Predicting the level of cognitive function from standing balance and salivary biomarkers

Purpose: Brief cognitive assessments (BCAs) like the Montreal Cognitive Assessment (MoCA) are useful in examining the cognitive function of older adults with memory complaints. However, BCAs can be perceived as stressful, have a high ceiling effect, influenced by language barriers and the level of education of patients suffering from memory decline or confusion, and there is a need to explore more objective ways of identifying cognitive impairment. Methods: One-hundred and twenty-four older adults (age = 74.4 ± 6.6 years) completed six 30-seconds standing balance trials on a force platform alternating between eyes open and eyes closed conditions, performed in four fixed-sequenced foot positions (natural, 30 cm between-feet, Romberg, semi-tandem). Relative (intraclass correlation coefficients [ICC]) and absolute (standard error [SEM]) reliability of anteroposterior (AP) and mediolateral (ML) centre of pressure (COP) path lengths were determined in each condition and foot position. The acceptability of each foot position and the MoCA was assessed using a five-point Likert-scale ranging from ‘very relaxing’ to ‘very stressful’. Conventional AP and ML COP path lengths were used to predict MoCA scores in a multiple linear regression whilst controlling for age. The frequency components of AP and ML COP movements were further analysed with discrete wavelet transforms (DWT) to determine if they could statistically predict MoCA scores and differentiate between MoCA categories with an acceptable accuracy using age and timed-up-and-go (TUG) as covariates in a binary logistic regression. Saliva samples were also taken and analysed with an Enzyme-linked Immunosorbent Assay to determine salivary biomarkers (i.e., cortisol, interleukin-6 [IL-6] and interleukin-1β [IL-1β]) to investigate if they were statistically predictive of MoCA scores or if they could differentiate between MoCA categories. Salivary biomarkers and DWT COP metrics were combined in multiple linear and binary logistic regressions to determine if a more accurate screening tool for cognitive impairment can be developed. Results: The majority (54.8%) of participants were categorised by the MoCA as cognitively impaired, and 22% of participants (with 11.6% being cognitively impaired) perceived the MoCA as ‘very stressful’ or ‘stressful’. In eyes open and closed conditions and all foot positions, good-to-excellent relative reliability (ICC = 0.78-0.93) and comparable absolute reliability (SEM = 0.06-0.08 cm) were detected. The natural stance was the most acceptable foot position with 60% of participants rating it as ‘very relaxing’. Whilst controlling for covariates, the conventional AP and ML COP path lengths could not predict MoCA scores, however the DWT COP metrics showed that the vestibular frequency (i.e., vestibular control of AP sway in the Romberg stance) could statistically predict MoCA scores with slope coefficient = -0.50 (standard error = 0.16, p = 0.003), and the vestibular frequency and cerebellar frequency (cerebellar control of AP sway in the semi-tandem stance) individually statistically differentiated between cognitively normal and cognitively impaired participants. A diagnostic test using age, TUG, and the vestibular frequency had a sensitivity of 72.1%, specificity of 75.0% and an area under the curve of 0.79 (95% CI = 0.71, 0.87; p < 0.001). Furthermore, salivary cortisol (slope coefficient = -2.07 [standard error = 1.29, p = 0.046]), IL-6 (slope coefficient = -8.54 [standard error = 10.0, p = 0.049] and IL-1β levels (slope coefficient = -0.37 [standard error = 0.21, p = 0.038]) were significant statistical predictors of MoCA scores in the cognitively normal and mild cognitive impairment (MCI) categories. Lastly, salivary cortisol and IL-6 statistically differentiated between participants with a dementia-causing diagnosis (n = 23) and those without (n = 89), however the salivary biomarkers did not improve the diagnostic accuracy of the DWT COP metrics. Conclusion: Findings from this thesis confirm that some older adults perceive BCAs like the MoCA as stressful. This, together with previously known limitations of these tests, highlighted the need to find alternative ways of assessing cognitive function. All standing balance measures had good-to-excellent reliability and the frequency bands of standing balance measures analysed with DWT could be used to determine an individual’s cognitive function with an acceptable accuracy. Salivary biomarkers could also be used to differentiate between participants with and without a dementia-causing diagnosis, whilst statistically predicting cognitive function in cognitively normal and MCI participants. These findings provide the opportunity for future research to develop simple, objective tests that are more acceptable to those with suspected dementia to predict cognitive function and to screen for cognitive impairment.