In an attempt to classify the multitude of mental disorders he encountered in his work, Emil Kraepelin adopted the term ‘dementia praecox’ to label a condition characterised by early psychosis and cognitive deterioration (Reference HoenigHoenig, 1983). Although Bleuler renamed the disease schizophrenia in 1911, emphasising his view of the disease as a lack of connection between a person's affect, thought and perception, he still viewed cognitive deficits as integral to the disorder (Reference GabrieleGabriele, 2000). The perceived importance of cognition in schizophrenia has since waxed and waned. Cognitive and negative symptoms, which were regarded as integral by both Kraepelin and Bleuler, were later overshadowed by the more easily observable and identifiable positive symptoms. The Research Diagnostic Criteria (RDC), which were designed to formalise the diagnosis of mental disorders, emphasised the Schneiderian symptoms, and this tradition has continued into DSM–III and IV (Reference AndreasenAndreasen, 1997). Although negative symptoms were added as criteria in DSM–IV, cognition is still not included in the formal criteria.
Although the move away from cognitive impairment as a focus in schizophrenia was initially motivated by enhancement of diagnostic reliability, it came to shape how the disease was viewed and investigated. However, a renewed interest in cognition has been evident recently, spurred in part by the strong empirical relationship between cognition and real-world functioning (Reference GreenGreen, 1996). Several studies have failed to demonstrate a significant correlation between positive symptoms and functional outcome (Reference GreenGreen, 1996), suggesting that a diagnostic and treatment focus on Schneiderian first-rank symptoms has sidelined key aspects of the disease.
IMPAIRED COGNITION AS A CORE FEATURE
A mounting body of evidence indicates that diminished cognitive ability is a core feature of schizophrenia. Severely impaired performance on cognitive tests (two standard deviations below the mean of healthy controls) in several cognitive domains is strong evidence for the importance of cognitive impairment in the disease (Reference Saykin, Gur and GurSaykin et al, 1991; Reference Harvey and KeefeHarvey & Keefe, 1997) Broad cognitive deficits, of moderate to severe magnitude, have been found in meta-analysis (Reference Heinrichs and ZakzanisHeinrichs & Zakzanis, 1998), large clinical trials (Harvey et al, Reference Harvey, Green and McGurk2003, Reference Harvey, Siu and Romano2004; Reference Keefe, Bilder and HarveyKeefe et al, 2006a ) and research studies (Reference Bilder, Goldman and RobinsonBilder et al, 2000; Reference Heaton, Gladsjo and PalmerHeaton et al, 2001; Reference Keefe, Goldberg and HarveyKeefe et al, 2004). Cognitive deficits have been shown to lack correlation with severity of positive symptoms and to be only mildly correlated with severity of negative symptoms (Reference Addington, Addington and Maticka-TyndaleAddington et al, 1991; Reference Gold, Queern and IannoneGold et al, 1999; Reference Keefe, Bilder and HarveyKeefe et al, 2006a ), indicating that impaired cognition is not an epiphenomenon of clinical symptoms.
Although some studies have indicated that a significant portion of people with schizophrenia test in the normal cognitive range (Reference Palmer, Heaton and PaulsenPalmer et al, 1997), strong evidence suggests that even these exhibit cognitive abilities below those expected if they did not have the disease. A study of monozygotic twins found that 80–95% of twins with schizophrenia scored below their unaffected twin (Reference Goldberg, Torrey and GoldGoldberg et al, 1993). Another study found that 98% of people with schizophrenia performed below the level predicted by estimates of their premorbid functioning based on level of parental education, compared with 42% of controls (Reference Keefe, Eesley and PoeKeefe et al, 2005).
Many early studies of the cognitive deficit in schizophrenia were of people who were either taking antipsychotics at the time of the study or had taken them in the past. However, several studies have since demonstrated cognitive deficits in people with first-episode schizophrenia who have never taken antipsychotics (Reference Saykin, Shtasel and GurSaykin et al, 1994; Reference Mohamed, Paulsen and O'LearyMohamed et al, 1999; Reference Bilder, Goldman and RobinsonBilder et al, 2000; Reference TorreyTorrey, 2002).
Unlike Schneiderian first-rank symptoms, cognitive deficits correlate highly with measures of functional outcome (Reference Velligan, Mahurin and DiamondVelligan et al, 1997; Reference Addington and AddingtonAddington & Addington, 1999; Reference Green, Kern and BraffGreen et al, 2000). In addition, the literature overwhelmingly supports a longitudinal correlation between cognitive ability at baseline and later assessments of functional outcome (Reference Green, Kern and HeatonGreen et al, 2004; Reference Carlsson, Nyman and GanseCarlsson et al, 2006), suggesting that cognitive deficits are a key and perhaps limiting factor in rehabilitation of people with schizophrenia. The overwhelming evidence supporting neurocognitive deficits as a core feature of schizophrenia and predictive of functional outcome has spurred the United States National Institute of Mental Health (NIMH) to target such deficits for pharmacological intervention (Reference Marder and FentonMarder & Fenton, 2004).
SEPARATE DOMAINS OF DEFICIT V. GENERAL DEFICIT
A crucial consideration for practical assessment of cognition in schizophrenia centres around whether the cognitive deficit is best described as broad or is more pronounced in specific domains. A large factor-analytical study involving the Wechsler Adult Intelligence Scale (WAIS–III) and Wechsler Memory Scale (WMS–III) batteries found that the performance of 1250 healthy controls was best described by a model composed of six separable domains of cognition: verbal comprehension, perceptual organisation, auditory memory, visual memory, working memory and processing speed (Reference Tulsky and PriceTulsky & Price, 2003). Although a meta-analysis of 22 studies of cognitive performance in populations with schizophrenia reported a strikingly broad deficit spanning all domains (Reference Heinrichs and ZakzanisHeinrichs & Zakzanis, 1998), some theories have focused on specific domains of impairment, such as working memory (Reference Goldman-RakicGoldman-Rakic, 1994), verbal memory (Reference Saykin, Shtasel and GurSaykin et al, 1994), and executive functions (Reference Goldberg, Weinberger and BermanGoldberg et al, 1987). However, these results may reflect differences in test sensitivity as opposed to true differential ability across domains.
Although some studies have emphasised differential impairment across domains, others indicate that cognitive performance (Reference Keefe, Bilder and HarveyKeefe et al, 2006a ) and cognitive deficits (Reference Dickinson, Iannone and WilkDickinson et al, 2004) exhibited by people with schizophrenia are largely mediated through a single common factor, suggesting a generalised cognitive impairment. This ongoing debate has implications for the aetiology of the disease (whether underlying brain abnormalities are local or global) as well as intervention strategies.
TOOLS FOR MEASURING COGNITION
A great many tasks have been developed to assess cognition and various batteries comprised of these tasks have been employed in research with populations with schizophrenia (Table 1). The most appropriate battery for a given study will depend upon the questions being addressed, study settings and available resources. A long battery comprised of many different tests has the disadvantage that missing data will be increased (Reference Keefe, Goldberg and HarveyKeefe et al, 2004). In addition, attrition rates and missing data may be higher in those with the most impairment, thus skewing results. Also, in many settings extensive batteries are impractical because of the time requirements placed on staff members administering the tests. However, a longer test battery usually will increase the ability of the data to measure multiple domains of cognition. Thus if the research question involves the efficacy of a treatment intervention for improving cognition, a composite score from a small battery might be sufficient and allow for a larger number of participants to complete the study while requiring fewer staff resources. If, however, the research question involves relative strengths and weaknesses of various cognitive domains, a thorough battery composed of multiple tasks in each domain may be required. If the research question is primarily focused on one cognitive domain, a brief general battery in combination with multiple tests of the domain of primary interest might be most appropriate.
Advantages and disadvantages of selected cognitive batteries
| Battery | Advantages | Disadvantages |
|---|---|---|
| WAIS-III and WMS-III | Long history of use allows comparison with previous studies, norms available | Lengthy, not designed specifically for research in schizophrenia |
| MCCB | Designed by panel of experts for research in schizophrenia, allows for domain scores with minimal testing, norms available | Many domain scores based on performance on one test |
| SCoRS | High face validity, easily administered by clinicians, minimal time requirements, demonstrated correlation with other measures of cognitive and functional outcomes | High subjectivity |
| UPSA | High face validity, proxy test of real-world functioning, minimal time requirements | Domain-level analysis not possible |
| RBANS | Minimal time requirements, small practice effects, performance correlated with that on WAIS-III and WMS-III. | Lacks measures of important cognitive domains in schizophrenia, significant ceiling effects |
| BACS | Minimal time requirements, designed for use in schizophrenia research, high correlation with composite scores from more extensive batteries, minimal practice effects, available in nine languages, norms available | Domain-level analysis is minimal |
| BCA | Extreme brevity of administration, high correlation with extensive cognitive battery and functional outcome measures | Domain-level analysis minimal |
| Computerised batteries | Automatic administration reduces rater error | Validity of many tests has not been examined |
| Psychophysiological tasks | Much of underlying neurobiological circuitry is known, afford quick assessment of schizophrenia-related endophenotypes | Relationship between improvement in these tasks and functional outcome unknown |
WAIS-III, Wechsler Adult Intelligence Scale - III; WMS-III, Wechsler Memory Scale - III; MCCB, MATRICS Consensus Cognitive Battery; SCoRS, Schizophrenia Cognition Rating Scale; UPSA, UCSD Performance-Based Skills Assessment; RBANS, Repeatable Battery for the Assessment of Neuropsychological Status; BACS, Brief Assessment of Cognition in Schizophrenia; BCA, Brief Cognitive Assessment
Wechsler Adult Intelligence and Memory Scales
The Wechsler Adult Intelligence Scale (WAIS; Reference WechslerWechsler, 1997a ) and Wechsler Memory Scale (WMS; Reference WechslerWechsler, 1997b ) have long been the most widely employed batteries of assessment for IQ and memory in healthy populations. However, the WAIS–III alone requires approximately 100 min for completion in a mixed clinical population (Reference Ryan, Lopez and WerthRyan et al, 1998). For studies of populations with schizophrenia, researchers using these batteries have tended to reduce the number of sub-tests administered to reduce demands on the patients and staff. Blyler et al (Reference Blyler, Gold and Iannone2000) used regression analysis to determine the four tests covering all four domains of functioning assessed by the WAIS–III that would best account for the variance in full-scale IQ in a sample of 41 out-patients with schizophrenia. They found that a shortened version of the WAIS–III, consisting of the sub-tests information, block design, arithmetic and digit symbol took only 30 min to administer and accounted for 90% of the variance in the full-scale IQ of the schizophrenia patients. Because of its brevity, the shortened version of the WAIS may have utility as a routine measure of cognition in clinical practice.
MATRICS Consensus Cognitive Battery
As part of the NIMH Measurement and Treatment Research to Improve Cognition in Schizophrenia (MATRICS), the MATRICS Consensus Cognitive Battery (MCCB) was constructed to provide a standard battery for the assessment of cognition-enhancing drugs (Reference Nuechterlein and GreenNuechterlein & Green, 2006). From more than 90 tests nominated for inclusion, a final battery of ten tests covering seven domains of cognitive functioning was chosen with a view to practicality of administration, high test–retest reliability, small practice effects, small ceiling effects and demonstrated relationship to functional outcome. The domains and relevant tests which comprise the final battery are: speed of processing (Brief Assessment of Cognition in Schizophrenia – Symbol Coding; Category Fluency; Animal Naming; Trail Making Test Part A), attention/vigilance (Continuous Performance Test – Identical Pairs), working memory (WMS–III Spatial Span; University of Maryland Letter-Number Span), verbal learning (Hopkins Verbal Learning Test – Revised), visual learning (Brief Visuospatial Memory Test – Revised) reasoning and problem-solving (Neuropsychological Assessment Battery – Mazes) and social cognition (Mayer–Salovey–Caruso Emotional Intelligence Test – Managing Emotions). The MCCB is a relatively concise battery (taking just 65 min to administer) that none the less allows for measurement of cognition at a domain-specific level. Because the final battery was co-normed with 300 community controls across five academic sites, an individual's results can be normalised against this same control group for all seven domain scores as well as a composite score. The MCCB has a computerised scoring system that produces t-scores and percentiles corrected for age and gender. Although the MCCB was designed to assess effects of pharmaceutical interventions on cognition in schizophrenia, the battery is suitable for use in cognitive remediation and non-intervention studies of people with schizophrenia. Although the MCCB requires more time to administer than the shortened WAIS–III, it has the potential to provide a more detailed assessment of a patient's cognitive performance.
Schizophrenia Cognition Rating Scale
The Schizophrenia Cognition Rating Scale (SCoRS; Reference Keefe, Poe and WalkerKeefe et al, 2006b ) is an 18-item interview-based assessment which covers all the cognitive domains tested in the MCCB, except social cognition, and takes approximately 12 min to complete. It is administered separately to the patient and to an informant (family member, friend, social worker, etc.) The interviewee is asked to rate the patient's level of difficulty in performing various cognitive functions on a 4-point scale, with 4 being the most difficulty and 1 being the least. Upon completion of the 18 items, the interviewee is asked to give a global rating of the patient's cognitive functioning on a scale of 1–10. After the interview has been administered to both the patient and the informant, the interviewer ranks the patient on all 18 items, and gives a global score based on the responses of both the patient and informant as well as the interviewer's observations of the patient.
Initial assessments of SCoRS results have demonstrated high interrater reliability (Reference Keefe, Poe and WalkerKeefe et al, 2006b ). The administrator's global rating was shown to be the single SCoRS measure that correlated most significantly with measures of cognition (BACS; Brief Assessment of Cognition in Schizophrenia; Reference Keefe, Goldberg and HarveyKeefe et al, 2004), performance-based assessment of function (UPSA; Reference Patterson, Goldman and McKibbinPatterson et al, 2001) and real-world assessment of function (Independent Living Skills Inventory (ILSI); Reference Menditto, Wallace and LibermanMenditto et al, 1999). Step-wise regression analysis demonstrated that the interviewer's global rating accounted for significant variance in real-world functioning as measured by the ILSI beyond that explained by results from the BACS and the UPSA (Reference Keefe, Poe and WalkerKeefe et al, 2006b ). Because the SCoRS assessment is based on patient and informant reports, it has high face validity.
In addition to its utility as a coping measure with the MCCB in drug trials, the SCoRS is ideally suited for use in the clinic and may thus increase awareness of cognitive deficits in the diagnosis and treatment of people with schizophrenia. Because patient scores have been found to account for little variance in cognitive performance, functional capacity or real-world functioning scores beyond that accounted for by informant ratings (Reference Keefe, Poe and WalkerKeefe et al, 2006b ), it is possible that informant ratings alone could be collected when an informant has sufficient contact with the patient. The assessment time could then be limited to 15 min. In addition, the SCoRS should be a familiar procedure for clinicians, who should require significantly less training than for batteries involving less familiar cognitive testing procedures. Interrater reliability of the SCoRS should be established before it is used for clinical or research purposes.
UCSD Performance-Based Skills Assessment
The University of California San Diego Performance-Based Skills Assessment (UPSA) was developed as a proxy of real-world functioning that is implemented in role-play. The UPSA measures daily living skills by recreating, in a clinical environment, situations a patient is likely to encounter in the real world. The tasks fall into five categories of functional skills: household chores; communication; finance; transportation; and planning recreational activities. The assessment is relatively brief, requiring an average of 30 min to administer and has demonstrated high interrater reliability (Reference Patterson, Goldman and McKibbinPatterson et al, 2001). Initial assessment found that UPSA scores for people with schizophrenia and schizoaffective disorders correlated significantly with both negative symptom severity and cognitive impairment, but not positive symptom severity (Reference Patterson, Goldman and McKibbinPatterson et al, 2001). Owing to the high face validity imparted by the ‘real-world’ nature of the assessment, the MATRICS committee adopted the UPSA as a second co-primary measure for studies of cognition in schizophrenia. The high degree of face validity might increase the cooperation of patients when it is used as a routine clinical assessment.
Repeatable Battery for the Assessment of Neuropsychological Status
The Repeatable Battery for the Assessment of Neuropsychological Status (RBANS; Reference RandolphRandolph, 1998) is a brief (45 min) assessment originally designed to test cognitive performance in older patients which has shown utility in providing reliable assessment of cognitive performance in populations with schizophrenia (Reference Wilk, Gold and BartkoWilk et al, 2002; Reference WeberWeber, 2003). The performance of people with schizophrenia on the RBANS is highly correlated with performance on the much longer WAIS–III and WMS–III batteries (Reference Gold, Queern and IannoneGold et al, 1999; Reference Hobart, Goldberg and BartkoHobart et al, 1999). Because it was designed to be administered repeatedly, the RBANS does not suffer from large practice effects. However, because the battery was developed to test for dementia, it is comprised largely of tests of memory, language and visual perception, and may suffer from ceiling effects on some sub-tests when used in people with schizophrenia. The battery also lacks measures of motor, executive and working memory performance, cognitive domains thought to be important in the cognitive impairment observed in schizophrenia. Despite these omissions, the RBANS is an appealing tool for the assessment of cognition in routine clinical practice owing to its relative brevity.
Brief Assessment of Cognition in Schizophrenia
The Brief Assessment of Cognition in Schizophrenia (BACS; Reference Keefe, Goldberg and HarveyKeefe et al, 2004) retains the positive attributes of the RBANS (brevity of administration and scoring, repeatability and portability) and more completely assesses the extent of cognitive impairment over multiple domains thought to be affected by schizophrenia (Table 2). The BACS, available in nine languages, requires approximately 30 min to complete and is devised for easy administration and scoring. The battery is specifically designed to measure treatment-related changes in cognition, and has alternate forms, thus minimising practice effects. The battery includes brief assessments of executive functions, verbal fluency, attention, verbal memory, working memory and motor speed, and generates a composite score that is calculated by summing z-scores derived by comparisons with a normative sample of 400 healthy controls. Its reliability, validity and comparability of forms has been established empirically (Reference Keefe, Goldberg and HarveyKeefe et al, 2004). The composite score has high test–retest reliability in people with schizophrenia and healthy controls (intraclass correlation coefficients>0.80). The BACS composite score has been shown to be as sensitive to the cognitive deficits of schizophrenia as a standard 2.5-hour battery (Reference Keefe, Goldberg and HarveyKeefe et al, 2004) and is highly correlated (r=0.84, P<0.001) with the composite score derived from the CATIE neurocognitive test battery (Reference Keefe, Sweeney and GuKeefe et al, 2007). The BACS also has clear functional relevance, as the composite score is strongly related to functional measures such as independent living skills (r=0.45), performance-based assessment of functioning (r=0.56) and interview-based assessments of cognition in people with schizophrenia (r=0.48) (Reference Keefe, Poe and WalkerKeefe et al, 2006c ). The BACS is well suited to routine clinical administration when a quick assessment of overall cognitive functioning is required.
Classification of BACS tests according to MATRICS neurocognitive domain 1
| MATRICS neurocognitive domain | BACS tests |
|---|---|
| Processing speed | Verbal fluency |
| Category instances (supermarket items) | |
| Letter fluency (F and S words) | |
| Processing speed | Token Motor Task |
| Symbol coding | |
| Reasoning and problem-solving | Tower of London Test |
| Verbal memory | List learning |
| Working memory | Digit sequencing |
BACS, Brief Assessment of Cognition in Schizophrenia; MATRICS, Measurement and Treatment Research to Improve Cognition in Schizophrenia
1. Keefe et al (Reference Keefe, Sweeney and Gu2007)
Brief Cognitive Assessment
An even shorter battery is the Brief Cognitive Assessment (BCA; Reference Velligan, DiCocco and Bow-ThomasVelligan et al, 2004), which was designed to assess cognition in people with schizophrenia in 15 min. Initial assessment of the battery has indicated good test–retest reliability and strong correlation (r=0.72; P<0.0001) with an extensive 2-hour battery (Reference Velligan, DiCocco and Bow-ThomasVelligan et al, 2004). The two batteries showed similar correlations with measures of functional ability. Normative data are available allowing adjustments for practice effects when performing repeated assessments. The extreme brevity of the BCA makes it a strong candidate for routine clinical administration.
Computerised batteries
A recent development in cognitive assessment for clinical trials is the availability of computerised test batteries that allow direct data transfer to study databases. These methods minimise rater error and reduce the costs for human quality assurance. However, many of these methods have not been fully validated and therefore results must be evaluated carefully.
PSYCHOPHYSIOLOGICAL TASKS
Assessment of psychophysiological tasks is appealing because much of the underlying neurobiological circuitry has been uncovered in animal and human studies. However, care must be taken when inferring an aetiological basis or treatment strategy for schizophrenia from the performance of patients on these tasks. The outstanding question remains whether interventions that improve the performance of patients on these tasks would have any effect on the whole of cognition or on functional outcomes. Regardless of their utility in uncovering neurobiological underpinnings of the disease, psychophysiological tasks are particularly appealing for use in genetic studies as tools to quickly assess an endophenotype that might reflect a specific genotypic vulnerability to schizophrenia. Some of the most utilised psychophysiological tasks are briefly described below.
Eye movements
Several eye movement abnormalities have been associated with schizophrenia. Two of the most prominent are abnormalities in smooth-pursuit eye movements, in which the patient is instructed to maintain foveation of a smoothly moving target, and antisaccade performance, in which the patient is instructed to make a mirror image saccade away from a suddenly appearing visual cue.
Antisaccade
Lesion studies in non-human primates have demonstrated the importance of the dorsolateral prefrontal cortex for inhibiting reflexive prosaccades in the antisaccade paradigm (Reference Fukushima, Fukushima and MiyasakaFukushima et al, 1994). Likewise, converging evidence has suggested that the dorsolateral prefrontal cortex is compromised in people with schizophrenia (Reference Bunney and BunneyBunney & Bunney, 2000). A review of patients' performance on the antisaccade task strongly indicates a significant elevation in erroneous prosaccades that is stable over time (Reference Everling and FischerEverling & Fischer, 1998). This was recently replicated in a seven site study by the Consortium on the Genetics of Schizophrenia in which the antisaccade performance of 143 people with schizophrenia was compared with that of 195 controls (Reference Radant, Dobie and CalkinsRadant et al, 2007). All sites found a significant difference in the number of errors (reflexive prosaccades) made by the two groups. In addition, first-degree relatives of people with schizophrenia have demonstrated higher reflexive saccade rates than unrelated controls (Reference Clementz, McDowell and ZisookClementz et al, 1994), suggesting that the endophenotype reflects a genetic vulnerability to schizophrenia.
Smooth-pursuit eye tracking
Decreased pursuit gain has long been viewed as a characteristic impairment in people with schizophrenia. However, it was recently shown that people with affective disorder displayed an indistinguishable smooth-pursuit gain (Reference Kathmann, Hochrein and UwerKathmann et al, 2003). Likewise, unaffected relatives of the two groups did not differ in their pursuit gain deficiencies. These results argue against the utility of smooth-pursuit gain as a phenotypic marker reflecting a genotype specific to schizophrenia. However, high rates of catch-up saccades (Reference Sweeney, Clementz and HaasSweeney et al, 1994) and anticipatory saccades (Reference Rosenberg, Sweeney and Squires-WheelerRosenberg et al, 1997) in the smooth-pursuit paradigm appear to be specific to schizophrenia and may offer phenotypic measures for genetic studies.
Prepulse inhibition and P50
Prepulse inhibition and P50 are both measures of pre-attentive processing that display impairment in people with schizophrenia and have been thoroughly researched in animal models. However, impairment in both of these paradigms is fairly widespread over various psychiatric populations, thus decreasing the utility of these measures as an endophenotype for schizophrenia (Reference BartBart, 2004)
CONCLUSIONS
Cognitive impairment is returning to prominence in the conceptualisation and practical assessment of schizophrenia, and is being considered for inclusion in the ICD and DSM diagnostic criteria. A wide variety of batteries currently exist for assessing cognitive impairment in relation to the disease, and cognitive outcome measures are essential as the field moves forward. However, current options are lacking in several regards. Notably, most cognitive outcome measures tend to suffer from low face validity: it is not obvious to patients or caregivers that improvements in performance on these batteries would make a difference to the patient's quality of life. Therefore, it is essential in many circumstances that an appropriate test of functional outcome is co-administered with a cognitive battery. Interview-based measures and functional proxy measures are based upon measures of intuitive value to patients and clinicians, and are amenable to use in the clinic, where cognitive assessments are far from routine because of the onerous nature of administering most batteries. Also, although modern cognitive batteries strive to include tasks with low practice effects, this confound is substantial. Thus, trials involving repeated assessments of patients require comparison with comparable control groups that have experienced the same testing procedure. Another consideration in the choice of cognitive testing batteries is that differential sensitivity to between-group or over-time differences across individual tasks can cause spurious findings of differential impairment across domains of cognitive functioning. The parsing of cognitive functioning into various domains is weakened by the lack of domain specificity for many cognitive tests (Reference KeefeKeefe, 1995). Despite these shortcomings, cognitive outcome measures have reshaped our view of schizophrenia and will be essential for identifying its aetiology as well as designing more effective interventions.
eLetters
No eLetters have been published for this article.