Dementia is a devastating condition carrying huge social and healthcare burdens (Alzheimer’s and Dementia, 2021), and will increase with the population aging. Without effective treatment for dementia, mild cognitive impairment (MCI), a transitional stage between normal aging and dementia that can be treated and even reverted (Petersen et al., 2018), becomes a valuable window to intervene in the progress of dementia. However, symptoms of MCI are often subtle and few patients seek medical assistance before their condition worsens, Thus, identifying MCI patients in the community population would be a precondition for the MCI intervention. The current screening methods for MCI are mostly for clinical settings (Connolly et al., 2011), and effective and convenient community screening tool for MCI is still in short supply.

Motor markers of cognitive changes have gained growing attention in recent years (Montero-Odasso et al., 2009; Beauchet et al., 2013). People with cognitive impairment often demonstrate gait disturbances, and the underlying mechanism points to brain regions and networks shared between gait motor control and cognitive processes (Montero-Odasso et al., 2014). A slowing in gait velocity was found associated with cognitive impairment and progression to dementia (Montero-Odasso et al., 2005; Waite et al., 2005). Moreover, studies have suggested that gait disturbances may precede the evident cognitive symptoms in dementia (Buracchio et al., 2010; Kikkert et al., 2016). Therefore, gait tests, characterized by low cost and convenience, could be a potential screening tool for cognitive impairment. A seminal study in 1997 stated that older adults who stopped walking while talking had a higher risk of falls (Lundin-Olsson et al., 1997). Since then, the dual-task gait test, a test in which participants perform walking and a secondary cognitive task concurrently (Al-Yahya et al., 2011), has become a research topic to explore the relationship between gait and cognition. Our recent meta-analysis demonstrated that the dual-task gait test exhibited higher sensitivity in MCI detection compared with a simple walking test (Yang et al., 2020). Because of the limited attentional capacity in people with compromised cognitive abilities, interference in the performance of one or both tasks would occur when an extra task was added (Jayakody et al., 2020).

There are various forms of dual-tasking and many cognitive tasks have been used in the dual-task paradigms. These tasks differentiate from each other in their nature, content and task difficulty. More importantly, different cognitive tasks have shown varying effects in people with MCI (Hunter et al., 2018). Currently, there is no consensus about which cognitive task should be paired with walking or which has a better discriminating ability for MCI (Montero-Odasso et al., 2019). According to Al-Yahya et al. (2011), cognitive tasks can be divided into five categories: reaction time, discrimination and decision-making, mental tracking, working memory, and verbal fluency, which relate to different cognitive domains. Several tests are commonly used in dual-tasks for older adults with MCI, including serial subtractions by 1 or 7 (mental tracking) (Montero-Odasso et al., 2017; Cullen et al., 2018; Hunter et al., 2018), naming animals (verbal fluency) (Montero-Odasso et al., 2017; Cullen et al., 2018; Hunter et al., 2018), and delayed recall (working memory) (Tseng et al., 2014; Nascimbeni et al., 2015). Discrimination and decision-making tests, however, are rarely used. Studies demonstrated that increasing the complexity of the cognitive task could bring greater sensitivity for MCI detection (Bahureksa et al., 2017; Yang et al., 2020). However, raising the difficulty too much could hurt specificity. The task needs to be challenging enough so that participants are working at or near the limit of their ability to uncover deficits (Hunter et al., 2018). On the other hand, MCI patients could exhibit impairment in different cognitive domains, including executive function, attention, language, memory, and visuospatial skills (Petersen et al., 2014). People always show different proficiency in different domains, so only testing one domain could conceal the deficit of other domains. Thus, developing a comprehensive task with appropriate complexity would improve the discrimination ability of the task greatly.

Picture naming tasks, which belong to the discrimination and decision-making tasks, are wildly used to characterize cognitive impairment. This task involved the function of visual perception, semantic processing, word retrieval, and oral naming, containing multiple cognitive processes (Lin et al., 2014). The multiple processes could test multiple cognitive domains (semantic memory, executive function, language, attention). In this study, we chose animal pictures because people are reasonably well knowledgeable about animal names; hence the influence of gender, age, and education could be minimal. In this way, we have developed an AniP-DT gait test, pairing the animal picture naming task with normal walking to form a dual-task gait test. The animal picture naming task required the participants to discriminate and name the animals in pictures, impairment in any domain (visual perception, semantic processing, word retrieval, and oral naming) could result in interference in motor performance. Therefore, using the AniP-DT gait test to assess motor-cognitive interaction would detect mild impairment in different cognitive processes, and improve the MCI detection capabilities.

This study developed a novel dual-task gait test (AniP-DT) that contained an animal picture naming task to test the MCI detecting ability in Chinese older adults. The MCI group had deteriorated gait performance compared to the cognitively intact group in both single and dual-task. Both single and dual-task gait velocities were significantly associated with MoCA scores even after adjustment of covariates. AniP-DT gait speed exhibited higher specificity in MCI detection in older adults. We found older adults whose AniP-DT gait velocity was less than 65.78 cm/s could be considered to have possible MCI in community settings. Adjustion of age and education could further improve the discrimination ability of AniP-DT gait speed. Our results suggested that gait velocity in AniP-DT could act as a behavioral marker to detect MCI.

Researchers have attempted to develop other motor-cognitive dual-task tests to help detect cognitive impairment (Klotzbier and Schott, 2017; Nielsen et al., 2018; Osuka et al., 2020). For instance, Latorre Román et al. (2020) designed a complex gait test (CGT) which was conducted on a 4 m × 6 m ground area and involved several obstacles when participants were walking. Among the discrimination and decision-making tasks, Trail-Walking Task (TWT) has also been used in MCI detection. Klotzbier and Schott (2017) used three TWT conditions with increased difficulty in a 4 m × 4 m area to differentiate MCI from cognitively healthy controls. Considering that a screening tool should be reliable, easy to use, and applicable to as many people as possible, we chose straight-path walking as the motor task, which has the lowest environmental and professional requirements, and chose animal picture naming as the cognitive task to ensure the scope of application for people are reasonably well knowledgeable about animals. The animals chosen in our picture naming test are familiar to most Chinese people. The whole test paradigm could be performed in a variety of settings including hospital hallways, local community centers, and parks, by personnel with simple training. On average, the test time for one person is less than 3 min.

In line with the previous findings (Montero-Odasso et al., 2014; Tseng et al., 2014), dual-task gait speed slowing was more prominent in our study and AniP-DT gait speed showed great discrimination ability of MCI from cognitively intact controls. The results suggested that the AniP-DT gait test performed well in MCI detection (AUC = 0.814), with a sensitivity of 62.8% and a specificity of 91.2%. In the complex gait test (CGT) of Latorre Román et al. (2020), obstacle negotiation could test participants’ executive function and exhibit a good diagnostic ability of MCI (AUC = 0.768). In the physical function test of Abe et al. (2022), models incorporating gait speed also showed good distinguish efficacy for MCI (AUC = 0.79), with a sensitivity of 73% and a specificity of 70%. The results from Klotzbier et al. showed that one of the TWT tests, the TWT-3, which has the highest cognitive load, showed a better differentiation (AUC = 0.860), the sensitivity of this task reached 100%, and the specificity was 66.67%. With a relatively close differentiation (AUC 0.814 vs. AUC 0.860), the sensitivity (62.8 vs. 100%) and specificity (91.2 vs. 66.67%) in our test and the results from Klotzbier et al. were nearly opposite. This could be due to the increased complexity of the task since TWT-3 involved participants’ cognitive flexibility, inhibition, and working memory (Klotzbier and Schott, 2017). Due to the stigma attached to dementia and cognitive impairment, screening and diagnostic decisions should be made with caution, thus particularly notable within this study is the high specificity of this test, which would be valuable in the detection of cognitive impairment in a large population. Further, we have determined the cut-off value of the gait speed. Older adults who walked slower than the cut-off value should have further evaluation of their cognition.

In widely used cognitive scales like MMSE or MoCA, age and years of education are often considered as the two factors greatly affect the performance of the participants. Xie et al. (2019) recruited MCI and cognitively normal participants in the community with matched demographic information (age, gender and education level). When adding these two parameters in our models, the ROC in our models also showed improved results, confirming the necessity of considering education and age in cognitive assessments. We think that the reason why the ajustment of age improved the differentiation ability of our test is that it is commensurate that gait velocity decreases with age, while the adjustment of education years increased the AUC of our models because we used the results from a Chinese version of MoCA, which is adjusted with educational level, as the reference in this study.

In summary, the AniP-DT gait test, which combined animal picture naming and straight-path walking, exhibited good discrimination ability in MCI detection in Chinese older adults in the community with a cutoff point of 65.78 cm/s. Although further exploration in a larger population is still needed, the testing paradigm showed great potential as a community screening tool for MCI in older adults, which could help identify MCI patients, thereby preventing or delaying their progression into dementia.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by the Medical Ethics Committee of Tongji Medical College, Huazhong University of Science and Technology, Wuhan, Hubei, China (s906). The patients/participants provided their written informed consent to participate in this study.

QY, YW, CT, JZ, and MY: acquisition of data. YW, QY, CT, MY, and JL: analysis and interpretation of data. YW, QY, and CT: drafting of the manuscript. All authors contributed to study conception and design, critical revision of the manuscript for important intellectual content, and approved the submitted version.