Study selection

The EMBASE (n = 217), Medline (n = 145) and PsycINFO (n = 151) searches returned 513 items. After removal of duplicates, 395 records remained. Initial screening of title, abstract and keywords excluded a further 376 articles, leaving 19 items. After applying the exclusion criteria, 14 studies were included in the review. See Appendix 2 for a flow chart of the study inclusion. The second author was consulted to clarify the inclusion of two studies^{Reference Bul, Doove, Franken, der Oord, Kato and Maras19,Reference Wols, Lichtwarck-Aschoff, Schoneveld and Granic20} and the exclusion of one.

Study characteristics

The study characteristics and participants of the included studies are presented in Table 1. The significant outcomes and qualitative results of the studies included are provided in Table 2. The 14 included studies were conducted in various locations ranging from Europe (Belgium,^{Reference Bul, Doove, Franken, der Oord, Kato and Maras19,Reference Bul, Kato, der Oord, Danckaerts, Vreeke and Willems21} France,^{Reference Flaudias, Zerhouni, Chakroun-Baggioni, Pires, Schmitt and De Chazeron22,Reference Maurin, Girod, Consolini, Belzeaux, Etain and Cochet23} Ireland,^{Reference Flaudias, Zerhouni, Chakroun-Baggioni, Pires, Schmitt and De Chazeron22} The Netherlands,^{Reference Bul, Doove, Franken, der Oord, Kato and Maras19–Reference Bul, Kato, der Oord, Danckaerts, Vreeke and Willems21,Reference Semkovska and Ahern24,Reference Schoneveld, Lichtwarck-Aschoff and Granic25} Portugal,^{Reference Gamito, Oliveira, Lopes, Brito, Morais and Rebelo26} Spain^{Reference García-Baos, D'Amelio, Oliveira, Collins, Echevarria and Zapata27,Reference Lado-Codesido, Pérez, Mateos, Olivares and Caballero28} ) to Asia (Turkey^{Reference Gülkesen, Işleyen, Cinemre, Samur, SŞ and Zayim29}) and the USA.^{Reference Sanchez, Brown, Kocher and DeRosier30,Reference Anguera, Gunning and Areán31} A total of 1167 participants were included (591 in intervention groups and 576 in control groups), with a total sample size ranging from 22 to 174 participants. Four studies targeted anxiety.^{Reference Wols, Lichtwarck-Aschoff, Schoneveld and Granic20,Reference Schoneveld, Lichtwarck-Aschoff and Granic25,Reference Sanchez, Brown, Kocher and DeRosier30,Reference Schoneveld, Malmberg, Lichtwarck-Aschoff, Verheijen, Engels and Granic32} Three studies focused on attention-deficit hyperactivity disorder (ADHD).^{Reference Bul, Doove, Franken, der Oord, Kato and Maras19,Reference Bul, Kato, der Oord, Danckaerts, Vreeke and Willems21,Reference García-Baos, D'Amelio, Oliveira, Collins, Echevarria and Zapata27} Depression was targeted in two studies.^{Reference Semkovska and Ahern24,Reference Anguera, Gunning and Areán31} Schizophrenia was the subject of two studies.^{Reference Lado-Codesido, Pérez, Mateos, Olivares and Caballero28,Reference Gülkesen, Işleyen, Cinemre, Samur, SŞ and Zayim29} Alcohol use disorder was targeted in two studies.^{Reference Flaudias, Zerhouni, Chakroun-Baggioni, Pires, Schmitt and De Chazeron22,Reference Gamito, Oliveira, Lopes, Brito, Morais and Rebelo26} Bipolar disorder was the focus of one study.^{Reference Maurin, Girod, Consolini, Belzeaux, Etain and Cochet23} Six studies were aimed at children,^{Reference Bul, Doove, Franken, der Oord, Kato and Maras19–Reference Bul, Kato, der Oord, Danckaerts, Vreeke and Willems21,Reference Schoneveld, Lichtwarck-Aschoff and Granic25,Reference García-Baos, D'Amelio, Oliveira, Collins, Echevarria and Zapata27,Reference Sanchez, Brown, Kocher and DeRosier30,Reference Schoneveld, Malmberg, Lichtwarck-Aschoff, Verheijen, Engels and Granic32} seven studies focused on adults^{Reference Flaudias, Zerhouni, Chakroun-Baggioni, Pires, Schmitt and De Chazeron22–Reference Semkovska and Ahern24,Reference Gamito, Oliveira, Lopes, Brito, Morais and Rebelo26,Reference Lado-Codesido, Pérez, Mateos, Olivares and Caballero28,Reference Gülkesen, Işleyen, Cinemre, Samur, SŞ and Zayim29} and one study was aimed at older adults.^{Reference Anguera, Gunning and Areán31} Six studies compared serious games with treatment as usual (TAU) or no intervention ^{Reference Bul, Doove, Franken, der Oord, Kato and Maras19,Reference Bul, Kato, der Oord, Danckaerts, Vreeke and Willems21,Reference Gamito, Oliveira, Lopes, Brito, Morais and Rebelo26,Reference Lado-Codesido, Pérez, Mateos, Olivares and Caballero28–Reference Sanchez, Brown, Kocher and DeRosier30} ; four studies compared them with a game;^{Reference Flaudias, Zerhouni, Chakroun-Baggioni, Pires, Schmitt and De Chazeron22,Reference Semkovska and Ahern24,Reference García-Baos, D'Amelio, Oliveira, Collins, Echevarria and Zapata27,Reference Schoneveld, Malmberg, Lichtwarck-Aschoff, Verheijen, Engels and Granic32} and four studies compared them with a therapy, such as CBT or a psychoeducational programme.^{Reference Wols, Lichtwarck-Aschoff, Schoneveld and Granic20,Reference Maurin, Girod, Consolini, Belzeaux, Etain and Cochet23,Reference Schoneveld, Lichtwarck-Aschoff and Granic25,Reference Anguera, Gunning and Areán31} The serious games were played in various settings: six studies were conducted at home,^{Reference Bul, Doove, Franken, der Oord, Kato and Maras19,Reference Bul, Kato, der Oord, Danckaerts, Vreeke and Willems21,Reference Semkovska and Ahern24,Reference García-Baos, D'Amelio, Oliveira, Collins, Echevarria and Zapata27,Reference Gülkesen, Işleyen, Cinemre, Samur, SŞ and Zayim29} three at school,^{Reference Wols, Lichtwarck-Aschoff, Schoneveld and Granic20,Reference Schoneveld, Lichtwarck-Aschoff and Granic25,Reference Schoneveld, Malmberg, Lichtwarck-Aschoff, Verheijen, Engels and Granic32} two at hospital,^{Reference Flaudias, Zerhouni, Chakroun-Baggioni, Pires, Schmitt and De Chazeron22,Reference Gülkesen, Işleyen, Cinemre, Samur, SŞ and Zayim29} two at a clinic^{Reference Gamito, Oliveira, Lopes, Brito, Morais and Rebelo26,Reference Lado-Codesido, Pérez, Mateos, Olivares and Caballero28} and two were unclear.^{Reference Maurin, Girod, Consolini, Belzeaux, Etain and Cochet23,Reference Anguera, Gunning and Areán31}

Table 1 Characteristics of the randomised controlled trials that are included in the review

GRADE, Grading of Recommendations, Assessment, Development and Evaluations; HRSD, Hamilton Rating Scale for Depression; PHQ-9, Patient Health Questionnaire; WHODAS-II, World Health Organization Disability Assessment Schedule II; TOVA, Test of Variables of Attention; PST, problem-solving therapy; ADHD, attention-deficit hyperactivity disorder; TAU, treatment as usual; AST, alcohol Stroop test; OCDS, Obsessive Compulsive Drinking Scale; MMSE, Mini-Mental Examination Test; FAB, Frontal Assessment Battery; ILFE, I am Learning Facial Expressions; RMV-SV, Reading the Mind in the Voice – Spanish Version; MARS, Medication Adherence Rating Scale; ALQ, Achieved Learning Questionnaire; PECK, Personal Experiences Checklist; SCAS, Spence Children’s Anxiety Scale; CBT, cognitive–behavioural therapy.

Table 2 Significant outcomes and qualitative results of the studies

CBT, cognitive–behavioural therapy; ADHD, attention-deficit hyperactivity disorder; RMV-SV, Reading the Mind in the Voice – Spanish Version.

GRADE rating

Table 3 shows the GRADE rating of each study. The GRADE criteria allocation was done by using the five factors of likely risk of bias, precision, consistency, directness and lack of publication bias. Specifically, tools for risk of bias and precision were investigated for. Presence of four or more factors gave a score of ‘high’, three factors gave a score of ‘moderate’, two factors gave a score of ‘low’ and one or no factors gave a score of ‘very low’. Three studies were rated as high.^{Reference Maurin, Girod, Consolini, Belzeaux, Etain and Cochet23–Reference Schoneveld, Lichtwarck-Aschoff and Granic25} Six studies were rated moderate.^{Reference Bul, Doove, Franken, der Oord, Kato and Maras19–Reference Bul, Kato, der Oord, Danckaerts, Vreeke and Willems21,Reference Lado-Codesido, Pérez, Mateos, Olivares and Caballero28,Reference Sanchez, Brown, Kocher and DeRosier30,Reference Anguera, Gunning and Areán31} Three studies were rated as low.^{Reference Flaudias, Zerhouni, Chakroun-Baggioni, Pires, Schmitt and De Chazeron22,Reference García-Baos, D'Amelio, Oliveira, Collins, Echevarria and Zapata27,Reference Schoneveld, Malmberg, Lichtwarck-Aschoff, Verheijen, Engels and Granic32} Two studies were rated as very low.^{Reference Gamito, Oliveira, Lopes, Brito, Morais and Rebelo26,Reference García-Baos, D'Amelio, Oliveira, Collins, Echevarria and Zapata27}

Table 3 GRADE rating criteria and individual study scores

GRADE, Grading of Recommendations, Assessment, Development and Evaluations.

The results have been divided into subheadings of the mental disorders targeted in the studies.

Anxiety

Four studies targeted symptoms of anxiety or social skills in children. Three used the same serious game MindLight.^{Reference Wols, Lichtwarck-Aschoff, Schoneveld and Granic20,Reference Schoneveld, Lichtwarck-Aschoff and Granic25,Reference Schoneveld, Malmberg, Lichtwarck-Aschoff, Verheijen, Engels and Granic32} Two of these studies^{Reference Wols, Lichtwarck-Aschoff, Schoneveld and Granic20,Reference Schoneveld, Lichtwarck-Aschoff and Granic25} were the same, but one had different outcomes measured at a later time.^{Reference Wols, Lichtwarck-Aschoff, Schoneveld and Granic20}

MindLight is a three-dimensional, third-person neurofeedback video focused on a boy saving his grandmother from evil forces at her scary mansion (see video of game trailer: https://www.youtube.com/watch?v=buNaErarLts). In all three studies the game was played at school, after hours. MindLight was defined as a goal-oriented and problem-solving type of serious game.

In one of the MindLight studies, the games were played in an unspecified number of groups of 7–19 children (aged 8–13 years, 46% male, 68 in the intervention group, 66 in the control group), who were all supervised by two research assistants. The rationale for the age group of children targeted was not provided. The sessions were 1 h long, two times a week, and they had five sessions in total. The rationale for the dose was not provided. The control group played Max and the Magic Marker, a puzzle platform video game involving a small vulnerable boy overcoming in-game challenges (see video of game trailer: https://www.youtube.com/watch?v=1MPGb7akIpc). An active control was used to ensure that attention, motivation, behavioural activity and expectations do not account for improvements in the intervention group. Both games are valid for the population chosen, as they are designed for children, and MindLight specifically targets children with anxiety symptoms. The participants and assistants running the sessions were not blinded. The randomisation of participants was stratified by gender and school grade, and was performed by an independent researcher using the SPSS random number generator. There was no presence of a conflicts of interest section in the paper, so it is unclear whether there were any or not. The outcomes were measured using questionnaires on anxiety symptoms (parent and child version of Spence Children’s Anxiety Scale; SCAS), average time spent playing video games, game expectations and game evaluations (appeal to self, appeal to others, relevance, flow and difficulty).^{Reference Schoneveld, Malmberg, Lichtwarck-Aschoff, Verheijen, Engels and Granic32}

In this first study when compared with the avatar game, both MindLight and the control group showed improvements in anxiety symptoms at the 3-month follow-up (P < 0.05). The control group also was rated as more appealing to themselves at post-treatment (P < 0.001) and follow-up (P < 0.05), more appealing to others at follow-up (P < 0.05) and more likely to induce feeling of flow at post-treatment (P < 0.05).^{Reference Schoneveld, Malmberg, Lichtwarck-Aschoff, Verheijen, Engels and Granic32}

The other two studies played MindLight in an unspecified number of groups of five to ten children (aged 7–12 years, 40.8% male, 86 in intervention groups, 88 in control groups) supervised by Master's students for 1 h a week, for 6 weeks. The rationale for the age group of children targeted was not provided. The control group did CBT as Coping Cat (an effective CBT programme for anxious children) for eight sessions after school, in groups of four to seven children led by two psychologists.^{Reference Wols, Lichtwarck-Aschoff, Schoneveld and Granic20,Reference Schoneveld, Lichtwarck-Aschoff and Granic25} The rationale for the dose was not provided. The participants and assistants running the sessions were not blinded. The randomisation of participants was stratified by gender and school grade, and was performed by an independent researcher using the SPSS random number generator. One of the authors of these two studies is the founder of the PlayNice Institute, the executive producer of the MindLight game.^{Reference Wols, Lichtwarck-Aschoff, Schoneveld and Granic20,Reference Schoneveld, Lichtwarck-Aschoff and Granic25}

The outcomes from the second study using MindLight were measured with questionnaires on total anxiety (parent and child SCAS), personalised anxiety, total hours per week playing video games, programme expectations and programme ratings (appeal to self, appeal to others, relevance, anxiety inducing and difficulty).^{Reference Schoneveld, Lichtwarck-Aschoff and Granic25}

Compared with online CBT, there was child- and parent-reported decreases in anxiety levels in children for up to 6 months. MindLight was also rated as non-inferior to the CBT control, suggesting that it is a sufficient alternative to traditional CBT. However, the CBT control was rated significantly more relevant to daily life.^{Reference Schoneveld, Lichtwarck-Aschoff and Granic25} The scale of difficulty of MindLight as a game was rated well by children; it was at the middle of the scale, suggesting that it has a good balance of learning and challenge.^{Reference Schoneveld, Lichtwarck-Aschoff and Granic25}

These two studies^{Reference Wols, Lichtwarck-Aschoff, Schoneveld and Granic20,Reference Schoneveld, Lichtwarck-Aschoff and Granic25} were the same, but one had different outcomes measured at a later time.^{Reference Wols, Lichtwarck-Aschoff, Schoneveld and Granic20} The outcomes from the third study using MindLight included how pre-test anxiety scores were related to in-game play behaviours during the first play session, and whether changes in in-game play behaviours from the first to last play session predicted changes in anxiety symptoms at the 3-month follow-up. The coding was done by research assistants that were blinded to the hypotheses.^{Reference Wols, Lichtwarck-Aschoff, Schoneveld and Granic20}

There were significant improvements in anxiety symptoms at the 3-month follow-up (P < 0.05) for children with engaged in-game play behaviours. There were significantly predicted higher anxiety symptoms at the 3-month follow-up (P < 0.01) for children with avoidant/safety in-game play behaviours.^{Reference Wols, Lichtwarck-Aschoff, Schoneveld and Granic20}

The fourth study used the serious game ‘Adventures aboard the S.S.GRIN’ in children with social skills challenges (aged 7–11 years, 60% male, 33 in the intervention group, 36 in the control group) (see video of game trailer: https://www.youtube.com/watch?v=fHEJs8gr7mk). The rationale for the age group of children targeted was not provided. The game was developed from skills taught in an evidence-based, in-person social skills programme,^{Reference DeRosier33} and involves using a personalised avatar to interact with non-playable characters on the sailing ship S.S. GRIN. There was in-game feedback, hints and prompts based on performance. Weekly episodes were released for a total of 9 weeks, and players had 1 week to complete each episode. The rationale for the dose was not provided. The game was specifically designed for children with social skills challenges. Adventures aboard the S.S.GRIN was defined as a goal-oriented and problem-solving type of serious game. The relevant subscales of the Behaviour Assessment System for Children, Second Edition (BASC-2) norm-referenced screening system for measuring behavioural and emotional strengths and weaknesses was used to identify children struggling with socioemotional skills. The rationale for the dose was not provided. The control group was placed on a waiting list, but it was not made clear why this decision was made.^{Reference Sanchez, Brown, Kocher and DeRosier30}

The participants and the assistants running the sessions were not blinded. The randomisation of participants was stratified by age, gender, race, ethnicity and BASC-2 subscale scores. The authors were employees of the 3C Institute, which may benefit financially from game sales. The outcomes were measured using questionnaires on social literacy (Achieved Learning Questionnaire); social self-efficacy (Self-Efficacy and Outcome Expectancy Measure); social anxiety (Social Anxiety Scale for Children – Revised); social satisfaction (Loneliness and Social Dissatisfaction Scale); bullying victimisation (Personal Experiences Checklist); and bullying perpetration (bullying others subscale of the California Bullying Victimisation Scale).^{Reference Sanchez, Brown, Kocher and DeRosier30}

This game appeared to be effective as, compared with the control group, there were improvements in social literacy (P = 0.000), social satisfaction (P = 0.039), social anxiety (P = 0.047) and bullying victimisation (P = 0.001). However, the control group was a waiting list and an active control was not used, meaning that the results could just reflect the effect of gaming.^{Reference Sanchez, Brown, Kocher and DeRosier30}

ADHD

Three studies targeted symptoms related to ADHD in children.^{Reference Bul, Doove, Franken, der Oord, Kato and Maras19,Reference Bul, Kato, der Oord, Danckaerts, Vreeke and Willems21,Reference García-Baos, D'Amelio, Oliveira, Collins, Echevarria and Zapata27}

Two papers used the same serious game ‘Plan-It Commander’^{Reference Bul, Doove, Franken, der Oord, Kato and Maras19,Reference Bul, Kato, der Oord, Danckaerts, Vreeke and Willems21} (see video of game trailer: https://www.youtube.com/watch?v=8FoSDUJS8eQ&ab_channel=%26ranj). These two papers performed the same study, but the later paper separately analysed a subgroup from the original data at a later time.^{Reference Bul, Doove, Franken, der Oord, Kato and Maras19} Plan-It Commander is an online adventure game designed for children with ADHD, and was developed by healthcare professionals, researchers and game experts to improve domains of daily life functioning, with a primary focus on time management, planning/organising and cooperation skills. The game had a closed social community of players, and had a competitive element where players could gain badges and gifts. The study focused on children aged 8–12 years, 81% were male and there were 88 in the intervention group and 82 in the control group. The rationale for the age group of children targeted was not provided. The game was played at home and players were instructed to play for a maximum of 65 min, approximately three times a week for 10 weeks. The rationale for the dose was not provided. The game was specifically designed for children with ADHD. Plan-It Commander was defined as a goal-oriented and problem-solving type of serious game. The control group received TAU; it was not made clear why this decision was made.^{Reference Bul, Doove, Franken, der Oord, Kato and Maras19,Reference Bul, Kato, der Oord, Danckaerts, Vreeke and Willems21}

The participants were not blinded, and it was not possible to blind researchers and teachers. The randomisation of participants was done on a 1:1 ratio by a prespecified, computer-generated randomisation list, and the allocation was stratified by study site and gender. The authors reported that there were no conflicts of interest. The primary outcomes for the first study were online questionnaires done by parents on reported time-management skills, planning/organising skills (the subscale Plan/Organize of the Behavior Rating Inventory of Executive Function, parent version) and cooperation skills (the subscale Cooperation of the Social Skills Rating System, parent version).^{Reference Bul, Kato, der Oord, Danckaerts, Vreeke and Willems21}

The outcome of the second study was to use a virtual twins analysis to identify subgroups that were likely to get a high benefit from the treatment compared with the estimated treatment effect of the total group.^{Reference Bul, Doove, Franken, der Oord, Kato and Maras19}

Plan-It Commander showed improvements in multiple domains: parent-rated time-management skills (P = 0.004), parent-reported working memory (P = 0.02), and responsibility skills (P = 0.04), suggesting that it is an effective intervention. However, there was a higher drop-out rate in the intervention and those who dropped out had higher ADHD severity scores. This suggests that this serious game may not be suitable for children with more severe ADHD.^{Reference Bul, Kato, der Oord, Danckaerts, Vreeke and Willems21}

Furthermore, in the follow-up study, it was found that there were significant improvements in planning and organising skills in two subgroups: girls, and boys with low hyperactivity and higher conduct disorder scores.^{Reference Bul, Doove, Franken, der Oord, Kato and Maras19}

The other study used the serious game RECOGNeyes in children (age 8–15 years, 64% male, 28 total participants). The rationale for the age group of children targeted was not provided. This game involved the player using eye-tracking software to catch snowflakes while avoiding fire in six subgames. The game was installed on provided laptops and played at home. The game was played for 30 min, three times a week for 3 weeks. The rationale for the dose was not provided. The game is specifically designed for training attention in people with ADHD. RECOGNeyes was defined as a cognition training type of serious game. The control group played the same game but using a mouse instead of eye-tracking software. The reason for this was not fully explained.

There was no description of blinding within the study. The description of the randomisation was unsatisfactory as the paper only stated that the groups were balanced for comorbidities and gave no more details. Two of the authors were affiliated with Braingaze SL, the eye-tracking software company, and one of these authors received money from Braingaze SL to carry out the study. The outcomes were assessed using a Frog task to assess the probability of ADHD, severity of ADHD, hyperactivity index, impulsivity index and performance parameters (number of errors, reaction time, and number and duration of gaze fixations). A word recognition task assessed the dyslexia index and performance parameters (number of errors, reaction time, and number and duration of gaze fixations). A questionnaire with a three-point scale was used to assess usability and enjoyability. It was not stated if this was a validated questionnaire.^{Reference García-Baos, D'Amelio, Oliveira, Collins, Echevarria and Zapata27}

The results from the serious game RECOGNeyes that used eye-tracking software appeared to show that it was effective; however, the study was given a ‘low’ GRADE score and there were many elements of the study that were not sufficiently described. This serious game could still have the opportunity to be shown to be effective in future studies with a more robust study design.^{Reference García-Baos, D'Amelio, Oliveira, Collins, Echevarria and Zapata27}

Depression

Two studies targeted depression.^{Reference Semkovska and Ahern24,Reference Anguera, Gunning and Areán31}

One study used the game Project: EVO on a mobile or iPad, which involved guiding a character thorough an immersive environment and responding to targets. Project: EVO was defined as a goal-oriented and problem-solving type of serious game. The game was played by adults (mean age 68 years, s.d. 6.3, 27% male, 12 in the intervention group, ten in the control group) and they were instructed to play for 20 min, five times a week for 1 month. The 4-week duration of the intervention was chosen because it was based upon the previous duration used by the authors’ previous study that the serious game in this study was derived from.^{Reference Anguera, Boccanfuso, Rintoul, Al-Hashimi, Faraji and Janowich34} The duration of the trial was based on a previous similar study. The game was not designed for this population and was originally designed for children. The control group received problem-solving therapy, an intervention that helps patients with goal-directed behaviour, for 8 weeks; the rationale for this choice was not clear. There was no description of blinding within the study. The randomisation was done by a random number generator. The authors reported that there were no conflicts of interest.

The outcomes were collected by research assistants on depression symptoms (Hamilton Rating Scale for Depression and Patient Health Questionnaire-9), physical disability (Charlson Comorbidity Index), sustained attention (test of variables of attention), working memory (delayed recognition task), a basic response time task and a trait adjective task to assess negativity bias.^{Reference Anguera, Gunning and Areán31}

The serious game showed improvements in working memory, with and without distraction (P < 0.05), but comparable improvements to the control in reported disability (P = 0.002) and depressive symptoms post-treatment and at the 8-week follow-up (P < 0.05). There was, however, full adherence to the intervention.^{Reference Anguera, Gunning and Areán31}

The second study use a game called RehaCom, a computerised neurocognitive remediation therapy in adults (aged 31–65 years, 18% male, 11 in the intervention group, 11 in the control group). The game had six procedures, all with multiple levels of difficulty: divided attention (1 and 2), verbal memory, figural memory, shopping and plan a day. The players downloaded the game on their personal computers and played at home. The control group played games that required selective attention, strategy and remembering cues. Both groups had to play for 1 h, four times a week for 5 weeks, and were sent random reminders to complete their sessions. This study followed the same research protocol as a previous study of one of the authors.^{Reference Semkovska, Lambe, Lonargáin and McLoughlin35} The reason for the choice of control was not fully explained. The game is designed for neurocognitive remediation therapy. RehaCom was defined as a cognition training type of serious game.

The participants were not blinded but the researchers were blinded. The group allocation was stratified by gender and generated by a pre-programmed command with the statistical package R. The authors reported that there were no conflicts of interest. The outcome for this study was neurocognitive functioning, which was assessed by a standardised battery of validated neuropsychological tests (the Digit Symbol Substitution to assess psychomotor processing speed; the d2 Test of Attention for divided attention; the Digit Span Forward for auditory attention; the Digit Span Backward for verbal working memory; the Logical Memory I and II for verbal learning and retention, respectively; the Rey–Osterrieth Complex Figures for visual learning (immediate recall) and retention (delayed recall); and the following Delis–Kaplan Executive Function System subtests for the assessment of the stated executive functions: Verbal Fluency (three consecutive categories) for self-regulation under external constraints, Fluency Switching for mental flexibility, Towers for planning and 20-Questions for abstract thinking.^{Reference Semkovska and Ahern24}

It is important to note that one study solely focused on neurocognitive functioning,^{Reference Semkovska and Ahern24} which is just one of the several deficits seen in patients with depression. The serious game in this study, RehaCom, showed multiple improvements in the intervention group compared with the control group: divided attention (P = 0.016), verbal working memory(P = 0.003), planning (P = 0.001), long-term verbal memory (P < 0.001) and mental flexibility (P = 0.014). Allowing this serious game to be played at home, when it is most convenient and potentially comfortable for the patient, could lead to improved adherence in these patients.^{Reference Semkovska and Ahern24}

Schizophrenia

Two studies targeted symptoms of schizophrenia.^{Reference Lado-Codesido, Pérez, Mateos, Olivares and Caballero28,Reference Gülkesen, Işleyen, Cinemre, Samur, SŞ and Zayim29}

One study used a game called I am Learning Facial Expressions (ILFE) in adults (mean age 37.3 years, s.d. 9.2, 38% male, 18 in the intervention group, 14 in the control group). This included eight serious games with a range of levels. The player was shown different images of people’s faces and had to correctly identify the shown emotion. Those without a computer at home played using a hospital computer, and if patients forgot to play then they were reminded. They were instructed to play for 1 h, at least two times a week for 1 month. The rationale for this dose was not provided. The control group is assumed to be TAU, but was not sufficiently described. The game is designed with a consideration for some of the common characteristics of schizophrenia. ILFE was defined as a cognition training type of serious game. The reason for the chosen dosage was not made clear. There was no mention of blinding in the study. The group allocation was randomised. The authors reported that there were no conflicts of interest. The primary outcome was correct identification of facial expressions.^{Reference Gülkesen, Işleyen, Cinemre, Samur, SŞ and Zayim29}

The serious game ILFE appeared to be effective at improving the correct identification of facial expression compared with the control (P < 0.001). However, the reliability of this study is questionable, as it was given a GRADE score of ‘very low’.^{Reference Gülkesen, Işleyen, Cinemre, Samur, SŞ and Zayim29}

The second study used the serious games VOICES in adults (mean age 40.9 years, s.d. 12.1, 52% male, 27 in the intervention group, 26 in the control group). The game involved the patient being played a phrase and selecting the emotion conveyed. The patients attended a centre and played in a quiet room with trained personnel there to show them how to use the game. They played for approximately 30 min, two times a week for 1 month, and each session increased in difficulty. The rationale for this dose was not provided. The control received TAU; the reason for this was unclear. The game is designed to train emotions. VOICES was defined as a cognition training type of serious game. The reason for the chosen dosage was not made clear. The participants were not blinded but the researchers were blinded. The group allocation was done by a computer-generated randomisation list. There was no presence of a conflicts of interest section in the paper, so it is unclear whether there were any or not.

The primary outcome was emotion recognition using the Reading the Mind in the Voice – Spanish Version test.^{Reference Lado-Codesido, Pérez, Mateos, Olivares and Caballero28}

The other study, which used the serious game VOICES, appeared to be effective as there was a significant improvement from the pre-test to the post-test scores (P = 0.009), and also a significant improvement compared with the control (P < 0.001).^{Reference Lado-Codesido, Pérez, Mateos, Olivares and Caballero28}

For adults with schizophrenia, it is important to note that one study solely focused on the correct identification of facial expressions^{Reference Gülkesen, Işleyen, Cinemre, Samur, SŞ and Zayim29} and the other study focused on emotion recognition from spoken words.^{Reference Lado-Codesido, Pérez, Mateos, Olivares and Caballero28} These two outcomes measured will reflect only two of the several deficits seen in patients with schizophrenia.

Alcohol dependence

Two studies targeted symptoms of alcohol dependence.^{Reference Flaudias, Zerhouni, Chakroun-Baggioni, Pires, Schmitt and De Chazeron22,Reference Gamito, Oliveira, Lopes, Brito, Morais and Rebelo26}

One study used a game based on a cognitive–behavioural modification programme on a tablet device in adults (intervention group mean age 48.5 years, s.d. 12.63; control group mean age 43.73 years, s.d. 7.86; 91% male; 18 in the intervention group; 15 in the control group). The game involved a series of pictures with four images shown at once, of which three were alcohol-related images and one was not. The patients needed to choose the unrelated picture as quickly as possible. The game was played in groups of four to five for 30 min, two times a week for 3 weeks. The rationale for the dose was not provided except for one reference to a previous study.^{Reference Fadardi and Cox36} This was compared with a control group that used a card-pairing memory game, which was used to avoid a potential effect of ‘gaming’ in the intervention. The game is designed for attentional bias training in patients with alcohol use disorder, and it was defined as a cognition training type of serious game. The reason for the chosen dosage was not made clear. There was no mention of blinding in the study. The groups were randomly allocated. The authors reported that there were no conflicts of interest. In this study, the outcomes were capacity to inhibit and appropriate automatic semantic response (Hayling Sentence Completion Test), attentional bias toward alcohol (alcohol Stroop test), attention control (classical Stroop test), severity of alcohol usage (Alcohol Use Disorders Identification Test) and craving (Obsessive Compulsive Drinking Scale).^{Reference Flaudias, Zerhouni, Chakroun-Baggioni, Pires, Schmitt and De Chazeron22} There were improvements in the intervention group in capacity to inhibit appropriate automatic semantic response (P = 0.006), craving (P < 0.001) and attentional bias (P = 0.001), but there were also improvements in the control group in attentional control (P = 0.047) and craving (P = 0.003). The GRADE score for this study was ‘low’, which makes the reliability of the results questionable.^{Reference Flaudias, Zerhouni, Chakroun-Baggioni, Pires, Schmitt and De Chazeron22}

The second study used several serious games designed as a cognitive stimulation programme in adults (mean age 45.45 years, s.d. 10.31, 90% male, 22 in the intervention group, 23 in the control group). The difficulty increasing with each level and the games were designed to stimulate attention, memory, decision-making, language, processing speed, strategic planning, perception and spatial vision. Some of the games were developed by the team and some were available commercially, but were set up as a cognitive stimulation programme. The game was defined as a cognition training type of serious game. The games were played for 45–50 min, three times per week for 1 month, for ten sessions in total. The rationale for the dose was not provided. The control was placed on a waiting list and received TAU. The reason for the choice of control was not fully explained. There was no mention of blinding in the study. The groups were randomly selected. There was no presence of a conflicts of interest section in the paper, so it is unclear whether there were any or not. The primary outcomes for this study were general cognitive ability (Mini-Mental State Examination) and frontal lobe functions (Frontal Assessment Battery).^{Reference Gamito, Oliveira, Lopes, Brito, Morais and Rebelo26} The study showed improvements in frontal lobe functions (P = 0.001), but again the GRADE score for this study was ‘very low’, which makes this significant result very unreliable.^{Reference Gamito, Oliveira, Lopes, Brito, Morais and Rebelo26}

Bipolar disorder

One study targeted symptoms of bipolar disorder by using the game BIPOLIFE in adults (intervention group mean age 46.58 years, s.d. 14.01; control group mean age 41.44 years, s.d. 7.62; 37% male; 20 in the intervention group; 21 in the control group). This game is centred on an avatar with bipolar disorder in everyday situations, with the aim being for the player to regulate the avatar's mood. The participants were told to play as often and for as long they wanted to during a 1-month period, and there were in-game messages to prompt the player. Before the game, the patients received a 12-week psychoeducational programme. The control group received TAU and a 12-week psychoeducational programme. The rationale for the dose was not provided. The reason for the choice of control was not fully explained. The game is designed for patients with bipolar disorder. BIPOLIFE was defined as a goal-oriented and problem-solving type of serious game.

The participants were not blinded, but the researchers were blinded. The allocation was done randomly and assigned with a 1:1 ratio, and the randomisation sequence was centralised and computed in blocks of four for each level of stratification by the study statistician, in an order unknown by the investigators. The study was funded by AstraZeneca, who were involved in the BIPOLIFE game development. The primary outcome of this study was medication adherence rate (Mediation Adherence Rating Scale), which was recorded by trained psychologists or psychiatrists.^{Reference Maurin, Girod, Consolini, Belzeaux, Etain and Cochet23}

In the intervention there was significantly higher absolute variation in medication adherence (P = 0.03) and improvements in attitude toward psychiatric medications (P = 0.002) after 1 month, but this was no longer significant at the 4-month follow-up. These results should be reliable as the study was given a GRADE score of ‘high’. Encouraging positive thoughts and behaviours around BIPOLIFE as a treatment could lead to improved medication adherence in patients.^{Reference Maurin, Girod, Consolini, Belzeaux, Etain and Cochet23}