Effectiveness of Psychological Intervention Methods for the Management of Positive and Negative Symptoms in Schizophrenia Patients: A Systematic Review and Network Meta-analysis

Abstract

Purpose of the Review:

There is a dearth of research on determining the optimal psychological intervention approach for symptom-specific management in schizophrenia. Hence, the current study aimed to assess the efficacy of therapeutic strategies in the management of various symptom subtypes, such as positive symptoms (PS) and negative symptoms (NS), hallucinations, and delusions in schizophrenia.

Collection and Analysis of Data:

Six electronic journal databases, namely Web of Science, Scopus, EBSCOhost, PubMed, Ovid, and Taylor and Francis were searched up to November 2023 using Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) for Network Meta-Analyses (NMA) guidelines. Articles were identified using predefined keywords and subsequently screened by pre-specified inclusion and exclusion criteria. Data from the included randomized controlled trials (RCTs) were extracted for PS, NS, hallucinations, and delusions, both for pre- and post-trial conditions. The effect size and standard errors of each trial were computed using the “metafor” library, and further NMA was performed using the “netmeta” package of R software version 4.2.1. The analysis was conducted on 100 RCTs involving 7,036 patients with schizophrenia undergoing eight different psychological interventions and treatment-as-usual (TAU). Meta-cognitive training (MCT+), social skills training (SST), cognitive behavioral therapy (CBT), and cognitive remediation therapy (CRT) were found to be effective in treating both PS and NS of schizophrenia. However, virtual reality therapy (VRT) was found to be effective only in the management of hallucinations and overall PS, and MCT+ and CBT were found to be effective treatment measures in the management of delusions.

Conclusions:

Psychotherapies were found to be more effective compared to TAU, except acceptance and commitment therapy (ACT) and supportive psychotherapy (ST).

PROSPERO Registration ID: CRD42022385392.

Schizophrenia is a major psychotic disorder that causes functional impairment, ¹ poor quality of life, ² unemployment, ³ and poor adherence to treatment, ⁴ in most patients. Due to the dominance of the biological model, antipsychotics remain the principal treatment choice but carry a risk of adverse treatment effects that require rigorous and resource-intensive monitoring. ⁵ Mean-while, developments in the field of social psychiatry and theoretical formulations of stress vulnerability models have led to the introduction of numerous intervention methods, not as an alternative treatment but as adjunctive psychological treatment for the improvement and management of psychotic symptoms. ⁶

Research across the globe has demonstrated that the use of antipsychotic drugs combined with psychological interventions is the most resource-efficient way to manage the symptoms of patients with schizophrenia. ⁷ For instance, cognitive behavioral therapy (CBT) has proven efficacious in treating positive symptoms (PS),^8,9 negative symptoms (NS), and functional impairments,^10,11 and leads to fewer dropouts than standard care. ¹² Cognitive remediation therapy (CRT) is efficacious in improving day-to-day life functioning in patients with schizophrenia^13,14 and reducing NS. ¹⁵ Similarly, supportive psychotherapy (ST) is effective in treating NS. ⁹ Many recent randomized controlled trials (RCTs),^16–18 and systematic reviews ¹⁹ have used virtual reality therapy (VRT)-based psychological interventions and shown their efficacy in treating auditory verbal hallucinations,^17,18 paranoia, ¹⁸ NS, ¹⁶ and social skills impairment ¹⁹ in schizophrenia and related disorders.

Recent meta-analytical studies,^20–24 have reported many pairwise comparisons of the efficacy of psychological therapies in the management of clinical symptoms in patients with schizophrenia. As network meta-analytic (NMA) studies have added methodological advantages over meta-analysis,^25,26 a few NMA studies,^27–29 have compared the effectiveness of various psychotherapeutic methods with conventional pharmacotherapy in managing PS,^27,29 NS, general psychopathology, ²⁷ and preventing relapse in patients with psychosis and related conditions. ²⁹ However, none of the NMA-based studies have examined the subtypes of PS, such as hallucinations and delusions, as treatment outcome measures. Although hallucinations and delusions are associated with each other, they occur on a continuum. ³⁰ Both can occur independently or often follow distinct longitudinal courses, may persist or remit independently over time, and respond differently to treatment.^31–33 It is evident that the content of hallucinations provides raw data for the development and preservation of delusions^30,34; however, other factors, such as social, emotional, and somatic, etc., also contribute substantially to the genesis and maintenance of delusions. ³⁰ Therefore, it is essential to identify the therapeutic approaches that are most effective in managing hallucinations and delusions.

Therefore, there is a need to undertake a systematic literature review (SLR) and NMA to generate symptom-specific effectiveness profiles for different psychological intervention methods for treating patients with schizophrenia. Hence, the current NMA intended to evaluate the efficacy of various psychotherapeutic approaches in the management of PS, that is, hallucinations and delusions, and NS in schizophrenia patients.

Methods

The present SLR and NMA were conducted as per the Preferred Reporting Items for Systematic Reviews and Meta-Analyses for Network Meta-Analyses (PRISMA-NMA) guidelines. ³⁵

Data Sources and Search Strategy

The present SLR was conducted by the first and second authors independently using various journal hosting sites, such as PubMed, Web of Science, EBSCOhost, Ovid, Scopus, and Taylor and Francis Online, until November 25, 2023. Predefined search keywords were used with Boolean operators (i.e., AND and OR). The search terms used were (Cognitive Therapy OR Behavior Therapy OR Cognitive Behavior Therapy or CBT OR Cognitive Remediation OR Cognitive Retraining OR Cognitive Rehabilitation OR Meta-cognitive Training OR MCT OR Supportive Psychotherapy OR Psychological Intervention OR Social Skills Training OR Virtual Reality OR Avatar Therapy OR Augmented reality OR Acceptance and Commitment Therapy OR Intervention OR Psychological Intervention) AND (Schizophrenia OR Schizoaffective Disorder OR Schizophrenia Spectrum Disorder OR Psychotic Disorder OR Psychosis) AND (Randomized Control Trial* OR Clinical Trial* OR RCT) AND (Psychosis OR Psychotic OR Delusion* OR Voice* OR Paranoia OR Hallucination* OR Positive Symptom* OR Negative Symptom*). Additionally, the complete reference lists of recently published SLRs and meta-analyses were searched manually. Each research record was initially evaluated based on the title and abstract. Subsequently, the entire text was assessed using the preset inclusion and exclusion criteria. In instances of disagreements, the same reviewers independently extracted data, and resolutions were reached through reviewer consensus.

Inclusion and Exclusion Criteria

The following inclusion criteria were used for the selecting potential articles estimating the efficacy of different psychological intervention techniques in the management of psychotic symptoms both PS and NS in schizophrenia patients: (a) sample comprised with patients diagnosed with Schizophrenia, Schizophrenia Spectrum Disorders, Schizoaffective Disorder, Delusion Disorders, Psychosis not otherwise specified either as per ICD-(10 and 11) or DSM (IV & 5); (b) participants subjected to psychological intervention methods, such as cognitive therapy (CT), CBT, meta-cognitive training (MCT+), CRT, VRT, ST, etc., (c) RCT used a pre-test post-test control group design; (d) psychotic symptoms (e.g., hallucination, delusion, NS) were assessed as both baseline and outcome measures, (e) peer-reviewed articles in the English Language.

The following exclusion criteria were used to provide homogeneity in the treatment group: (a) participants diagnosed with bipolar disorders, depressive disorder, organic psychosis, substance induced psychotic disorders, and personality disorders, (b) review and meta-analytical studies, (c) studies conducted in child and adolescent population (d) only cognitive functioning, including attention and concentration, memory, executive functioning, etc., as outcome measures. (e) Studies that did not provide post-test outcome assessment.

Data Extraction

Relevant information, such as the author’s name, publication year, country, research design, types of intervention used, sample size, psychometric tools to measure psychopathology, frequency of sessions, duration of therapy, and pre- and post-assessment measures of psychopathology for both the treatment and treatment-as-usual (TAU) groups, was extracted from the included articles. Furthermore, essential data extracted from the studies were entered into Microsoft Excel sheets based on the chosen outcomes, namely, PS, NS, hallucinations, and delusions. Among all the included studies, the Positive and Negative Syndrome Scale (PANSS)-Positive, Scale for the Assessment of Positive Symptoms (SAPS), and Brief Psychiatric Rating Scale (BPRS)-Positive were used as measurement tools for PS. In contrast, PANSS-Negative, Scale for the Assessment of Negative Symptoms (SANS), and BPRS-Negative were used to measure NS. The Psychotic Symptom Rating Scales (PSYRATS)-Hallucinations and PSYRATS-Delusions subscales were used to quantify the severity of hallucinations and delusions, respectively.

Additionally, many RCTs included in the present NMA used therapeutic approaches such as relaxation, befriending, and psychoeducation as control conditions. These approaches are typically unstructured, lack targeted mechanisms, involve minimal therapist contact, and are low-intensity and low-patient engagement. Due to these shared features, these therapeutic approaches have been grouped as control conditions in the present NMA. Previous NMA-based studies^28,29 have also included befriending, medication, and family support as control conditions.

Risk of Bias Assessment

Potential biases in the included studies were measured using the Risk of Bias (RoB) 2.0 tool. ³⁶ The Cochrane Collaboration designed and developed this tool specifically for RCT-based studies. The tool encompasses biases associated with the randomization process, deviations from planned interventions, incomplete outcome data, outcome measurement, and the reporting of results. Studies were judged as low, with some concerns and high RoB for each of the five domains, as well as for the judgment of overall bias. The RoB of all the included studies in the present NMA is depicted in Supplementary Figures S1 and S2.

Statistical Analysis

The NMA was performed using the R 4.2.1 software along with two packages, namely “metafor,” ³⁷ and “netmeta.” ³⁸ All analyses were performed with a two-tailed significance level set at 5%. First, the effect sizes and variances of all studies were computed using the cmicalc() function of the “metafor” package. As most studies reported the means and standard deviations of outcome measures for the treatment and TAU groups in pre- and post-intervention conditions, the effect size was calculated using the “dppc2” method proposed by Morris. ³⁹ Estimation of the dppc2 and variance requires a correlation between pre- and post-test scores, which has not been reported in any of the included studies. Therefore, a moderate correlation coefficient of 0.5 was used to estimate the dppc2 and the variance using the Morris method. ³⁹ A correlation value of 0.5 provides a more realistic estimate with a moderate deviation. Effect sizes were adjusted to Hedges g by J-correction for a sample size of less than 20 reported in studies. ⁴⁰ Additionally, a few studies did not report mean and SD under pre- and post-treatment conditions but reported Cohen’s d and 95-CI values; therefore, these were used for the estimation of the standard error. ⁴¹ Few studies reported mean gain or mean change scores in pre-post conditions, along with standard deviations, which were further converted to Cohen’s d and standard error by the conversion convention provided by Wilson and Lipsey. ⁴² Furthermore, the “netmeta” package was used to perform pairwise comparisons, both direct and indirect, across the network. The “netmeta” package uses “TE” and “seTE” as the generic inputs for effect size and standard error, respectively, for pairwise comparison of treatment and TAU groups. ⁴³ All computations were performed using a random-effects model. In the present study, TAU was used as a reference group for comparison with other treatment groups. To visually represent the effectiveness of “TAU” in comparison to other intervention methods, a forest plot was created using the “forest” function. ³⁸ To evaluate potential publication bias in the current NMA, the funnel () function was applied, and the symmetry of the resulting plots was tested using Egger’s regression method. A function netgraph () was used to create a network diagram representing the layout of all interventions for each outcome. The diagram contains nodes representing interventions and lines reflecting direct comparisons made between each intervention. ³⁸ Furthermore, the netheat () function was used to draw heat maps that illustrated inconsistencies in the network. ³⁸ Lastly, the netrank() function with “surface under the cumulative ranking” (SUCRA) method was used to identify the best intervention among all interventions to alleviate the particular outcome. ³⁸

Results

Search Results

As indicated in the flowchart (Figure 1), the preliminary search yielded a total of 2,421 titles from PubMed, 2,646 from EBSCOhost, 318 from Ovid, 3,453 from Web of Science, 191 from Taylor and Francis Online, and 1,267 from Scopus. Following a comprehensive screening of titles, 384 articles were identified as potentially relevant to the present study. Subsequently, articles were shortlisted based on full-text review. Of the 162 articles shortlisted through full-text screening, 43 were identified as review or meta-analytical studies and were excluded from the analysis. After applying the inclusion and exclusion criteria, 100 articles were finally retained for the present meta-analysis (Supplementary Table S1).

Figure 1. Flow Chart for the Illustration of the Sequential Process of Conducting Searches for the Present NMA.

Table 1 shows the standardized mean difference (SMD) along with the 95% confidence interval (CI) and z-score of each outcome measure of all eight psychological treatment methods compared with TAU. The network plot (Figure 2) displays pairwise comparisons of multiple interventions with TAU for each outcome measure, that is, PS, NS, hallucinations, and delusions. The included studies in the present NMA were conducted by various researchers across different parts of the globe at varying points in time. Further, variability was observed among the included studies regarding the selection of samples, measurement tools, and the number and frequency of therapeutic sessions. Hence, a random-effects model was chosen for the com-putation. Additionally, network heterogeneity was observed in all four networks, which was an additional rationale for selecting a random-effects model for the NMA.

Table 1.

SMD, 95% CI, and Z-score for Each Intervention Compared with Treatment-as-usual.

Interventions	Positive Symptoms		Negative Symptoms		Hallucinations		Delusions
	SMD [95% CI]	Z-score	SMD [95% CI]	Z-score	SMD [95% CI]	Z-score	SMD [95% CI]	Z-score
ACT	–0.09 [–0.65; 0.47]	–0.31	0.27 [–0.39; 0.93]	0.80	××		××
CBT	–0.33 [–0.47; –0.18]	–4.49**	–0.27 [–0.43; –0.12]	–3.50**	–0.19 [–0.43; 0.06]	–1.51	–0.33 [–0.55; –0.12]	–3.03**
CRT	–0.14 [–0.25; –0.02]	–2.36*	–0.31 [–0.43; –0.18]	–4.71**	–0.16 [–0.91; 0.58]	–0.43	0.15 [–0.36; 0.67]	0.58
CT	–0.12 [–0.60; 0.37]	–0.47	–0.26 [–0.80; 0.27]	–0.96	××		××
MCT+	–0.46 [–0.63; –0.29]	–5.19**	–0.39 [–0.64; –0.14]	–3.05**	–0.21 [–0.71; 0.30]	–0.80	–0.36 [–0.65; –0.06]	–2.39*
SST	–0.31 [–0.55; –0.07]	–2.55**	–0.42 [–0.73; –0.12]	–2.70**	0.18 [–0.72; 1.07]	0.39	–0.08 [–0.81; 0.64]	–0.22
ST	–0.12 [–0.32; 0.08]	–1.17	–0.06 [–0.31; 0.20]	–0.43	0.00 [–0.36; 0.37]	0.03	–0.30 [–0.65; 0.06]	–1.64
VRT	–0.60 [–1.14; –0.06]	–2.17*	–0.29 [–0.79; 0.21]	–1.12	–0.80 [–1.33; –0.28]	–3.02**	–0.39 [–1.12; 0.34]	–1.04
Heterogeneity	Q = 116.33, df = 70, tau = 0.20; I2 = 39.8%		Q = 153.51, df = 72, tau = 0.27; I2 = 53.1%		Q = 49.88, df = 19, tau = 0.33; I2 = 61.9%		Q = 51.07, df = 23, tau = 0.28; I2 = 55%

SMD: Standardized Mean Difference, CI: Confidence Interval, ACT: Acceptance and Commitment Therapy, CBT: Cognitive Behavior Therapy, CRT: Cognitive Remediation Therapy, CT: Cognitive Therapy, MCT+: Meta-cognitive training, ST: Supportive Therapy, SST: Social Skills Training, VRT: Virtual Reality Therapy, TAU: Treatment-as-Usual.

*Significant at 0.05 level, ** Significant at 0.01 level.

“××” shows no studies have reported data on that particular outcome measure and treatment.

Figure 2. Network Graph for Visual Representation of Treatment Comparisons for (a) Positive Symptoms, (b) Negative Symptoms, (c) PSYRATS Hallucination, (d) PSYRATS Delusion. The Thickness of Lines Represents the Number of Trials Comparing Each Treatment Pair, and the Size of Nodes Represents the Number of Studies Providing Data for Each Treatment.

Positive Symptoms

From a total of 78 studies, eight interventions were compared, which assessed PS in both pre- and post-intervention conditions in patients with schizophrenia (Figure 3, Tables 1 and 2). VRT was the most effective method for treating PS compared with TAU (SMD: –0.60, SUCRA = 0.88), followed by MCT+ (SMD = -0.46, SUCRA = 0.86). Social skills training (SST) and CBT were significantly superior to TAU. They achieved SUCRA scores exceeding 50, indicating that these three therapeutic approaches are likely to be about half as effective as the best hypothetical intervention for treating PS of schizophrenia (Figure 3, Tables 1 and 2). Moreover, CRT was also notably effective in decreasing PS, even though the SUCRA scores were under 50. Acceptance and commitment therapy (ACT), CT, and ST were not significantly different from TAU, indicating similar therapeutic effectiveness to that of TAU. The I ² = 39.8% indicates low heterogeneity present among the studies assessing the efficacy of various psychological intervention techniques to treat PS of schizophrenia (Table 1). Egger’s test yielded a non-significant result (p = .13), suggesting no evidence of publication bias in the studies representing the treatment of PS (Supplementary Figure S3). Except for the CT, ST, and TAU sub-networks, the heat plot showed consistency throughout the network. It permitted the use of the transitivity assumption for indirect comparisons between study designs (Supplementary Figure S4).

Figure 3. Forest Plot Representing the Treatment Effects of Psychological Intervention Methods on the Severity of Positive Symptoms, Negative Symptoms, Hallucinations, and Delusions in Schizophrenia Patients.

Table 2.

Rank Order of Psychological Intervention Methods and Their SUCRA Values Compared to TAU.

Intervention Methods	Positive Symptoms		Negative Symptoms		Hallucinations		Delusions
	SUCRA (Random)	Rank	SUCRA (Random)	Rank	SUCRA(Random)	Rank	SUCRA(Random)	Rank
ACT	0.34	7	0.09	9	NC	NC	NC	NC
CBT	0.66	3	0.58	5	0.57	3	0.70	2
CRT	0.34	6	0.64	3	0.50	4	0.13	7
CT	0.37	5	0.56	6	NC	NC	NC	NC
MCT+	0.86	2	0.77	2	0.61	2	0.74	1
SST	0.63	4	0.82	1	0.23	7	0.38	5
ST	0.31	8	0.25	7	0.31	5	0.63	4
VRT	0.88	1	0.59	4	0.97	1	0.70	3
TAU	0.10	9	0.18	8	0.31	6	0.22	6

SUCRA: The surface under the cumulative ranking curve, NC: Not compared, ACT: Acceptance and Commitment Therapy, CBT: Cognitive Behavior Therapy, CRT: Cognitive Remediation Therapy, CT: Cognitive Therapy, MCT+: Meta-cognitive training, ST: Supportive Therapy, SST: Social Skills Training, VRT: Virtual Reality Therapy, TAU: Treatment-as-usual.

Negative Symptoms

Eight psychological interventions from 80 studies examined NS in patients with schizophrenia in both pre- and post-intervention conditions (Table 1). SST emerged as the most effective treatment method for NS (SMD = -0.42, SUCRA = 0.82), followed by MCT+ (SMD = -0.39, SUCRA = 0.77). CBT and CRT were statistically significantly different compared with TAU and had SUCRA scores above the 50 level, indicating that these three therapeutic approaches are likely to be about half as effective as the best hypothetical intervention for treating NS of schizophrenia (Figure 3, Tables 1 and 2). CT was also found to be a more effective treatment method than TAU, with an SUCRA score of 0.56, although the SMD was not statistically significant. ST, VRT, and ACT were not statistically significant treatment effects compared with TAU (Table 2). The total heterogeneity score for this network was statistically significant, and a moderate I² score indicates that the NMA explains 53.1% of heterogeneity among the studies representing psychological interventions for the management of NS in schizophrenia (Table 1). The heat plot shows consistency throughout the network with minor exceptions in the ST-TAU-CBT and TAU-SST-MCT+ sub-networks, validating the use of the transitivity assumption for the direct and indirect comparisons among different study designs (Supplementary Figure S4). The significant result of Egger’s test (p = .03) indicated detectable publication bias in the studies evaluating NS treatment (Supplementary Figure S3).

Hallucinations

The analysis included a total of 25 studies that examined six psychological interventions using PSYRATS-H to measure hallucinatory symptoms in schizophrenia patients. VRT was the only treatment method that produced better treatment effectiveness compared to TAU (SMD = -0.80, SUCRA = 0.97). SST ranked lower than TAU, with a SUCRA score of 0.23 (Table 1 and Table 2). Despite MCT+, CBT, and CRT having SUCRA scores above 50 levels, the SMDs of these methods did not differ significantly compared with TAU, indicating similar therapeutic effectiveness compared to TAU in the management of hallucination (Table 2 and Figure 3). ST produced almost similar treatment effectiveness compared with TAU and ranked SUCRA scores of 0.31. Cochran’s Q value being greater than the degrees of freedom and an I² statistic of 61.9% indicates a moderate degree of heterogeneity present among studies estimating the efficacy of different psychological approaches for managing hallucinations (Table 1 and Figure 3). The heat plot shows consistency throughout the network, supporting the use of the transitivity assumption for indirect comparisons among study designs (Supplementary Figure S4). The non-significant outcome of Egger’s test (p = .19) indicated an absence of publication bias in the studies addressing hallucination treatment (Supplementary Figure S3).

Delusions

Twenty-nine studies have attempted to treat delusions in schizophrenic patients using six intervention methods—all the included studies employed PSYRATS-D to rate delusions both in pre-intervention and post-intervention conditions. The NMA indicates (Table 1 and Table 2) that MCT+ emerged as the most effective treatment method compared to TAU (SMD = -0.36, SUCRA = 0.74). CBT emerged as the second most effective treatment measure to treat delusions. Further, VRT and ST are ranked with SUCRA scores above 50 levels, but SMDs did not differ significantly compared with TAU (Table 2). The total heterogeneity score (Cochran’s Q = 51.07, df = 23, p < .01) was found to be statistically significant. Additionally, an I² score of 55% indicates a moderate degree of variability accounted for by the NMA due to heterogeneity across studies measuring the effectiveness of psychological interventions for managing delusions in schizophrenia (Table 1 and Figure 3). The heat plot shows consistency in the network, supporting the use of the transitivity assumption for direct and indirect comparisons between psychological interventions (Supplementary Figure S4). Egger’s test yielded a p value of .42, providing no statistical indication of publication bias in research on the treatment of delusions (Supplementary Figure S3).

Sensitivity Analysis

We imputed a correlation value of 0.5 to estimate the effect sizes and variances using the Morris method. Therefore, as per the recommendations of Borenstein et al. ⁴¹ and Higgins et al., ⁴⁴ a sensitivity analysis was conducted to evaluate the stability of the current results with different imputed r values of 0.25 and 0.75. Findings of sensitivity analysis revealed that no significant differences in SMD, z-scores, and p values were observed in all different imputed r conditions (i.e., 0.25, 0.50, and 0.75) for all the outcomes except for variations in I ² (Table 3). The MCT+, VRT, SST, CBT, and CRT were also found to be the most effective treatment methods to manage PS, and SST, MCT+, CBT, and CRT produced significant treatment effects compared with TAU in the management of NS. Similar to the primary network analysis, VRT was the only intervention method found compelling in the management of hallucinations, and CBT and MCT+ were found effective in the management of delusions. The majority of interventions were ranked similarly based on their SUCRA scores, and the findings with different imputed correlations were aligned with those of the primary analysis.

Table 3.

Sensitivity Analysis of Imputed Pre-post Correlations (r = 0.25 and r = 0.75) for Evaluating the Efficacy of Psychological Interventions on Symptom Management in Schizophrenia.

Therapy	Imputed Correlation	Positive Symptoms		Negative Symptoms		Hallucinations		Delusions
		SMD [95% CI]	Z–score	SMD [95% CI]	Z–score	SMD [95% CI]	Z–score	SMD [95% CI]	Z–score
ACT	0.25	–0.09 [–0.67; 0.49]	–0.30	0.27 [–0.41; 0.95]	0.78	××		××
	0.75	–0.09 [–0.63; 0.45]	–0.32	0.27 [–0.36; 0.90]	0.84	××		××
CBT	0.25	–0.33 [–0.48; –0.18]	–4.40**	–0.27 [–0.43; –0.11]	–3.32**	–0.20 [–0.45; 0.04]	–1.62	–0.33 [–0.55; –0.12]	–3.01**
	0.75	–0.32 [–0.45; –0.18]	–4.67**	–0.28 [–0.42; –0.13]	–3.82**	–0.17 [–0.41; 0.07]	–1.42	–0.34 [–0.55; 0.13]	–3.14**
CRT	0.25	–0.14 [–0.26; –0.03]	–2.40*	–0.31 [–0.45; –0.18]	–4.58**	–0.17 [–0.91; 0.57]	–0.44	0.15 [–0.35; 0.66]	0.60
	0.75	–0.13 [–0.24; –0.02]	–2.36*	–0.30 [–0.42; –0.18]	–4.97**	–0.16 [–0.92; 0.59]	–0.42	0.15 [–0.37; 0.67]	0.58
CT	0.25	–0.10 [–0.62; 0.43]	–0.36	–0.23 [–0.82; 0.35]	–0.78	××		××
	0.75	–0.15 [–0.57; 0.28]	–0.68	–0.32 [–0.79; 0.16]	–1.31	××		××
MCT+	0.25	–0.46 [–0.64; –0.28]	–4.96**	–0.40 [–0.67; –0.13]	–2.92**	–0.20 [–0.71; 0.31]	–0.76	–0.34 [–0.63; –0.05]	–2.27*
	0.75	–0.45 [–0.61; –0.29]	–5.54**	–0.38 [–0.61; –0.15]	–3.26**	–0.22 [–0.71; 0.28]	–0.85	–0.37 [–0.66; –0.08]	–2.49**
SST	0.25	–0.34 [–0.58; –0.10]	–2.76**	–0.45 [–0.76; –0.13]	–2.77**	0.18 [–0.69; 1.06]	0.42	–0.06 [–0.75; 0.63]	–0.18
	0.75	–0.28 [–0.51; –0.05]	–2.39*	–0.40 [–0.69; –0.11]	–2.72**	0.17 [–0.76; 1.10]	0.35	–0.09 [–0.85; 0.67]	–0.24
ST	0.25	–0.14 [–0.35; 0.06]	–1.39	–0.06 [–0.33; 0.20]	–0.47	0.02 [–0.34; 0.38]	0.11	–0.29 [–0.63; 0.05]	–1.69
	0.75	–0.08 [–0.27; 0.11]	–0.85	–0.05 [–0.29; 0.19]	–0.39	–0.01 [–0.38; 0.35]	–0.08	–0.30 [–0.66; 0.06]	–1.62
VRT	0.25	–0.61 [–1.19; –0.02]	–2.04*	–0.30 [–0.82; 0.23]	–1.11	–0.79 [–1.31; –0.27]	–2.99**	–0.39 [–1.07; 0.29]	–1.11
	0.75	–0.58 [–1.07; –0.10]	–2.36 *	–0.27 [–0.74; 0.19]	–1.15	–0.82 [–1.34; –0.30]	–3.08*	–0.39 [–1.17; 0.39]	–0.98
Heterogeneity	0.25	Q = 89.39, df = 70, tau = 0.15; I2 = 21.7%		Q = 117.48, df = 72, tau = 0.24; I2 = 38.7%		Q = 34.77, df = 19, tau = 0.28; I2 = 45.3%		Q = 35.38, df = 23, tau = 0.23; I2 = 35%
	0.75	Q = 191.16, df = 70, tau = 0.24; I2 = 63.4%		Q = 250.94, df = 72, tau = 0.29; I2 = 71.3%		Q = 92.41, df = 19, tau = 0.37; I2 = 79.4%		Q = 94.40, df = 23, tau = 0.33; I2 = 75.6%

SMD: Standardized Mean Difference, CI: Confidence Interval, ACT: Acceptance and Commitment Therapy, CBT: Cognitive Behavior Therapy, CRT: Cognitive Remediation Therapy, CT: Cognitive Therapy, MCT+: Meta-cognitive training, ST: Supportive Therapy, SST: Social Skills Training, VRT: Virtual Reality Therapy, TAU: Treatment-as-Usual.

*Significant at 0.05 level, ** Significant at 0.01 level.

“××” shows no studies have reported data on that particular outcome measure and treatment.

Meta-regression Analysis

Meta-regression analyses were performed across various therapeutic interventions to assess whether therapy duration and session frequency affected therapeutic outcomes (Supplementary Table S2). These analyses were prompted by the substantial heterogeneity observed in the NMA (I² = 0.00%–51.69%). None of the moderator tests reached statistical significance, indicating that therapy duration and the number of sessions did not significantly impact treatment outcomes except for pairwise comparisons between CBT versus TAU. The R² values for all models ranged between 0.00% and 90.45%, further supporting the limited influence of these variables as moderators. Meta-regression analyses for direct pairwise comparisons involving ST versus TAU, VRT versus TAU, SST versus TAU, and ACT versus TAU were not feasible to be conducted due to a limited number of studies (fewer than three) for each comparison. For the efficacy of pairwise comparison between CBT versus TAU, 7.12% of heterogeneity was accounted for by duration of sessions and number of sessions. Therefore, these further emerged as significant moderators in determining the effectiveness of CBT.

Discussion

The present SLR and NMA assessed the efficacy of nine psychological interventions in the management of different symptoms and overall functioning of 7,053 patients with schizophrenia across 100 RCTs published over the past two decades. The present study derives its uniqueness by focusing on the assessment of symptom-specific efficacy of therapeutic interventions. Therefore, pairwise comparisons were conducted separately for overall PS, NS, hallucinations, and delusions of schizophrenia patients.

The findings indicate that MCT+ is effective in the management of overall PS, NS, and delusions in schizophrenia patients. However, MCT+ did not emerge as an effective process in managing hallucinations. As MCT+ is an eclectic approach comprised of three therapeutic approaches, procedures, namely, CRT, CBT, and psychoeducation, to treat the patients.^45,46 MCT+ focuses on the different components of metacognition, including belief flexibility, source-monitoring, self-reflectivity and perspective-taking, to correct cognitive biases,^47–49 but operates at the level of the object rather than the level of the meta-cognitive processing typically addressed by MCT+. ⁵⁰ As patients with schizophrenia have severe source-monitoring deficits, ⁵¹ which leads to a misperception of external reality as intrinsic, which is also a known cause of hallucination. Because MCT+ does not address meta-cognitive processes, this treatment approach may not be effective in treating hallucinations. The MCT+ also aims to modify and change cognitive distortions by promoting belief flexibility, thereby reducing delusions. ⁴⁸ MCT+ also increases self-awareness and goal-directed behavior, which helps to reduce NS. ⁵² Improved metacognition promotes better perspective-taking, which in turn enhances motivation, emotional involvement, and social functioning. However, its effects on NS are modest and may require additional intervention. ⁵²

It is noteworthy that self-reflectivity has been found to predict current and future NS and to act as a mediator between cognitive impairment and reduced emotional expression. ⁵³ Because CRT and CBT are an integral part of MCT+, it leads to self-reflectivity and reduces cognitive dysfunctions, which in turn leads to an improvement in NS in schizophrenia. ²⁰

Furthermore, psychoeducation helps the patients to learn more scientific facts about the disorder, leading to considerable progress in patients’ social adjustment and a general reduction in psychopathology. ⁵⁴ Furthermore, the present findings are in concordance with existing empirical evidence and meta-analytical findings wherein MCT+ is effective in the management of PS^22,55 as well as NS of patients with psychotic disorders.^22,54

The VRT is an emerging form of therapy in which an artificial environment is created. The patients are instructed, guided, and monitored through artificially constructed social situations to control hallucinatory voices and challenge the beliefs of the patients.^56,57 Avatar therapy is a form of VRT in which management strategies are implemented through the avatar of patients. ⁵⁸ The voices and implemented settings of these Avatars are designed by the perceptual and thought disturbances of the patients in a controlled and graded manner. ⁵⁷ In the present study, VRT is found to be effective in the management of hallucinations and overall PS. However, this therapeutic method is not found to be suitable for the management of NS and delusions in patients. Our findings align with outcomes reported in prior meta-analyses wherein almost all VRTs have been found effective in the management of PS such as paranoia, auditory hallucination, and delusions of schizophrenia patients, ⁵⁹ but significant dropouts in the management of NS. ¹⁶ Lack of motivation to participate, ⁶⁰ and expressive and cognitive difficulties, ⁶¹ are predominantly found among the patients having NS, which may further restrict them in active involvement in various therapeutic tasks that may be potential reasons behind the ineffectiveness of VRT in the management of NS. Although VRT offers immersive and experiential settings for challenging delusional beliefs more effectively than standard CBT, it appears less suited for targeting NS, such as anhedonia, avolition, and social amotivation. This may be due to the lack of gamified or tailored tasks specifically designed to engage and remediate these core motivational and affective impairments. ⁵⁹ Further, Novo et al., ⁶² highlighted a key concern that lies in the traditional assessment methods (e.g., PANSS and SANS) for measuring NS, which may inadequately capture the nuanced deficits of NS, particularly in virtual social contexts; therefore, VRT has shown limited effectiveness in managing NS in schizophrenia.

Results indicate that CBT was also effective in the management of PS, NS, and delusions in schizophrenia patients. CBT is commonly used with psychotic patients to alleviate the discomfort associated with hearing voices and to diminish the sense of control attributed to the voices by addressing core beliefs regarding their power and omnipotence. ⁶³ Patients also learn to identify, gain control, and challenge beliefs about the voices. ⁶⁴ CBT for psychosis addresses the NS by reducing pressure and immediate expectations, identifying the motivating factors, which subsequently help in meaningful engagement in activities. ⁶⁴ Findings from this study corroborate earlier studies demonstrating the efficacy of CBT in the treatment of PS.^8,54,65 and NS of schizophrenia patients. ⁸ CBT targets maladaptive beliefs and cognitive distortions associated with psychotic experiences, which may lead to a reduction in distress associated with psychotic symptoms. ⁶⁴ It also facilitates emotional regulation and symptom management through techniques such as cognitive restructuring and behavioral experiments, which help reduce the intensity of delusions and hallucinations in schizophrenia patients.^64,66,67 Further, by promoting reality testing and balanced thinking, CBT enables individuals to gain insight into their experiences and cope more adaptively. ⁶⁷ Furthermore, CBT emphasizes engagement in meaningful activities, which supports improvements in social and emotional functioning and contributes to the reduction of NS. ⁶⁸ These combined approaches make CBT a comprehensive and effective intervention for improving the overall functioning of schizophrenia patients.

In SST, symptoms and functioning of the patients are targeted by using the concepts of social learning and interpersonal skills. ⁶⁹ The present NMA also indicated SST to be effective in the management of PS and NS of schizophrenia patients. Similar findings have also been reported by Turner et al. ⁷⁰

It is a well-established fact that impairments in attentional processes, memory functioning, and higher order functioning, etc., are positively associated with functional impairment and overall NS of schizophrenia patients. ⁷¹ Therefore, it would be reasonable to infer that improvement in cognitive functioning would further lead to improvement in the symptoms and consequent overall functioning of patients. Additionally, CRT is an active therapist-based behavioral training that includes different cognitive exercises targeted to improve the cognitive functioning of schizophrenia patients in a structured way. ⁷² In the present study, CRT was found to be effective in treating generally PS and NS of schizophrenia patients. However, this therapeutic method fails to show effectiveness in treating hallucinations and delusions. Our results are in agreement with earlier studies, which also found that the CRT is an effective therapeutic procedure for managing PS ²¹ and NS^21,73,74 in schizophrenia patients. However, the present finding is contradictory to the meta-analytical findings of Lejeune et al. ⁷⁴ who have reported that CRT did not produce significant effectiveness on the recovery of PS of schizophrenia patients. Several factors contribute to these contradictory findings. In the meta-analytical study by Lejeune et al. ⁷⁴ a total of 73 CRT-based studies were included, five of which used a retraining module developed by the authors themselves, and the empirical validity of those modules was not predetermined. Further, the majority of the included studies employed computer-based CRT modules, where patients had limited opportunities to interact with the therapist. Additionally, in most of the included studies, participants in the control group were engaged in other adjunct therapeutic procedures.

ST is often taken as an active control group in RCTs; however, it has not proven beneficial for any symptom domains in schizophrenia patients when compared to TAU. The ACT is considered a third-generation form of CBT used to reduce the stress and worry associated with psychotic symptoms and uses a range of strategies targeted to experiential avoidance. ⁷⁵ Although a previous SLR by Yildiz et al. ⁷⁶ has reported ACT as a promising technique to manage psychosis. In the present NMA, we did not find ACT as a significantly effective treatment method in the management of various symptoms of schizophrenia. Similar findings have also been reported previously by Brown et al., ²⁴ who have also reported that ACT contributed significantly less in managing the symptoms as compared to other forms of psychotherapy. Although ACT was effective in improving acceptance of symptoms and reducing avoidance behaviors, it did not lead to substantial variations in the frequency or severity of hallucinations and delusional thinking. ⁷⁷ As the implementation of ACT mainly focused on valued living of the patients without focusing on symptoms, ⁷⁸ and it promotes flexibility and awareness of thoughts in the patients, which results in a reduction of symptoms. ⁷⁹ ACT is mainly used to reduce distress caused by psychotic symptoms by promoting the process of defusion, where the patients are asked to conceptualize and observe themselves by separating voices from thoughts.^78,79 Metaphors are used to understand fusion versus defusion to reduce the feeling of being enslaved by voices. Since a larger proportion of schizophrenia patients have significant cognitive impairment, and it is difficult to use these metaphors effectively, this may also be a potential reason for the ineffectiveness of this approach in treating PS and NS in schizophrenia patients.

The NMA findings further reveal considerable heterogeneity and no evidence of publication bias among the included studies. This could be attributed to the inclusion of a hundred articles using different assessment tools across different regions over the last 20 years. The present NMA is one of the first attempts to analyze the symptom-specific efficacy of different psychotherapy interventions on schizophrenia patients. Although the study benefits from incorporating a broad spectrum of RCTs across six major journal hosting sites, the current NMA has certain limitations. For instance, it measures effectiveness based solely on data from the pre- and post-assessment phases of the intervention. The only way to ensure the longevity of an intervention program is to compare post-assessment and follow-up results, which is not performed in the present NMA.

Furthermore, in all RCTs, patients received antipsychotic medications in addition to various psychological intervention methods. Almost all studies did not provide detailed descriptions of the type and dosage of the antipsychotics used to treat patients. Therefore, the potential role of antipsychotic medication type and dosage in the recovery of patients was not seen in this study, which could be a prime limitation of this study. Although it was hypothesized that participants in both the treatment and control groups would receive care from the same medical team, randomization ensured comparable conditions of pharmacological treatment across the study arms. Finally, different treatment approaches like relaxation, psychoeducation, and befriending are considered active control conditions. They are arbitrarily combined into the TAU group because adding more arms to an NMA can cause loop inconsistencies in the network that further violate transitivity assumptions for both direct and indirect comparisons between study designs. Therefore, this could be a possible limitation of the present NMA.

Conclusions

Schizophrenia disorder falls on a spectrum because the symptoms are varied, and several etiological factors are responsible for its development and maintenance. Therefore, comparing and understanding the effectiveness of different therapeutic interventions helps to make an informed clinical decision for the integrated biopsychosocial management of schizophrenia. This study represents one of the initial efforts to perform a symptom-specific NMA assessing the efficacy of diverse psychological therapies in managing schizophrenia.

Findings suggest that MCT+, SST, CRT, and CBT are found to be effective in the management of PS and NS in schizophrenia patients. VRT was found to be effective for the management of hallucinations and overall, PS. Further, MCT+ and CBT were found to be effective treatment measures in the management of delusions. Lastly, these therapeutic procedures are recommended to be implemented as an adjunct to conventional pharmacotherapy. The study also highlights the need to assess specific other components, such as a follow-up assessment for more conclusive findings.