Physical and Cognitive Function Assessment to Predict Postoperative Outcomes of Abdominal Surgery

Abstract

Background:

Current evaluation methods to assess physical and cognitive function are limited and often not feasible in emergency settings. The upper-extremity function (UEF) test to assess physical and cognitive performance using wearable sensors. The purpose of this study was to examine the 1) relationship between preoperative UEF scores with in-hospital outcomes; and 2) association between postoperative UEF scores with 30-day adverse outcomes among adults undergoing emergent abdominal surgery.

Methods:

We performed an observational, longitudinal study among adults older than 40 years who presented with intra-abdominal symptoms. The UEF tests included a 20-sec rapid repetitive elbow flexion (physical function), and a 60-sec repetitive elbow flexion at a self-selected pace while counting backwards by threes (cognitive function), administered within 24-hours of admission and within 24-hours prior to discharge. Multiple logistic regression models assessed the association between UEF and outcomes. Each model consisted of the in-hospital or 30-day post-discharge outcome as the dependent variable, preoperative UEF physical and cognitive scores as hypothesis covariates, and age and sex as adjuster covariates.

Results:

Using UEF physical and cognitive scores to predict in-hospital outcomes, an area under curve (AUC) of 0.76 was achieved, which was 17% more sensitive when compared to age independently. For 30-day outcomes, the AUC increased to 0.89 when UEF physical and cognitive scores were included in the model with age and sex.

Discussion:

Sensor-based measures of physical and cognitive function enhance outcome prediction providing an objective practicable tool for risk stratification in emergency surgery settings among aging adults presenting with intra-abdominal symptoms.

Background

The growing aging population in the United States is leading to an increase in emergency surgical procedures among geriatric patients, with abdominal surgery being amongst the most common procedures [1,2]. Furthermore, adults over the age of 65 had the highest incidence rate of emergent hernia repairs, 71.9 per 100,000 person-years in 2010 [2]. Regardless of improvements in surgical and anesthetic practices, geriatric patients undergoing major emergency surgery have a higher risk of adverse health-related outcomes such as surgical complications, excessive length of stay (LOS), hospital readmission, and increased morbidity and mortality [3–6]. Wright et al. recently examined postoperative complications in patients median age 57.0 years undergoing abdominal emergency surgeries and reported that 14.4% experienced surgical complications while in the hospital and 14.7% were admitted to the emergency department (ED) within 30 days [6]. Another study showed that 37% of 400 abdominal surgical patients experienced surgical complications, 20% died within 30 days, and among those older than 85 years old mortality was as high as 51% after one year [3].

Physiological conditions, age, chronic comorbidities, and other risk factors are commonly used as predictors for adverse outcomes among surgical patients. Based on these factors, preoperative risk evaluation tools have been developed to identify high-risk patients for adverse outcomes; for example, the American Society of Anesthesiology (ASA), Physical Status Classification System, and Charlson Comorbidity Index (CCI) [7,8]. However, current risk-stratification methods are mainly focused on limited organ-related diagnosis and can frequently be subjective to the physician decisions [9–11]. Evidence has demonstrated that physical frailty, defined as a low physiological reserve and vulnerability to poor resolution of homeostasis, improves prediction of surgical outcomes in older adults, and independent to the aforementioned risk factors [10–13]. In addition to measures associated to physical function, recent studies have shown that cognitive impairment contributes to an increased risk of excessive LOS, hospital readmission, and death [14,15]. Although recommendations call for screening of older adults for cognitive impairment, preoperative risk stratification that includes objective cognitive measures are not well-established [11]. There is a significant challenge in providing optimal care for geriatric patients undergoing surgery, especially for those who experience cognitive deficits.

Several frailty assessment methods (e.g., frailty index (FI) and comprehensive geriatric assessment (CGA)) have been implemented for risk stratification. Although FI, CGA, and several cognitive screening tools, such as Montreal Cognitive Assessment (MoCA), are valuable evaluation instruments, they are often subjective and time consuming [16], which makes them inapplicable for emergency settings. We have previously validated an upper-extremity function (UEF) test to assess physical and cognitive function, which uses kinematics and kinetics parameters of upper-extremity performance [17–21]. UEF has the potential to be applied in such emergency settings, since it can be completed within 5–10 minutes and it is applicable for bed bound patients.

The UEF physical score is based on slowness, weakness, flexibility, and exhaustion, while performing elbow flexion tasks; UEF physical score was validated based on physical frailty, utilizing the FI as the gold standard [17]. Additionally, UEF provides a cognitive score incorporating dual-task performance, which was validated based on clinically diagnosed cognitive impairment (Alzheimer’s disease and amnestic mild cognitive impairment) in older adults [20,21]. The goals of the current study were to: 1) assess the relationship between preoperative UEF physical and cognitive function scores with in-hospital outcomes; and 2) examine the association between postoperative UEF physical and cognitive function scores with 30-day adverse outcomes. We hypothesized that a worse UEF physical and cognitive function score will be associated with an increased risk of in-hospital and 30-day adverse outcomes. Further, we expected that UEF physical and cognitive function scores would be equal or more predictive of in-hospital and 30-day outcomes compared to demographic characteristics and clinical measures (i.e., ASA, Life-Space Assessment (LSA), and MoCA scores). Finally, we examined the hypothesis that whether adding cognitive function to physical assessment improve prediction of such adverse outcomes.

Methods

Study Design and Participants

We performed an observational 30-day longitudinal study among admitted patients at Banner University Medical Center Tucson (BUMC-T). Aging adults presenting with intra-abdominal symptoms requiring emergency surgery were recruited between August 2018 and October 2019. Participants were approached twice during their hospital stay, once before the surgical treatment within 24 hours of admission (first visit), and once after surgery 24 hours before discharge (second visit). Cerner Millennium, Banner Health’s Power Chart was used to identify eligible patients. Types of abdominal surgery for participant recruitment included emergent cholecystectomy, appendectomy, laparoscopic cholecystectomy, exploratory laparotomy, and any type of hernia repair. The inclusion criteria included: 1) aged 40 years or older; 2) ability to comprehend study instructions; and 3) present with intra-abdominal symptoms including mesenteric ischemia, appendicitis, bowel obstruction, sigmoid volvulus, biliary disease, cholelithiasis, and diverticulitis. Participants were excluded if: 1) diagnosed with a disease associated with severe motor performance deficits (e.g., Parkinson’s disease or stroke); 2) severe upper-extremity disorders (e.g., elbow bilateral fractures or rheumatoid arthritis); and 3) known severe dementia or other major psychiatric disorders. The pilot study was approved by the University of Arizona Institutional Review Board (IRB). Before participating in the study, a written informed consent according to the principles expressed in the Declaration of Helsinki was obtained from each participant [22]. Participating patients were informed about the benefits, possible negative consequences, and their right to withdraw from the study at any time without any repercussion.

Clinical Measures

The MoCA questionnaire was administered during the second visit to assess a range of cognitive abilities [23], and the scores were adjusted for the level of education and age based on previous work [24]. For participants that were not able to complete the MoCA assessment, the Eight-item Interview to Differentiate Aging and Dementia (AD8) was administered by a trained researcher. LSA was administered to measure the participants’ mobility and degree of independence [25]. ACS The National Surgical Quality Improvement Program (NSQIP) was used to calculate the estimated risk of outcomes and expected LOS [26]. A thorough review of the participants’ electronic medical chart was done to collect individual demographic characteristics, ASA classification, type of surgical procedure, and comorbidities. These characteristics were used as input variables of the ACS NSQIP risk estimation program [26]. Functional capacity was assessed by each researcher by asking participants to flex and extend their arms simulating the UEF test. Failure to complete task, patients were disqualified from the study.

UEF Measures

The UEF test was used to assess physical and cognitive function. Two wearable motion sensors (LEGSys tri-axial gyroscope sensors, sampling frequency = 100 Hz, BioSensics LLC, Cambridge, MA) were applied to the wrist and upper-arm of the dominant arm using elastic bands to measure elbow angular velocity. If a dominant arm was not available (e.g., because of intravenous line or bruising), then the opposite arm was used. Participants were asked to perform the series of the function tests twice, once during the first visit and once during the second visit. Each series of UEF testing involved one 20-second rapid elbow flexion for the physical function test, and one 60-second self-selected normal pace elbow flexion for the cognitive function tests. All protocol assessments were administered by trained researchers and, for consistency, all participants were given the same instructions.

The physical and cognitive function scores for a given participant were determined by the summation of points given based on UEF single and dual-task performance, respectively. Where points were assigned based on variable comparisons to previously determined ranges. The UEF physical function score (range: resilient=0 - extremely frail=100) was calculated through the fast elbow flexion for 20 seconds, while participants repetitively, fully extended and flexed their elbow as fast as possible. The upper-extremity kinematics and kinetics were measured to obtain outcome parameters representing slowness, weakness, exhaustion, and flexibility [17,18]. Such parameters included; 1) speed (mean value of the elbow angular velocity range), 2) flexibility (mean value of the elbow flexion range), 3) moment (mean moment on elbow estimated from moment of inertia, elbow angle, and gravitational force of arm, 4) speed variability (coefficient of variation of the angular velocity range), 5) speed reduction (difference in the angular velocity range between the last and the first five seconds of elbow flexion as a percentage of initial angular velocity range), and 6) flexion number (number of elbow flexion) [17].

The UEF cognitive function score (range: cognitive normal=0 - cognitively impaired=100) was assessed based on motor function variability within dual-task performance, which involved UEF motor task and a cognitive task of counting backwards by three (dual-task). During the test, each participant performed a 60-second self-selected pace and consistent elbow flexion while counting backwards. We decided this based on previous work, where self-selected pace while performing elbow flexion provided better prediction of cognitive status than rapid flexion [27]. Cognitive function scores were calculated utilizing UEF parameters, including: 1) range of motion variability (coefficient of variation of the elbow range of motion - ROM), 2) flexion variability (variability between flexion cycles based on coefficient of variation of peak velocity values), and 3) flexion number. These parameters were selected based on our previous work to represent accuracy and agility in performing the motor function during dual-tasking [20]. More details can be found in previous work, regarding calculation of cognitive score for dual-task [20].

Adverse Outcomes

In-hospital outcomes identified in this study included excessive LOS, after surgery complications (falls, increase in acuity to ventilation/intubation, pressure ulcers, intensive care unit (ICU) visit, or similar complications), and death as variables associated with frailty and aging [26, 28–29]. The actual LOS was obtained from each participant’s medical chart, while the expected LOS was calculated utilizing the ACS NSQIP model [26]. Through chart reviews, all risk factors including ASA classification were accounted for when utilizing the ACS NSQIP model. This model enabled the acquisition of the predicted LOS, and subsequently the difference between actual LOS and predicted LOS was considered as a dichotomous measure of excessive LOS. A dichotomous variable (1 = in-hospital adverse outcome, 0 = no in-hospital adverse outcome) indicated whether any of the above measures occurred, including excessive LOS, surgical complication, or death.

A 30-day follow-up phone call was administered by trained research staff to determine adverse 30-day outcomes for each participant, which included: readmissions defined at 30 days and 30-day death [26]. Readmission included ED visits because of reasons associated with the surgical procedure. Similar to in-hospital outcomes, a single dichotomous variable was created to indicate the presence of adverse 30-day outcomes, including death or readmission within 30-days.

Statistical analysis

We used univariate analysis of variance (ANOVA) to evaluate the differences in demographic characteristics and clinical measures between two groups with and without adverse outcomes for in-hospital and 30-day outcomes; chi-square (χ²) tests were used to assess differences in sex among groups. Multivariable ANOVA models were used to assess differences in UEF scores between groups with and without adverse outcomes. The analysis repeated once for in-hospital and once for 30-day outcomes. In each multivariable model, the dichotomous variable of adverse outcome plus age, sex, and BMI were considered as independent variables.

Further, we evaluated associations between preoperative UEF scores and in-hospital outcomes using multiple logistic regression, which consisted of the following: In-hospital outcome (adverse or not adverse) as the dependent variable, preoperative UEF physical and cognitive scores as hypothesis covariates, and age and sex as adjuster covariates. A stepwise procedure based on Akaike information criterion (AIC) values was performed to select the predictive covariates. For each predicting model, the area under the curve (AUC) with 95% CI was calculated using receiver operator characteristics (ROC) curves. Similar approach was used to develop logistic models to determine the association between postoperative UEF scores (as well as clinical measures including ASA, MoCA, LSA, and ACS NSQIP Score) with 30-day outcomes.

Finally, we used Pearson correlation to assess the relationships between UEF physical scores, UEF cognitive scores, NSQUIP scores, ASA class, LSA score, and MoCA. All data was analyzed using the JMP statistical program (version 14.2.0, copyright 2018 SAS institute Inc), and statistical significance was indicated when p<0.05.

Results

Participants

A total of 100 patients were recruited for this study (age=61.3±11.3; BMI = 28.38±8.11; and 49% male). Among 100 participants, 90 and 50 participants completed the preoperative and postoperative UEF assessment, respectively, and 11 used the non-dominant arm; 50 (50%) participants were lost to follow-up. Consequently, only those that completed the UEF assessments were included in the subsequent data analysis. Among 90 participants within the first visit, 51 experienced at least one in-hospital outcome as shown in Table 1. Among those with adverse in-hospital outcomes, forty-four (86%) had an excessive LOS, five (10%) required ICU visit, and there were two (4%) in-hospital deaths (Table 1). Out of 50 participants that completed the second visit, 14 (28%) were readmitted to the hospital 30 days after being discharged (Table 1). We were unable to determine readmission rates for those lost to follow-up; even though, deaths were assessed in this group through electronic medical chart, no adverse outcomes were determined.

Table 1.

Adverse outcomes by group

Parameter	In-hospital Adverse Outcome (n=90)	Post-discharge Adverse Outcome (n=50)
Number, n (% of total)	51 (57%)	14 (28%)
Excessive LOS, n (% of group)	44 (86%)	-
In-hospital Death, n (% of group)	2 (4%)	-
In-hospital Complications	11 (25%)	-
Falls, n (% of group)	0 (0%)	-
Pressure Ulcer, n (% of group)	2 (4%)	-
Ventilation, n (% of group)	2 (4%)	-
Higher Acuity in ICU, n (% of group)	5 (10%)	-
30-Day Death, n (% of group)	-	0 (0%)
30-day Readmissions, n (% of group)	-	14 (28%)

LOS: length of stay, n: total, ICU: intensive care unit

Association between UEF Scores and In Hospital and 30-Day Outcomes

All demographic data for each in-hospital and 30-day outcome group, including physical and cognitive frailty scores are reported in Table 2 and 3, respectively. There was a significant difference in age and physical and cognitive function scores (p<0.05) for in-hospital outcome groups. Similarly, there were significant differences in UEF physical and cognitive scores among 30-day outcome groups (p<0.01), but not for age (p=0.15, Tables 2 and 3). Preoperative UEF physical and cognitive scores were 42% and 58%, respectively, larger among those with adverse in-hospital outcomes compared to those who did not experience any adverse outcomes during their hospital stay (Table 2, Figure 1). Similarly, postoperative physical and cognitive scores were 41% and 107% larger among those with adverse 30-day outcomes, compared to those without.

Table 2.

Mean values and standard deviation of demographic data and UEF scores for patients with and without in-hospital adverse outcomes.

Variable	In-hospital Adverse Outcome (n = 51)	In-hospital No Adverse Outcome (n = 39)	p-value
Male, n (% of the group)	27 (30%)	14 (15.56%)	0.11
Age, years (SD)	63.67 (9.95)	57.59 (12.71)	0.01
Stature, cm (SD)	169.01 (11.86)	167.84 (12.65)	0.65
Body mass, kg (SD)	79.34 (18.52)	83.60 (23.23)	0.34
BMI, kg/m2 (SD)	28.21 (8.42)	29.55 (7.19)	0.43
White, n (% of the group)	39 (43.33%)	26 (28.89%)	0.33
Physical Function, 0–100 (SD)	53.76 (24.76)	37.87 (21.07)	<0.01*
Cognitive Function, 0–100 (SD)	39.95 (35.89)	25.27 (27.62)	0.05*

SD: standard deviation, BMI: body mass index,

^*significant difference at p<0.05

Table 3.

Mean values and standard deviation of demographic data and UEF scores for patients with and without 30-day post discharge adverse outcomes.

Variable	30-Day Adverse Outcome (n=14)	30-Day No Adverse Outcome (n=36)	p-value
Male, n (% of the group)	9 (18)	19 (38)	0.47
Age, years (SD)	58.50 (11.37)	63.56 (10.78)	0.15
Stature, cm (SD)	170.47 (13.12)	169.02 (12.93)	0.72
Body mass, kg (SD)	80.48 (21.02)	83.42 (19.24)	0.64
BMI, kg/m2 (SD)	27.99 (7.95)	29.49 (7.78)	0.55
White, n (% of the group)	10 (20)	29 (58)	0.49
Physical Function, 0–100 (SD)	75.93 (14.83)	53.78 (27.46)	0.01*
Cognitive Function, 0–100 (SD)	57.69 (33.16)	27.80 (29.81)	<0.01*

SD: standard deviation, BMI: body mass index,

^*significant difference at p<0.05

Figure 1.

Differences in age and UEF scores between: A) groups with in-hospital adverse outcomes and without any in-hospital adverse outcomes, and B) groups who experienced 30-day adverse outcomes and experienced no 30-day adverse outcomes. Mean values and standard errors are presented.

Figure 2 and Table 4 show the ROC analysis results from logistic models for predicting adverse in-hospital and 30-day outcomes. For in-hospital outcomes, the AUC for age and sex separate univariate models were 0.65 and 0.59, respectively. Including both UEF physical and cognitive scores into the same model to predict in-hospital outcomes, improved the AUC to 0.76. Based on these models, sensitivity of in-hospital outcome prediction was improved by 63%, when physical and cognitive function parameters were added to demographic information for predicting in-hospital outcomes (Table 4). Likewise, the AUC for age and sex was 0.63 and 0.54 for each independent model predicting 30-day outcomes. For 30-day outcomes, the AUC increased to 0.89 when UEF physical and cognitive scores were included in the model with age and sex (Table 4). Sensitivity of 30-day outcome prediction was improved by 96%, when physical and cognitive scores were added to 30-day outcome prediction (Table 4). Results from separate logistic regression models for each physical and cognitive function score revealed that each were significantly associated with in-hospital and 30-day outcomes, where physical function had the greatest effect (Table 5).

Figure 2.

ROC Curves for predicting A) in-hospital outcomes and B) 30-day outcomes using demographic characteristics and UEF scores.

Table 4.

Nominal logistic models for in-hospital and 30-day outcomes.

In-hospital outcome prediction
Parameter	Specificity	Sensitivity	AUC
Age	0.75	0.43	0.65
Sex	0.75	0.37	0.59
Physical Function	0.75	0.53	0.69
Cognitive Function	0.75	0.38	0.61
Age, Sex, Physical and Cognitive Function	0.75	0.70	0.76
30-day post-discharge outcome prediction.
Parameter	Specificity	Sensitivity	AUC
Age	0.75	0.46	0.63
Sex	0.75	0.29	0.54
Physical Function	0.75	0.5	0.74
Cognitive Function	0.75	0.69	0.75
Age, Sex, Physical and Cognitive Function	0.75	0.90	0.89

AUC: area under the curve, Sensitivity is presented for defined a defined specificity of 0.75

Table 5.

Multivariable ordinal logistic prediction models for in-hospital and 30-day outcomes

In-hospital outcome prediction model (Sensitivity = 0.70 ; Specificity = 0.75 ; AUC =0.76)
	Estimate	Standard Error	χ2	p-value		Lower 95%	Upper 95%
Intercept	4.10	1.48	7.72	<0.01*		1.35	7.19
Age	−1.10	0.81	1.84	0.18		−2.74	0.47
Sex	−0.39	0.25	2.49	0.11		−0.89	0.09
Physical Function	−0.02	0.01	2.95	0.09		−0.04	<0.01
Cognitive Function	−0.05	0.02	5.26	0.02*		−0.10	−0.01
30-day outcome prediction model (Sensitivity = 0.90; Specificity = 0.75 ; ROC = 0.89 )
Intercept	−3.81	2.92	1.7	0.19	−10.18		1.67
Age	−0.03	0.02	2.11	0.15	−0.08		0.01
Sex	−6.32	2.33	7.38	0.01*	−11.97		−2.49
Physical Function	0.16	0.06	7.28	<0.01*	0.06		0.31
Cognitive Function	−0.48	0.53	0.83	0.36	−1.61		0.51

ROC: Receiver operating characteristic,

^*significant difference at p<0.05

UEF Scores vs Current Risk Assessments

Results showed that the AUC for the model including ASA classification, age, and sex was 0.67 for predicting in-hospital outcomes (more information on clinical measure analysis can be found in supplementary material II). The AUC only increased slightly to 0.69 when both the ASA classification and MoCA score were included in the same predicting model. The AUC for predicting 30-day outcomes improved slightly to 0.74 when the LSA score were added along with age and sex.

When predicating 30-day readmission, the UEF cognitive function model had the highest AUC with 0.74, followed by UEF physical function and ACS NSQIP independent models with an AUC of 0.69 and 0.61, respectively. When adjusted for age and sex, the AUC for the UEF model (physical and cognitive scores) increased to 0.79, while the ACS NSQIP model only improved to 0.66.

Discussion

In-Hospital Adverse Outcomes

As hypothesized, our study showed that worse preoperative UEF physical and cognitive performances were observed in participants who experienced in-hospital adverse outcomes compared to those who did not (Figure 1). Additionally, UEF physical and cognitive scores were independent predictors of excessive LOS, surgical complications (e.g., pressure ulcers), and death in this cohort of patients aged 40 years or older undergoing emergency abdominal surgery. ASA classification remains one of the most common risk stratification methods in surgical patients; however, it is often inconsistent and inaccurate for predicting the risk of in-hospital adverse outcomes [30–34]. For instance, Stonelake et al. reported that the ASA classification overestimated mortality risk following a laparotomy procedure [32,34]. Across the same sample of participants, our findings suggest a stronger association between UEF parameters and in-hospital outcomes with an AUC of 0.83, compared to ASA with an AUC of 0.67 (Table 4 and supplementary material II). The observed advantage of UEF testing over ASA may be because preoperative risk scores, such as ASA classification, do not incorporate individualized physiological or cognitive components. The UEF test allows for individualized assessment based on each patient’s functional status, regardless of comorbid conditions.

Post-Discharge Adverse Outcomes

As predicted, postoperative UEF physical and cognitive scores were 41% and 58% higher, respectively, among participants who experienced adverse outcomes within 30 days of discharge (Table 3). Multiple frailty assessment methods have been recognized to improve prediction of discharge adverse outcomes [33, 35]. LSA, for example, has been utilized to measure frailty based on mobility function. According to Fathi et al., LSA scores were associated with greater odds of post-discharge outcomes, such as hospital readmission and death [35]. Furthermore, Makary et al. showed that AUC improved from 0.68 to 0.72 when frailty was incorporated into the predictive model for 30-day mortality with sex, race, and modified FI [29]. Based on our results, AUC curves demonstrated an improved accuracy of 20% in predicting post-discharge outcomes, when the predictive model including UEF physical and cognitive function scores, age, and sex was compared to the similar model including LSA scores (Table 4, supplementary II). Nevertheless, more common frailty measures, such as the FI, were not feasible to preform among the current sample of bedbound aging adults with poor physical condition; therefore, we were not able to compare the outcome prediction validity of UEF with comparison with the Fried phenotype and other frailty measures requiring gait assessment. The UEF test can be administered within minutes of the medical procedure, which may provide a more precise assessment for each patient compared to risk evaluation tools that requires subjective questionnaires such as LSA.

Physical vs. Cognitive Function

Recent evidence suggests that the co-occurrence of physical and cognitive function improves accuracy for detecting patients at risk for developing postoperative adverse outcomes [15, 36–37]. Physical frailty is associated with executive function and attention domains of cognitive impairments and, a decline in executive function domain is related to poor physical performance during dual-task activities [38]. Additionally, cognitive impairments share a number of casual mechanisms with physical performance, including hormones, nutrition, chronic inflammation, and cardiovascular diseases [39]. Although kinematic parameters were incorporated into each of the UEF physical and cognitive indexes, due to the type of assessment for each measure (e.g., strength vs. accuracy of motor task), there were weak but significant correlations between UEF physical and cognitive scores (r²=12% and r²=15%, p<0.01, for in-hospital and 30-day outcomes respectively). This in turn suggest simultaneous occurrence of comorbid physical and cognitive impairments in this cohort as we observed significant correlation between these two scores; however, there are some distinct measures incorporated within each test and the difference between the level of physical and cognitive deficits may be different between participants. Accordingly, cognitive impairment assessment here improved the predictive validity of frailty for adverse health outcomes [40]. In our study, the AUC improved by approximately 15% on average when both UEF performance scores were included in the predictive model compared to physical function alone.

The growing significance of physical and cognitive decline as predictors for poor health outcomes has led to the application of their assessments in clinical settings, especially among older adults. Previous studies have used separate and independent methods to assess physical and cognitive performance [41–44]. Multiple assessment tools exist to measure physical function such as functional gait assessment (FGA), Short Physical Performance Battery (SPPB), Timed Up and Go (TUG), and 6/10-Minute Walk Test (6/10MWT). These frequently used tools are based on mobility performance, measured by gait speed, endurance, and balance tests, where the patient is required to complete a series of movement activities (e.g., walking a certain distance/time or getting up a chair) [45–47]. FGA, TUG, and 6MWT are inexpensive, quick to administer in clinical settings, and easy to interpret [48, 49]. However, these tools are less accurate and inconsistent compared to more complex methods such as SPPB that incorporate multiple physical performance measures [48, 49]. Nevertheless, SBBP is time-consuming and requires skilled personnel to properly administer the test and can be limited to non-bedbound patients [46]. UEF assessment can, however, be easily applied to patients with restricted mobility function within minutes.

Emerging research utilizes dual-task performance, which requires to complete motor and cognitive tasks simultaneously, to assess cognitive function [50–52]. For example, patients are asked to walk for 10 min, while completing a cognitive task (e.g., counting backwards or naming the months backwards) [50–52]. However, these major motor assessments could exclude frail patients or those with lower-extremity disabilities. There are other common screening tools used for cognitive function such as Trail Making Test A/B, Symbol Digit Modality Test (SDMT), Mini Mental State Examination (MMSE), and MoCA [41–44]. Although these cognitive screening tools are time efficient and largely validated; they can be biased. Our study integrates physical and cognitive function using a quick and practical test to provide a more robust risk stratification tool for surgical patients.

Clinical Significance

Postoperative complications can lead to severe health consequences, such as disability, functional and cognitive decline, and premature death [53]. However, there is a significant challenge in providing optimal value-based care for adults in emergency settings. Over 50% of adults 65 and older have three or more chronic diseases [54–56] with higher rates of exacerbation, hospitalization, mortality, and healthcare utilization [57]. However, the value of objective physical function assessment and added burden of cognitive impairment for predicting health complications remains understudied. Previous studies have suggested that UEF assessment could predict adverse outcomes after a major emergency surgery by assessing physical function [58]. In continuation of our research, we proposed a multidimensional approach to simultaneously assess physical and cognitive function to improve adverse outcome prediction among surgical patients.

The UEF test is easily performed and could be integrated into clinical care to be tracked over time, similar to blood pressure measurement. From a clinical viewpoint, UEF performance can inform procedures of planning service provisions and resource allocation, which will contribute to the health-care costs for both during the hospital stay and after discharge periods [59,60]. Consequently, geriatric patients undergoing surgery can benefit from individualized evaluation, including physical and cognitive function, for reduction of postoperative risk for in-hospital and 30-day outcomes.

Limitations

We acknowledge some limitations within this study. First, we were not able to include 50% of participants for 30-day post-discharge analysis due to missing data. There were several obstacles for postoperative measurement acquisition, one of which involved the nature of a second visit that must be completed after surgery but before discharge, leading to limited available time for postoperative measurements. Other obstacles included; 1) participants were discharged right after surgery, 2) participants were discharged at night when trained researchers were not available, 3) participants were under anesthesia, 4) participants were not completely recovered, and 5) participants refused to continue with the second part of the study. Nevertheless, incorporating UEF assessment as a routine measure in surgical settings can optimize completion rate. Secondly, data loss occurred during 30-day phone call and self-report follow-ups due to disconnected phone lines, insufficient alternative contacts, and email surveys with no response. Yet, we were able to cross-reference 30-day outcomes for patients who reported to be admitted to BUMC by completing retrospective chart review. Initial assessment of this loss-to-follow-up group was not statistically different compared to the group with follow-up (Figure 3). For future research, integrating retention strategies such as offering monetary incentives for completing follow-up could reduce loss to follow-up rates. Moreover, there was missing information in the electronic medical chart such as time of surgery or date of discharge that was necessary to calculate variables of interests and ACS NSQIP estimated risk scores. Several participants were not able to complete the MoCA assessment due to vision issues; however, the AD8 was administered for them. Third, we were not able to perform any analysis on discharge disposition because most participants were discharged home without any assistant. Only two participants were discharged to a nursing home and assisted living facility, and their postoperative UEF scores were not collected for reason previously mentioned. However, previous studies have shown an association between UEF scores and discharge disposition [58,61].

Figure 3.

Differences in UEF and normalized clinical scores between participants lost to follow up and those who were not.

Lastly, our study excluded an important population that might benefit from UEF testing, including individuals with known severe dementia or other psychiatric disorders. However, we still included those with mild to moderate cognitive impairments, since the UEF cognitive score has previously been validated among older adults with early-stage Alzheimer’s disease and other types of dementia [20, 62]. Nevertheless, the UEF test could not be administered in patients with severe cognitive impairments since such patients would not be able to comprehend the testing instructions.

Future Implementation of UEF Test in Surgical Settings

Implementing a novel approach in a clinical setting always comes with some obstacles, especially in emergency and surgical environments. Initially, individuals who could potentially administer the UEF test were trained. Even though, the UEF test requires minimal guidance and expertise for proper conduction, it still requires a five-to-10-minute training session. Of note, due to its objective assessment platform, nurses and physicians in emergency and critical care can easily get trained and perform the protocol for both pre- and post-operative measurement. Therefore, the UEF test could efficiently be implemented as a generic tool for assessing physical and cognitive function, with potentials to be substitute with more extensive assessment tools currently implemented (e.g. MoCA and Six-Minute Walk Test), especially for pre-operative situation where the time limitation is a critical concern. Secondly, the availability and accessibility of new medical equipment could be a concern for some hospitals and clinics. The UEF equipment uses low-cost wearable and reusable sensors which could be available as low as $200 (17), however, it might not be accessible for small and resource-limited settings. We acknowledge there are common barriers to consider when using sensor-based technology such as calibration, reliability, loss of signal, or signal processing among others [63]. However, there has been previous research assessing the accuracy and reliability of UEF sensor data [17–21,58,62], which suggest UEF can be a suitable alternative for risk assessment of surgical patients in busy clinical settings

Conclusion

Pre-existing physical and cognitive deficits contribute to postoperative complications among the aging population. In emergency settings, effective risk assessments should be implemented among geriatric patients. We provide supportive evidence that sensor-based objective measures for risk stratification can be utilized in a surgical setting among aging adults to guide further preoperative management and targeted care to optimize patients’ outcomes.