Introduction

Schizophrenia is a heterogeneous disorder, with some patients manifesting prominent negative symptoms and cognitive impairment [1, 2], features that are intimately linked to outcomes in this illness [3–5], while others experience mild or even relative absence of these symptoms. One long-standing approach to dissecting the heterogeneity of schizophrenia has been to parse a subgroup of patients who experience prominent, persistent, and primary negative symptoms, dubbed the “deficit” subtype of schizophrenia [6]. Several lines of investigation have supported the position that this subtype is, in fact, separable from the non-deficit subtype, including evidence of a unique profile of neuropsychological impairment [7–9].

One of the earliest studies examining cognition in the deficit syndrome found that these patients were more impaired on a global cognition measure compared with non-deficits patients [10]. Moving beyond more severe generalized cognitive impairment, later work aimed at discerning whether a different profile of cognitive test results emerged for deficit versus non-deficit patients, which if found would further support the validity of this syndrome. Indeed, some studies support this contention, finding selective impairments among deficit patients on tests putatively sensitive to frontoparietal functioning [11, 12]. To synthesize the existing findings, Cohen and colleagues meta-analyzed these data and found evidence for a moderate generalized cognitive impairment characterizing deficit syndrome patients [13]. Notably though, the studies included in the meta-analysis mostly evaluated a small number of deficit patients, typically less than 30, with two exceptions. The first of these studies included 25 adult and 94 geriatric deficit patients and found these patients had poorer performance on a global cognition measure and in individual cognitive domains [14]. The second study, which assessed 58 deficit patients, also found general cognition impairments among this group of patients, but did not find evidence for differential performance on specific cognitive tests [12].

Since the aforementioned meta-analysis, several additional studies on the topic of cognitive performance in patients classified as having the deficit syndrome have been reported. One of the larger studies, involving 143 patients with deficit schizophrenia and 123 non-deficit patients, found evidence for global cognitive impairments among the deficit syndrome patients [15]. Another recent and relatively large study with 153 deficit patients reported that these patients scored worse on a measure of verbal memory, but this was not the case for other cognitive tests [16]. Along similar lines, an additional study evaluated 30 deficit patients and identified worse performance only on the Wisconsin Card Sorting Test, a different cognitive domain [17]. In addition, a study examining 29 deficit patients noted inferior performance among these individuals in several, but not all, cognitive tests [18]. Yet another study with 26 deficit patients found inferior performance among these patients on fluency and memory tests [19], while an investigation involving 88 deficit patients reported evidence of generalized cognitive impairments among this subgroup [20]. The finding of global cognition deficits has also been found in five other studies, the first with 15 deficit schizophrenia patients [21], the second and third each with 27 deficit patients [22, 23], the fourth with 29 deficit patients [24], and the other evaluating 51 patients classified as having the deficit syndrome [25]. Some contradictory evidence also exists. Specifically, one study with 15 deficit patients failed to find a significant performance difference compared to non-deficit patients on a measure of global cognition [26]. Another study with 17 first-episode deficit patients found evidence for global cognitive deficits, but this finding was not observed in the 52 chronic deficit schizophrenia patients evaluated [27]. These findings, along with the previous meta-analysis, support the notion of more severe generalized cognitive impairments in patients with deficit schizophrenia compared to non-deficit patients. The notion of a differential or unique profile of cognitive impairment in deficit patients remains without substantial support.

The present study sought to further examine the neuropsychological substrates of deficit schizophrenia and compare the degree of performance differences in deficit versus non-deficit patients. We were also interested in potential cognitive differences between deficit and non-deficit patients when the latter group experienced prominent, albeit not primary or persistent, negative symptoms. Lastly, we evaluated the utility and reliability of cognitive test results in differentiating deficit patients from non-deficit schizophrenia patients. Based on the published literature, we hypothesized that deficit patients would evidence moderate impairment in global cognition compared with non-deficit patients.

Methods

Study design and participants

Data from the Clinical Antipsychotic Trial of Intervention Effectiveness (CATIE) study for chronic schizophrenia were utilized in the present study. Details of the study design and rationale [28], as well as primary findings [29], have been presented elsewhere. The purpose of the CATIE study was to compare the effectiveness of atypical and conventional antipsychotic medications through a randomized controlled trial. In the first phase of the trial, 1493 patients were randomized to receive olanzapine, perphenazine, quetiapine, risperidone, or ziprasidone under double-blind conditions and were followed for up to 18 months or until treatment was discontinued for any reason [28]; those who discontinued their initially assigned medication were eligible to receive other antipsychotics and continue in the trial.

The present study utilizes data from the 6- and 12-month visits. Data from the pre-randomization visit were not examined to limit the impact of changes in clinical status and to evaluate a relatively more stable group of patients. The 6-month visit of the CATIE study therefore serves as baseline for the present investigation.

The study inclusion criteria have been reported previously [28]. Briefly, participants were eligible if they were between the ages of 18 and 65 years and had a diagnosis of schizophrenia confirmed using the Structured Clinical Interview for DSM-IV Axis I Disorders [30]. Participants were excluded from the study if they had a diagnosis of schizoaffective disorder, mental retardation, pervasive developmental disorder, delirium, dementia, amnesia, or other cognitive disorders; had only one episode of schizophrenia; were pregnant or breast-feeding; or had a serious and acutely unstable medical condition.

The study was approved by the institutional ethics review board at each site, and written informed consent was obtained from the patients or their legal guardians.

Instruments and procedure

The neuropsychological tests that comprised the CATIE cognitive battery have been described in detail in a previous report [31]. Scores on individual tests were converted to z-scores and combined to construct the following cognitive domain scores: (1) processing speed, calculated as the average of the Revised Wechsler Adult Intelligence Scale Digit Symbol Test, Grooved Pegboard, and the mean of the Controlled Oral Word Association Test and category instances; (2) working memory, as measured by the average of the Letter–Number sequencing test and a computerized test of visuospatial working memory; (3) verbal memory, assessed with the average of three trials from the Hopkins Verbal Learning Test; (4) vigilance, as evaluated by the d-prime summary score from the Continuous Performance Test; and (5) reasoning, measured by an average score from the Revised Wechsler Intelligence Scale for Children Mazes test and a computerized version of the Wisconsin Card Sorting Test [32]. These are all domains assessed by the MATRICS cognitive consensus battery [33]. The five domain scores were then averaged to create a cognitive composite score [32].

Social cognition was evaluated using the Facial Emotion Discrimination Task (FEDT) [31, 34]. The FEDT is designed to evaluate emotional processing, which is one key component of social cognition [35].

Psychopathology was evaluated using the Positive and Negative Syndrome Scale (PANSS) [36] and the Calgary Depression Scale for Schizophrenia (CDSS) [37]. The PANSS is a 30-item clinician-administered instrument that evaluates the severity of a range of positive, negative, and general psychopathology symptoms. The CDSS is a 9-item clinician-administered instrument that was specifically designed to assess depressive symptoms in schizophrenia, separate from negative symptoms [38, 39].

Extrapyramidal symptoms (EPS) were evaluated using the abbreviated Simpson-Angus Rating Scale (SAS) [40, 41], which contains six items rated on a scale from 0 to 4 (higher scores denoting more severe side effects). The specific items include gait, arm dropping, shoulder shaking, elbow rigidity, wrist rigidity, and tremor.

The Heinrichs–Carpenter Quality of Life Scale (QLS) [42] was used to measure deficit symptom psychopathology and community functioning. Specifically, items from the intrapsychic foundation subscales were used to evaluate the severity of specific deficit symptoms. The items from the remaining subscales of the QLS including the instrumental role functioning, interpersonal relations, and common objects were summed and used as a measure of community functioning [43].

Identifying deficit syndrome patients

Individuals were classified as deficit/non-deficit based on previously published criteria for the deficit syndrome [6]; patients with schizophrenia are classified as having the deficit syndrome if they experience at least two negative symptoms with moderate-to-severe severity, and if these symptoms are persistent and cannot be accounted for by secondary influences. The Schedule for the Deficit Syndrome (SDS) [44] is considered the “gold standard” deficit syndrome classification tool. Although the present study did not employ this instrument, we created a proxy using a combination of other instruments, as has been done in previous work [14, 45]. Specifically, we created a proxy SDS. The SDS evaluates the severity and persistence of negative symptoms, and estimates whether these symptoms are idiopathic. All three criteria must be met for classification as deficit schizophrenia; patients must have persistent and primary negative symptoms.

The SDS evaluates the severity of six negative symptoms: restricted affect, diminished emotional range, poverty of speech, curbing of interests, diminished sense of purpose, and diminished social drive [44]. These items were substituted with items from the PANSS and QLS. Specifically, restricted affect was evaluated using the “blunted affect” item from the PANSS; diminished emotional range with the “anhedonia” item from the QLS, and in addition the absence of marked hostility and anxiety as rated on the PANSS; poverty of speech with the “lack of spontaneity and flow of conversation” item from the PANSS; curbing of interests with the “curiosity” item from the QLS; diminished sense of purpose with the “motivation” item from the QLS; and diminished social drive from the “passive/apathetic social withdrawal” item from the PANSS. If two or more of these were present with moderate-to-severe severity (i.e., score >3 on the PANSS or score <4 on the QLS), the severity criteria of the SDS were met. Patients were followed prospectively, and psychopathology was reassessed after 6 months. If the severity criteria remained true at this follow-up, negative symptoms were deemed persistent. Lastly, as there was no clinical judgement of whether negative symptoms were primary or potentially secondary to other factors, as is done in the SDS, we enforced additional criteria. Specifically, we excluded individuals with other severe symptoms that could potentially, but not necessarily, confound the assessment of negative symptoms. These criteria encompassed exclusion of individuals with moderate-to-severe depression, defined as a total score >6 on the CDSS [46], as well as individuals who experienced EPS, operationalized as having a score of >1 on any item of the SAS. Patients with marked suspiciousness, defined as having a PANSS item score >4, were also excluded. These exclusion criteria were evaluated both at baseline and at the follow-up visit. Individuals meeting the severity and persistence criteria, and not meeting any of the exclusion criteria, were classified as deficit syndrome patients.

It is noteworthy that we opted to apply the aforementioned criteria to classify patients as deficit/non-deficit, rather than using the Proxy for the Deficit Syndrome (PDS) criteria [47]. This was done for several reasons. PDS scores are calculated by subtracting the severity of blunted affect by the severity of affective symptoms, including depression, anxiety, tension, and hostility. This measure therefore incorporates the rating of one negative symptom and the lack of dysphoria (i.e., a potential source of secondary negative symptoms). However, this approach does not tap into “amotivational” negative symptoms that are measured on the SDS [44], and has found to characterize a portion of deficit schizophrenia patients [48]. Also, the PDS does not take into consideration other possible sources of negative symptoms. One example being EPS, which might confound the assessment of blunted affect; indeed, in the present sample, those with at least moderate severity of blunted affect had significantly higher ratings of EPS (t 651 = 4.26, p < 0.001). It is also worth mentioning that the PDS was designed to be used in instances where the SDS was not available or not feasible to be used [47]. Given the availability of side effect data and prospective follow-up of patients in the present study, we chose to develop a measure that more closely aligned with the SDS and therefore incorporates more facets of the deficit syndrome criteria [6]. However, for comparison, we also repeat our main analyses using the PDS to define deficit schizophrenia, and the findings are similar (Supplemental Tables 1 and 2).

Statistical analyses

Patients with deficit syndrome were compared with non-deficit patients in terms of performance on neurocognitive tests using both univariate and multivariate statistical tests. First, potential group differences on the composite cognition score were examined using a one-way analysis of variance model (ANOVA). Because the performance scores on the social cognition measure were highly skewed [32], potential group differences on this measure were evaluated using the Mann–Whitney nonparametric U test. Next, differences on individual cognitive domains were examined using a multivariate ANOVA model, followed by univariate one-way ANOVAs for completeness. We then explored the independent predictive value of individual cognitive domain scores in determining deficit status using logistic regression modeling. This was followed by a discriminant function analysis, which was designed to shed further light on the value of cognitive test results in classifying patients into deficit versus non-deficit groups.

Next, we compared deficit patients to only those non-deficit patients who also experienced negative symptoms at baseline, although these symptoms were not necessarily persistent or idiopathic in nature. These two groups were compared on measures of cognitive test performance using ANOVAs.

Cohen’s d values are presented as measures of effect size [49]. For reference, a Cohen’s d value of 0.5 (i.e., a medium effect size) translates to approximately 33 % non-overlap in scores between groups, while a value of 0.2 (i.e., a small effect size) translates to approximately 14.7 % non-overlap or 85.3 % overlap, between groups [49]. Statistical tests were considered significant at a p value of <0.05 (two-tailed). All analyses were conducted in SPSS Statistics version 20 (IMB Corp., Armonk, NY, USA).

Results

Patient characteristics

Sociodemographic and clinical characteristics of the sample are presented in Table 1. Symptom and neurocognition data were available for 657 participants. Of these, 144 individuals (21.9 %) experienced prominent, primary, and enduring negative symptoms and were therefore classified as having the deficit subtype of schizophrenia. The construct validity of the deficit classification is supported by the notion that these patients had significantly more severe negative symptoms compared to non-deficit patients [7], but at the same time did not differ in severity of positive symptoms (Table 1); deficit patients also experienced less severe depressive symptoms and poorer community functioning. Deficit patients also had significantly higher PDS scores than non-deficit patients at baseline (F 1,654 = 31.67, p < 0.001) and follow-up (F 1,604 = 34.55, p < 0.001).

Table 1 Sociodemographic and clinical characteristics of sample
Full size table

Cognitive performance in deficit schizophrenia

Deficit patients had significantly poorer performance on cognitive tests as measured by the composite score (Table 2). The effect size of these differences was notably modest to moderate. Deficit patients also demonstrated poorer performance relative to non-deficit patients on individual cognitive domains taken together (Wilk’s λ = 0.85, F 5,608 = 21.96, p < 0.001). Follow-up univariate tests revealed that deficit syndrome patients with schizophrenia had lower scores in every cognitive domain compared with non-deficit patients (Table 2). These between-group differences remained even after controlling for level of education and sex (all p’s < 0.05). Between-group differences for each cognitive measure are presented in Table 3; the largest group difference effect size emerged for the category fluency measure, whereas one of the smallest effect sizes was observed for the mazes test.

Table 2 Cognitive performance scores in deficit versus non-deficit schizophrenia
Full size table
Table 3 Performance on individual cognitive test measures in deficit versus non-deficit schizophrenia
Full size table

We next turned to the question of prediction (i.e., whether neuropsychological test scores can reliably differentiate deficit versus non-deficit patients). This was accomplished using two complementary methods. First, we constructed a logistic regression model predicting deficit syndrome status using cognitive domain scores as predictors. The overall model was significant despite relatively poor fit (χ 2 = 31.22, df = 5, p < 0.001; Negelkereke R 2 = 0.08), and the processing speed score (Wald’s χ 2 = 5.74, OR 0.70, CI [0.53, 0.94], p = 0.02) and the verbal learning score emerged as significant predictors of deficit status (Wald’s χ 2 = 5.55, OR 0.75, CI [0.59, 0.95], p = 0.02; other predictors, all p’s >0.05). Next, a discriminant function analysis was conducted. The first discriminant function was significant (χ 2 = 31.00, df = 5, p < 0.001; canonical correlation coefficient = 0.23), and in line with the logistic regression analysis, processing speed and verbal learning scores demonstrated the largest loading on this function (0.61 and 0.53, respectfully; all others <0.20). However, examination of the classification results revealed that this model had less than ideal (i.e., 62.3 %) overall accuracy in predicting individuals who met criteria for deficit syndrome, with approximately 37.7 % of participants being misclassified (false-positive rate = 38.3 %; false-negative rate = 35.4 %). These results, taken together with the effect size of the between-group univariate effects, suggest that deficits patients do not reliably perform poorer on cognitive tests.

Comparing deficit patients with non-deficit patients with negative symptoms

As the severity of negative symptoms has been found to be associated with cognitive performance in patients with schizophrenia [32, 50–52], we wanted to examine whether the observed differences in cognitive test results between deficit versus non-deficit patients would hold when the non-deficit comparison group also presented with negative symptoms, albeit not persistent or primary. Two hundred and thirty-five individuals in the non-deficit group experienced prominent negative symptoms at baseline (i.e., met the severity criteria for deficit syndrome). The clinical characteristics of these two groups are shown in Table 4; notably, the two groups did not differ in terms of negative symptom severity. When comparing these individuals with deficit patients, no group difference was found on the measure of global cognition (Table 5). Examining each individual domain of cognition revealed that only within the verbal learning domain was there differential performance, with deficit patients performing worse (Table 5); however, this difference would not survive correction for multiple comparison testing. Furthermore, each of these between-group differences failed to reach statistical significance after controlling for sex (all p’s > 0.05).

Table 4 Characteristics of deficit schizophrenia patients versus non-deficit patients with negative symptoms
Full size table
Table 5 Cognitive performance scores in deficit schizophrenia patients versus non-deficit patients with negative symptoms
Full size table

Discussion

The present investigation explored the question of whether individuals identified as having the deficit subtype of schizophrenia were neurocognitively distinct from non-deficit patients. Consistent with previous work [13], we found evidence for a generalized cognitive deficit characterizing patients with deficit schizophrenia, such that these patients scored on average approximately 0.4 SD below that of non-deficit patients. However, it is noteworthy that the magnitude of this effect is relatively modest, translating to over 70 % overlap in scores between groups [49]. Furthermore, cognitive test scores were not able to reliably distinguish patients into deficit versus non-deficit groups, which further underscores the modest difference and large overlap in scores between groups. Taken together, these results suggest that the deficit subtype of schizophrenia is not markedly distinct from non-deficit schizophrenia in terms of neurocognitive performance.

We found evidence for poorer cognitive test performance among patients classified as having the deficit syndrome versus a heterogeneous group of schizophrenia not classified as such. This is consistent with findings from a previous meta-analysis [13] and more recent studies [15]. However, only one study to our knowledge has examined potential differences between individuals with prominent, primary, and enduring negative symptoms (i.e., deficit syndrome) with those not meeting criteria for the deficit syndrome, but also manifesting prominent negative symptoms, albeit not idiopathic in nature or stable over time. This study contrasted 29 deficit patients with 12 non-deficit patients with negative symptoms and failed to find any marked difference between these two groups on a measure of global cognition [53]. Our findings echo these results in a larger sample, with deficit patients not demonstrating markedly different scores on the composite measure of cognition relative to non-deficit patients with prominent negative symptoms. These results suggest that previous findings of group differences between deficit and non-deficit schizophrenia patients on measures of cognitive test performance may have been driven by differential burden of negative symptoms between groups, rather than factors specifically linked to idiopathic and persistent negative symptoms that characterize the deficit syndrome.

Future studies aiming to delineate markers specific to deficit schizophrenia may consider contrasting this group with a less heterogeneous sample of non-deficit patients. This can be accomplished for example by excluding non-deficit patients with prominent negative symptoms. Repeating our discriminant function analysis while excluding this ambiguous group did in fact result in a slightly better fitting model with more deficit patients being classified correctly (data not shown). However, caveats of this approach should also be acknowledged. For instance, taking this approach now makes unclear whether findings between deficit and non-deficit groups are reflective of pathology unique to the deficit syndrome characterized as it is by idiopathic and enduring negative symptoms or to negative symptoms more broadly defined without regard to their etiology or persistence over time.

Our findings argue against the notion that cognitive test results can distinguish deficit versus non-deficit subtypes of schizophrenia, at least when measured using standard neuropsychological tests. Thus, findings emerging from cognitive tests do not convincingly support the notion of the deficit syndrome representing a unique and distinct subgroup of schizophrenia. Our results, along with those from previous neuropsychological studies [13], are more in line with the notion that deficit syndrome represents a quantitatively more severe form of schizophrenia that is not necessarily qualitatively distinct. This is not to say that other findings from different methodologies such as neuroimaging do not support this notion. In fact, several studies have found evidence for distinct neuropathology characterizing deficit schizophrenia versus non-deficit schizophrenia [54–61]. These findings, along with those from other lines of investigation, have so far supported the notion of deficit schizophrenia being a distinct subtype [7–9]. Our results, though, suggest that neurocognitive findings are not among those that can reliably parse this subgroup, at least as currently defined. It remains possible that only a portion of deficit schizophrenia patients demonstrates a unique profile of cognitive impairment.

The present investigation has limitations that should be acknowledged. First, participants in the present study were entering into a treatment trial, which may have introduced some degree of bias in the patients represented. Second, we employed a proxy identification method to classify individuals into deficit versus non-deficit schizophrenia, rather than using the gold standard SDS [44]. Although our proxy identification method mirrored the criteria of the SDS in the specific symptoms evaluated, stability criteria, and examination of primary versus secondary etiologies of symptoms, it remains that this method has not been validated. However, having the deficit group score significantly higher on the PDS serves as a preliminary validation [47]. Third, the present study focused mainly on neurocognition and evaluated social cognition using only a single test. Fourth, the proxy identification tool employed in the present study, like the SDS, was designed to be restrictive in classifying patients as having the deficit syndrome. That is, ambiguous cases are classified as non-deficit in order to parse a more homogenous subgroup of deficit patients. It remains that at least some of these ambiguous cases may actually share pathology with deficit syndrome patients. However, as we continue to define deficit syndrome patients based on presentation of clinical signs and symptoms, rather than basing this classification on an objective (biological) marker, such assertions remain conjecture and if true will continue to obfuscate the boundaries demarcating subtypes of illness. It would be interesting going forward to examine pathology within particular neural circuits across patients with schizophrenia and test whether aberrant functioning within a specific circuit can distinguish deficit versus non-deficit forms of schizophrenia.

In the present study, we replicated previous findings that deficit patients tend to perform poorer on cognitive tests than non-deficit patients; group differences emerged for all neuropsychological measures evaluated (Tables 2, 3). That is, we found evidence for a generalized cognitive deficit among individuals with deficit syndrome schizophrenia versus non-deficit patients, and our findings do not support the notion of differential neuropsychological impairment among deficit patients. Furthermore, we failed to find convincing evidence that deficit schizophrenia patients are cognitively distinct from non-deficit schizophrenia patients who also experience negative symptoms. Therefore, we suggest that the moderate generalized cognitive deficit found among deficit schizophrenia patients is more closely linked with negative symptoms rather than primary and enduring negative symptoms per se (i.e., the deficit syndrome). These findings suggest that we must turn to other tools that are more proximal to underlying pathology to better parse subgroups of patients within schizophrenia, at least with respect to the deficit syndrome. It will be important to discern going forward whether more specialized neuropsychological tests or measures derived from other methodologies can distinguish those schizophrenia patients with prominent, persistent, and idiopathic deficit symptoms versus those who present with negative symptoms more broadly defined (i.e., with putatively differing underlying pathology).