Introduction
Background
Major depressive disorder (MDD) represents one of the most common mental health conditions on a global level, showing a lifetime prevalence of over 20% in high-income countries and interesting more than 300 million people worldwide [Reference Hasin, Meyers, Saha, Ruan and Stohl1–Reference Moreno-Agostino, Wu, Daskalopoulou, Hasan, Huisman and Prina3]. It also represents an essential source of disability, producing a significant socio-economic burden that extends well beyond direct healthcare costs [Reference Greenberg, Chitnis, Louie, Suthoff, Chen and Maitland4].
Cognitive impairment represents a central component of MDD [Reference Barlati, Lisoni, Nibbio, Baldacci, Cicale and Ferrarin5–Reference Rock, Roiser, Riedel and Blackwell7]. Despite the high level of heterogeneity of clinical presentations of the disorder, cognitive impairment affects a large proportion of people with MDD, often reaching moderate levels of severity and emerging from the first depressive episode [Reference Ahern and Semkovska8–Reference Varghese, Frey, Schneider, Kapczinski and de Azevedo Cardoso10]. This impairment frequently persists after the remission of the affective episode and appears to increase in severity over time with relapses and with age [Reference Dotson, McClintock, Verhaeghen, Kim, Draheim and Syzmkowicz11, Reference Semkovska, Quinlivan, O’Grady, Johnson, Collins and O’Connor12], involves both “hot” and “cold” cognitive functions [Reference Dam, Stenbæk, Köhler-Forsberg, Ip, Ozenne and Sahakian13–Reference Salehinejad, Ghanavati, Rashid and Nitsche15] and appears to be composed of state, trait, and scar elements [Reference Barlati, Lisoni, Nibbio, Baldacci, Cicale and Ferrarin5, Reference Allott, Fisher, Amminger, Goodall and Hetrick16, Reference Ronold, Schmid, Oedegaard and Hammar17]. Processing speed appears to be the most heavily interested domain, showing also a state-dependent effect during affective episodes, but attention, memory, and executive functions are also involved [Reference Hammar, Ronold and Rekkedal6, Reference Morey-Nase, Phillips, Bryce, Hetrick, Wright and Caruana18, Reference Snyder19].
Cognitive deficits in MDD represent one of the core determinants of impairment in social, interpersonal, and occupational functioning and also have a significant negative impactive on subjective quality of life [Reference Årdal, Lund and Hammar20–Reference Woo, Rosenblat, Kakar, Bahk and McIntyre24]. As such, they are becoming a treatment target of increasing importance and interest [Reference Maj, Stein, Parker, Zimmerman, Fava and De Hert2].
Cognitive remediation (CR) is a training-based psychosocial intervention targeting cognitive performance to provide significant and durable improvement in psychosocial functioning [Reference Wykes, Bowie and Cella25]. Based on the structured repetition of cognitive exercises and the development of novel strategies with the help of an active and trained therapist [Reference Bowie, Bell, Fiszdon, Johannesen, Lindenmayer and SR26], CR interventions have been shown to provide substantial positive effects on cognitive performance and on psychosocial functioning in people living with schizophrenia [Reference Barlati, Nibbio and Vita27–Reference Yeo, Yoon, Lee, Kurtz and Choi31] which are durable over time [Reference Vita, Barlati, Ceraso, Nibbio, Durante and Facchi32]. As such, it currently represents the psychosocial intervention with the highest level of recommendation in the European Psychiatric Association guidance for the treatment of cognitive impairment in schizophrenia [Reference Vita, Gaebel, Mucci, Sachs, Barlati and Giordano33].
CR interventions appear as a promising treatment also in MDD: recently, four meta-analyses have specifically investigated this topic [Reference Goldberg, Kuslak and Kurtz34–Reference Thérond, Pezzoli, Abbas, Howard, Bowie and Guimond37]. The first meta-analysis included 8 studies and 268 adults with MDD and reported positive treatment effects on most cognitive domains [Reference Thérond, Pezzoli, Abbas, Howard, Bowie and Guimond37]. A meta-analysis included 19 studies for a total of 944 participants with MDD or clinically relevant and depressive symptoms and included both a superiority and a noninferiority analysis: CR emerged as effective in treating depressive symptoms and noninferior to standard interventions for depression [Reference Sociali, Borgi, Pettorruso, Di Carlo, Di Natale and Tambelli36]. Another meta-analysis included studies on individuals with affective disorders for a total of 22 randomized controlled trials, 4 of which of individuals with bipolar disorder, and 993 participants: it reported small to moderate positive treatment effects on cognitive performance and depressive symptoms [Reference Goldberg, Kuslak and Kurtz34]. Finally, another meta-analysis included only MDD studies and included 21 trials and 978 participants: significant positive effects were observed on all outcomes, medium-sized on cognitive performance, and small-sized in depressive symptoms and daily functioning [Reference Legemaat, Semkovska, Brouwer, Geurtsen, Burger and Denys35].
While the results of these works converged in many aspects and showed that CR can represent an effective intervention in MDD, they also highlighted the current limitations in the available evidence: the certainty in the evidence assessed by Legamaat et al. with the GRADE procedure was very low for depressive symptoms and daily functioning, and low for cognitive performance [Reference Legemaat, Semkovska, Brouwer, Geurtsen, Burger and Denys35].
Besides the high level of heterogeneity in participants’ characteristics, interventions, and measures included in the trials, most available studies included only small samples and were conducted in a single research center. Moreover, subjective experiences of participants and patient-reported outcomes measures, which are becoming increasingly important in a regulatory and in a treatment personalization perspective [Reference Correll, Ismail, McIntyre, Rafeyan and Thase38], were seldom considered in available trials.
In this perspective, a large multicentric randomized controlled trial, adopting standardized and validated tools, including patient-reported outcomes, could offer valuable insights into cognitive, depressive, and psychosocial outcomes in MDD from both scientific and clinical perspectives.
Aims
The aims of the present single blind, multicentric randomized controlled trial were to assess the effects of a computerized CR intervention delivered by an active and trained therapist compared to an active control condition (computer games) in adults living with MDD undergoing a Major Depressive Episode (MDE) on cognitive performance, depressive symptoms, and psychosocial functioning. The main research hypothesis was that CR could provide significant improvements in all these outcomes and that the positive effects would be observed in both external, investigator-rated measures and in subjective, participant-reported measures.
Methods
Study design and participants
The present trial included nine different centers for the recruitment of participants: I) Department of Mental Health and Addiction Services, ASST Spedali Civili di Brescia and Department of Clinical and Experimental Sciences, University of Brescia, which also acted as the coordinating center; II) Department of Psychiatry, University of Campania Luigi Vanvitelli, Naples; III) Department of Clinical and Experimental Medicine, University of Foggia; IV) Department of Medical Sciences and Public Health, Psychiatry Section, University of Cagliari; V) Department of Clinical Neurosciences, IRCCS San Raffaele, Milan; VI) Department of Mental Health, ASL Lecce; VII) AOU San Luigi Gonzaga, Orbassano and Department of Neurosciences, University of Turin; VIII) Ospedale Santa Maria della Misericordia and Division of Psychiatry, Clinical Psychology and Rehabilitation, Department of Medicine, University of Perugia; IX) Department of Neurosciences, Imaging and Clinical Sciences, “Gabriele D’Annunzio” University of Chieti-Pescara.
Centralized training for the delivery of the CR and the control intervention and for the administration and rating of all outcome measures for researchers of all participating centers took place before the start of the study. Regularly scheduled online meetings were held during the course of the study between the coordinating center and all participating centers to assess and resolve potential issues.
Participants were consecutively recruited in each center with the following inclusion criteria: I) diagnosis of MDD according to DSM-5 criteria [39] confirmed with the Structured Clinical Interview for DSM-5- Clinical Version (SCID-5-CV) [Reference First, Williams, Karg and Spitzer40]; II) age between 18 and 60 years; III) clinical diagnosis of current MDE of mild or moderate severity according to DSM-5 criteria; IV) education of at least 8 years; V) consent to participate in the study, attested by a signed written informed consent form.
Exclusion criteria were: I) history of neurological disorders or severe traumatic head injury; II) diagnosis of intellectual disability; III) history of manic or hypomanic episodes, schizophrenia spectrum disorders or other psychotic disorders; IV) MDE with psychotic features and/or treatment resistance; V) history of alcohol or substance abuse in the previous 6 months; VI) participation in any type of CR intervention in the previous 6 months; VII) concurring pharmacological treatment with mood stabilizers, and first or second generation antipsychotics (benzodiazepines or Z-drugs low dose treatments were accepted); VIII) pregnancy.
Both outpatients and inpatients in psychiatric rehabilitation settings were considered for inclusion, as long as they met all inclusion criteria and no exclusion criteria. Participants were consecutively enrolled in each center from the study start (April 2021) to the recruitment of the targeted number of participants for each center, based on the results of a power analysis conducted prior to the start of the study.
All individuals who were considered for enrollment received a detailed explanation of the study aims and procedures. The subject agreed to participate in the study, provided explicit consent, and signed a dedicated written informed consent form.
The study was approved by the Local Ethical Committee (registration code NP 3173, approved April 29, 2020) of the coordinating center and of other participating centers and conducted in accordance with the Code of Ethics of the World Medical Association and the Declaration of Helsinki.
All necessary precautions were adopted to maintain patients’ anonymity and data confidentiality.
Participants were considered as drop-outs if they suspended antidepressant pharmacological treatment for more than 5 days and expressed intention of discontinuing it or if they did not participate in the active or control interventions for 2 consecutive sessions and expressed intention of discontinuing it.
Central randomization using a random number table was carried out in order to allocate participants to the CR intervention group or to the control condition. The randomization was performed by a researcher who was not involved in the delivery of the interventions, in the assessment of participants and in the data analysis.
Interventions
Cognitive remediation intervention
The CR intervention (CR group) was delivered with a computerized approach, using the Cogpack program Version 6.0 (Marker Software®, Mannheim, Germany). The program is composed of several cognitive exercises, including domain-specific exercises as well as non-domain-specific exercises focusing on different domains at once. It also includes exercises that focus on language and calculation skills. The exercises can be adapted for individual participants, with the software automatically setting the difficulty level on the basis of the participant’s performance during the session.
The exercises were delivered with the assistance and feedback of dedicated therapists who were specifically trained to deliver the intervention for the present study. Therapists were trained in centralized seminars that included lectures and practical exercises and were supported throughout the duration of the study with regularly scheduled online meetings.
Therapists had an active role in the delivery of the cognitive exercises, providing instruction and feedback to participants, closely monitoring their progress, and fostering the development of novel cognitive strategies.
The cognitive exercises were also integrated into a structured rehabilitation program that included discussion sessions dedicated to the development of cognitive strategies, with the aim of improving the transfer of cognitive gains into real-world functioning.
The CR intervention was administered two times a week, in 45-minute sessions, for 12 weeks and a total of 24 sessions. Missed sessions that were not related to study dropout were individually rescheduled.
Control intervention – computer games
The control intervention, included to assess the role of placebo and nonspecific effects, consisted of the use of computer games (CG group). Available games were “Minesweeper,” “Solitary” and “Pinball,” according to the participant’s choice. Games were played in a setting that included the presence of the therapist, but in this instance, the therapist did not play any active role, and no discussion session on cognitive strategies was provided.
The control intervention was also scheduled two times a week, in 45-min sessions, for a duration of 12 weeks and a total of 24 sessions.
At the conclusion of the study and of the dedicated assessments, participants allocated to the control group were invited to receive the CR intervention.
Measures
All participants were assessed at baseline (T0) and at the conclusion of the intervention (T1) with validated measures of cognitive performance, clinical symptoms, and psychosocial functioning.
Assessors were specifically trained to conduct the assessment and were kept blind as regard to the group allocation of participants for the whole duration of the study. Participants were explained and explicitly reminded not to disclose their group allocation to assessors.
Performance-based cognitive measures included the California Verbal Learning Test (CVLT) [Reference Delis, Kramer, Kaplan and Ober41], measuring the domain of verbal learning and memory; the Digit Symbol Substitution Test (DSST) [Reference Wechsler42] measuring mostly processing speed; Trail Making Test A (TMT-A) and B (TMT-B), and B-A (TMT-B-A) [Reference Reitan43] measuring processing speed, working memory, and executive functions/cognitive flexibility, respectively. Participant-rated measures included the Perceived Deficits Questionnaire – Depression 5 items (PDQ-D-5) to assess subjective cognitive dysfunction in people with depression [Reference Sullivan, Edgley and Dehoux44].
Clinical severity measures included the Montgomery–Åsberg Depression Rating Scale (MADRS) [Reference Montgomery and Åsberg45] and the Clinical Global Impression – Severity scale (CGI-S) [Reference Guy46] for clinician-rated measures and the Beck Depression Inventory, Second Edition (BDI-II) [Reference Beck, Steer and Brown47] for participant-rated measures. Real-world psychosocial functioning was measured with the Personal and Social Performance scale (PSP) [Reference Morosini, Magliano, Brambilla, Ugolini and Pioli48].
The inter-rater reliability was evaluated for all measures and for all assessors of each participating center after the centralized training using Interclass Correlation Coefficient (ICC): good to excellent agreement between raters was observed for all measures (ICC ranging from 0.62 to 0.91).
Statistical analyses
All individuals recruited in the study and randomized to one of the treatment arms (CR group or CG group) who completed the initial assessment were included in the analyses according to their allocation, regardless of their completion of the study, to implement an Intention-To-Treat (ITT) approach.
Baseline socio-demographic (age, gender, years of education) and clinical characteristics were compared between groups using t-tests for continuous variables and χ 2 tests (or Fisher’s Exact test where appropriate) for categorical variables to identify potential differences: variables that emerged as significantly different were introduced as covariates in the outcome analyses to avoid potential confounding factors.
Drop-out rates were compared between the intervention groups to identify potential sources of bias. Baseline values for all outcome measures were also compared between study completers and dropouts.
Outcomes were assessed using mixed models for repeated measures (MMRM), considering baseline (T0) and end-of-treatment (T1) scores and group allocation (CR or CG) to implement an ITT approach. Statistically significant Time x Group interactions were considered as indicating a significant treatment effect. Simple effects for each treatment arm were also investigated.
p-Values <0.05 were considered significant. Analyses were performed using Jamovi version 2.2.5.
Between-group effect sizes were calculated as Cohen’s d, considering both baseline and end-of-treatment values and baseline standard deviations, and taking into account group allocation and size [Reference Morris49].
Statistical analyses were carried out maintaining blinding as regard to participants’ group allocation to avoid potential sources of bias.
Results
A total of 142 individuals were screened, and 101 participants were allocated to the CR group (n = 52) or to the CG group (n = 49). A total of 81 participants, 45 in the CR group and 36 in the CG group, completed the study. The complete CONSORT Flow Diagram of the study is reported in Figure 1.
CONSORT flow diagram.
Baseline socio-demographic and clinical characteristics of the total sample and between-groups comparisons are reported in Table 1. No significant between-groups differences were observed for any baseline socio-demographic or clinical characteristic.
Sample characteristics and baseline comparisons
Abbreviations: MDE, major depressive episode; SNRI, serotonin norepinephrine reuptake inhibitors; SSRI, selective serotonin reuptake inhibitor; VOR, vortioxetine.
a Other: amitriptyline, buproprione, clomipramine, trazodone.
Comparisons between study completers and dropouts are reported in Supplementary Table S1. No significant between-group differences in dropout rates were observed (χ2 = 2.713, p = 0.100). No differences were observed between completers and dropouts regarding baseline variables, except for psychosocial functioning as measured by the PSP (t = 2.296, p = 0.024).
Study outcomes are reported in Table 2. The CR intervention produced significantly superior results, compared to the CG control condition, in clinician-rated depressive symptoms as measured by the MADRS (p = 0.023, d = 042) and the CGI-S (p = 0.025, d = 0.39) and in subjective depressive symptoms as measured by the BDI-II (p = 0.005, d = 0.45). The CR intervention also produced significantly superior results in the working memory domain as measured by the TMT-B test (p = 0.004, d = 0.34), in the executive functions/cognitive flexibility domain with a moderate effect size, as measured by the TMT-B-A (p = 0.020, d = 0.43), and in subjective cognitive impairment as measured by the PDQ-5 (p = 0.006, d = 0.48). No significant treatment effects were observed as regards the verbal memory domain as measured by the CVLT, the processing speed domain as measured by the DSST and the TMT-A, and in real-world psychosocial functioning as measured by the PSP.
Study outcomes (mixed model repeated measures)
Abbreviations: BDI-II, Beck Depression Inventory, Second Edition; CGI-S, Clinical Global Impression - Severity scale; CVLT, California Verbal Learning Test; DSST, Digit Symbol Substitution Test; MADRS, Montgomery–Åsberg Depression Rating Scale; PDQ-5, Perceived Deficits Questionnaire – Depression 5 items; PSP, Personal and Social Performance scale; TMT, Trail Making Test. Bold font highlights significant p-values (p<0.05).
Discussion
The present multicentric randomized controlled trial aimed to assess the effects of a CR intervention in people living with MDD undergoing a MDE.
The CR intervention showed a good feasibility profile, with a dropout rate of 13.5%: this is in line with that of other evidence-based psychosocial interventions for MDD [Reference Cooper and Conklin50, Reference Solomonov and Barber51], as well as with that of CR interventions delivered to individuals with other diagnoses, such as Schizophrenia Spectrum Disorders [Reference Vita, Barlati, Ceraso, Deste, Nibbio and Wykes52].
The CR intervention produced significant improvements, compared to the CG control intervention, in several outcomes of interest. In particular, a larger reduction in the severity of depressive symptoms, both measured by raters and measured by the participants themselves, was observed in the CR group. The observation that CR produces significant, small-to-moderate-sized improvements in depressive symptoms is in line with the results of previous, smaller studies and of recent meta-analyses [Reference Goldberg, Kuslak and Kurtz34, Reference Legemaat, Semkovska, Brouwer, Geurtsen, Burger and Denys35]. The fact that this improvement is also recognized by participants, as attested by the scores of self-reported instruments, represents an encouraging finding, further confirming that CR can represent an effective intervention to treat this domain.
As regards the effects on cognitive performance, significantly small-to-moderate-sized positive effects were observed in the working memory and executive functions domains. On the contrary, no significant effect was observed in the processing speed and verbal memory domain. Again, these findings are mostly in line with those reported in the most recent meta-analytic investigation, where a positive effect was observed in the working memory and executive functions, but not in the processing speed, attention, or visual memory domains [Reference Goldberg, Kuslak and Kurtz34]. It might well be that CR interventions produce in people living with MDD undergoing a MDE improvements that are more evident in some cognitive domains, and are not capable of producing significant improvements in other domains, contrary to what can be observed in the context of CR in Schizophrenia Spectrum Disorders [Reference Lejeune, Northrop and Kurtz29, Reference Vita, Barlati, Ceraso, Nibbio, Ariu and Deste30]. Developing CR interventions that are specifically designed to also improve cognitive domains such as processing speed, which are among those that are more heavily interested in MDD [Reference Hammar, Ronold and Rekkedal6, Reference Morey-Nase, Phillips, Bryce, Hetrick, Wright and Caruana18, Reference Snyder19], might represent an interesting perspective for future research.
However, considering the heterogeneity and the prevalence of cognitive impairment in depression and the substantial negative impact that it produces on the daily lives of people with MDD, where it represents one of the main predictors of functional disability [Reference McIntyre, Cha, Soczynska, Woldeyohannes, Gallaugher and Kudlow53–Reference Cambridge, Knight, Mills and Baune56], even providing small improvements in selected domains could represent a substantial benefit for participants.
As regards subjective cognitive deficits, in the present study CR produced significant, small-to-moderate-sized improvements compared to the control condition. This is an interesting and novel finding: in fact, in a recent meta-analytic investigation, a small positive effect on this outcome was observed, but the analysis did not reach statistical significance [Reference Goldberg, Kuslak and Kurtz34]; this might be due to the small number of included studies (n = 4) and the small sample size of each included work. Again, this finding is very encouraging, attesting that CR interventions can provide improvements that are perceived and are considered meaningful for participants: in a patient-centered and person-first approach [Reference Correll, Ismail, McIntyre, Rafeyan and Thase38], this outcome is of utmost relevance.
Conversely, no significant improvement was observed in real-world psychosocial functioning. This is a somewhat unexpected finding that could be due to different reasons. First and foremost, the lack of a follow-up observation could largely explain this result: in fact, the transfer of cognitive and clinical improvements to psychosocial functioning usually requires more time to be observed, and in samples of subjects with Schizophrenia Spectrum Disorders, longer follow-up observations are related to increased psychosocial functioning improvements [Reference Vita, Barlati, Ceraso, Nibbio, Durante and Facchi32]. Moreover, it should also be pointed out that participants who dropped out of the study showed lower baseline psychosocial functioning: this issue could have a negative impact on the results observed for this specific outcome. Finally, it should also be highlighted that part of the present study took place during the COVID-19 pandemic and this issue, which had an enormous impact on the lives of all people and in particular on those of people with mental health conditions and their daily functioning [Reference Moreno, Wykes, Galderisi, Nordentoft, Crossley and Jones57, Reference Robinson, Sutin, Daly and Jones58], could have negatively impacted the transfer of cognitive gains into real world functional outcomes due to the peculiar context of the pandemic itself, with limited opportunities for novel personal interactions and developments. In any case, to better explore the impact of CR interventions on the psychosocial functioning of people with MDD, future studies should include consistent follow-up observation periods after the conclusion of the active treatment phase.
Strengths of the present study include its multicentric nature and the consequent sample size, which allows for more reliable results compared to previous individual studies, and the inclusion of participant-rated measures for both clinical and cognitive outcomes, which is becoming an increasingly important element also from a regulatory perspective [Reference Baandrup, Rasmussen, Mainz, Videbech and Kristensen59].
A limitation of the present work is the lack of follow-up observations, which could have allowed a better evaluation of treatment impact on functional outcomes and to assess the durability of positive effects. Another limitation is the lack of direct and specific assessment of participants’ real-world outcomes: while psychosocial functioning represents an outcome of the present work and was assessed with a validated scale, instruments such as the Specific Level of Functioning [Reference Schneider and Struening60] could allow a more nuanced and detailed evaluation of participants’ functioning.
Future perspectives include the development of CR interventions tailored specifically for the characteristics of individuals living with MDD, both during MDEs and during remission phases, the inclusion of follow-up observations even for long periods of time, the inclusion of specific measures of real-world functional outcomes, and the use of novel technologies, such as telemedicine and virtual reality to help deliver evidence-based interventions. In a clinical perspective, a wider dissemination of CR intervention in real-world, day-to-day mental health services could represent an important asset to help treat essential elements of MDD, such as cognitive impairment, providing also improvements that are perceived and appreciated directly by participants [Reference Maj, Stein, Parker, Zimmerman, Fava and De Hert2].
In conclusion, the results of the present work highlight that CR is an effective intervention in people with MDD, capable of improving clinical outcomes and cognitive performance both in a clinician-rated and in a subjective manner. In this perspective, CR should be considered more consistently in MDD treatment recommendations and guidelines as an evidence-based intervention and should be delivered in a more widespread manner in mental healthcare and rehabilitation services.
Supplementary material
The supplementary material for this article can be found at http://doi.org/10.1192/j.eurpsy.2025.10073.
Data availability statement
Data from the present study is available upon reasonable request to the Corresponding Author.
Acknowledgements
All authors who contributed to this paper are listed as authors. No professional medical writer was involved in any portion of the preparation of the manuscript.
Financial support
The study was promoted and conducted as an independent study partially supported by Lundbeck Italia.
Competing interests
The authors declare none.
Open access
Comments
No Comments have been published for this article.