Frailty is associated with worse executive function and higher cerebral blood velocity in cognitively healthy older adults: a cross-sectional study

Abstract

Frailty is characterized by an increased vulnerability to adverse health events. Executive function impairment is an early sign of progression towards cognitive impairments. Whether frailty is associated with executive function and the associated mechanisms are unclear. We test the hypothesis that higher frailty is associated with worse executive function (Trail Making Test) and if aerobic fitness, prefrontal cortex oxygenation (ΔO₂Hb), or middle-cerebral artery velocity (MCAv) impact this association. Forty-one (38 females) cognitively health older adults (70.1 ± 6.3 years) completed a Trail task and 6-min walk test. Prefrontal cortex oxygenation was measured during the Trail task (via functional near-infrared spectroscopy) and MCAv in a sub-sample (n=26, via transcranial Doppler). A 35-item frailty index was used. Frailty was independently, non-linearly related to trail B performance (Frailty²: β=1927 [95% CI: 321–3533], p = 0.02), with the model explaining 22% of the variance of trail B time (p = 0.02). Aerobic fitness was an independent predictor of trail B (β=−0.05 [95% CI: −0.10–0.004], p = 0.04), but age and ΔO₂Hb were not (both, p > 0.78). Frailty was positively associated with the difference between trails B and A (β=105 [95% CI: 24–186], p = 0.01). Frailty was also associated with a higher peak MCAv (ρ = 0.40, p = 0.04), but lower ΔO2Hb-peakMCAv ratio (ρ = −0.44, p = 0.02). Higher frailty levels are associated to worse Trail times after controlling for age, aerobic fitness, and prefrontal oxygenation. High frailty level may disproportionately predispose older adults to challenges performing executive function tasks that may manifest early as a compensatory higher MCAv despite worse executive function, and indicate a greater risk of progressing to cognitive impairment.

Supplementary Information

The online version contains supplementary material available at 10.1007/s11357-023-00991-3.

Introduction

Frailty is conceptualized as a state of vulnerability to adverse outcomes and is a consequence of cumulative multisystem deterioration that results in more frequent use of healthcare services, as well as a decrease in quality of life, an increase in disability, and an increased risk of death [1, 2]. Observational studies consistently document a negative association between frailty and scores on the Mini-Mental State Examination (MMSE), which is a marker of cognitive health used to screen for general cognitive impairment [3]. Impairments in executive functions represent early indicators in the development of major neurocognitive disorders and diseases [4, 5]. Understanding which factors are associated to executive function performance among cognitively healthy older adults may be useful for preventing the development of cognitive decline.

It is well-established that higher aerobic fitness is associated with better executive function and lower frailty levels [6, 7]. Specifically, aerobically fit older adults demonstrate faster reaction times in neuropsychological tests of executive function than their less aerobically fit counterparts [8, 9]. Faster reaction times, which generally translate into better executive function and lower frailty, might be attributed to greater prefrontal cortex oxygenation [8, 9]. It is unclear however if better prefrontal cortex oxygenation results from more effective extraction of oxygen from local blood flow, or whether it occurs because of an increased delivery of blood to the brain (cerebral blood flow).

Altogether, even if previous studies suggested relationships among aerobic fitness, brain oxygenation, cerebral blood flow, frailty, and executive functions, evidence to date has been presented in fragmented pieces. Assessing relationships among these variables concurrently could help shed light on the mechanistic underpinnings of aging-related frailty and cognitive declines, as well as their potential co-development. Therefore, the purpose of this study was to (1) investigate the relationship between a frailty index and executive function in cognitively healthy older adults, and (2) document whether this relationship is impacted by prefrontal cortex oxygenation, MCAv, and/or aerobic fitness. We test the hypothesis that higher frailty would be associated with worse executive function. We also test the hypothesis that prefrontal cortex oxygenation, MCAv, and aerobic fitness explain the anticipated frailty-executive function relationship.

Methods

Participants

Forty-one community-dwelling older adults (n=38 females) consented to participate in the present study. Based on a general correlational effect size of r=0.59 observed between aerobic fitness and trail B time in older adults [9], it was estimated that a minimum of 17 participants were needed assuming a two-tailed, α=0.05, and power of 80% (G*Power 3.1 [10]). Participants were recruited through local community programs (e.g., rotary club, club d’age d/or, dames de l’Acadie), social media, and through word of mouth. All participants were right-handed and had normal-to-corrected vision. No participants were smokers or had a history of neurological or psychiatric disorder, color blindness, surgery with general anesthesia in the past six months, involuntary tremors, epilepsy, or drug/alcohol problems. One participant was taking medication for depression, one participant was on estrogen therapy, 17 were taking medication a statin, and 15 were taking medication for high blood pressure. All participants scored >23 on the MMSE. The protocol was reviewed and approved by the Institutional Research Ethics Board at the Université de Sherbrooke and was conducted in accordance with recognized ethical standards and national/international laws.

Experimental design

All participants in this cross-sectional study completed a functional and a cognitive assessment during two, 60-min sessions; a minimum of 48 h was required between sessions. For the first session, participants signed the consent form and completed questionnaires on their health status, followed by measurement of heart rate (via radial pulse), blood pressure (via sphygmomanometer and brachial cuff), and other anthropometrics (i.e., height and weight via stadiometer). The first session concluded with a cognitive test (Trail Making Test described below). During the cognitive test and rest, cerebral oxygenation and cerebral blood velocity were simultaneously measured with functional near-infrared spectroscopy (fNIRS) (OctaMon, Artinis Medical Systems, Netherlands) and bi-lateral Transcranial Doppler (Multigon Industries Inc., New York, USA). During the second session, participants completed a functional assessment (also described below). The testing was always completed in this order. Participants were asked to refrain from vigorous exercise 24 h prior to testing and were also asked to consume a typical breakfast on the morning of testing to maintain their usual routine.

Frailty index

We utilized a 34-item frailty index using gathered data as previously described in [11]. Items in the frailty index broadly represented the following domains: sociodemographic, physical function, comorbidities, self-estimated health/emotions, activities of daily living, anthropometrics, and cognition. Items are presented in Supplemental Table 1. To be included in a frailty index, items must be become worse with age, associated with adverse health outcomes, and cover a range of physiological systems [11]. A minimum of 30 items is required [11]. Each item was coded as 0 (no deficit) or 1 (deficit). Interval or ordinal variables were coded as a proportion of complete deficit (e.g., self-rated health has 5 options: excellent=0, very good=0.25, good=0.5, fair=0.75, poor=1). Each participants’ frailty index was calculated as the total deficits accumulated ÷ the number of total deficits measured (e.g., 3.5/34 = 0.11). A value closer to 1.00 indicates a higher (worse) degree of frailty.

Executive function

The Trail Making Test is a well-established cognitive task [12]. All participants completed part A (trail A) prior to part B (trail B). Trail A was used to measure processing speed. Participants were given the following Trail Making Test instructions for part A: “Please take the pencil and draw a line from one number to the next, in order. Start at 1 [point to the number], then go to 2 [point], then go to 3 [point], and so on. Please try not to lift the pen as you move from one number to the next. Work as quickly and accurately as you can.” Participants were encouraged to correct their errors, and this was included in the total time to complete. The speed at which all numbers were connected was measured in seconds (s). Trail B was used to measure cognitive flexibility or switching ability. In trail B, participants were given the same instructions as trail A but had to alternate between numbers in ascending order and letters in alphabetical order (1-A-2-B-3-C, etc.). The time to complete trail B was also measured in seconds. Prior to the standard administration of this test, participants were given a short practice trial before taking each test.

Cerebral oxygenation

During the cognitive tests, a multi-channel fNIRS device (OctaMon, Artinis Medical Systems, Netherlands) was used to measure relative changes in concentrations of oxygenated hemoglobin (ΔO₂Hb) in the brain. The fNIRS device had two sets of 1×4 multichannel probe holders, which consisted of 8 light-emitting sources and 2 light-absorbing detectors arranged alternately at an inter-probe distance of 3.5cm (wavelengths of 760 and 850 nm) (see Supplemental Figure 1). This configuration resulted in 4 channels per set (total of eight channels). Optode position was set and covered the participant’s prefrontal brain area according to the international 10/20 positioning system. The concentration ΔO₂Hb was calculated using the difference in absorbance based on the Beer-Lambert law. Because continuous-wave technology does not measure optical path lengths [13], only changes in concentration of O₂Hb relative to baseline (last minute of a 5-min resting period prior to the Trail Making Test) could be inferred assuming both a path length factor and partial volume. Continuous NIRS measurements were recorded during Trail test (parts A and B) with the average relative concentration of O₂Hb recorded for each participant. NIRS data was acquired at 10 Hz and filtered with a Savitzky-Golay smoothing algorithm before analysis. All data analyses were completed in the Oxysoft analysis software.

Four neighboring channels were combined to form left and right prefrontal cortex areas, defined as the region of interest. The regions included the right prefrontal cortex (channels 1, 2, 3, and 4) and the left prefrontal cortex (channels 5, 6, 7, and 8). We also measured the cerebral oxygenation in the right dorsolateral (channels 1 and 2), the left dorsolateral (channels 5 and 6), the right ventrolateral (channels 3 and 4), and left ventrolateral (channels 7 and 8) individually. Data was averaged over every task component (rest, trail A, trail B) and normalized to express the magnitudes of change from the baseline period. The baseline period took place immediately before the cognitive assessment; participants sat quietly and were asked to close their eyes and eliminate extraneous thoughts to establish a 60-s baseline of NIRS data (last 60 s of a 5-min rest period).

Middle cerebral artery velocity

MCAv was measured bilaterally in all participants during the cognitive testing using a Transcranial Doppler (RobotoC2MD system, Multigon Industries, Yonkers, NY). Insonation of the left and right middle cerebral artery was performed on the trans-temporal window using two 2-MHz probes secured in place with an adjustable headset. All hemodynamic parameters were analyzed offline using a data recorder, which was attached to the transcranial Doppler system (Neurovision, Multigon Industries, Yonkers, NY). Peak middle cerebral artery velocity (MCAv_peak) was collected from both the right and left side for each condition. Peak values from the left and right side were averaged, or when one side was unavailable (n=6), unilateral MCAv was used. Peak was selected as the largest MCAv response, whereas time-weighted mean MCAv is more impacted by MCAv during diastole (time-weighted mean = 1/3×systole + 2/3×diastole).

Six-minute walk test

Prior to the functional assessment, heart rate, blood pressure, and anthropometrics (i.e., heigh and weight via stadiometer) were determined. Aerobic fitness was assessed via the 6-min walk test (6MWT), which was completed according to standardized procedures provided by the American Thoracic Society [14]. The 6MWT was conducted in an open gym, and the course was marked by red cones placed 30 m apart. The cumulative distance covered over the 6 min was recorded to the nearest centimeter.

Statistical analysis

All data were assessed for normality using a Shapiro-Wilk test. Non-normally distributed data were analyzed using non-parametric tests.

Separate ordinary least squares regression analyses were conducted on trail A and trail B time. To determine whether the relationships were non-linear, alternative models were explored and included either: (1) linear and quadratic terms, or (2) linear, quadratic, and cubic terms. For both Trail times, the quadratic term resulted in a statistically improved model, whereas the cubic term did not further improve the model. Final multiple regression models examined whether frailty, frailty², 6MWT distance, ΔO₂Hb for the task, and age were associated with trail A or trail B times.

The relationship between frailty with the difference between trail A and trail B (trail B–trail A time) was examined after controlling for age. We explored whether separate locations of NIRS-derived ΔO₂Hb during trail A and trail B were associated with frailty via Spearman rank correlations.

In our sub-sample with Doppler (n=26), we examined whether frailty was associated with peak MCAv responses and the ratio of prefrontal cortex oxygen extraction to peak oxygen delivery (ΔO₂Hb/Peak MCAv) during the Trail task. Separate analyses were conducted for trail A and trail B. All statistics were completed in SPSS Version 28.0 (IBM, NY). Statistical significance was accepted as p<0.05. All data are presented as means ± standard deviations.

Results

Participants were 70.1 ± 6.3 years of age and a mild frailty level (average: <0.10) (Table 1).

Table 1.

Participant characteristics and physical function outcomes

Variable	Participants (n=41)
Age (years)	70.1 ± 6.3 (61–86)
Males, females	3, 38
Height (cm)	159.2 ± 6.7 (147–174)
Weight (kg)	86.1 ± 8.2 (72–94)
Body mass index (kg/m2)	31.84 ± 5.92 (22–44)
Resting heart rate (beats/min)	68 ± 11 (58–92)
Systolic blood pressure (mmHg)	134 ± 15 (109–162)
Diastolic blood pressure (mmHg)	79 ±8 (65–96)
MMSE score	28.8 ± 1.1 (26–30)
Trail A time (s)	31.7 ± 12.0 (14–82)
Trail B time (s)	67.3 ± 18.9 (36–132)
Trail B–trail A time (s)	35.6 ± 16.4 (9–71)
6MWT (distance)	625 ± 131 (120–845)
Frailty Index	0.088 ± 0.060 (0.015–0.235)

Continuous data are presented as means ± SD (minimum–maximum) and categorical variables as n (proportion). MMSE, Mini-Mental State Exam; 6MWT, 6-min walk test

The relationship between frailty and trail A time was non-linear, with a quadradic frailty term improving the association beyond a linear term (F-change: p = 0.02), but no further improvement with a cubic frailty term (Fig. 1A). The model examining the predictors of trail A time is presented in Table 2. The total model predicted 24% of the variance (p = 0.04), with frailty and frailty² being independent predictors of trail A, whereas 6MWT and age were not (all, p > 0.15; Table 2). When the change in prefrontal ΔO₂Hb during trail A was added to the model, the overall model was not statistically significant (p = 0.07), although frailty and frailty² remained as independent predictors (both, p<0.048)

Fig. 1.

Relationships between frailty with trail A time (panel A), trail B time (panel B), and trail B minus trail A time (panel C). The ordinary least squares regression model strength and unstandardized coefficient for frailty is presented for each panel. The relationship was non-linear for trail A and trail B, but linear for the difference in Trail times. Frailty was determined via a 35-item index. Data are presented for n=41

Table 2.

Multiple regression model predicting time on the trail A and trail B task

Predictor	Unstandardized β (95% CI)	Standard error	p-value
Trail A
Frailty (ratio)	−251 (−496, −5)	120.90	0.045
Frailty2 (ratio)	1090 (34, 2146)	520.79	0.04
6MWT (distance)	−0.01 (−0.04, 0.03)	0.02	0.70
Age (years)	0.46 (−0.18, 1.11)	0.32	0.15
Constant	12.2 (−47.0, 71.5)	29.21	0.68
Trail B
Frailty (ratio)	−353 (−726, 20)	183.69	0.06
Frailty2 (ratio)	1927 (321, 3533)	791.24	0.02
6MWT (distance)	−0.05 (−0.10, −0.004)	0.02	0.04
Age (years)	−0.14 (−1.14, 0.86)	0.49	0.78
ΔO2Hb (a.u.)	1.11 (−3.14, 5.37)	2.09	0.60
Constant	118.1 (27.6, 208.6)	44.57	0.01

Data are presented as unstandardized β [95% confidence intervals]

The relationship between frailty and trail B was non-linear to the quadratic term (Fig. 1B), with the overall model explaining 22% of the variance of trail B time (p = 0.02). Frailty² and 6MWT (negative) distance were independent predictors of trail B time, whereas age and the change in prefrontal ΔO₂Hb were not (both, p > 0.78). The relationship between frailty and trail times was non-existent at a frailty level of 0.02–0.13, but steeply increased after a frailty level of 0.13 (Fig. 1). Given that frailty level was positively, non-linearly related to both trial tasks, we examined the difference between trail B and trail A and observed a positive linear relationship between frailty and trail B–trail A (Fig. 1C), this remained after controlling for age (frailty: p = 0.01).

There was no association between total NIRS-derived ΔO₂Hb during trail A or trail B and our frailty index (Fig. 2). This observation remained when examining ΔO₂Hb responses of the right-side only, left-side only, dorsolateral only, or ventrolateral only with our frailty index (all, p > 0.15).

Fig. 2.

Relationship between frailty with functional near-infrared spectroscopy derived changes in prefrontal cortex oxygenated hemoglobin from baseline (ΔO₂Hb) during trail A (panel A) and trail B (panel B). No relationships were observed between frailty and ΔO₂Hb during either Trail task as assessed via Spearman Rank Correlations. Frailty was determined via a 35-item index. Data are presented for n=41

In the sub-sample with MCAv (Supplemental Table 2), frailty was positively, linearly correlated with peak MCAv during trail A and trail B (Fig. 3), which remained for trail B when expressed as a change in MCAv from rest to task for trail B (p = 0.04), but not trail A (p = 0.09). Null findings were observed if mean, instead of peak MCAv was used. Frailty was negatively associated with the ratio of ΔO₂Hb/peak MCAv during trail B (p = 0.02), but not trail A (p = 0.11; see Fig. 3).

Fig. 3.

Relationship between frailty with peak middle cerebral artery velocity (MCAv) and ΔO₂Hb÷MCAv ratio during trail A (panels A–B) and during trail B (panels C–D). Positive relationships were observed between frailty with peak MCAv during trail A and trail B (via Spearman Rank Correlations). A negative relationship was observed between frailty and the extraction-delivery ratio during trail B only (panel D). Frailty was determined via a 35-item index. Data are presented for a sub-sample of n=26

Discussion

The purpose of this study was to examine the relationship between frailty and executive function in cognitively healthy older adults and clarify the association by exploring the underlying physiological mechanisms, including aerobic fitness, cerebral oxygenation, and MCAv. Consistent with our hypothesis, higher frailty was associated with worse executive function, with a non-linear, steeper worsening of trail B performance exhibited by more frail participants, but a null relationship at lower frailty levels. The relationship remained after controlling for chronological age and aerobic fitness and was not explained by intra-trail task changes in prefrontal oxygenation. Rather, in a sub-sample with transcranial Doppler, higher frailty was associated with a larger peak in cerebral blood velocity response during the trail task, which resulted in a lower ΔO2Hb-peakMCAv ratio. These cross-sectional results may provide direction for interventional studies and insight into the mechanisms linking frailty and executive function.

The trail making task reflects processing speed and cognitive flexibility domains within executive function [15]. We demonstrated that frailty was non-linearly related to both processing speed (trail A) and cognitive flexibility (trail B). Frailty may have a minor or null impact on cognition among the lowest frailty levels (e.g., <0.10), but as deficits accumulate, we observed a steeper, exponential impact of frailty on cognition. Importantly, this impact of frailty remains after controlling for other well-established predictors of cognition, such as age, prefrontal oxygenation [16], and aerobic fitness [9]. By examining the differences in Trail tasks, frailty was linearly related to this isolated executive function performance that remained after controlling for age. Given that a pre-requisite for items to be included in a frailty index is that they are age-related [11], this supports that frailty specifically is an important predictor of executive function after controlling for chronological age. Frailty has been promoted as a proxy for biological aging [17, 18] in that it describes the accumulated of age-related physiological damages and may be a more relevant marker of health than chronological age. Accordingly, the non-linear relationship may suggest that older adults, who exhibit a higher biological age or accelerated aging phenotype, may be more susceptible to worse executive function compared to older adults who are able to prevent the development of age-related health concerns. Strategies that mitigate frailty or potentially reverse frailty (e.g., proper exercise, cognitively engaging tasks, disease management) [6, 19] may be efficacious in attenuating the decline in cognition processes that accompany age. Moreover, our observations clearly indicate that those who have accumulated more health deficits are more likely to exhibit worse executive function despite being absent of cognitive impairments. This is not trivial, as it might predispose these individuals to additional health problems. In other terms, people with health problems are more likely to develop more problems. These observational findings may reflect a “tipping point” whereby frailty management is needed to help cognitively healthy older adults reduce their risk of developing multiple comorbidities and/or progressing to dementia.

The exercise-cognition literature supports that higher aerobic fitness and engaging in physical activity improve executive function [20, 21], with neuroimaging studies documenting that this improvement may be due to greater prefrontal cortex oxygenation [8, 9]. In the covariate-adjusted model including prefrontal oxygenation, higher 6MWT distance was independently associated with faster trail B time and thus executive function. Accordingly, other, non-cerebral oxygenation-based mechanisms (e.g., brain-derived neurotropic factor, insulin growth factor 1 [22]) may better link exercise performance and executive function. Of note, executive functions are complex and encompass multiple cognitive processes and our study is specific to the executive functions tested during a commonly used Trail Making Task [12]. Despite a clear reduction in hemodynamic activity of the prefrontal cortex among older adults versus young adults [23], prefrontal cortex oxygenation was not associated with frailty level, despite higher frailty being associated with slower trail task times and a lower ΔO2Hb-peakMCAv ratio during the executive trail task. Alternatively, there may be compensations of other brain areas, including the temporal, parietal, and cerebellar region that may also play a major role in executive functions [24]. Of note, the Irish Study of Longitudinal Ageing failed to observe a relationship between deficit-based frailty and global cognition [25], indicating that frailty may be most related to complex executive functions. As well, the model implemented explained ~20–25% of the variance in Trail task times, indicating that other factors beyond frailty are also primary contributors to executive function. While cross-sectional, our fNIRS observations suggest that it is not the extraction of oxygen in the prefrontal cortex that explains the impaired executive function among frailer older adults, but this should be confirmed with interventional studies that reduce frailty or cohort studies that longitudinally track the impact of increasing frailty on cerebrovascular outcomes and cognitive performance.

Interestingly, frailty level was positively associated with MCAv, whereby participants with worse frailty exhibited larger MCAv responses despite similar prefrontal cortex oxygenation. This observation conflicts with previous work documenting that older adults classified as frail (via frailty phenotype) and with mild cognitive impairment exhibited lower middle cerebral artery reactivity and MCAv than healthy older adults [26]. Such inconsistencies may be due to the frailty phenotype being a measure of “physical frailty” [27] in contrast to the health-deficit model implemented in the present study. Rather than producing a continuous ratio, the frailty phenotype has a pre-defined set of criteria that categorize people as non-frail, pre-frail, and frail. These two measures have distinct theoretical and measurement differences, preventing findings from one tool to be synonymous with another [28]. Alternatively, between-study differences in cognitive health are plausibly a greater factor. The Compensation-Related Utilization of Neural Circuits Hypothesis (CRUNCH model) suggests that age-related overactivation is compensatory, but as task demand increases, older adults reach their capacity [29]. In contrast, older adults with slower walking gaits exhibited smaller changes in MCAv in response to a 2-back cognitive challenge versus counterparts with faster walking gaits [30]. While frailty was not measured per se, these conflicting reports indicate a need for interventional work aimed at understanding whether reducing frailty levels improves cognition. Nevertheless, the frailer, but cognitively healthy older adults included in our study may have required greater blood resources to be directed towards the prefrontal cortex, despite worse reaction times and a matched oxygenation to their less frail counterparts. While aging is associated with impairments in cerebromicrovascular function [31–33], our null impact of prefrontal cortex oxygenation that also controlled for age on the trail making task outcomes is inconsistent with these prior studies in this specific sample and suggests that cerebromacrovascular function (or MCAv) may be a greater factor in the frailty-trail relationship. Its plausible that worsening neurovascular coupling, or the link between neuronal activity and cerebral blood flow on a beat-by-beat basis, is implicated. Specifically, the older adults with higher levels of frailty exhibited slower reaction times and a corresponding lower neuronal activity coupled with an augmented MCAv response compared to older adults with lower frailty levels. Future research including beat-by-beat analyses and standardized neurovascular coupling methodology [34] are needed to confirm this. With advancing frailty, may be an impaired ability to further deliver and extract optimal nutrients may manifest as impaired executive function. This aligns with the neural efficiency hypothesis [35] in that less frail persons likely exhibited less neuronal activity as a proportion of the total blood delivered when performing cognitive tasks because they require less effort in comparison to more frail older adults. This more efficient delivery and/or extraction of oxygen among less frail persons may be due to superior cerebrovascular regulation or improved endothelial function of cerebral vessels that warrants future study. The inclusion of multiple imaging techniques provides important mechanistic information regarding frailty and future work examining the link between frailty and cognition is encouraged to consider both fNIRS and transcranial Doppler.

Our study may be limited by a smaller sample size with MCAv, but a statistically significant correlation was observed, indicative of sufficient statistical power to answer our research question. Second, the study was cross-sectional in design and therefore cannot establish causality. Third, the external validity of our data is impacted by our sample composed of cognitively healthy older adults and, therefore, should not be extrapolated to those with cognitive impairments. As well, our sample is primarily comprised of females (n=38/41) and may not apply to males who typically have lower frailty levels at the same age [36]. While we did not specifically aim to include primarily females in research, future recruitment strategies should aim to target both males and females to discern the impact of biological sex on the relationship between cognition and frailty. Fourth, the fNIRS and transcranial Doppler methods do not reflect absolute oxygen values since fNIRS relying on changes and transcranial Doppler unable to provide middle cerebral artery diameter, and thus, actual blood flow. Finally, the ΔO2Hb-peakMCAv ratio is a proxy of the oxygenation of the prefrontal cortex relative to the amount of blood delivered, but not all blood flowing through the middle cerebral artery is directed to the prefrontal cortex. Nevertheless, our study is strengthened by including behavioral, physiological tests, and a rigorous statistical analysis. In addition, this is among the first study trying to understand the frailty-executive function relationship in a group of older adults cognitively healthy.

Among cognitively healthy older adults, higher frailty level was non-linearly associated with worse executive function. More frail older adults exhibited larger cerebral blood velocity responses but similar prefrontal oxygenation during a cognitive task. High frailty level may disproportionately predispose older adults to challenges performing executive function tasks that may indicate a greater risk of progressing to cognitive impairment.

Funding

This project was supported by a CIHR Catalyst Grant - Official Language Minority Communities in Health Research (#472388)

Data Availability

Data from this study is available on reasonable request to said.mekari@usherbrooke.ca.

Declarations

Conflict of interest

The authors declare no competing interests.