A large body of research has demonstrated an important interdependence between gait and cognition in elderly people noting that slow motor performance is associated with cognitive impairment and dementia [1–5]. Walking is a complex learned task that becomes automatic for most people from early childhood onwards. However, there is evidence that cognitive control of gait becomes increasingly important in older adults[6, 7]. Since a seminal study demonstrated that the inability to maintain a conversation while walking is a marker for future falls in older adults[8], walking while performing a secondary task (dual-task paradigm) has become a classic way to assess the relationship between cognition and gait Because the dual-task paradigm is a realistic proxy for daily living activities that seniors may perform at home, it is growing in interest and application in clinical research settings.

Although previous studies have reported good test-retest reliability using electronic walkways while assessing gait in older people without performing a secondary task [9–11], the reliability of quantitative gait assessment under dual-task conditions among those with cognitive problems has not been established. One of the challenges faced when assessing cognitively impaired older adults is their potential inability to perform dual-tasking properly, thereby increasing measurement error. Therefore, our objective was to determine the reliability of quantitative gait assessment under both single and dual-task conditions in people with mild cognitive impairment (MCI) using a one-week space between assessments. This timeframe is appropriate for research purposes in this population, since gait variables are stable over short intervals. Establishing the reliability of quantitative gait assessment under dual-task conditions among those with cognitive problems is an important first step in validating this methodology for future longitudinal studies.

Gait variability results are expressed as CoV in Table 2. Of the six parameters analyzed, only step and stride time have significantly increased during dual-task condition and with considerable stride-to-stride variability. In fact, the CoV for stride time increased from 6.36 on sG to 11.02 on dG in week-one assessment. As shown in Figure 1, under dual task conditions, there was an increase in the CoV for the variables related to time: step, stride, and double support time. For these time-related variables, the CoV increased by 63.13%, 73.27%, and 36.82% respectively from sG to dG conditions, while the CoV for the remaining gait variables increased by less than 6%.

Reliability results are shown in Table 2. Under single task conditions, re-test reliability was excellent with the ICCs for six gait variables being 0.83 or higher with the exception of double support time, which was 0.80. Reliability results under dual-task conditions were also excellent with ICCs for the six gait measures were 0.93 or higher.

This study established the test-retest reliability of gait assessment under single and dual-task conditions in older adults with MCI. There is excellent reliability in both conditions over a one-week time span. Reliability of gait assessment in older adults has been reported previously[11, 20, 28]; however, to our knowledge, this is the first report which determined the reliability under dual-task conditions in mildly cognitively impaired older people. Due to the growing importance of dual-task paradigm in current research of cognitive decline, falls, and dementia [29–31], our results are of particular relevance since the presence of cognitive impairment in our population did not preclude performance of dual-tasking while gait assessment.

The ICCs for the six variables analyzed under single and dual-task were higher than 0.75, showing excellent clinical reliability for the assessment of these spatial and temporal gait parameters.

One interesting finding of our study is that the time required to perform steps and strides significantly increased under the dual-task condition. This is consistent with previous studies, which have demonstrated that these parameters have a greater dependence on brain cortical control than other gait parameters[25, 31]. In addition, step and stride time showed a greater variability under dual-task conditions when compared with the other parameters analyzed (Table 2, Figure 1). This finding is particularly relevant given that gait variability under dual-tasking has been demonstrated to be an early predictor of future falls[8] In our sample of people with Mild Cognitive Impairment, gait variability significantly increased under dual-task conditions. While gait variability is minimal under dual-task conditions for the general population, high gait variability is associated with Parkinson disease, Alzheimer's disease and other dementias, a variety of types of dementias and is a predictor of future falls[25, 30, 31]. Since our results confirm the reliability of this assessment in persons with MCI, this type of assessment may be an effective early measure of detecting individuals with MCI who are at higher risk of future falls. Increased gait variability represents a more unstable walking pattern with less rhythmicity. We postulate that this altered gait pattern is a type of "arrhythmia" of the gait and may correlate to other health outcomes such as future dementias, falls or other comorbidities.

Particularly, our results provide support to apply this methodology in people with cognitive problems. Although gait has long been considered as primarily an automatic motor task, emerging evidence suggests that this view may be overly simplistic[32]. Cortical brain control may play a key role in the regulation of even routine walking, specifically through attention. Attention is a necessary cognitive resource for maintaining normal walking and attentional deficits are independently associated with postural instability, impairment in performing daily living activities, and future falls[24]. Specifically, dual-tasking cost has been traditionally related to the prefrontal cortical regions[33]. These brain regions are crucially involved in the mediation of the division of attention and executive function. Functional neuroimaging studies showed correlations between dual-task performance with increase activity in prefrontal areas, cingulate, parietal and premotor areas[34, 35]. Therefore, we postulate that those regions may have a control on gait in older individuals. In line with previous studies, our results support the hypothesis that occupying these areas with concurrent cognitive processing (dual-tasking) may result in a brain resource limitation that affects gait in people with MCI.

Strengths of our study include the use of a well-defined population that met strict criteria for MCI, and the use of established and validated measures of cognition using a sophisticated quantitative gait assessment. The small sample size of our study, however, is a potential limitation, as it may not represent the full spectrum of MCI population. Despite these limitations, this study provides evidence that quantitative gait analysis while dual-tasking among older adults with MCI is a reliable methodology.

In older adults with MCI, assessment of quantitative gait variables using an electronic walkway was highly reliable under single and dual-task conditions. In line with previous studies conducted in elderly with normal cognition, variability of time-related gait parameters increased while dual-tasking in our sample. The presence of cognitive impairment did not preclude performance of dual-tasking gait assessment in our sample; therefore, this methodology can be reliably used in older people with MCI.