Efficacy of Cognitive Remediation in Schizophrenia: A Short Review of Its Variable Effects According to Cognitive Domain ()
1. Introduction
Relative to the positive and negative symptoms of schizophrenia, the cognitive deficits associated with the condition have a stronger relationship with social outcomes in daily life, including those involving work and school [1] . In recent years, cognitive interventions (cognitive remediation therapies) have been implemented and their results reported. As a means of improving participants’ daily functioning, numerous interventions have been designed to improve multiple cognitive domains, and the number of papers reporting related randomized controlled trials (RCTs) has also increased [2] [3] .
Intervention studies have been conducted using a variety of methods that differ in terms of terminology, approach, and targeted cognitive domains. Some focus on specific cognitive functions, such as attention [4] [5] and executive processing [6] , while others target a range of these functions simultaneously [7] or examine social cognition [8] [9] . In addition, there are now integrative programs that pair cognitive remediation with activities such as social skills training [10] .
Several papers have reviewed the degree to which cognitive remediation therapy improves cognition. Overall, these reviews have shown a medium effect size. McGurk et al. [11] performed a meta-analysis of 26 RCTs, calculated effect sizes across various outcomes, and reported effect sizes of 0.41, 0.36, and 0.28 for cognitive performance, psychosocial functioning, and symptoms, respectively. Wykes et al. [12] investigated the effects of cognitive remediation in a meta-analysis of 40 studies and reported a medium effect size of 0.45 for improved cognition and a very similar effect size of 0.42 for improved functioning. They also reported that the effect of cognitive remediation therapy on function was significantly larger when it was carried out in conjunction with other types of psychiatric rehabilitation. Cognitive remediation is used with increasing frequency as a therapeutic intervention intended to improve social function, employment, and quality of life [13] - [15] . Medalia and Saperstein [16] summarized studies published since 2011, with a focus on improvement of social function. They concluded that “cognitive remediation is most likely to impact functional outcome when individuals are given opportunities to practice the cognitive skills in real-world settings”. In this study, we summarized the cognitive outcomes reported in cognitive remediation research conducted subsequent to the reviews of McGurk et al. [11] and Wykes et al. [12] . Further, while referring to these reviews, we aimed to determine whether outcomes varied consistently according to cognitive domain in the existing research and how their effect size should be judged.
Summary of recent Findings incognitive remediation therapy Research
In this study, we examined trends in recent cognitive remediation research conducted subsequent to the meta- analyses of McGurk et al. [11] and Wykes et al. [12] . While there are many types of cognitive remediation therapy, no reviews focusing exclusively on the effectiveness of computer-assisted cognitive remediation have been conducted thus far. Therefore, we identified studies conducted between 2009 and 2013 to examine computer-assisted cognitive remediation and analyzed the literature describing their major results and trends.
2. Methods
2.1. Literature Sampling
We performed our literature search using the PubMed online database. We initially searched for literature containing all of our search terms, which were “cognitive,” “remediation” or “enhancement” or “rehabilitation,” “training” or “therapy,” and “schizophrenia.” We then reduced the results to those that met the following criteria: 1) the training made use of computers; 2) at least one cognitive domain was measured both pre- and post- intervention; and 3) a randomized controlled trial was conducted. Sixteen publications met these criteria [2] [16] - [30] . We then examined the cognitive outcomes described for these 16 studies.
2.2. Methods of Evaluating Efficacy and Effect Size
We used a 16-study sample to verify cognitive efficacy. Cognitive domains were categorized according to the grouping described by the Measurement and Treatment Research to Improve Cognition in Schizophrenia consensus panel: 1) attention/vigilance; 2) speed of processing; 3) verbal working memory; 4) verbal learning and memory; 5) visual learning and memory; 6) reasoning/problem solving; and 7) social cognition. Nonverbal working memory was only addressed by one of the 16 studies and was therefore omitted.
We examined the efficacy of cognitive remediation therapy in each of the seven cognitive domains described above, according to three criteria: 1) six or more studies examined the domain (1/3 or more of the 16-study sample); 2) at least half of the studies reported significant improvement; and 3) the median effect size was 0.3 or larger. On the basis of these criteria, we classified the effects of cognitive remediation therapy into one the following four categories: Category A: “effective,” in which all three criteria were met; Category B: “likely effective,” in which two of the three criteria were met; Category C: “possibly effective,” in which one of the three criteria were met; And Category D: “cannot claim efficacy,” in which none of the three criteria were met. The efficacy of cognitive remediation for each domain was evaluated according to these categories. It should be noted that some studies found significant improvements but did not report effect sizes; therefore, median effect sizes were calculated using the studies for which they were reported [2] [17] [18] [27] [30] . If the only literature available for a particular domain did not report effect sizes, the domain was considered to have failed Criterion 3 (“the median effect size was 0.3 or larger”).
3. Results
Using the methods described above, we tabulated the efficacy and effect sizes of computer-assisted cognitive remediation for each cognitive domain, with the following results (see Table 1).
1) Attention/vigilance: 12 of the 16 studies [2] [17] [18] [20] [21] [23] - [28] [30] assessed this domain, seven of which [2] [17] [23] [25] - [28] reported significant improvement. The median effect size recorded in the literature was 0.50 (0.35 - 0.65); therefore, we assigned this domain to Category A: “effective.”
2) Speed of processing: 13 of the 16 studies [2] [18] - [27] [30] [31] assessed this domain, four of which [22] [23] [27] [30] reported significant improvement. The median effect size recorded in the literature was 0.30; therefore, we assigned this domain to Category B: “likely effective.”
3) Verbal working memory: 14 of the 16 studies [17] - [30] assessed this domain, eight of which [18] [23] [25] - [30] reported significant improvement. The median effect size recorded in the literature was 0.51 (0.44 - 0.58); therefore, we assigned this domain to Category A: “effective.”
(4) Verbal learning and memory: 15 of the 16 studies [2] [17] - [19] [21] - [31] assessed this domain, nine of which [2] [18] [19] [21] [23] [24] [27] [28] [31] reported significant improvement. The median effect size recorded in the literature was 0.66 (0.18 - 1.13); therefore, we assigned this domain to Category A: “effective.”
(5) Visual learning and memory: 6 of the 16 studies [2] [19] [23] [25] - [27] assessed this domain, three of which [2] [23] [27] reported significant improvement. The median effect size recorded in the literature was 0.26; therefore, we assigned this domain to Category B: “likely effective.”
(6) Reasoning/problem solving: 14 of the 16 studies [2] [17] - [25] [27] [28] [30] [31] assessed this domain, 10 of which [2] [17] - [19] [23] [24] [27] [28] [30] [31] reported significant improvement. The median effect size recorded in the literature was 0.79 (0.28 - 1.29); therefore, we assigned this domain to Category A: “effective.”
(7) Social cognition: 2 of the 16 studies [18] [19] assessed this domain, one of which [18] reported significant improvement. However, because effect size was not recorded in the literature, this domain was considered to have failed Criterion 3; therefore, we assigned the domain to Category C: “possibly effective.”
4. Discussion
As shown above, of the seven cognitive domains considered, attention/vigilance, verbal working memory, verbal learning and memory, and reasoning/problem solving were assigned to Category A (effective). Category B (likely effective) contained the speed of processing and visual learning and memory domains, and Category C (possibly effective) contained the social cognition domain. No domains were assigned to Category D (cannot claim efficacy). We can therefore conclude that effects were demonstrated for intervention across all seven cognitive domains.
These results are very similar to those of the meta-analysis conducted by Wykes et al. [12] ; however, while meta-analysis showed strong improvement in social cognition, our trend analysis only found possible efficacy for this domain.
We theorized that this may have occurred as so few studies in our review measured social cognition. Our results also showed somewhat larger effect sizes than those reported by Wykes et al. [12] across all domains. The effect sizes in our study were calculated solely from the medians of those reported for the studies that measured effect size, and served as nothing more than reference values. However, the present study specifically considered research examining computer-assisted cognitive remediation, and these results suggested that computer-
![]()
![]()
Table 1. Study of the effect size of cognitive function domain.
![]()
Table 2. Study of the effect size of cognitive function domains.
*Summarizes the article on cognitive remediation for 2009-2013.
assisted cognitive remediation may be more effective relative to the various other types of remediation.
Comparing the degrees of improvement in the different domains, the largest effect size in our review was that of reasoning/problem solving (ES 0.79), followed by those of verbal learning and memory (ES 0.66), verbal working memory (ES 0.51), attention/vigilance (ES 0.50), and speed of processing (ES 0.30). In Wykes et al.’s meta-analysis [12] , the largest effect size was that of reasoning/problem solving (ES 0.572), followed by those of verbal learning and memory (ES 0.410), verbal working memory (ES 0.346), speed of processing (ES 0.258), and attention/vigilance (ES 0.250). Further, in McGurk et al.’s meta-analysis [11] , the largest effect size was that of verbal working memory (ES 0.52), followed by those of speed of processing (ES 0.48), reasoning/problem solving (ES 0.47), attention/vigilance (ES 0.41), and verbal learning and memory (ES 0.39; see Table 2). In McGurk et al.’s review [11] , outcomes were similar across all cognitive domains. In contrast, our review and that conducted by Wykes et al. [12] found great improvement in the reasoning/problem solving, verbal learning and memory, and verbal working memory domains, while the effect sizes for attention/vigilance and speed of processing indicated small improvement. In our review and the meta-analysis conducted by Wykes et al. [12] , the research results analyzed were more recent relative to those considered by McGurk et al. [11] and should reflect the efficacy of computer-assisted cognitive remediation more clearly. In short, our results showed improved cognitive performance subsequent to cognitive remediation therapy using computer programs, with particularly promising outcomes for reasoning/problem solving, verbal learning and memory, verbal working memory, and attention/vigilance.
NOTES
*Corresponding author.