale, ranging from 1 to 100 with 10-point divisions. Each division is characterised by a descriptive statement in behavioural terms, and the clinician rates subjects based on their lowest level of functioning in the previous week. Higher scores indicate greater health. Rating is made on the basis of all information available to the clinician (e.g. through interviewing, informant reports, behavioural observation, etc.).
Multidimensional scales
Health of the Nation Outcome Scales (HoNOS) [46]	Developed as a routine measure of mental health consumer outcomes. It consists of 12 scales scored from 0 to 4. A total score is calculated using the sum of the 12 items. There are also four subscales rated: behaviour; impairment; symptoms; and social skills. Rating is made on the basis of all information available to the clinician (e.g. through interviewing, informant reports, behavioural observation, etc.).	Twelve scales that rate problems with: overactive, aggressive, disruptive or agitated behaviour; suicidal thoughts or behaviour; non-accidental self-injury; problem drinking or drug taking; cognitive problems; physical illness or disability; hallucinations and delusions; depressed mood; other mental and behavioural problems; social relationships; activities of daily living; living conditions; and occupation and activities.
Mental Health Inventory (MHI) [50]	This is a 38-item scale completed either as a self-report or as part of an interview. Each item has a descriptive statement concerning a particular symptom or state of mind, and subjects report to what extent they have experienced this event in the past month.	The items allow a global Mental Health Index score to be derived; psychological distress and wellbeing scores; and six subscales of anxiety, depression, loss of behavioural/emotional control, general positive affect, emotional ties, and life satisfaction.
Other scales of interest
Early Signs Scale (ESS) [51]	Patient self-report instrument consisting of four subscales. There are 34 items on a scale ranging from 0 to 3, with high scores indicating more problems. Typical completion time is less than 5 minutes.	Subscales include anxiety, depression, disinhibition and incipient psychosis.
Insight Scale (IS) [54]	Patient self-report instrument consisting of eight items.	Three subscales of re-labelling symptoms, awareness of illness and need for treatment.
Knowledge About Schizophrenia Questionnaire (KASQ) [55]	This is a 25-item multiple choice test.	Assesses patients' knowledge about the diagnosis of schizophrenia, its prevalence, aetiology, course, prognosis, medical and non-medical treatment, stress factors and legal issues.

Symptom measurement

Brief Psychiatric Rating Scale

The Brief Psychiatric Rating Scale (BPRS) is a widely used measure of symptoms, developed originally by Overall and Gorham [14], and also available in an expanded version (BPRS-E) [15]. The scale has most frequently been used in initial assessment and outcome evaluation in a variety of psychiatric disorders [16–18], as well as research into the structure of psychotic symptoms [19–22]. The BPRS may be used at regular intervals for research purposes, typically ranging from one day to one month. The areas covered provide a broad perspective across numerous areas of subject behaviour and symptoms. However, although the reference group against which ratings are made are normal controls, the scale may suffer from a floor effect with early psychosis or ‘prodromal’ subjects; that is, the scale may not be sensitive to the relatively mild fluctuations in symptoms of possibly ‘prodromal’ patients. In addition, there is evidence that the lack of firm definitions for the BPRS items may result in nonstandardised use in different centres [16]. Nevertheless, overall the BPRS has very good psychometric properties. It is strongly correlated with equivalent items on the Scale for the Assessment of Positive Symptoms (SAPS) (r > 0.88) [16] and Scale for the Assessment of Negative Symptoms (SANS) (r > 0.84) [16], (r = 0.89) [21], suggesting it has excellent concurrent validity. The BPRS has also been found to have good internal consistency (α = 0.69 and 0.68 for the positive and negative syndrome scales, respectively) [23].

Scale for the Assessment of Negative Symptoms

The SANS [24] was designed to assess negative symptoms in patients with psychotic disorders [18], most notably schizophrenia. It has also been used to investigate the theoretical structure of psychotic symptoms [19,25]. The SANS provides a detailed assessment of patients' negative symptoms. However, it is possible that the use of the scale in primary prevention studies will suffer from floor effects, as most ‘prodromal’ individuals will cluster around scores of 0 or 1. The SANS has been widely used and has good psychometric properties. Its concurrent validity is demonstrated through its high correlation with the BPRS [21], as described earlier.

Scale for the Assessment of Positive Symptoms

The SAPS [26] was developed as a companion measure to the SANS, and focuses on delusions, hallucinations and thought disorder. It has also been used with the SANS to investigate the theoretical structure of psychotic symptoms [19,25]. The SAPS provides a detailed assessment of the positive symptoms of psychosis, but may suffer from similar floor effects as the SANS when used in ‘prodromal’ individuals. Along with the SANS, the scale has been used extensively, and has sound psychometric properties.

Positive and Negative Syndrome Scale

The Positive and Negative Syndrome Scale (PANSS) is designed to assess the positive and negative symptoms associated with schizophrenia [27]. In addition, the scale aims to measure the relative contributions of positive and negative symptoms to the patient's clinical presentation. The scale has been used in studies of the theoretical negative-positive symptom dichotomy in schizophrenia [28]. The investigators reported excellent reliability and strong criterion and construct validity [29]. In one study, the PANSS demonstrated consistently better interrater reliability and predictive validity than the BPRS [23]. Assessments of the interrater reliability of the PANSS have shown intraclass correlation coefficients (ICC) of 0.72, 0.80 and 0.56 for the positive, negative and general psychopathology scales, respectively. In addition, the scale demonstrates concurrent validity with the SAPS and SANS (r = 0.70 and 0.81 for the positive and negative scales, respectively) [30].

Positive and Negative Syndrome Scale for children and adolescents

The Positive and Negative Syndrome Scale for children and adolescents (Kiddie-PANSS) was developed as an instrument based on the adult PANSS, but for use in the assessment of children and adolescents with psychiatric disorders [31]. Owing to the age group for which the scale is designed, it is possible that the Kiddie-PANSS may be useful in primary prevention or ‘prodromal’ studies, which frequently target younger age groups than studies of first episode or chronic schizophrenia. The investigators reported that the Kiddie-PANSS demonstrated good interrater reliability, with ICC of 0.69, 0.89 and 0.52 for the positive, negative and general psychopathology scales, respectively. The instrument was also successful in distinguishing patients with schizophrenia from those with other psychiatric conditions.

Manchester Scale

The Manchester Scale (MS), also known as the Psychiatric Assessment Scale (PAS) [32], has been used to assess positive and negative symptoms in subjects with schizophrenia. It has also been used in conjunction with the Global Assessment Scale (GAS) to provide a comprehensive assessment and evaluation of treatment progress in terms of symptoms and general functioning [32,33]. A strength of the MS is that it can be used to obtain regular ratings to map progress over time. It has sound psychometric properties, correlating with comparable items on the BPRS, SAPS and SANS, suggesting acceptable concurrent validity [33].

Symptom Checklist-90-Revised

The Symptom Checklist-90-Revised (SCL-90-R) is a standardised self-report measure in which patients rate 90 common symptoms on a five-point Likert scale [34]. The instrument consists of 10 clinical scales of somatisation, depression, anxiety, obsessive-compulsiveness, hostility, interpersonal sensitivity, phobic anxiety, paranoid ideation and psychoticism. In addition, three global symptom indices are also assessed. While the author of the scale reported acceptable validity and reliability, others have suggested that certain populations are not adequately identified by the scale. For example, in patients with obsessive-compulsive disorder the scale was found to have poor divergent and criterion-related validity [35]. In patients with substance abuse plus psychiatric disability, the dimensional structure of the SCL-90-R was found to vary across genders [36]. The norms of the scale for adolescents have also been called into question [37], and in a study of malingering using the scale, 100% of untrained volunteers were able to fake psychopathology at a level comparable to psychiatric patients [38]. These findings suggest that the SCL-90-R may not have sufficient specificity to assist in accurately identifying early psychosis patients.

Role functioning measurement

Quality of Life Scale

The Quality of Life Scale (QLS) was developed for use both as an outcome measure and a measure of change for patients with schizophrenia [39]. The measure systematically elicits detail on a wide range of functioning, focusing on deficit symptoms rather than positive symptoms. The use of an additional instrument is therefore necessary to monitor positive symptoms. The researchers reported strong interrater reliability when users were given appropriate training in the use of the instrument (intraclass correlation coefficients 0.91 for intrapsychic foundations, 0.94 for interpersonal relations, 0.97 for instrumental role, 0.94 for commonplace objects and activities, and 0.94 for total score).

Role Functioning Scale

The Role Functioning Scale (RFS) [40] was designed as an outcome measure for people with chronic mental disability and established schizophrenia [10]. Although it has not been extensively reported in the literature [10], it includes most of the dimensions of role functioning relevant to patients of mental health services [11]. It has been reported that the scale has excellent internal consistency (α = 0.86) [40], adequate test–retest and interrater reliability, and adequate construct and criterion validity, and correlates negatively with the Health of the Nation Outcome Scales (HoNOS) (r = −0.58, p < 0.01) [10]. The scale also has modest convergent validity with the GAS [41].

Life Skills Profile

The Life Skills Profile (LSP) was developed to assess functioning of patients with chronic severe mental illness [42]. The scale is focused towards community-based rather than inpatient services. Owing to its focus on chronic mental illness and established schizophrenia [10], the utility of the measure is uncertain for schizophrenia prevention studies other than tertiary prevention. However, the psychometric properties of the LSP are sound, with excellent item reliability and adequate test–retest reliability and internal consistency (α = 0.93 for the total score; subscales range from α = 0.67 to α = 0.88). The scale also demonstrates adequate content and criterion validity, as well as sensitivity to change [10].

Global scales

Global Assessment Scale

The Global Assessment Scale (GAS) has been frequently used for psychiatric inpatient planning and evaluation [43]. It has also been used in conjunction with the MS to provide an overall picture of functioning [33], and to monitor treatment progress [44]. Owing to the structure of the scale, it is theoretically applicable to a wide range of functions, from those suffering from severe mental illness-related disability to those with superior functioning in most areas. It also takes into account a broad range of areas, not just psychiatric symptoms. Therefore, it is useful as a summary statement of functioning and could be applied to studies of schizophrenia prevention. Its wide use may be partly attributable to its relative simplicity, ease of coding and speed. However, although the scale is useful in recording change and monitoring overall progress, it is unable to monitor specific symptom change over time, and cannot provide the detail that may be achieved with multidimensional rating scales. The scale has been reported to have modest convergent validity with the RFS [41], and good interrater reliability and concurrent validity. The GAS was modified for inclusion in the DSM-III-R as the Global Assessment of Functioning (GAF) scale, which was again modified for the DSM-IV as the SOFAS, where it is used to evaluate adaptive functioning (axis V) [45].

Multidimensional scales

Health of the Nation Outcome Scales

The HoNOS was developed for use as a routine measure of mental health consumer outcomes [46]. It is widely used as a clinical instrument to measure patients on a range of dimensions. A study by Amin and colleagues, which followed first-onset psychosis patients after 3 years, reported that the HoNOS demonstrated strong concurrent validity through significant correlations with other scales, such as the SANS and GAF [47]. Their study also found moderate to good interrater reliability (intraclass correlation coefficients ranged from 0.47 to 0.85 for the subscales, and 0.59 overall). Similarly, the Victorian field trial of the HoNOS reported acceptable interrater reliability (intraclass correlation coefficient of 0.71) [48]. However, Bebbington and coworkers have suggested that the HoNOS has poor interrater reliability, possibly related to the training and experience of professionals using the scales [49]. In their study, no item had a κ value above 0.40 (‘fair’).

Mental Health Inventory

The Mental Health Inventory (MHI) was originally designed as a measure of mental health for the general population, although it has also been used in studies of mental health in a wide range of physical disorders [50]. Stedman et al. noted that the instrument takes little time to administer, and collects information about a wide range of areas [10]. However, as the scale has been used mostly in nonpsychiatric populations, its relevance for people with severe mental illness is uncertain [10], as is its applicability in early or ‘prodromal’ schizophrenia. The psychometric properties of the instrument are favourable, with sound test-retest reliability (r = 0.97) and internal consistency (α = 0.97), and adequate content and criterion validity.

Other measures of interest

Early Signs Scale

The Early Signs Scale (ESS) [51] has been used in studies attempting the early identification of relapse in schizophrenia [52,53]. Jorgensen reported that the scale had excellent predictive validity, and sound sensitivity and specificity [53]. Given these factors, the scale would appear to be extremely useful for studies in secondary prevention. The measure also takes little time and is simple to administer.

Insight Scale

The Insight Scale (IS) was designed as a self-report measure to investigate patient levels of insight [54]. The psychometric properties of the scale were investigated in a population of patients with nonaffective psychoses. Results suggested that the IS demonstrated excellent test–retest reliability (total scale r = 0.90), as well as sound construct, concurrent and criterion validity. The IS was also found to be sensitive to change, and may therefore be used as a longitudinal measure of progress over time from the ‘prodromal’ period onwards.

Knowledge about Schizophrenia Questionnaire

The Knowledge About Schizophrenia Questionnaire (KASQ) was designed to evaluate patients' knowledge regarding schizophrenia and its management [55]. Psychometric evaluation revealed excellent reliability as assessed by internal consistency (α = 0.89 and α = 0.85 in two samples) and test–retest (r = 0.83, p < 0.005) procedures. In addition, the measure was found to be sensitive to change. However, the test was specifically designed to evaluate a particular psychoeducation program, and may not be readily applicable to prevention studies.

Discussion

The above constitutes a summary of selected outcome measures that are potentially useful in schizophrenia prevention studies. Many of the instruments presently available to researchers have been designed for people with an established diagnosis of schizophrenia, and tend to have been developed with groups of mentally ill patients. Therefore, it is likely that the scales will not adequately account for the experience of the premorbid or ‘prodromal’ individual who is not experiencing symptoms of psychosis other than, perhaps, in attenuated form. It is also likely that floor effects of the scales will limit the utility of such instruments in this context.

We could find no scale suitable for evaluating the validity of categories used to allocate people to groups at putative ‘high risk’ of psychosis, which would be relevant in the primary prevention arena [3,4,7]. Yet evaluating the validity of the ‘high risk’ categories is vital if the goal of primary prevention of schizophrenia is to be achieved. In the absence of ‘high risk’ categories of demonstrable reliability and validity, the efficacy of primary prevention strategies cannot be properly tested. This is because, if putative ‘high risk’ patients do not proceed to a psychotic episode, it will be impossible to know whether this is due to successful intervention, lack of predictive validity in the determination of risk status, or failure to follow the patients through the period of highest likelihood for the onset of psychosis. Therefore, it is imperative that the ‘high risk’ criteria be empirically validated, and that measures that focus specifically on the assessment and longitudinal monitoring of premorbid or ‘prodromal’ patients be developed and tested.

A particular problem in developing a measure of ‘high risk’ of schizophrenia is that none of the known risk factors, several of which were listed in the introduction, are specific to schizophrenia. Similarly, none of the state or trait features that define the ‘high risk’ categories currently in clinical use [3] are specific to schizophrenia. Other conditions in which such risk indicators are commonly found premorbidly include affective psychoses; other psychoses; some personality disorders and disturbances of conduct; substance use disorders; learning disorders; and even some neuroses. Furthermore, the substantial overlap between the phenomenology (and other features) of schizophrenia and many other psychiatric syndromes is well known. It is possible that the state of our current knowledge in this area may accurately reflect the true nature of psychosesand related disorders, and that the present nosological system in psychiatry may not map well onto risk factors or antecedent factors. If this is the case, it may be time to take a broader view of primary prevention and focus not just on schizophrenia but on the spectrum of severe mental disorders that share common risk factors.

While the goal of primary prevention is yet to be realised, secondary and tertiary prevention of schizophrenia is far less complex and may be adequately measured by instruments presently available. Although most of the scales have been developed for use in populations with established psychotic illness, they have also been used in studies of first-episode psychosis. In particular, the measures of symptoms appear to be useful in evaluating early treatment of patients experiencing a first episode of psychosis [56,57]. In addition, there have been indications in the literature of the development of quantitative scales, which are of particular relevance to working with preventing relapse in first episode patients, such as the ESS [51] and the IS [54].

Measurement of outcome in the context of secondary and tertiary prevention is not beset by the conceptual and methodological problems that presently apply in the domain of primary prevention. Fifty years ago there was very little written about recovery from schizophrenia. Nowadays, recovery is a more commonplace event and is widely reported in the outcome literature. The concept of any form of prevention of schizophrenia 50 years ago would have been virtually unthinkable. Yet, today we are in a strong position to prevent much of the morbidity, disability and suffering associated with this illness, provided that interventions of demonstrated effectiveness in secondary and tertiary prevention are broadly implemented. Beyond this, the goal of primary prevention appears more remote. Advances in neuro-imaging such as high resolution structural MRI and functional MRI, electrophysiology, neuropsychology and, perhaps, genetic screening, may help to bring that goal closer to realisation.