Longitudinal Association Between Generalized Anxiety Disorder and Cognitive Change: A Meta-Analysis
Article information
Abstract
Objective
This study aimed to systematically evaluate the longitudinal relationship between generalized anxiety disorder (GAD) and cognitive function changes. Specifically, we examined whether GAD is an independent risk factor for sustained cognitive decline and explored the potential effects of cognitive behavioral therapy (CBT) on cognitive and emotional outcomes.
Methods
We conducted a systematic search of PubMed, Embase, Web of Science, and the Cochrane Library for eligible longitudinal studies and randomized controlled trials published up to January 20, 2025. Meta-analyses were performed using RevMan 5.4.1. Effect sizes were estimated using mean differences (MD) and 95% confidence intervals. Heterogeneity was assessed with the I2 statistic. Sensitivity analyses and publication bias assessments were also conducted.
Results
A total of 13 studies were included. The results showed no significant differences between GAD patients and controls on memory and attention subscores of the dementia rating scale (p>0.05). Total error scores also showed no significant difference. Subgroup analyses indicated that GAD patients receiving CBT demonstrated greater improvement in clinical symptoms than those receiving non-cognitive interventions (MD=-4.39, p=0.0004). Beck Depression Inventory score differences were statistically significant, while Penn State Worry Questionnaire score differences were not.
Conclusion
This meta-analysis suggests that while GAD does not show consistent evidence of sustained cognitive decline, CBT may be associated with improvement in emotional symptoms. By integrating longitudinal data with treatment outcomes, this study offers preliminary insights into the potential indirect cognitive benefits of psychotherapy in GAD.
INTRODUCTION
Generalized anxiety disorder (GAD) is a chronic psychiatric condition characterized by excessive and persistent worry and tension, with a lifetime prevalence of approximately 5%–6% in the general population [1-5]. GAD is frequently associated with cognitive symptoms such as impaired concentration, memory decline, and difficulty in decision-making, which can significantly affect patients’ quality of life, social functioning, and treatment adherence [6-8]. Although GAD is primarily categorized as an anxiety disorder, its cognitive impairments have received growing attention in both clinical and research settings.
In recent years, research on the relationship between GAD and cognitive function has increased, particularly regarding domains such as attention, executive function, and memory. Several studies have shown that GAD patients perform poorly on various neuropsychological tests, possibly due to heightened vigilance, attentional bias, and emotional interference [9,10]. Cross-sectional studies have reported significant deficits in working memory and executive functioning among individuals with GAD [11], while functional magnetic resonance imaging studies have revealed abnormal activity in cognitive-related brain regions such as the prefrontal cortex and hippocampus [12-14]. However, most existing studies adopt cross-sectional designs, which limit the ability to determine whether cognitive impairments are state-related or indicative of a progressive decline over time.
Longitudinal studies aim to explore the dynamic association between GAD and cognitive changes from a causal perspective, but the findings remain inconclusive [11,15,16]. Some prospective studies suggest that GAD may be an independent risk factor for cognitive decline, while others have not found statistically significant associations. These discrepancies may stem from heterogeneity in sample characteristics, inconsistent cognitive assessment tools, and variations in follow-up duration. Moreover, there is limited evidence on whether interventions such as cognitive behavioral therapy (CBT) can impact cognitive outcomes in GAD patients.
Therefore, this meta-analysis seeks to integrate current longitudinal research to clarify whether individuals with GAD are at risk for sustained cognitive decline. We also evaluate whether CBT, beyond improving anxiety symptoms, may offer cognitive benefits. This study applies rigorous inclusion criteria, quality assessment, and analytical methods including subgroup and sensitivity analyses, aiming to provide evidence-based insights for clinical intervention and long-term cognitive health management.
METHODS
Literature search
This systematic review and meta-analysis was registered with the International Prospective Register of Systematic Reviews (PROSPERO; registration number: CRD420251061423).
We conducted a comprehensive computer-based search of several authoritative medical and biomedical databases, including PubMed, Embase, Web of Science, and the Cochrane Library. The search was limited to studies published in English, with the data updated to January 20, 2025. The English search terms included: “generalized anxiety disorder,” “GAD,” “cognitive function,” “cognitive impairment,” “longitudinal,” “prospective,” “follow-up,” and “cognitive decline.” The detailed search strategy is provided in Supplementary Table 1.
Inclusion and exclusion criteria
Inclusion criteria: 1) Participants were clinically diagnosed with GAD. 2) The study was designed as a randomized controlled trial (RCT), cross-sectional, or cohort study. 3) The study reported on the relationship between GAD and cognitive impairment.
Exclusion criteria: 1) Non-original research, such as reviews or conference abstracts. 2) Studies with low quality, duplication, incomplete or unclear data. 3) Studies not involving the relationship between GAD and cognitive impairment.
Study selection and data extraction
Two independent researchers performed the initial screening based on titles and abstracts to exclude irrelevant or ineligible studies. Full texts of potentially eligible articles were then reviewed to ensure inclusion criteria were fully met. Discrepancies were resolved through discussion or arbitration by a third reviewer. Data extraction was independently conducted by both researchers and included basic study characteristics, participant demographics, and outcome indicators. All extracted data were cross-verified to ensure accuracy and consistency, with disagreements resolved through discussion or expert consultation.
Quality assessment
Two independent reviewers assessed the risk of bias for the included studies using the Cochrane Risk of Bias Tool for randomized trials, covering five domains: randomization process, deviations from intended interventions, missing outcome data, measurement of the outcome, and selection of the reported results. Each domain was rated as “low risk,” “unclear,” or “high risk.” Disagreements were discussed or referred to a third reviewer.
Statistical analysis
Literature management was performed using NoteExpress 3.2, and data extraction was done using Excel 2003. Meta-analysis was conducted with RevMan 5.4.1 software (Cochrane Collaboration). Heterogeneity was evaluated using the Q test (p-value) and I2 statistics. If p>0.10 or I2 ≤50%, a fixed-effects model (FEM) was used; otherwise, a random-effects model (REM) was applied. Effect sizes were expressed as odds ratios (OR) or mean differences (MD) with 95% confidence intervals (CI), and results were visualized using forest plots. Sensitivity analyses were conducted to test result stability, and publication bias was assessed with funnel plots. A two-sided α level of 0.05 was used for statistical significance.
RESULTS
Search results
A total of 1,005 records related to GAD and cognitive impairment were initially retrieved from the English-language databases. After removing 61 duplicates, 944 articles remained. Of these, 235 were excluded based on titles and abstracts due to obvious irrelevance. Abstract screening removed another 482 articles, leaving 227 for full-text review. After a detailed evaluation, 214 studies were excluded for various reasons. Ultimately, 13 studies were included in the meta-analysis. The flowchart of the screening process is shown in Figure 1, and the basic characteristics of the included studies are listed in Table 1.
Flowchart of study selection.
Basic characteristics of included studies
Quality assessment of included studies
Quality assessment revealed that most included studies adhered strictly to randomized controlled trial (RCT) design principles. Randomization procedures were clearly described in the majority of studies, ensuring comparability between intervention and control groups at baseline. However, double-blind designs were applied in only a few studies. Selective reporting was generally unclear, and five studies had unclear risk of other biases, while eight showed evidence of additional biases (Figures 2 and 3). Overall, the included studies met our methodological standards.
Risk of bias summary.
Risk of bias graph.
Meta-analysis results
Dementia rating scale–memory score
Two included studies reported dementia rating scale–memory scores for both GAD patients and controls. Heterogeneity analysis showed I²=37% and p=0.21, indicating no significant heterogeneity between studies; therefore, a FEM was used. The meta-analysis revealed no statistically significant difference between GAD patients and controls in memory scores (MD=-0.63, 95% CI [-1.44 to 0.18], p=0.13) (Figure 4).
Forest plot comparing dementia rating scale–memory scores between the two groups. GAD, generalized anxiety disorder; SD, standard deviation; CI, confidence interval.
Dementia rating scale–attention score
Two studies reported dementia rating scale–attention scores. Heterogeneity analysis showed I²=0% and p=0.36, indicating no heterogeneity; thus, a fixed-effects model was applied. The results demonstrated no significant difference between the two groups in attention scores (MD=-0.06, 95% CI [-0.83 to 0.70], p=0.87) (Figure 5).
Forest plot comparing dementia rating scale–attention scores between the two groups. GAD, generalized anxiety disorder; SD, standard deviation; CI, confidence interval.
Total error score
Two studies reported total error scores. Heterogeneity analysis showed substantial heterogeneity (I²=84%, p=0.01), and a REM was used. The pooled analysis indicated no statistically significant difference in total error scores between GAD patients and controls (MD=-0.13, 95% CI [-9.29 to 9.03], p=0.98) (Figure 6).
Forest plot comparing total error scores between the two groups. GAD, generalized anxiety disorder; SD, standard deviation; CI, confidence interval.
Clinical effectiveness
To further explore the longitudinal association between GAD and cognitive changes, patients receiving cognitive therapy were designated as the experimental group, while those receiving non-cognitive therapy were used as controls. Five studies reported clinical outcomes. Heterogeneity analysis showed I²=58% and p=0.05, suggesting moderate heterogeneity; thus, a REM was adopted. Results showed that patients receiving cognitive therapy had significantly better clinical outcomes than controls (MD=-4.39, 95% CI [-6.82 to -1.95], p=0.0004) (Figure 7).
Forest plot comparing clinical effectiveness between cognitive and non-cognitive therapy groups. CBT, cognitive behavioral therapy; SD, standard deviation; CI, confidence interval.
Penn State Worry Questionnaire score
To examine the association between GAD and cognitive changes, three studies reported Penn State Worry Questionnaire (PSWQ) scores. Heterogeneity was high (I²=90%, p<0.0001), and a REM was applied. The meta-analysis showed no significant difference in PSWQ scores between the two groups (MD=-1.82, 95% CI [-12.42 to 8.78], p=0.74) (Figure 8).
Forest plot comparing Penn State Worry Questionnaire scores between the two groups. CBT, cognitive behavioral therapy; SD, standard deviation; CI, confidence interval.
Beck Depression Inventory score
To further explore the link between GAD and cognitive changes, two studies reported Beck Depression Inventory (BDI) scores. Heterogeneity analysis showed I²=0%, p=0.77, indicating no heterogeneity; therefore, a FEM was used. The results showed a statistically significant difference in BDI scores between the groups (MD=-2.30, 95% CI [-4.03 to -0.57], p=0.009) (Figure 9).
Forest plot comparing Beck Depression Inventory scores between the two groups. CBT, cognitive behavioral therapy; SD, standard deviation; CI, confidence interval.
Publication bias
Publication bias was assessed for all outcome indicators. Funnel plots appeared largely symmetrical, suggesting the absence of significant publication bias (Figure 10).
Funnel plots for publication bias assessment. A: Dementia rating scale–memory score. B: Dementia rating scale–attention score. C: Total error score. D: Clinical effectiveness. E: Penn State Worry Questionnaire. F: Beck Depression Inventory. SE, standard error; MD, mean difference.
Sensitivity analysis
To evaluate the robustness and reliability of the meta-analysis results, we conducted sensitivity analyses. By removing one study at a time and recalculating the pooled effect size, we assessed each study’s impact on the overall results. The sensitivity analysis showed that excluding any single study did not significantly alter the direction of the effect size, and the CIs remained stable. This indicates that the model was robust (Figure 11).
Forest plot of sensitivity analysis. CBT, cognitive behavioral therapy; SD, standard deviation; CI, confidence interval.
DISCUSSION
This meta-analysis included 13 longitudinal studies to systematically assess the association between GAD and changes in cognitive function, while also exploring the potential impact of CBT on relevant outcomes. The findings can be summarized as follows. First, GAD patients did not differ significantly from healthy controls in key cognitive domains such as memory (dementia rating scale–memory) and attention (dementia rating scale–attention). Second, although the total error score exhibited substantial heterogeneity, the pooled effect showed no significant difference between groups. The observed heterogeneity in outcomes such as the total error score (I²=84%) may be attributed to differences in cognitive assessment tools, population age ranges, and intervention durations across studies. Third, CBT was associated with a significant reduction in emotional symptoms in GAD patients, particularly in terms of overall clinical efficacy and depression scores (BDI). Fourth, results from sensitivity analyses and funnel plots indicated that the overall conclusions of this study are robust and not substantially affected by publication bias. Taken together, although GAD is widely associated with subjective complaints such as attention deficits and memory impairment, current longitudinal empirical evidence is insufficient to support the hypothesis that GAD leads to sustained cognitive decline over time. On the other hand, cognitive behavioral interventions remain clinically valuable in improving anxiety and depressive symptoms.
Our results differ from some cross-sectional studies. Prior research has shown that GAD patients often exhibit marked deficits in working memory, executive functioning, and attention control; however, these studies are largely based on single-timepoint assessments and do not reflect temporal trends in cognitive changes [17,18], Notably, we observed that CBT appears effective in reducing comorbid depressive symptoms in GAD patients, a finding that aligns with previous literature [19,20]. Domínguez-Pérez et al. [21] also reviewed the potential neurobiological mechanisms underlying CBT’s effectiveness in treating anxiety disorders. However, we did not find evidence that CBT directly improves objective cognitive indicators such as memory or attention, suggesting that any cognitive benefits of CBT, if present, may be secondary and mediated by emotional symptom improvement rather than direct enhancement of cognitive function.
To our knowledge, this study is among the few meta-analyses to explore the longitudinal relationship between GAD and cognitive function. Compared to cross-sectional studies, our analysis integrates long-term follow-up data, enhancing the scientific basis for causal inference. Moreover, we are the first to analyze clinical efficacy, emotional outcomes, and cognitive indices side-by-side, in an effort to uncover the potential indirect impact of emotional interventions on cognitive outcomes. This may offer a novel perspective for the long-term management of GAD and lays an evidence-based foundation for the development of personalized intervention strategies.
From a clinical standpoint, our findings support the view that systematic follow-up to monitor cognitive changes in GAD patients may not be an urgent need. In contrast, emotional management and psychotherapeutic approaches should remain the primary intervention focus. Clinicians evaluating the functional status of GAD patients should pay particular attention to issues such as depressive comorbidity, social withdrawal, and sleep disturbances, rather than overinterpreting cognitive decline.
Despite its novelty, this study has several limitations. First, the limited number of included studies and the variability in both cognitive assessment instruments and follow-up schedules may have reduced the overall comparability and statistical power of our findings. Second, cognitive outcomes were assessed with diverse instruments and at non-uniform followup time points, which may limit harmonization and partially attenuate the observed effects. Third, we were unable to conduct an individual participant data meta-analysis, which limited our ability to perform subgroup analyses (e.g., by age, gender, or education level) to identify moderators of cognitive outcomes. Finally, because covariate adjustment varied among the included studies, residual confounding from factors such as depressive symptom burden, psychotropic medication use, education level, or medical comorbidities could not be fully excluded.
Future research should prioritize the use of standardized cognitive assessment tools (e.g., Montreal Cognitive Assessment, Trail Making Test, n-back tasks), extend follow-up periods, and include various anxiety disorder types as comparators to better delineate GAD’s cognitive profile across subdomains. Additionally, further investigation is needed into the mechanisms and pathways through which interventions—such as pharmacotherapy, mindfulness, or meditation—may influence cognition. Exploring the potential mediating role of emotional improvement in cognitive enhancement could lead to more precise and effective interventions for functional impairment in GAD patients.
Conclusion
This meta-analysis provides preliminary insights into the longitudinal association between GAD and cognitive change. Although the current evidence does not show consistent support for sustained cognitive decline in individuals with GAD, our findings suggest that CBT may be associated with improvements in emotional symptoms such as anxiety and depression. These observations should be interpreted cautiously given the limited sample size, methodological heterogeneity, and potential residual confounding. Overall, the results offer a suggestive basis for future longitudinal and interventional studies exploring the indirect cognitive implications of CBT in GAD.
Supplementary Materials
The Supplement is available with this article at https://doi.org/10.30773/pi.2025.0299.
Search strategy
Notes
Availability of Data and Material
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.
Conflicts of Interest
The authors have no potential conflicts of interest to disclose.
Author Contributions
Conceptualization: Jie Li, Nan Jiang, Zhuozhao Zheng. Data curation: Jie Li, Zhuozhao Zheng. Formal analysis: Jie Li, Nan jiang, Yujie Wang. Investigation: Jie Li, Benqi Zhao. Methodology: Jie Li, Leiyu Shi, Nan Jiang, Yujie Wang. Project administration: Leiyu Shi, Zhuozhao Zheng. Resources: Leiyu Shi, Nan Jiang, Zhuozhao Zheng. Software: Jie Li, Yujie Wang. Supervision: Zhuozhao Zheng. Validation: Leiyu Shi. Visualization: Nan Jiang. Writing—original draft: Jie Li. Writing—review & editing: Jie Li, Benqi Zhao, Zhuozhao Zheng.
Funding Statement
None
Acknowledgments
None