|Home | About | Journals | Submit | Contact Us | Français|
Individuals with schizophrenia have consistently been found to exhibit cognitive deficits, which have been identified as critical mediators of psychosocial functional outcomes. Recent reviews of cognitive remediation (CRT) have concluded that these deficits respond to training. This multi-site community study examined 40 individuals with schizophrenia who underwent cognitive remediation using the Neuropsychological Educational Approach to Remediation1 (NEAR). Assessments using the same neuropsychological tests and measures of psychosocial outcome were made at four time points: baseline, before start of active intervention, end of active intervention and 4 months after end of active intervention. Dose of antipsychotic medication remained constant throughout the study period. After participating in NEAR, individuals showed significant improvements in verbal and visual memory, sustained attention and executive functioning. This effect persisted 4 months after the treatment ceased. The average effect size was mild to moderate. Social and occupational outcomes also improved from baseline to post-treatment, which persisted 4 months later. Our findings replicate those of previous studies that suggest that NEAR is effective in improving cognition in individuals with schizophrenia in a naturalistic and ecologically valid setting. Further it extends such findings to show a generalisation of effects to social/occupational outcomes and persistence of effects in the short term.
Core to schizophrenia are cognitive impairments in attention, memory, and executive functions.1–3 These impairments are relatively independent from other symptom domains such as positive and negative psychotic symptoms4 and have been found to predict functional outcome5 as well, if not better, than do negative symptoms.6,7 Therefore, in order to improve outcome, treatment efforts should target cognitive impairments in addition to symptomatology.
Antipsychotic medications developed to date have had limited effect on cognitive functions in schizophrenia.8 Alternatively, a number of cognitive remediation programs have been developed, and randomized controlled studies have shown positive effects on cognition.9,10 One cognitive remediation approach is known as the Neuropsychological Educational Approach to Remediation (NEAR),11 an evidence-based approach to cognitive remediation that utilizes a set of carefully crafted instructional techniques that reflect an understanding of how people learn best. The NEAR program involves a combination of “drill and practice” exercises and teaching strategies to improve cognitive functioning.11 It utilizes commercially available educational software to create a rich learning environment that is intrinsically motivating and rewarding.11
Using NEAR, Medalia et al12 found that a group of 14 acutely ill psychiatric inpatients who received 6 hours of problem solving remediation showed significantly greater improvement in verbal problem solving (as measured by the Comprehension Test) than did another group of 14 individuals with psychiatric illness who just received typing instruction.12 There were also concurrent improvements in ability to cope as measured by the Symptom Checklist (Positive Symptom Distress Index). Medalia et al13 conducted another study of 18 individuals with chronic schizophrenia who participated in 10 sessions of NEAR, specifically targeting problem solving. Relative to a control group of individuals with schizophrenia who received treatment as usual and no remediation, those who participated in NEAR showed significant improvements in problem solving as reflected in better scores after treatment on an ecologically valid measure of problem solving from the Independent Living Scale. Follow-up studies have shown that gains in problem solving seen over the course of the cognitive remediation program persisted for up to 4 weeks after the program ended.14
Therefore, the NEAR program appears to hold promise in improving cognitive functions in those with schizophrenia. However, key questions remain about the treatment effect in individuals with first-episode schizophrenia. The research conducted to date often used more chronic older populations. First-episode psychosis patients generally have had a more recent decline in cognitive skills from premorbid functioning and therefore would benefit from an early intervention program that specifically aims to remediate cognitive deficits and enhance recovery and prevent relapse. In addition, studies of NEAR to date have used single cognitive measures to evaluate outcome and limited measures of psychosocial functioning. Therefore, there are some unanswered questions about the generalization of improvements to other cognitive domains and also psychosocial domains. The durability of NEAR's effects beyond 4 weeks is also unclear.
This study sought to evaluate the effects of NEAR on cognition immediately after treatment and 4 months after cessation of treatment in a community setting. It sought to use a wide range of cognitive measures in those with first-episode and chronic schizophrenia. It also aimed to examine the effects of NEAR on functional outcome measures including those related to social and occupational functioning, quality of life, and self-esteem.
The following were hypothesized:
The multisite study initially recruited 69 individuals (41 males, 28 females) with schizophrenia, schizophreniform disorder, or schizoaffective disorder via referral from case managers within the Sydney West and Northern Sydney and Central Coast Area Health Services. There were 3 early intervention outpatient centers, 2 chronic inpatient rehabilitation centers, and 4 chronic community outpatient centers.
Once contact was established with the referring center, a site inspection by the investigators was conducted to review the setup of computers and treatment room, the availability of an adequate range of computer software, adequate trained staff to complete rating scales and administer treatment, and adequate area to conduct neuropsychological assessments on site in order to establish treatment veracity across centers. In addition, to determine whether there was compliance with treatment principles, the investigators of this study reviewed the treatment plans and strategies recorded by facilitators in their progress notes. The centers that participated in this study met all the criteria mentioned above.
Inclusion criteria were as follows:
Informed consent was obtained from all subjects. This research project was approved by the ethics committees from Sydney West Area Health Service, Northern Sydney and Central Coast Area Health Service, and Macquarie University. The trial was registered with the Australian New Zealand Clinical Trials Registry (ACTRN12605000202662).
Figure 1 shows the consort diagram. Twenty-nine participants (42.0%) dropped out at different points of the study. There were a number of reasons including inpatient admission (N=2), work commitments (N=4), discharge from hospital (N=1), lost contact with center (N=5), withdrew consent (N=2), deceased (N=2), clinician unavailable (N=2), wait-list (N=7), loss of motivation to attend cognitive remediation therapy (CRT) (N=4). Comparison of participants who dropped out of the study and those who remained on core clinical, demographic, and neuropsychological variables revealed no significant differences.
The final sample consisted of 40 participants. Twenty-two were randomized to treatment group (57.1%) and 18 to wait-list group (42.9%). Participants were aged between 17 and 50 years with an average of 11 years of education (see table 1).
Participants from both the wait-list and immediate treatment groups were assessed at (see figure 2) baseline (time 1), posttreatment (time 3), and 4-month follow-up (time 4) by a research assistant (master's level) for all cognitive tests. Additionally the wait-list group received a second baseline assessment prior to commencement of treatment (pretreatment, time 2). Clinicians were trained in the administration of rating scales used in the battery and after attaining acceptable interrater reliability rated subjects on symptom rating scales.
Assessment included the following measures.
Alternate forms of tests were used for TMT, RAVLT, and Rey Figure and sorting subtest.
The study had a randomized wait-list control design. Participants who enrolled in the study were randomized to either a group that was immediately treated or to a group that “waited” for 15 weeks before being treated. Both groups completed the above-mentioned questionnaires and participated in cognitive assessment. After 20–30 sessions of treatment (if in the immediate treatment group) or 15 weeks of waiting (if in the wait-list control), individuals were tested again with the same measures. Individuals were tested again after a further 15 weeks to determine if any benefits persisted. This design allowed all individuals to be treated eventually. It allowed for a comparison between a period of treatment and a period of no treatment. It also permitted measurement of change in performance before and after treatment and also the persistent of such effects 4 months later.
Participants were randomized in permuted blocks of 4 patients into immediate treatment or wait-list control. The research assistant (D.S.) was responsible for the generation of the allocation sequence and was blind to this assignment at the time of treatment allocation. However, subsequent to this, the study was not blinded, and those clinicians administering the intervention and those assessing the outcome were aware of group assignment. There was no masking in this study.
Based on the treatment protocol described in the treatment manual,25 participants attended two 1-hour sessions each week for 10–5 weeks, making a maximum total of 30 sessions. The treatment program consisted of participants working on various computer software programs with guidance/assistance from a therapist. The computer software programs were specifically selected to promote improvement and practice in specific cognitive skills that were identified as being deficient on neuropsychological testing as identified by performance more than 1 SD below the normative mean for each respective cognitive measure. The participants worked on the computer software programs at their own pace and over the treatment period. Each participant built a repertoire of software programs that addressed the following cognitive domains: sustained attention, processing speed, memory, and executive functions. The therapist guided the process using questions to enhance meta-cognition and information processing during activity performance. The therapist's goal was to facilitate the development of cognitive functions by monitoring subjects’ progress, providing instruction and feedback as necessary, and facilitating a positive learning experience. All facilitators received the same training workshops on NEAR conducted by the same 2 experienced clinicians (M.A.R.H., P.W.) trained originally by Dr Alice Medalia. Treatment was conducted on site at the various treatment centers following a satisfactory site inspection by trained clinicians in CRT. Random inspection of CRT session notes confirmed treatment fidelity across centers. The treatment sessions conformed to the treatment plan in terms of cognitive domains covered and time spent and captured the theoretically active ingredients important in enhancing learning and self-determination. In order to help ensure treatment fidelity, all clinicians were supervised fortnightly—monthly over the course of the trial. All rating scales measuring psychopathology were completed by the treating clinician and collated by the research assistant. All scores were entered into a database and were checked by 2 investigators who were blind to the group membership.
Raw scores were used for all measures except for DKEFS sorting for which scaled scores were used. For all cognitive and psychosocial measures, repeated-measures analysis of variance (ANOVA) was used to compare the following:
The Bonferroni method was used to adjust for multiple comparisons of outcomes with a mean correlation of 0.41. Therefore, the α level was set at .02. To determine the magnitude of change, this study calculated effect sizes (Cohen d).
To determine whether cognitive changes seen were clinically significant, this study adopted method of Jacobson and Truax26 to calculate reliable change indices for each cognitive measure.
where σ=SD, r=test-retest correlation.
Reliability coefficients from published norms that matched the general demographics of the experimental sample were used. The 68% confidence interval was used to determine reliable change.
The demographic and clinical characteristics of the group are presented in table 1. At baseline, there were no differences between treatment and wait-list groups on any of the demographic, clinical, or cognitive variables. There were no statistically significant changes from baseline to posttreatment in terms of medication dose.
Repeated-measures ANOVA revealed no significant changes in all the cognitive test scores from baseline to pretreatment (table 3). These results suggest no evidence of significant practice effects.
Inspection of cognitive scores (see table 3) revealed significant improvement in scores from baseline to posttreatment after correction for multiple comparisons. This was evident in RAVLT delayed recall (F=4.93, P < .02), Rey delayed recall (F = 37.78, P < .02), CCPTd, (F = 19.85, P < .02), and TMTb (F = 4.58, P < .02). The effect sizes for these changes were 0.18, 0.26, 0.65, and 0.28, respectively.
The change seen in Rey delayed recall, CCPTd, and TMTb was reliable at 68% confidence using formula of Jacobson and Truax23 for determining reliable change.
Repeated-measures ANOVA (table 4) revealed no significant interaction between time (pretreatment to posttreatment) and group (wait-list and immediate treatment) on any of the cognitive measures. However, effect sizes ranged from 0.08 (RAVLT delayed recall) to 0.77 (Rey delayed recall) to 0.90 (CCPTd and sorting).
Inspection of individual performances across the cognitive domains revealed that up to 38% (15/40) of the sample made reliable changes based on formula of Jacobson and Truax.23
From baseline to posttreatment, the PANSS, CDS, LSP-39, WHOQOL-Bref, and Rosenberg Self-Esteem Scale scores did not change significantly (table 2). In contrast, the SOFAS score increased from baseline to posttreatment significantly (F=8.64, P < .005). The effect size for this change was 0.14. There were positive trends for improvement in the LSP-39 and the Rosenberg Self-Esteem ratings. SOFAS change scores from pre- to posttreatment were significantly correlated with treatment-related changes on Rey delayed recall (r=0.36, P < .05) but were not correlated with those changes in RAVLT delayed recall (r = 0.03, P > .05), CCPTd (r=−0.03, P > .05), or TMTb (r=0.05, P > .05).
Comparison of performance at posttreatment and 4-month follow-up revealed no significant changes on any of the cognitive or psychosocial variables. In fact, there was a trend toward continued improvement in RAVLT total learning. These results suggest that the effects seen immediately posttreatment are at least durable up to 4 months later.
This randomized controlled multisite study has demonstrated that the NEAR technique is associated with broad cognitive improvement after 15 weeks of cognitive remediation. This cognitive improvement was accompanied by a significant improvement in psychosocial functioning.
Practice effect factors were minimized by the use of alternate forms of cognitive tests. The observed improvements were greater than practice effects usually seen on repeat testing as indicated by reliable change analyses. This is further supported by data showing that changes seen in those treated were greater than those made by a subgroup of individuals who received no NEAR treatment between 2 separate testing occasions.
This provides the first evidence to our knowledge of the effectiveness of NEAR in individuals with schizophrenia. This is also the first examination to date of the NEAR approach using an extensive battery of cognitive and psychosocial outcome measures with a longer posttreatment follow-up. The present study found improvements in visual and verbal memory, sustained attention, and set shifting that are core cognitive deficits in schizophrenia.5 The cognitive performances (particularly for sustained attention and visual memory) at posttreatment normalized to within 1 SD of the usual reference age group (20–29 years old). These effects were maintained for up to 4 months after treatment ceased. The average effect sizes were low to moderate and somewhat compatible with those obtained in previous CRT trials in schizophrenia.27 Trends toward improvement were also seen in most other cognitive domains. Further, the effects of NEAR generalized to yield an improvement in social and occupational functioning, which previously had only been measured in a small group of studies in CRT. The group of individuals in this study improved on SOFAS scores from 54 at baseline to 60–61 at posttreatment, which reflect movement from being rated as having “moderate difficulty” to “some difficulty … but generally functioning well,” which is a clinically meaningful change. Because the participants had been on a stable dose of medication for at least 2 months prior to entering the study and there were no significant changes in medication dose over the period of the study including the 4-month follow-up, this is unlikely to be due to changes in medication. Further, there were trends toward improvement in LSP and self-esteem. The latter is consistent with feedback given by participants about enhanced sense of self-worth while attending the CRT program.
The findings are in line with those found by Medalia et al28 and also other CRT work by Ueland and Rund29 that has shown that targeted repeated programs can partially remediate attention functions in schizophrenia. The results are also similar to work of Medalia et al30 in relation to observed improvements in verbal memory following NEAR. The group's performance on posttreatment moved from the impaired range to the low average range. The present results also extend Medalia's work to show NEAR's additional benefit to visual memory.
A number of limitations may potentially limit conclusions of the study. First, the sample size was small. The small size of each of the groups (wait-list, N=18, and immediate treatment group, N=22) meant that the repeated-measures (between groups) analyses yielded nonsignificant improvements in most cognitive domains, probably reflecting reduced power. On a positive note, effect sizes of some of the changes were relatively large (up to 0.90) suggesting that larger numbers within each group would have yielded more significant findings. Another limitation is that the raters were not blind to the treatment group, and therefore, the resulting increase in psychosocial ratings from pre- to posttreatment could be due to bias. Analysis of correlations between changes in subjective SOFAS ratings and objective measures of cognitive functioning yielded a significant correlation in only one domain. In addition, the functional assessment was brief. A more detailed assessment of function and how these improvements transferred to successful completion of actual daily activities would be important. Fourth, the design of the study only allowed for a short-term follow-up. Follow-up of function over 12 months would determine whether effects are sustainable and generalize to real-world functions. It may be that the full gains to be obtained from this treatment will require further treatment using more traditional psychosocial rehabilitation techniques. Further work needs to examine whether the gains found with cognitive remediation therapy are synergistic, complement, or replace these forms of treatment.
Alternative study designs are necessary to further support the utility of cognitive remediation in schizophrenia. This may include randomized controlled trials focusing on (a) demonstration of cognitive gains that extend to actual performance of activities of daily living; (b) blinding raters to treatment group membership; (c) its longer term durability (>4 months); (d) adequacy of the social control condition that matches for the therapist's time, contact, and enthusiasm; and (e) examination of the dose of treatment is required. Replication of findings in early psychosis with larger sample sizes is warranted. The applicability of NEAR to other clinical populations with similar cognitive deficits should also be explored. Recent research has shown promising results for cognitive remediation in anorexia nervosa31 and depression.32
It has been hypothesized that in part CRT works by training basic processes “via proliferation and refining of basic neural connections.”33 Indeed, a functional imaging study by Wykes et al33 has demonstrated enhanced neural activation postcognitive remediation. Future efficacy studies that incorporate functional neuroimaging methodology may help address such questions regarding the underpinnings of cognitive remediation.
In sum, these preliminary observations (1) support the NEAR framework (a readily available, motivating, time-effective group intervention) to guide cognitive interventions in individuals with schizophrenia, (2) provide information about which specific cognitive functions could be targeted by this cognitive intervention, (3) demonstrate that the remediation of basic cognitive deficits does appear to persist and that these effects appear to generalize to social and occupational functions.
Eli Lilly Australia “Answers that Matter” Mental Health Award fund; Perpetual Trustees JS Love Trust; Trust fund of the Schizophrenia Fellowship of New South Wales.
The authors of this article gratefully acknowledge the contribution of clinicians from the Northern Sydney and Central Coast Area Health Service and Sydney West Area Health Service. Thank you to Dr David Cairns (Macquarie University) for providing statistical advice. We are also grateful to Professor Alice Medalia for training and ongoing supervision.