Schizophrenia Bulletin vol. 31 no. 4 pp. 910–921, 2005 doi:10.1093/schbul/sbi035 Advance Access publication on July 27, 2005 Negative Symptoms and Specific Cognitive Impairments as Combined Targets for Improved Functional Outcome Within Cognitive Remediation Therapy Kathryn E. Greenwood1,2, Sabine Landau3, and Til Wykes2 Key words: schizophrenia/cognition/mechanisms/ community function/neurocognitive/interventions 2 Department of Psychology, Institute of Psychiatry, De Crespigny Park, London SE5 8AF; 3Biostatistics and Computing, Institute of Psychiatry, London Introduction The association between cognition and functioning in schizophrenia has been widely reported in the last decade,1–2 yet the positive effects of remediation programs on cognition do not translate directly to improved real-life functioning.3 The cognitive mechanisms that should be targeted to achieve the best outcomes are, as yet, only loosely defined,2 and certain groups of people, such as those with negative symptoms, may respond better than others.4 Indeed both cognition and negative symptoms are associated with poor community function in schizophrenia, yet the role of negative symptoms on functioning is contentious.1, 5–15 Previous studies have considered the strong independent associations between cognition and community functioning, while symptoms have been assumed to contribute only indirectly through their link with cognition or through overlaps with definitions of functioning.1, 5 The independent contribution of symptoms to community functioning has not been fully explored, however, due to the confounding of negative symptoms with poor general cognition.16–18 The area is also complicated by the range of different cognitive measures, the use of global and questionnaire measures of functioning, and the scarcity of theoretical frameworks.1–2 Broad global domains such as IQ, executive function (Wisconsin Card Sorting Test), vigilance, memory, and more recently working memory have each been associated with poor function.1, 5–12 Indeed composite executive function scores, in particular, have been associated with daily living skills, occupation, and community function, as assessed through indirect questionnaire or global measures.1 The use of these global measures adds little, however, to our understanding of the mechanisms by which poor cognition is related to poor function. Negative symptoms are associated with the same global cognitive impairments in IQ, executive function, vigilance, memory, and working memory that predict poor community function.17–26 Negative symptoms have, however, also been linked theoretically and empirically with specific Negative symptoms and poor cognition are both associated with poor functional outcome in schizophrenia. This poor functional outcome has been attributed to poor cognition rather than any independent contribution from symptoms. Identifying target cognitive processes and mechanisms that predict community function, and possible moderator effects of negative symptoms, will allow the development of cognitive remediation programs that are successful in improving functional outcome. A referred sample of 53 in- and outpatients with schizophrenia with general cognitive impairment (including 28 with severe negative symptoms) and 22 healthy controls, balanced for premorbid IQ, were compared cross sectionally on measures of community shopping skills, executive function, and working memory. Across the groups, there were direct relationships between community functioning and specific executive functions, and there were interactions between group membership and the types of associations found. Working memory was independently associated with accurate community functioning only in people with schizophrenia and negative symptoms. This association was not due to the sole presence of working memory impairment or just to negative symptoms. Poor community function is predicted both by specific cognitive impairments that are prominent in people with negative symptoms and through the moderating effect of negative symptoms on the working memory–community function relationship. This may reflect a synergistic association between symptoms and cognition: negative symptoms arise from cognitive impairment but also impact detrimentally on working memory functioning. Both cognitive processes and negative symptoms should be targeted in cognitive remediation to effect the greatest change in community functions. 1 To whom correspondence should be addressed; tel: þ44 (0) 207 919 2983, fax: þ44 (0) 207 919 2473, e-mail: k.greenwood@ iop.kcl.ac.uk. Ó The Author 2005. Published by Oxford University Press on behalf of the Maryland Psychiatric Research Center. All rights reserved. For permissions, please email: [email protected]. 910 Targeting Cognitive Mechanisms to Improve Function cognitive processes. These processes have included deficits in the generation of plans and strategies, the initiation of actions, and the use of immediate working memory, which result in a restriction in spontaneous activity.27–28 Strategy, response initiation, and working memory impairments may constitute core impairments in schizophrenia with negative symptoms.29 The strong association between negative symptoms and global poor cognition has confounded investigations of specific cognition– function relationships. This confounding has rendered equivocal the independent effect of negative symptoms on community function when cognition is included. Certainly, while symptoms have been associated with poor community function in some first-episode and chronic schizophrenia studies,11–15 others have described only limited or indirect links through the association of symptoms with poor cognition.1, 5 Cognitive and neurocognitive remediation programs do improve cognition.3, 30–31 Some studies have suggested selectivity in the response to cognitive remediation therapy, with better effects, for example, for people with negative symptoms of schizophrenia than for those without these symptoms.4 Few studies have looked at real-world outcomes, and those that have report mixed results.3 Improved real-life community, occupation, and daily living skills do occur following remediation programs. The best effects are limited to comprehensive programs that incorporate a real-life element, such as neurocognitive enhancement with work therapy.32 Certainly it is apparent that improving cognition does not translate simply to improved functional outcome. This study will elucidate the cognitive processes and mechanisms that result in poor community function in schizophrenia, particularly for people with negative symptoms. Executive functions and working memory are targeted because of their close links both to negative symptoms and to poor functioning in the community. If general cognitive functioning is the significant driver of poor community function, then people with schizophrenia matched for broad general cognition but not negative symptoms should differ from healthy controls but not from each other. If, however, the main predictor of community function is specific executive processes, then these poorer executive processes would be associated with poorer community function within each group. Finally if negative symptoms are important in the prediction of community function, then they should moderate the relationship between executive function and community function. Method Design The study was cross sectional, with the dependent variables (community function) and the independent vari- ables (executive function and general cognitive level) measured in participants who fell into 1 of 3 groups: these were healthy controls and 2 groups of participants with schizophrenia defined according to the presence or absence of negative symptoms. Participants Participants with DSM-IV schizophrenia were in- or outpatients. They were referred by mental health teams because they demonstrated some problems with cognition and/or functioning and were deemed suitable candidates for cognitive remediation therapy. Participants with schizophrenia were excluded if they demonstrated intact general memory (a score of 22 or more on the Rivermead Behavioural Memory Test)33 and intact general executive function (a score of 4 categories or more on the Wisconsin Card Sorting Test)34 or intact social behavior function (a score of 3 or less on the Social Behaviour Scale).35 This enabled the 2 schizophrenia groups to be balanced for general cognitive level. These 53 participants with DSM-IV schizophrenia and cognitive impairments were divided into 2 groups—28 participants with negative symptoms and 25 participants without negative symptoms— and were compared to 22 healthy control participants. Participants with negative symptoms additionally fulfilled criteria of at least 2 negative symptoms (score of 3–5) on the Positive and Negative Syndrome Scale (PANSS), including at least 1 of the core psychomotor poverty symptoms of blunted affect and lack of spontaneity and flow of conversation.36 Healthy control participants comprised a nonpsychiatric volunteer sample recruited from within local employment and sheltered work centers in order to facilitate the balancing of premorbid IQ and education levels across samples. All participants gave informed consent for participation in this study, and ethical approval for the work was obtained from the South London and Maudsley National Health Service Trust Ethical Committee. All participants were aged 18–65, with premorbid IQ (National Adult Reading Test–Revised)37 above 70, English as a first language, no neurological problems or head injury, and no current substance abuse. Measures Executive Function Measures. Component executive process and task selection were theoretically and empirically motivated. Executive measures were selected that were specifically impaired in participants with negative symptoms in a previous study or were theoretically predicted to be specifically impaired in people with negative symptoms.27, 29 Working Memory. Verbal working memory was assessed using the Letter Number Span Task (scaled score).38–39 Working memory was defined as the ability 911 K. E. Greenwood et al. to hold information in memory for conscious processing and manipulation.40 Response Initiation–Inhibition. Verbal response initiation–inhibition was evaluated using 2 measures. Response initiation was assessed using the Phonological Fluency Test (total number of correct words).41 Response inhibition was assessed using the Hayling task (error scaled score).42 Participants were required to complete sentences presented orally, with a single word that made no sense at all in the context of the sentence. Meaningful and related completions contributed to an error score that reflected poor inhibition. Strategy Use. Strategy use was assessed using 3 measures. Spatial strategy use was evaluated using the Key Search Task (profile score) from the Behavioural Assessment of the Dysexecutive Syndrome.43 Participants were required to draw a spatial search route to ensure finding lost keys within a paper representation of a field. A poorer profile score reflects a less organized search strategy. Verbal strategy use was assessed using measures derived from the phonological fluency task (percentage of words using a phonemic strategy, percentage of words using a semantic strategy, number of clusters, and mean cluster length).44–46 Phonemic cluster words included successive words that differed by 1 letter, began with the same letter sound or 2 letters, or rhymed. Semantic cluster words are successive words that are linked by a superordinate semantic category, represent 2 forms of the same word, or form a common phrase. Verbal strategy use was also assessed using a strategy assessment from the Hayling task (strategy score).42 The measure was the percentage of meaningless sentence completions that encompassed a strategy of either reporting items visible in the room or reporting successive semantically related completions. Community Function Measures. Community function was assessed directly using a supermarket shopping task, introduced by Hamera and Brown47–48 and shown to be reliable and ecologically valid in assessing real-life community living skills in schizophrenia. The task was adapted for the United Kingdom, and measures were taken to reflect those relevant executive processes thought to underlie performance. Participants were required to select the correct item, size, and cheapest alternative for each of 10 items presented on a shopping list. All normal shopping strategies, such as requesting help from staff, were allowed: Accuracy was assessed through the total number of correct items, correct sizes, and lowest-price items selected. Efficiency was evaluated using the time taken. Redundancy of effort was calculated using the number of aisles entered above the minimum required when using the most efficient route. 912 Strategy was measured by the number of items selected when using an ordered progression through the aisles without entering unnecessary aisles. Supermarkets were chosen from a single chain of stores with comparable store layout and selection of items. Participants were taken to a novel store in order to provide the greatest test of executive functions. Allocation of participants to stores was such that group, gender, IQ, and age distributions were similar across stores, and confounding of these effects with store effects was avoided. Statistical Analyses The main objective of the statistical analyses was to identify the differential effects of group membership and specific executive functions on community function. Demographics and Clinical Data. One-way analyses of variance and chi-squared tests (for sex, category of illness length, proportion of atypical and cholinergic medication use) were employed to investigate group differences in sociodemographic and clinical characteristics and cognitive exclusion criteria. Executive Functions. Group differences in executive functions were investigated using analyses of variance for continuous variables and Kruskall Wallis tests for discrete noncontinuous variables (Key Search profile score and Hayling error scaled score). The Hayling task verbal strategy score was arcsine transformed to approximate normality. Community Functions. Group differences in community functions were investigated using generalized linear models,49 with group as a between-subject factor and model distribution specified according to the predetermined nature of the data, which were a 30-trial binomial distribution with logit link for accuracy and an equivalent 10-trial distribution for the strategy measure, a Poisson distribution with log link for redundancy, and a normal distribution with identity link for efficiency. Model Building. The associations between community functions and cognitive measures were analyzed in 3 steps using generalized linear regression analyses and the distributions described above. Corrections were made for multiple group comparisons. Premorbid IQ was included throughout the analyses in order to investigate specific cognition–function relationships. Step 1: Determining Which Variables Should Be in the Model. Relationships were tested across the whole group Targeting Cognitive Mechanisms to Improve Function between community function measures and single executive and working memory impairments identified from the analysis of group differences. Step 2: Determining Which Executive Process–Group Interactions Should Be in the Model. To identify community functions that were differentially predicted by executive processes in particular groups, individual executive function–group interaction terms were entered into the regression analyses, in addition to the main effects of group and of executive processes from previous analyses. Potential moderator effects were thus identified. Step 3: Final Model With All Significant Process and Process–Group Interactions Included. The full regression model was constructed for each community function measure. All individual executive functions from step 1, and all group–cognitive function interactions identified at step 2 were included. For interactions between group and executive function, the associations between community measures and executive functions were further assessed for each group separately. Systematic Analysis of Significant Process–Group Interactions. A possible explanation of the moderating effect of the negative group could be that executive function affects community function increasingly for more severe levels of executive impairments (a threshold effect). This was investigated for accuracy and verbal working memory. The negative group was divided into 2 subgroups with different levels of impairment. This resulted in 12 participants with severe working memory impairment (at least 2 standard deviations below the average standard score) and 16 participants with moderate to intact working memory. If the level of impairment affected the association between working memory and community function, then this would suggest that the impairment factor had a moderating effect in the negative group. Investigations of group differences in all analyses were adjusted for multiple group comparisons using Holm’s Multistage Bonferroni procedure: the smallest p values were compared in series to 0.05/3, 0.05/2, or 0.05/1, depending on the number of remaining group comparisons.50 Demographic and executive functions were investigated using the Statistical Package for the Social Sciences (version 10).51 Community function and association analyses were investigated in STATA (version 7).52 total psychopathology. The groups were balanced for premorbid IQ, age, sex, and years of education and did not differ clinically or statistically on any other measure. Importantly, the schizophrenia groups were also balanced for IQ and broad memory, executive, and social behavior measures. Do the Schizophrenia Groups Differ on Executive Functions? Executive measures have been linked to negative symptoms and therefore should be specifically impaired in the negative group compared to the non-negative group and healthy controls. The groups differed in verbal working memory performance (F[2,72] = 8.67, p < .001), with the negative group being significantly impaired compared to controls (p < .001 compared to 0.05/3). The nonnegative group showed a trend toward greater impairment than controls (p = .026 compared to 0.05/2) and did not differ significantly from the negative group (p = .061 compared to 0.05/1). The groups also differed in verbal response initiation (F[2,72] = 3.17, p = .048), with a trend for impairment in the negative group relative to controls (p = .022 compared to 0.05/3) and a similar level of impairment in the non-negative group relative to controls (p = .044 compared to 0.05/2). Finally, the groups differed in spatial strategy use (X2[2] = 16.3, p < .001), with greater impairment in the negative group compared to each other group (for controls, p < .001 compared to 0.05/3; for non-negative, p = .007 compared to 0.05/2) but no difference between the non-negative group and controls (p = .36; see figure 1). There were no significant group differences for the remaining executive measures. Are There Differences in Community Function Between the Groups? People with negative symptoms have poor functional outcome and specific impairments in working memory, strategy, and initiation processes that may underlie this poor function. For these reasons, people with schizophrenia and negative symptoms were predicted to perform less well on community function measures than either the comparison schizophrenia group or healthy controls. Table 2 presents the group comparisons on community function measures adjusted for multiple group comparisons. Demographics Accuracy. The negative group was significantly less accurate in selecting items correctly than either the nonnegative schizophrenia group or healthy controls (see table 2). The nonnegative group was also significantly less accurate than controls. Overall, the negative participants were less than half as likely as controls to accurately select items. Sociodemographic, clinical, and cognitive data are presented in table 1. By definition, the schizophrenia groups differed due to group selection on negative, general, and Efficiency, Redundancy, and Strategy Measures. The negative group also completed the task less efficiently and Results 913 K. E. Greenwood et al. Table 1. Sociodemographic, Cognitive, and Clinical Characteristics Statistical Test Characteristic Control, n = 22 Non-negative, n = 25 Negative, n = 28 F/X2 df p-Value Age (years) 36.2 (12.9) 35.3 (10.6) 35.1 (8.8) .09 2,72 .92 Sex (m/f) 16/6 19/6 20/8 .15 2 .93 Premorbid IQ 89.8 (11.3) 91.1 (14.2) 91.4 (13.3) .10 2,72 .90 Education (years) 12.2 (1.62) 11.9 (1.32) 11.7 (1.36) .83 2,72 .44 Social Behavior Schedule Total Score NA 11.8 (9.4) 14.2 (12.2) .66 1,51 .42 Rivermead Behavioural Memory Test Profile Score NA 15.2 (4.2) 13.5 (4.7) 2.0 1,51 .16 Wisconsin Card Sorting Test Category Score NA 2.9 (2.2) 2.0 (1.9) 2.4 1,51 .13 Illness Length (1/2/3/4)* NA (3/7/2/8) (5/1/6/11) .73 3 .06 Medication Dose** NA 55.21 (69.7) 53.21 (29.1) .007 1,51 .94 % on Atypical Neuroleptics NA 76 89 .85 1 .29 Cholinergic Medication (yes) NA 5 3 .31 1 .45 Positive and Negative Syndrome Scale (PANSS) Positive Score NA 13.2 (4.7) 13.4 (4.8) .02 1,51 .91 PANSS Negative Score NA 12.7 (3.1) 19.4 (6.7) NA NA NA PANSS General Score NA 25.0 (6.0) 28.9 (6.9) NA NA NA PANSS Total Score NA 50.9 (11.1) 61.7 (14.2) NA NA NA Note: Where appropriate, value is the mean (with SD in parentheses). NA = not applicable either because the variable is not relevant to the control group or because the statistical test is different by design. *Category of illness length was defined as 1 = less than 1 year since first psychiatric contact, 2 = 1–5 years, 3 = 5–10 years, and 4 = greater than 10 years. **Medication dose was converted to a standard dose by calculating the total percentage of the maximum recommended dose53. more slowly, with increased redundancy and poorer strategy use, compared to both the non-negative group and controls. The non-negative group demonstrated significantly poorer strategy use than controls but showed less redundancy, in that they entered significantly fewer aisles above the minimum. Overall, controls were 4 times more likely and non-negative participants were twice as likely to use a strategy compared to negative participants. Fig. 1. Group Differences in Executive Functions. The graphs present the mean score and standard error bars for each group. 914 Note: OR = odds ratio, D = difference, IRR = incidence rate ratio. Significant group differences compared to adjusted significance level a = 0.05/3, 0.05/2, or 0.05/1, according to order of significance in the Multistage Bonferroni procedure. CI(1–a) denotes intervals at the (1–a) 3 100% confidence level. *N = 68 for redundancy; n = 52 for strategy. <.0001 44.54 .006 (CI 0.95 [.29, .81]) 2.76 OR = .49 .001 (CI 0.98 [.13, .37]) <.001 (CI 0.97 [.27, .72]) OR = .44 Strategy* Aisle Strategy (0–10) 3.66 OR = .22 6.52 2 <.0001 94.97 <.001 (CI 0.98 [.29, .81]) 8.75 OR = 2.63 <.001 (CI 0.97 [1.56,2.40]) .013 (CI 0.95 [.31, .47]) IRR = .73 Redundancy* Aisles Above Minimum 2.47 OR = 1.93 6.64 2 <.0001 23.06 .001 (CI 0.97 [4.2,19.1]) 3.40 D = 11.66 .21 D = 4.56 Efficiency Time (mins) 1.25 D = 16.22 4.57 <.001 (CI 0.98 [8.0,24.5]) 2 <.001 2 40.27 < .001 (CI 0.97 [.52, .84]) 3.17 .007 (CI 0.95 [.54, .91]) OR = .571 Accuracy Total Correct (0–30) 2.68 OR = .47 6.19 <.001 CI 0.98 [.35, .62] OR = .466 df X2 p z Effect Size p z Effect Size p z Effect Size Measure Negative Versus Control Non-negative Versus Control Table 2. Group Differences in Outcome Measures on the Supermarket Shopping Task Negative Versus Non-negative Group Effect p Targeting Cognitive Mechanisms to Improve Function Negative participants were likely to enter twice as many aisles above minimum as controls and nearly 3 times as many as non-negative participants (see table 2). Hence, despite similar IQ and general cognitive function, the negative group performed more poorly than the non-negative group on all community function measures. General cognitive function cannot, therefore, account for all the differences between groups in community function. The next set of analyses therefore investigated whether specific executive functions, which are impaired in the negative group, can account for these differences. Model Building Step 1: Which Executive Impairments Predict Community Function Impairments? Regressions of single executive function variables on community function measures were undertaken across the whole group to identify executive processes that could account for community function impairments. These are shown in table 3. The direction of the correlations across the groups is as predicted, such that the overall odds ratios predict that higher cognitive function is associated with better performance. All individual associations between cognitive and functional measures were significant. Step 2: Which Executive Impairments Interact With Group in Predicting Community Function? In order to test whether executive function measures interact with group, individual executive function–group interaction terms were entered into the regression models in addition to significant predictors from step 1. Working memory interacted with group membership for most investigated measures (accuracy: X2[2] = 22.5, p < .001; redundancy: X2[2] = 13.0, p = .0015; strategy: X2[2] = 9.2, p =.01) and is thus a particularly strong contender for a differential group effect in predicting community function. Premorbid IQ also interacted with group membership for 2 measures (accuracy: X2[2] = 7.9, p = .02; strategy: X2[2] = 6.4, p = .04), spatial strategy for 1 measure (redundancy: X2[2] = 6.2, p = .045), and response initiation for 1 measure (redundancy: X2[2] = 32.4, p < .001). These significant interactions between cognitive measures and group are indicated by a double asterisk in table 3. Step 3: Which Executive Functions Independently Predict Community Function Either Differentially Within a Particular Group or Within All Groups? The final model for each community function measure included both the significant predictor main effects from step 1 and the significant interaction terms from step 2. The results are presented in table 4. When a predictor effect interacted with group, this effect was assessed separately for each group. Two cognitive processes were each predictive of community functioning across the groups: poor working memory predicted less efficient community function, 915 916 6.96 .001** OR = 1.03 3.82 .001 and poor spatial strategy use predicted less accurate community function. These cognitive measures were predictive despite being independent from the real-life supermarket context. In addition, numerous cognitive processes were shown to be independently predictive of community function only within particular groups. Importantly, poor working memory was a highly significant independent predictor of inaccurate community function, and low premorbid IQ was a specific independent predictor of inefficiency only in the negative group. Good working memory also significantly predicted better aisle strategy use only in controls, although the same pattern was found in the negative group. Response initiation predicted redundancy in controls, with a trend also in the negative group, but the associations were in opposite directions. In the latter association, the same underlying relationship may have operated differently in the different groups. In the negative group, whose search strategies were poorer, better response initiation may have contributed to more active searches and less redundancy, but in controls, whose search strategies were already superior, better response initiation may have reflected higher self-motivated sustained attention resulting in longer searches and greater redundancy. There were no independent cognitive predictors of community function within the non-negative group. Are the Associations in the Negative Group due to the Greater Prevalence of Severe Executive Impairments in This Group? Note: OR = odds ratio, IRR = incidence rate ratio, D = difference. *N = 68 for redundancy; n = 52 for strategy. **Also interacts with group in predicting community function. .001 5.89 2.14 OR = 1.02 Strategy* Aisle Strategy (0–10) .03** OR = 1.46 OR = 1.22 <.001** –4.32 IRR = .98 <.001** –7.44 <.001** –3.62 OR = .89 <.001 –5.02 IRR = .98 Redundancy* Aisles Above Minimum .04 D = –.25 Efficiency Time (mins) –2.02 <.001** 5.86 OR = 1.02 Accuracy (0–30 correct) z Effect Size Shopping Measure IRR = .91 .009 –2.63 D = –.34 <.001 –4.82 .004 –2.87 D = –3.17 OR = 1.14 .004 6.99 OR = 1.28 D = –2.07 <.001 6.17 8.81 <.001** OR = 1.03 z Effect Size z Effect Size p Effect Size z p Working Memory (scaled score: 1–19) Spatial Strategy (profile score: 0–4) Premorbid IQ (National Adult Reading Test–Revised) Associated Executive, Memory, and IQ Factors Table 3. Model Building Steps 1 and 2—Executive and Premorbid Predictors of Community Function in Individual Analyses p Response Initiation (no. correct words) p K. E. Greenwood et al. The possibility of a threshold effect was investigated. Executive function may be related to community functioning only where these executive impairments are marked, as in the negative syndrome. Within the negative group, and for the selected community function measure (accuracy), the interaction between the impairment factor (marked versus moderate–intact working memory impairment group) and the working memory–community function relationship was not significant (z = 0.98, p = .33). Contrary to the case in a threshold model, the effect of working memory on community functioning was not stronger in the marked impairment subgroup compared to the moderate–intact subgroup. In fact, the odds of accurately selecting an item were estimated to increase by 22% for every unit change in working memory score in the moderate–intact group but only by 11% in the marked impairment group. Also counter to a simple threshold model, 50% of participants with marked impairment but no negative symptoms (n = 8) demonstrated good community functioning within half a standard deviation of controls and above the negative group mean. Indeed, the poorest working memory span in the absence of negative symptoms coincided with intact community Table 4. Final Model (Step 3)—Core Independent Predictors of Community Function Associated Executive, Memory, and IQ Factors Premorbid IQ (National Adult Reading Test–Revised) Shopping Measure Effect Size z Spatial Strategy (profile score: 0–4) p Effect Size .13 OR = 1.1 z Working Memory (scaled score: 1–19) p Effect Size .013 OR = 1.19 z Response Initiation (no. correct words) p Effect Size z p .48 <.001 (CI0.99 [1.08,1.3]) OR = .96 .77 .44 .52 .60 D = .005 .03 .98 Accuracy (0–30 correct) N OR = 1.01 1.5 2.48 C OR = 1.01 .54 .96 OR = 1.03 NN OR = 1.01 1.92 .06 OR = .98 .60 .55 D = 1.95 3.14 .002 1.4 Efficiency Time (mins) N D = .47 24.8 C D = .44 1.8 NN D = .07 .013 (CI0.99 [0.96, 0.02]) D = –1.05 .89 .37 .07 .43 .66 1.21 .23 Redundancy* Aisles Above Minimum N IRR = .99 IRR = 1.01 .89 IRR = .95 .16 IRR = .98 1.98 .048 .17 IRR = .97 .75 .45 IRR = 1.03 4.02 .001 (CI0.99 [1.01,1.05]) IRR = 1.08 1.19 .23 IRR = .99 .25 .81 IRR = .99 .69 .49 OR = 1.16 1.67 .095 OR = 1.27 1.79 .074 OR = .97 1.90 .058 .006 (CI0.99 [1.01, 1.63]) IRR = .87 NN Strategy* Aisle Strategy (0–10) N OR = 1.0 .24 .81 C OR = .98 .78 .43 OR = 1.29 2.74 NN OR = .98 1.45 .15 OR = 1.03 .42 .67 Note: N = negative group, C = control group, NN = nonnegative group, OR = odds ratio, D = difference, IRR = incidence rate ratio. Significant associations should be compared to adjusted significance level a = 0.05/3, 0.05/2, or 0.05/1, according to order of significance in the Multistage Bonferroni procedure. Significant associations are shaded grey. CI(1–a) denotes intervals at the (1–a) 3 100% confidence level. *N = 68 for redundancy; n = 52 for strategy. 917 Targeting Cognitive Mechanisms to Improve Function .14 1.39 C K. E. Greenwood et al. Model 1: Symptoms Independent e.g.14 Cognition Community Function Negative Symptoms Model 2: Symptoms Indirect e.g.5 Cognition Community Function Negative Symptoms Model 3: Symptoms as predictors of cognition e.g.5 Negative Symptoms Community Function Cognition Model 4: Synergistic interaction of negative symptoms and cognition in moderating community function Cognition Community Function Negative Symptoms Fig. 2. Models of Hypothetical Relationships Among Cognitive Function, Negative Symptoms, and Community Function. functioning above the control mean. Negative symptoms alone with intact working memory (n = 6) were also associated with good community functioning above the negative group mean and within half a standard deviation of the control group mean. Hence, working memory impairment in the absence of negative symptoms did not necessarily confer poor outcome, and many of those with marked working memory impairment alone performed well on the community functioning task. Discussion This study revealed 2 distinct and important mechanisms through which cognition and/or symptoms are related to poor community function in schizophrenia. These are, first, direct associations between specific cognitive impairments and community functions and, second, a synergistic interaction between negative symptoms and cognition in predicting community function. The negative group was associated both with specific verbal working memory and executive impairments and with impairments in a direct and ecologically valid measure of real-life community function when compared to non-negative schizophrenia and healthy controls. These impairments were independent of global cognitive deficits, as, on the whole, they 918 were not present in the non-negative group despite the equivalent IQ, general memory, and general executive function in this group. The results pave the way for a more refined consideration of the mechanism by which negative symptoms and executive processes contribute to community functioning. In the first mechanism, specific cognitive impairments were directly associated with specific aspects of poor community function. People who had trouble in an empirical task with drawing a spatial search strategy to locate a hypothetical lost item within a designated space were also less accurate in a real-life supermarket environment in locating and selecting items correctly. In addition, people who were less accurate at holding and manipulating information in working memory were less efficient and took significantly longer to complete the community function task. While numerous studies have demonstrated links between cognition and community function1–2 and even between cognition and real-life skills such as grocery shopping,48 this study is unique in demonstrating such close links between similar underlying processes on both empirical and real-life tasks. Both working memory and strategic processes would appear to be ideal candidates for cognitive remediation therapy, in order to enhance generalization to the real world. In terms of mechanisms derived from the literature (see figure 2), cognition and negative symptoms may contribute to community function independently (model 1).14 Cognition may contribute to community functioning directly, while symptoms are associated only indirectly through their link with cognition (model 2), or indeed the reverse may occur so that cognition only associates indirectly to community functioning through symptoms (model 3).5 The mechanism thus far is similar to that reported by previous researchers, in that a direct association is demonstrated across all groups between cognition and community function. This study extends current knowledge, however, by demonstrating that this mechanism occurs for cognitive predictors that are theoretically and empirically linked to negative symptoms. Since impairments in these processes are greater in people with negative symptoms, these people are significantly less skilled in the cognitive processes that underlie community functioning and so do less well (model 2 or 3). However, the second and perhaps more important mechanism is the previously undocumented role of negative symptoms as a moderator of the relationship between cognitive impairments (particularly working memory) and poor community functioning in schizophrenia. The moderator effect reflected the specific association between cognition and community function only in people with negative but not non-negative schizophrenia. The interaction between working memory and symptoms in the current study is not consistent with independent contributions from cognition and symptoms to community Targeting Cognitive Mechanisms to Improve Function function (model 1). Neither is it consistent with indirect contributions of symptoms or cognition to community function (models 2–3). Several mechanisms may explain the interaction between cognition and the negative symptom group. Working memory impairments may be particularly marked in schizophrenia with negative symptoms, and only these marked impairments then predict poor community function (an adaptation of model 2). This explanation seems unlikely, however, as there was no interaction between the severity of impairment (marked versus moderate– intact) and cognition on community function in the negative group. In fact, cognition was a somewhat stronger predictor of community function in the group with moderate–intact cognition. This suggests that, if anything, symptoms play a greater role where impairments are milder. So, the effect was not a simple threshold effect, since the most cognitively impaired people in the negative group did not perform the least well on community functions. Negative symptoms may, instead, bestow some additional contribution to the relationship between cognition and function found in this group. A mechanism for this interaction is suggested by a consideration of working memory capacity and function, which are governed both by general ability and by domain-specific experience. Domain-specific experience enhances working memory capacity in experts compared to novices, across such spheres as professional chess and academic performance.54–57 Experience may increase working memory capacity by providing a broader knowledge base and so promoting enhanced ‘‘chunking’’ and organization of domain-specific information in working memory through the development of new cognitive schemata.58 Reduced active participation across a varied array of community, social, and occupational experiences, as a result of negative symptoms, may create exactly this division in ‘‘expertise’’ between negative and non-negative schizophrenia and controls. This will, in turn, lead to even poorer domainspecific working memory and community function. The significant association between good working memory and good strategy use only in the control group may reflect the opposite ‘‘expert’’ end of this continuum. Hence, the current data are consistent with symptoms acting as a true moderator of the relationship between cognition and community functioning (model 4). It is proposed that a synergistic interaction between working memory and negative symptoms arrests the normal development of working memory mechanisms within particular domains, through the curtailing of appropriate domain-specific experiences. This synergy between symptoms and working memory impairment produces the poorest community functioning. It may also have the greatest impact where the core impairment is milder and where environmental effects can have a greater effect. This second mechanism provides further significant clinical implications. The interaction of negative symptoms with poor working memory may gradually expand the divide between novice and expert working memory, leading to a progressive disruption of community and other functional domains and a more chronic disorder. According to this mechanism, programs that address poor functional outcome should target both cognition and negative symptoms early in the disorder. A wider and more immersed experience of community function may break the synergistic link between negative symptoms and cognition to promote a more adaptive functional outcome. There are several limitations to the current study. Supermarket shopping is a narrow domain of community function, and the target processes for remediation may not generalize to other aspects of community living. Furthermore, the current cognitive processes predict baseline function. It is unclear whether change in these cognitive processes will predict change in functioning. Both cognitive processes and mechanisms are identified, however, which are specific to a particular target group and which can be investigated further. The dynamic nature of the interaction model, in particular, provides both a process and a mechanism by which change may occur. The study controlled for multiple group comparisons but not for multiple correlations between cognition and community function. This was deliberate in order to capture all valid mechanisms, but the possibility of Type I errors is thus greater. The findings require replication, but, nevertheless, many of the results including the moderating effect of negative symptoms on the working memory– community function relationship were particularly robust and were highly significant. In conclusion, this study provides strong evidence that influencing real-world community function in schizophrenia may require a more complex approach than simple remediation of cognition in general across all people. Rather, the study supports several distinct cognitive processes and mechanisms, moderated by negative symptoms, which may be active simultaneously and which predict the exceptionally poor community function in people with schizophrenia and negative symptoms. Targeting these processes and mechanisms in conjunction with their moderating symptoms will provide the greatest opportunity for cognitive remediation therapy to improve community function in the real world. Acknowledgments This study was supported by a grant, RFG 757, from the Department of Health. We would like to thank Dr. Pall Matthiasson, M.D., MRCPsych, Division of Psychological Medicine, Institute of Psychiatry, London, for his diagnosis and symptom assessments 919 K. E. Greenwood et al. for people with schizophrenia. Earlier forms of this work have been presented at the Society for Research in Psychopathology, San Francisco, September 2002; the International Congress of Schizophrenia Research and Sixth Biennial Mt. Sinai Conference on Cognition in Schizophrenia, Colorado Springs, Colo., April 2003; and the Fifth International Conference on Psychological Treatments for Schizophrenia, Oxford, September 2003. References 1. Green MF. What are the functional consequences of neurocognitive deficits in schizophrenia? Am J Psychiat 1996;153(3):321–330. 2. Green MF, Kern RS, Braff DL, Mintz J. Neurocognitive deficits and functional outcome in schizophrenia: are we measuring the ‘‘right stuff’’? Schizophrenia Bull 2000; 26(1):119–136. 3. Krabbendam L, Aleman A. Cognitive rehabilitation in schizophrenia: a quantitative analysis of controlled studies. Psychopharmacol 2003;169(3–4):376–382. 4. van der Gaag M, Kern RS, van den Bosch RJ, Liberman RP. A controlled trial of cognitive remediation in schizophrenia. Schizophrenia Bull 2002;28(1):167–176. 5. Velligan D, Mahurin R, Diamond P, Hazleton B, Eckert S, Miller A. The functional significance of symptomatology and cognitive function in schizophrenia. Schizophr Res 1997;25:21–31. 6. Velligan DI, Bow-Thomas CC, Mahurin RK, Miller AL, Halgunseth LC. Do specific neurocognitive deficits predict specific domains of community function in schizophrenia? J Nerv Ment Dis 2000;188(8):518–524. 7. Bellack AS, Gold JM, Buchanan RW. Cognitive rehabilitation for schizophrenia: problems, prospects, and strategies. Schizophrenia Bull 1999;25(2):257–274. 8. Lysaker PH, Bell MD, Zito WS, Bioty SM. Social skills at work: deficits and predictors of improvement in schizophrenia. J Nerv Ment Dis 1995;183(11):688–692. 9. Bryson G, Bell MD, Kaplan E, Greig T. The functional consequences of memory impairments on initial work performance in people with schizophrenia. J Nerv Ment Dis 1998;186(10):610–615. 10. Bilder RM, Goldman RS, Robinson D, et al. Neuropsychology of first-episode schizophrenia: initial characterization and clinical correlates. Am J Psychiat 2000;157(4):549–559. 11. Suslow T, Schonauer K, Ohrmann P, Eikelmann B, Reker T. Prediction of work performance by clinical symptoms and cognitive skills in schizophrenic outpatients. J Nerv Ment Dis 2000;188(2):116–118. 12. McGurk SR, Meltzer HY. The role of cognition in vocational functioning in schizophrenia. Schizophr Res 2000;45(3):175–184. 13. McGurk SR, Moriarty PJ, Harvey PD, et al. Relationship of cognitive functioning, adaptive life skills, and negative symptom severity in poor-outcome geriatric schizophrenia patients. J Neuropsych Clin N 2000;12(2):257–264. 14. Breier A, Schreiber J, Dyer J, Pickar D. National Institute of Mental Health longitudinal study of chronic schizophrenia: prognosis and predictors of outcome. Arch Gen Psychiat 1991;48(3):239–246. 920 15. Tamminga CA, Buchanan RW, Gold JM. The role of negative symptoms and cognitive dysfunction in schizophrenia outcome. Int Clin Psychopharm 1998;13(S3):S21–S26. 16. Crow T. Molecular pathology of schizophrenia: more than one disease process? Brit Med J 1980;280:66–68. 17. Yang P, Liu C, Chiang S, Chen Y, Lin TS. Comparison of adult manifestations of schizophrenia with onset before and after 15 years of age. Acta Psychiat Scand 1995;91(3):209–212. 18. Nibuya M, Kanba S, Sekiya U, et al. Schizophrenic patients with deficit syndrome have higher plasma homovanillic acid concentrations and ventricular enlargement. Biol Psychiat 1995;38(1):50–56. 19. Nieuwenstein MR, Aleman A, de Haan EH. Relationship between symptom dimensions and neurocognitive functioning in schizophrenia: a meta-analysis of WCST and CPT studies. J Psychiat Res 2001;35(2):119–125. 20. Franke P, Maier W, Hardt J, Hain C. Cognitive function and anhedonia in subjects at risk for schizophrenia. Schizophr Res 1993;10:77–84. 21. Moritz S, Krausz M, Gottwalz E, et al. Cognitive dysfunction at baseline predicts symptomatic 1-year outcome in firstepisode schizophrenics. Psychopathology 2000;33(1):48–51. 22. Stirling J, Hellewell J, Hewitt J. Verbal memory impairment in schizophrenia: no sparing of short-term recall. Schizophr Res 1997;25(2):85–95. 23. Pantelis C, Stuart GW, Nelson HE, Robbins TW, Barnes TR. Spatial working memory deficits in schizophrenia: relationship with tardive dyskinesia and negative symptoms. Am J Psychiat 2001;158(8):1276–1285. 24. Carter C, Robertson L, Nordahl T, Chaderjian M, Kraft L, O’Shora-Celaya L. Spatial working memory deficits and their relationship to negative symptoms in unmedicated schizophrenia patients. Biol Psychiat 1996;40:930–932. 25. Buchanan RW, Strauss ME, Kirkpatrick B, Holstein C, Breier A, Carpenter WT, Jr. Neuropsychological impairments in deficit vs nondeficit forms of schizophrenia. Arch Gen Psychiat 1994;51(10):804–811. 26. Bryson G, Whelahan HA, Bell M. Memory and executive function impairments in deficit syndrome schizophrenia. Psychiat Res 2001;102(1):29–37. 27. Frith C. The Cognitive Neuropsychology of Schizophrenia. Hove, U.K.: Laurence Erlbaum; 1992. 28. Goldman-Rakic P, Selemon L. Functional and anatomical aspects of prefrontal pathology in schizophrenia. Schizophrenia Bull 1997;23(3):437–458. 29. Greenwood KE. The nature and stability of executive impairments in schizophrenia. Ph.D. thesis, University of London; 2000. 30. Wykes T, Reeder C, Williams C, Corner J, Rice C, Everitt B. Are the effects of cognitive remediation therapy (CRT) durable? results from an exploratory trial in schizophrenia. Schizophr Res 2003;61(2–3):163–174. 31. Kurtz MM, Moberg PJ, Gur RC, Gur RE. Results from randomized, controlled trials of the effects of cognitive remediation on neurocognitive deficits in patients with schizophrenia. Psychol Med 2004;34(3):569–570. 32. Bell M, Bryson G, Greig T, Corcoran C, Wexler BE. Neurocognitive enhancement therapy with work therapy: effects on neuropsychological test performance. Arch Gen Psychiat 2001;58(8):763–768. 33. Wilson B, Cockburn J, Baddeley AD. The Rivermead Behavioural Memory Test. Bury St. Edmunds, U.K.: Thames Valley Test Company; 1985. Targeting Cognitive Mechanisms to Improve Function 34. Heaton R. A Manual for the Wisconsin Card Sorting Test. Odessa, Fla.: Psychological Assessment Resources; 1981. 35. Wykes T, Sturt E. The measurement of social behaviour in psychiatric patients: an assessment of the reliability and validity of the SBS schedule. Brit J Psychiat 1986;148:1–11. 36. Liddle P. The symptoms of chronic schizophrenia: a reexamination of the positive–negative dichotomy. Brit J Psychiat 1987;151:145–151. 37. Nelson H, Willison J. National Adult Reading Test (NART): Test Manual. 2nd ed. Windsor, U.K.: National Foundation for Educational Research; 1991. 38. Gold J, Carpenter C, Randolph C, Goldberg T, Weinberger D. Auditory working memory and Wisconsin Card Sorting Test performance in schizophrenia. Arch Gen Psychiat 1997;54:159–165. 39. Wechsler D. Weschler Adult Intelligence Scale. 3rd ed. San Antonio, Tex: Psychological Corporation; 1997. 40. Shallice T, Burgess P. The domain of supervisory processes and temporal organization of behaviour. Philos T Roy Soc B 1996;351(1346):1405–1411. 41. Spreen O, Benton A. Neurosensory Centre for the Comprehensive Examination of Aphasia, (NCCEA) Revised Edition. Victoria, Canada: Neuropsychological Laboratory: University of Victoria; 1977. 42. Burgess PW, Shallice T. The Hayling and Brixton Tests. Bury St. Edmunds, U.K: Thames Valley Test Company Ltd.; 1997. 43. Wilson BA, Evans JJ, Emslie H, Alderman N, Burgess P. The development of an ecologically valid test for assessing patients with dysexecutive syndrome. Neuropsychol Rehabil 1998;8(3):213–228. 44. Roberts PM, Le Dorze G. Semantic organization, strategy use, and productivity in bilingual semantic verbal fluency. Brain Lang 1997;59(3):412–449. 45. Kessler J, Bley M, Mielke R, Kalbe E. Strategies and structures in verbal fluency tasks in patients with Alzheimer’s disease. Behav Neurol 1997;10(4):133–135. 46. Leggio MG, Silveri MC, Petrosini L, Molinari M. Phonological grouping is specifically affected in cerebellar patients: a verbal fluency study. J Neurol Neurosur Ps 2000;69(1):102–106. 47. Hamera E, Brown CE. Developing a context-based performance measure for persons with schizophrenia: the test of grocery shopping skills. Am J Occup Ther 2000;54(1):20–25. 48. Rempfer MV, Hamera EK, Brown CE, Cromwell RL. The relations between cognition and the independent living skill of shopping in people with schizophrenia. Psychiat Res 2003;117(2):103–112. 49. McCullagh P, Nelder JA. Generalized Linear Models. 2nd ed. London: Chapman and Hall; 1989. 50. Holm S. A simple sequentially rejective multiple test procedure. Scand J Stat 1979;6:65–70. 51. Statistical Package for the Social Sciences. Version 10. Chicago, Ill.: SPSS Inc.; 1995. 52. STATA: Statistics Data Analysis. Version 7.0. College Station, Tex.: Stata Corporation; 1984–2001. 53. Taylor D, Duncan D, McConnell H, Abel K, Kerwin R. The Bethlem and Maudsley NHS Trust Prescribing Guidelines. 4th ed. London, England: The Bethlem and Maudsley NHS Trust; 1997. 54. Chase WG, Simon HA. Perception of chess. Cognitive Psychol 1973;4:55–81. 55. Gobet F. Expert memory: a comparison of four theories. Cognition 1998;66:115–152. 56. Ericsson KA, Pennington N. The structure of memory performance in experts: implications for memory in everyday life. In: Davies GM, Logie RH, eds. Memory in Everyday Life: Advances in Psychology. Amsterdam, The Netherlands: Elsevier Science; 1993; 241–272. 57. Ericsson KA, Kintsch W. Long-term working memory. Psychol Rev 1995;102:211–245. 58. Bartlett FC. Remembering. Cambridge: Cambridge University Press; 1932. 921
© Copyright 2026 Paperzz