Gait is a multifactorial process regulated by multiple control mechanisms, with speed often used as its main quantitative indicator.1 Among the characteristics of aging, slower gait speed is an important marker of several physiological and functional changes2,3 and a strong predictor of adverse outcomes,4 such as falls,5 frailty,6 hospitalization,7 and mortality.4 For these reasons, gait speed has been proposed as the sixth vital sign3 and has been widely investigated.4

Similar to gait speed, cognitive function declines with age.8,9 Cognition is a fundamental capacity encompassing several domains, including attention; auditory, visual, and tactile perceptual functions; verbal and linguistic skills; visuospatial processing capacity; memory and learning; executive functions;10,11 and orientation.12 In older adults, cognitive impairment is a well-established predictor of negative health outcomes, including frailty and mortality.8,9 Consequently, interventions targeting cognitive function are recommended in major healthy aging guidelines.13

The relationship between cognitive function and gait speed during the aging process has been widely investigated.14 Most studies have considered gait speed as a predictor of cognitive impairment, whereas our study adopts the reverse perspective, examining cognition as the exposure. This approach addresses a relevant gap, since few large-scale, population-based studies have explored how cognitive impairment influences gait speed, particularly in middle-income countries.1,15–21 This issue is especially important given the growing recognition of neuropsychological influences on mobility.15

Gait involves complex cortical mechanisms, suggesting that distinct cognitive processes may contribute to walking performance.1 Therefore, clarifying these associations may help identify neural networks shared by motor and cognitive functions.1 Most evidence shows that impaired executive function17,19–21 is associated with slower gait in older adults. However, findings for other cognitive domains remain inconsistent.1,17,19,20 Nonetheless, there are plausible mechanisms linking them to gait. For instance, age-related structural changes in the temporal lobes—critical for memory processing—have been associated with slower gait.22,23

To date, all studies that have examined cognition as a predictor of gait speed have been conducted in older adults from high-income countries.1,16–21 In Brazil, the pronounced regional and socioeconomic disparities may further shape this relationship. Data from the ELSI-Brazil study reveal macroregional differences in cognitive performance, even after adjusting for demographic factors, with lower scores observed in the North and Northeast.24 These disparities likely reflect inequalities in education, healthcare access, mobility infrastructure, and opportunities for cognitive engagement, potentially leading to patterns that differ from those seen in high-income countries.24

Therefore, the objective of this study was to investigate the association between global and domain-specific cognitive function among Brazilian older adults. We hypothesized that impairments in global cognition and specific cognitive domains would be associated with slower gait speed among Brazilian older adults.

Of the 9412 participants in ELSI-Brazil, 6940 were eligible for this study. A total of 756 individuals were excluded due to missing gait speed and/or cognitive function data, and 1716 due to missing information on covariates (Fig. 1).

Fig. 1.

Flowchart of sample selection, ELSI-Brazil, 2015–16.

The characteristics of the participants are summarized in Table 1. Most of them were female (52.8%), aged 50 to 59 years (51.0%), self-identified as brown (45.3%), lived with partner (66.3%), had 1 to 4 years of formal education (37.0%), and were in the upper income tertile (37.7%). Regarding lifestyle behaviors, 40.2% were overweight, 45.4% never smoked, 68.2% did not consume alcohol, and 51.8% were insufficiently active. In addition, 81.2% had no multimorbidity, 67.4% had three or fewer depressive symptoms, and 77.8% presented normal handgrip strength.

The prevalence of impairments in cognitive function and gait is presented in Table 2. Global cognitive impairment was observed in 12.5% (95%CI: 11.1, 14.0) of participants. For specific cognitive domains, impairments were identified in immediate memory (13.6%, 95%CI: 12.3, 15.1), delayed memory (17.0%, 95%CI: 15.4, 18.6), verbal fluency (13.3%, 95%CI 11.8, 14.9), and temporal orientation (13.2%, 95%CI 12.1, 14.5). Mean scores and corresponding 95% CIs for each domain of cognitive function are also shown in Table 2. Slower gait speed was observed in 10.1% of participants (95% CI: 8.3, 12.1) (Table 2). Moreover, participants with slower gait speed had a mean of 0.4 m/s (95% CI: 0.3, 0.4), compared with 0.8 m/s (95% CI: 0.7, 0.8), among those with normal gait speed.

Table 3 presents the prevalence of slower gait speed among individuals with normal and impaired cognitive function. Overall, participants with impairments in global cognitive function, memory (immediate and delayed), verbal fluency, and temporal orientation exhibited a significantly higher prevalence of slower gait speed. Among these domains, impaired verbal fluency had the highest prevalence (18.7%; 95% CI: 15.6, 22.3).

Table 3 also shows that both crude and adjusted prevalence ratios indicated a significant association between global cognitive function and all individual cognitive domains with gait speed. In the adjusted model, impaired global cognitive function was associated with a 63% higher prevalence of slower gait speed (PR: 1.63; 95% CI: 1.30, 2.04). Among cognitive domains, verbal fluency (PR: 1.80; 95% CI: 1.45, 2.24) and delayed memory (PR: 1.46; 95% CI: 1.16, 1.83) demonstrated the strongest associations.

In addition, statistically significant interaction was observed between global cognitive function and educational level in relation to slower gait speed (F = 4.64; p = 0.001). No interactions were found with other sociodemographic, behavioral, or health-related variables. Marginal effects analyses showed that the association between cognitive impairment and slower gait speed varied across educational strata. Among individuals with ≥12 years of education, cognitive impairment was not significantly associated with slower gait speed (Δ probability = −0.05; 95% CI: −0.11, 0.01; p = 0.103). Similar results were observed for those with 9–11 years of schooling (Δ = 0.02; 95% CI: −0.04, 0.07; p = 0.562). In contrast, cognitive impairment was associated with a significantly higher probability of slower gait speed among individuals with 5–8 years of education (Δ = 0.06; 95% CI: 0.001, 0.11; p = 0.047), 1–4 years (Δ = 0.09; 95% CI: 0.04, 0.13; p < 0.001), and among those who never attended school (Δ = 0.10; 95% CI: 0.02, 0.17; p = 0.01). Post hoc contrasts confirmed statistically significant differences in marginal effects between lower educational levels and the highest educational categories, indicating non-parallel slopes. Fig. 2 presents the adjusted probabilities of slower gait speed according to global cognitive function and educational level.

Fig. 2.

Adjusted probabilities of presenting slower gait speed according to global cognitive function and educational level, ELSI-Brazil, 2015–16.

Bars represent adjusted predicted probabilities from survey-weighted Poisson regression models, and error bars indicate 95% confidence intervals. *p < 0.05 for differences between cognitive function groups within the same educational level.

The results of this study showed that impairments in global cognitive function and specific domains were associated with a higher prevalence of slower gait speed among Brazilian older adults. While executive function exhibited the strongest association with gait speed, memory also played an important role, in contrast to the weaker association observed for temporal orientation. These findings underscore the multifactorial cognitive contributions to gait performance.

Previous research has associated impaired global cognition16-18,21 and executive function18-21 with slower gait speed in aging. Recently, increasing attention has been given to the interaction between higher-level cognitive functions and gait, which is no longer recognized as an automated motor activity requiring minimal cognitive processing.15 Instead, gait is now understood as a complex task that demands energy, balance, and coordination across multiple systems, including the nervous system.44

In this context, executive function seems to play a critical role in preserving gait speed.19 It encompasses a range of higher-order cognitive processes that integrate and adapt information from multiple cortical sensory systems, primarily within the frontal lobes and associated neural networks. Executive function is essential for generating, monitoring, regulating, executing, and readjusting behaviors to achieve complex goals.15,45 With aging, structural and physiological changes in the frontal lobes may contribute to declines in certain domains of executive function, which in turn have been linked to reductions in gait speed.15,46 Nonetheless, further research is needed to clarify these associations.15

Executive function is challenging to operationalize and measure given the diversity of available definitions and the complexity of its components.47 Miyake et al. suggested three core subdomains of executive function, supported by neuroimaging and neuropsychological assessments: (1) task and mental process switching, (2) real-time monitoring and updating, and (3) inhibition of automatic responses.48 Other researchers have proposed additional subdomains, such as access to long-term memory, decision-making, processing speed, and others.45

Verbal fluency, the measure used in this study, is considered an indicator of executive function. It reflects the ability to retrieve information stored in long-term memory45 and involves organization, self-regulation, and working memory.49 Although relatively simple to administer, verbal fluency tasks engage complex cognitive processes and are sensitive to various types of brain damage. They can also serve as an early marker of cognitive decline, including dementia.49

The relationship between other cognitive function domains and gait speed in older adults remains inconclusive, with divergent findings reported across studies.1,16,17,19,20 For memory specifically, three studies1,16,21 found a significant association with gait speed, while two did not.19,20 Notably, the temporal lobes, which are central to memory processing,22 undergo age-related structural changes that have been linked to slower gait.23 Callisaya et al.19 suggested that differences in study design (cross-sectional vs. longitudinal) might explain inconsistencies across studies. This rationale seems insufficient, as two of the three studies reporting an association used a longitudinal design. Instead, methodological differences, such as variations in memory and gait assessments, are more likely to account for the inconsistent results.

To the best of our knowledge, temporal orientation, another cognitive function domain investigated in this study, has not been examined in isolation in previous research. The lower complexity of this domain may partly explain the weaker association with gait speed observed in the present study.

The results of the present study also indicated that educational level moderated the association between global cognitive function and gait speed, with higher education attenuating and lower levels of schooling strengthening this relationship. Previous research has shown the role of education in mitigating age-related cognitive50 and mobility impairments,51 although evidence specifically addressing gait outcomes remains limited. These findings suggest that education may protect mobility in older adults, highlighting the importance of expanding lifelong educational opportunities and developing targeted interventions to support those with lower levels of schooling.

The present study presents some limitations. Its cross-sectional design precludes causal inference, preventing clarification as to whether cognitive deficits contribute to slower gait speed or vice versa. Moreover, only four cognitive domains were assessed, and within executive function, only a single aspect was evaluated. As a result, the assessment may not fully reflect the complexity of cognitive functioning or capture all mechanisms potentially involved in gait regulation.

The strengths of this study include the use of a large, nationally representative sample of Brazilian older adults, and the adoption of objective, straightforward, and literature-supported measures of both cognition and gait. In addition, cognitive function was evaluated through well-established, literature-supported tests, with scores standardized within subgroups defined by age, education, sex, and race/skin color. This strategy, widely adopted in other epidemiological studies,31,32 helps control potential sociodemographic confounders that could influence test performance. Another factor motivating this choice was the lack of standardized cut-off points for all tests used in the Brazilian population. Furthermore, standardizing cognitive function scores ensured more precise comparisons and facilitated interpretation.52

Gait speed was also standardized into z-scores, an approach previously applied in older adults.53 Incorporating age, sex, and height in the standardization helped minimize their influence on test performance.29 Standard cut-off points for gait speed in older adults were not used due to the specific characteristics of the walking test in this study. The test was performed at participants’ homes over a three-meter course, without additional distance for acceleration or deceleration. As a result, the reference values reported by Moreira et al. (0.70 to 0.53 m/s in women and 0.68 to 0.48 m/s in men, depending on age) are likely lower than standard thresholds.27

Impairments in global cognition and specific domains, particularly verbal fluency and delayed memory, were associated with slower gait speed in older Brazilian adults. Educational level moderated this relationship, with higher levels of education attenuating and lower levels of education strengthening the association. These findings suggest that, when slower gait speed is observed in clinical settings, cognitive aspects, particularly verbal fluency and memory, should be considered among the possible contributing factors. Additionally, considering an individual’s educational background may improve our understanding of the multifactorial determinants of mobility in aging.

Danielle Soares Rocha Vieira: Methodology; Formal analysis; Writing – review & editing; final review.

Laís Coan Fontanela: Laís Coan Fontanela contributed to the writing of the manuscript and critically reviewed the final version for important intellectual content. She approved the final version to be published and agreed to be accountable for all aspects of the work.

Vanessa Pereira Corrêa Rampinelli: Vanessa Pereira Corrêa Rampinelli contributed to the writing of the manuscript and critically reviewed the final version for important intellectual content. She approved the final version to be published and agreed to be accountable for all aspects of the work.

Nathália Tamara Stedile: Nathália Tamara Stedile contributed to the conceptualization of the study, writing of the manuscript, and critical review of the final version. She approved the final version to be published and agreed to be accountable for all aspects of the work.

Cesarde de Oliveira: Cesar de Oliveira contributed to all stages of the study, including study conceptualization, methodology, data acquisition and analysis, manuscript writing, critical revision, final approval, and accountability for the integrity of the work.

Ione Jayce Ceola Schneider: Ione Jayce Ceola Schneider contributed to all stages of the study, including study conceptualization, methodology, data acquisition and analysis, manuscript writing, critical revision, final approval, and accountability for the integrity of the work.