Malaysia is rapidly becoming an aging society. The proportion of the population aged 60 years and older is projected to increase from 11.6% in 2024 to 17.3% by 2040, surpassing the 15% threshold commonly used to define an aging population [1]. This demographic transition reflects a global trend, as many high- and middle-income countries face similar shifts toward older age structures [2]. Based on these projections, Malaysia will officially be classified as an aging society by 2040.

Globally, the prevalence of mild cognitive impairment (MCI) has been estimated at 19.7% among individuals aged 50 years and older [3]. In Malaysia, previous studies have reported even higher estimates, with approximately 21.1% of older adults affected [4,5]. More recently, a study conducted in Kuantan, Malaysia, found the prevalence of MCI among adults aged 60 years and older to be 18.7% [6]. Despite these substantial prevalence rates, most individuals remain undiagnosed until later stages of cognitive decline, with only 8% of those in the early stage of MCI receiving a formal diagnosis [7]. These findings have important implications for independence, health care use, and caregiver burden, contributing to the rising societal impact of dementia [8].

Lifestyle-based interventions are increasingly recognized as promising strategies for delaying cognitive decline, particularly given the limited effectiveness of pharmacological treatments [9]. Aerobic exercise has been associated with improvements in brain structure and function through mechanisms such as enhanced cerebral blood flow and improved regulatory effects on the brain. Sustained aerobic exercise increases heart rate, boosts cerebral perfusion, and promotes glucose metabolism in the brain [10]. Among the various modalities, Zumba represents a culturally engaging form of exercise that combines rhythm, coordination, and social interaction. Participation in Zumba classes has been shown to improve quality of life and visuospatial memory among older adults [11,12]. Zumba has been proposed as a feasible public health strategy due to its enjoyable and socially engaging nature [13], and its increasing popularity within Malaysian community settings further underscores its potential for broader implementation.

Parallel to this, technology-driven cognitive training programs, including video game–based interventions, have emerged as scalable and accessible tools for older adults. Evidence from computerized and video game–based cognitive training indicates that overall effects on cognitive performance in healthy older adults are positive but small. Reported benefits vary substantially across cognitive outcomes and are strongly influenced by intervention design characteristics, including delivery format and training intensity [14]. Improvements in specific domains are inconsistent, and transfer to global cognition remains variable, underscoring the importance of task design and intervention structure. With increasing digital literacy among urban-dwelling Malaysian older adults, including those in the Klang Valley, such approaches are becoming increasingly feasible.

Emerging research suggests that combining physical and cognitive interventions may have synergistic effects, as aerobic exercise provides a neurophysiological foundation for plasticity, while cognitive training directly stimulates neural networks [15]. However, studies directly comparing these approaches, particularly in Asian populations, remain limited.

The present semirandomized controlled trial was designed to address these gaps in the Malaysian context. Older adults living in the Klang Valley were allocated into physical activity (Zumba), cognitive stimulation (video game), combined intervention, or control groups within community settings. This study evaluated the effects of the interventions on global cognitive function using the Montreal Cognitive Assessment (MoCA). We also explored changes across cognitive subdomains, examined whether attendance influenced outcomes, and identified participants who demonstrated meaningful improvements, defined as achieving the minimal clinically important difference of a ≥2-point increase in MoCA scores [16]. Thus, this study aimed to evaluate the effects of Zumba, video game–based cognitive stimulation, and combined interventions on global cognitive function among older adults in Klang Valley, Malaysia, and to examine the influence of adherence on cognitive outcomes, with the hypothesis that the combined intervention would produce the greatest improvements.

This semirandomized trial found that physical (Zumba), cognitive (video game), and combined interventions were associated with within-group improvements in global cognitive function among older adults in Klang Valley, Malaysia. The combined intervention demonstrated the largest magnitude of improvement, although between-group differences were not statistically significant. Patterns of improvement suggested that adherence may play an important role in achieving meaningful cognitive benefits.

Cognitive impairment and age-related decline substantially affect daily functioning and independence in older adults. With rapid global aging and increasing dementia prevalence, effective preventive strategies are urgently needed [15]. Pharmacological approaches show limited success and are often accompanied by adverse effects [24-26], prompting growing interest in nonpharmacological interventions targeting cognitive frailty [27-31]. Despite this progress, the comparative effectiveness of different intervention modalities and their combinations remains uncertain [9]. Our study addresses this gap by examining physical, cognitive, and combined interventions on global and domain-specific MoCA performance, providing insights into strategies to preserve cognitive health in later life.

At baseline, there were no significant differences across groups in age (P=.96), sex (P=.23), MoCA scores before intervention (P=.77), or hypertension prevalence (P=.64), indicating a well-balanced distribution among participants. Recently, our research team examined the association between C-reactive protein (CRP) levels and various demographic factors. The analysis demonstrated that CRP concentrations were significantly correlated with gender, ethnicity, and BMI (P<.05), but not with age or economic status [18]. CRP is an inflammatory biomarker that plays a pivotal role in the pathophysiology of age-related conditions, including cognitive decline.

However, a significant between-group difference was observed in educational attainment (P<.001), which is a well-established determinant of cognitive reserve and MoCA performance. The cognitive group had the highest proportion of participants with a bachelor’s degree, while the combination group included a relatively higher proportion with postgraduate education. This imbalance may have influenced both baseline cognitive performance and responsiveness to intervention, thereby complicating causal interpretation of between-group comparisons.

Although educational attainment differed across groups at baseline, the modest sample size and small subgroup numbers limited the feasibility of conducting covariate-adjusted or stratified analyses to account for potential education effects. Consequently, the observed intervention effects may have been influenced by baseline differences in cognitive reserve associated with education. Nevertheless, prior research suggests that cognitively engaging and socially meaningful interventions can benefit older adults across a broad range of educational backgrounds [32]. These considerations highlight the need to interpret our findings cautiously and the importance of larger, well-powered trials to separate intervention effects from educational influences.

Overall, the mixed ANOVA results showed a significant improvement in cognitive performance, as demonstrated in MoCA score from preintervention to postintervention (P=.008), indicating that engagement in structured group activities can enhance cognitive outcomes in older adults, regardless of intervention type [33]. The time × group interaction, based on the mixed ANOVA, was not significant (P=.64), suggesting that the magnitude of cognitive improvement was similar across intervention types. These findings were further supported by Kruskal-Wallis and Mann-Whitney analyses in both ITT and PP approaches, which showed no significant differences in MoCA change scores in between-group comparisons or in pairwise comparisons of each intervention with the control group. Consistent with Barnes et al [34], who reported that 12 weeks of physical, cognitive, or combined activities in inactive older adults did not differ from the control groups, these findings suggest that overall engagement or the total amount of activity may be more influential than the specific type of intervention. One possible explanation is that nonspecific factors, such as social interaction, may contribute to the cognitive benefits observed [33].

The absence of significant between-group differences warrants further consideration. Several methodological and contextual factors may account for this pattern. First, the relatively small and unequal group sizes in the PP analysis, particularly the physical group (n=10), reduced statistical power to detect modest effects. Second, the improvements observed in the control group might be due to practice effects from taking the MoCA test multiple times or from nonspecific benefits such as social interaction, as previously noted by Lam et al [33], as well as incidental contact with the research team during assessments. Third, the sample’s relatively high baseline MoCA scores (mean≈25) suggest a potential ceiling effect, limiting the magnitude of observable gains among cognitively healthy participants. Finally, interindividual variability in engagement and adherence may have contributed to heterogeneous treatment responses, attenuating between-group contrasts. These explanations are consistent with prior intervention studies showing that cognitive enrichment benefits can occur across multiple activity types, but group-level differences often emerge only with larger samples or higher training intensity [35-37].

Notably, the combination group showed numerically larger within-group improvements and more consistent effect sizes across analytic approaches, while the cognitive group also demonstrated significant gains in the PP analysis. Importantly, these significant improvements were observed only among participants who achieved at least moderate adherence (≥50% attendance), indicating that sufficient intervention exposure was necessary to elicit measurable gains in global cognition. However, between-group comparisons did not reveal statistically significant differences, indicating that cognitive improvements over time were broadly comparable across intervention conditions. Within this context, the relatively larger effect sizes observed in the combined intervention, reflected by large within-group effects in both the ITT (r=0.54) and PP (r=0.66) analyses, suggest a potential advantage of multimodal engagement when adherence is maintained. These patterns are consistent with previous reports indicating that combined physical and cognitive interventions may be associated with greater improvements in global cognition compared with single-modality approaches, as observed by Yu et al [35] and Han et al [36]. Mechanistically, dual-task or multitask training that integrates physical and cognitive elements has been proposed to support cognitive function through neurophysiological pathways, including reduced bilateral prefrontal cortical oxygenation [36], increased hippocampal volume [38], and improved white matter integrity [39].

The video game-based cognitive intervention demonstrated meaningful benefits in this study, particularly among participants who maintained at least moderate adherence in the PP analysis. This aligns with the meta-analysis by Lampit et al [14], which reported that computerized cognitive training produces small but significant improvements in cognitive performance among healthy older adults. Participants engaging more consistently with the cognitive intervention showed the largest gains, highlighting the importance of adherence in achieving measurable benefits. In contrast, the physical activity group showed minimal cognitive change across both analytic approaches, suggesting that a once-weekly Zumba session may not have provided a sufficient cumulative training dose to elicit substantial cognitive improvements.

Besides, Zhang et al [40] examined the effects of all FITT-VP (frequency, intensity, time [duration], type, volume, and progression) exercise variables on cognition across healthy populations, including older adults, and found that variations in FITT-VP influence cognitive benefits. Older adults showed the greatest improvements in global cognition, executive function, and memory. Yuan et al [31] similarly reported that exercise improved cognition in older adults with cognitive frailty, particularly when conducted 3 or more times per week, with each session lasting at least 45 minutes over 12 weeks or longer. Chia et al [17] also observed that participants with higher physical activity levels exhibited significantly greater median MoCA scores compared with less active participants. Collectively, these findings emphasize that exercise characteristics, including frequency, intensity, and session duration (time), are crucial determinants of cognitive outcomes.

The present findings are consistent with recommendations from the American College of Sports Medicine and prior meta-analyses. Northey et al [41] reported that moderate-intensity exercise lasting at least 45 minutes per session on most days of the week enhances global cognition in adults aged 50 years and older. Xue et al [42] similarly found that moderate-duration, moderate-intensity exercise is associated with significant cognitive improvements among older adults. Although these studies do not explicitly recommend the FITT-VP framework, their results indirectly support its core principles, particularly frequency, intensity, and time, as key determinants of cognitive benefit.

In this study, Zumba sessions were delivered once weekly for 60 minutes over approximately 19 weeks. While session intensity was appropriate, the low frequency likely limited cumulative training dose, providing a plausible explanation for the minimal cognitive gains observed. Applying FITT-VP principles as an interpretive framework highlights the importance of optimizing exercise frequency, duration, and intensity in future interventions to ensure sufficient training stimulus for measurable cognitive improvements among older adults. While each intervention component was matched in session duration, the combined arm received a higher total weekly exposure, and this difference in intervention “dose” was considered when interpreting between-group comparisons.

The study was further extended with subdomain analyses under both analytic approaches (ITT and PP) to explore domain-specific responsiveness to the different interventions. The results showed that the language domain, which included assessments of verbal, categorical, and phonological fluency [43], demonstrated the most consistent improvement across all intervention groups. This suggests that verbal fluency and communication-related cognitive processes are particularly sensitive to structured and socially interactive activities. These findings are also supported by previous studies, although the results varied depending on the intervention type. Nocera et al [44] reported that aerobic exercise improved verbal fluency, whereas cognitive training was associated with decreased or nonsignificant improvement in this domain. In contrast, Calatayud et al [45] demonstrated that a comprehensive cognitive stimulation program significantly enhanced language proficiency, verbal fluency, and overall cognitive function. These contrasting findings indicate that the type and structure of the intervention may differentially influence specific cognitive domains.

Significant gains in the visuospatial domain were observed in the combination group in the ITT analysis, while improvements in orientation emerged in the cognitive stimulation group and reached significance in the PP analysis. These patterns suggest that distinct cognitive mechanisms may underlie the effects of different stimulation modalities. Interventions that combine physical and cognitive elements may enhance visuospatial and integrative processing, whereas purely cognitive activities may strengthen orientation. Collectively, these subdomain-level findings highlight the importance of examining domain-specific outcomes beyond total MoCA scores, as they may reveal early indicators of intervention efficacy [15].

Exploratory analyses examining adherence showed that attendance rates were not significantly correlated with MoCA improvement. While weak positive associations were observed for video game and total attendance, regression and ROC analyses showed limited predictive value for cognitive improvement. This suggests that mere attendance frequency may not fully capture the quality or engagement level of participation, underscoring the complexity of dose-response relationships in behavioral interventions. Nevertheless, adherence remains clinically relevant, as participants with consistent engagement in cognitive or combined activities demonstrated the most robust improvements in the PP analyses.

From a clinical perspective, these findings highlight the feasibility of combined physical and cognitive interventions as a practical approach for promoting cognitive engagement among older adults. While between-group differences were not statistically significant, the consistent within-group improvements and favorable subdomain patterns observed in participants with higher adherence suggest that sustained participation in structured multimodal activities may be beneficial in real-world community settings.

In relation to previous literature, the present findings are consistent with meta-analyses reporting small-to-moderate improvements in global cognition following cognitive and combined interventions [15,46]. This study adds novel insights by directly comparing physical activity, cognitive stimulation, and combined training within a single cohort, employing both ITT and PP analytic frameworks and mapping MoCA subdomains to identify domain-specific responsiveness. Such findings contribute to growing evidence that multimodal engagement may support cognitive health in aging populations, although definitive conclusions regarding superiority over single-domain training require larger, adequately powered trials.

Several limitations should be acknowledged. The semirandomized allocation and lack of blinding may have introduced expectancy effects, although baseline characteristics were comparable across groups and standardized outcome assessments were used to mitigate potential bias. Additionally, the sample size was modest and uneven across groups, limiting statistical power and precision. The intervention frequency (once weekly) and duration may have been insufficient to elicit large cognitive changes. Cognitive outcomes were based solely on the MoCA, without complementary neuropsychological or neuroimaging measures. Moreover, the study lacked long-term follow-up, preventing evaluation of the persistence of cognitive gains. Furthermore, as participants were community-dwelling older adults from a single urban region, generalizability to rural or clinical populations is limited.

Besides, adherence thresholds for the PP analysis were pragmatically defined at ≥50% attendance. Sensitivity analyses using higher cutoffs (70%‐80%) yielded directionally similar results, but small subgroup sizes reduced statistical stability, highlighting the trade-off between stricter adherence definitions and analytic feasibility in nonpharmacological interventions. Pooling participants with no intervention exposure and those with partial exposure below the ≥50% adherence threshold increased heterogeneity and may have attenuated detectable between-group differences.

Finally, formal 95% CIs for nonparametric between-group effect sizes were not calculated due to methodological constraints with small, skewed samples. Consequently, effect sizes (r and η²) were used to indicate magnitude and direction rather than statistical precision, and between-group findings should be interpreted cautiously.

In summary, both cognitive and combined interventions produced meaningful improvements in cognitive performance among older adults, with the strongest effects observed when adherence was maintained. Although between-group differences were not statistically significant, the consistent within-group trends and domain-specific patterns suggest practical cognitive benefits of engaging in regular, structured multimodal activities. Future research with larger samples, longer intervention durations, and multidomain cognitive assessments is warranted to confirm these findings and further clarify the mechanisms underlying differential cognitive responsiveness.

This study demonstrates that participation in video game–based cognitive stimulation, physical activity, and combined interventions was associated with within-group improvements in global cognitive performance among community-dwelling older adults. PP analyses indicated that participants who attended at least 50% of the cognitive stimulation sessions showed moderate improvements, while those who adhered to at least 50% of both the physical and cognitive components in the combined intervention demonstrated numerically larger and more consistent gains. Domain-specific analyses suggested that language, visuospatial, and orientation abilities may respond differently to intervention type. Although between-group differences were not statistically significant, consistent within-group trends support the practical value of structured engagement in cognitive and combined activities. Future studies with longer durations, higher frequency or intensity of physical activity, larger sample sizes, and multidomain cognitive assessments are needed to clarify the mechanisms underlying intervention-specific cognitive responsiveness.