Frailty is associated with postoperative delirium but not postoperative cognitive decline in older non-cardiac surgery patients

Abstract

Background:

Postoperative cognitive dysfunction (POCD) and delirium are the most common perioperative cognitive complications in older adults undergoing surgery. A recent study of cardiac surgery patients suggests that physical frailty is a risk factor for both complications. We sought to examine the relationship between preoperative frailty and postoperative delirium and preoperative frailty and POCD after major noncardiac surgery.

Methods:

We performed a prospective cohort study of patients greater than 65 years-old having major elective noncardiac surgery with general anesthesia. Exclusion criteria were preexisting dementia, inability to consent, cardiac, intracranial or emergency surgery. Preoperative frailty was determined using the FRAIL scale, a simple questionnaire which categorizes patients as robust, prefrail, or frail. Delirium was assessed with the confusion assessment method for the intensive care unit (CAM-ICU) twice daily, starting in the recovery room until hospital discharge. All patients were assessed with neuropsychological tests (California Verbal Learning Test II, Trail Making Test, subtests from the Wechsler Adult Intelligence Scale, Logical Memory Story A, Immediate and Delayed Recall, Animal and Vegetable verbal fluency, Boston Naming Test, and the Mini-Mental Status Exam) prior to surgery and at 3 months afterwards.

Results:

178 patients met inclusion criteria; 167 underwent major surgery and 150 were available for follow up 3 months after surgery. The median age was 70 years old. 31 patients (18.6%) tested as frail, and 72 (43.1%) prefrail prior to surgery. After adjustment for baseline cognitive score, age, education, surgery duration, American Society of Anesthesiology (ASA) status, type of surgery, and gender, patients who tested frail or prefrail had an estimated 2.7 times the odds of delirium (97.5% confidence interval 1.0– 7.3) when compared to patients who were robust. There was no significant difference between the proportion of POCD between patients who tested as frail, prefrail or robust.

Conclusions:

After adjustment for baseline cognition, testing as frail or prefrail with the FRAIL Scale is associated with increased odds of postoperative delirium, but not postoperative cognitive dysfunction after noncardiac surgery.

INTRODUCTION:

As life expectancy increases, the number of older adults undergoing surgical procedures has also increased. Postoperative delirium is the most common postsurgical complication in older adults and occurs in 14–60% of older surgical patients.^1–3 Risk factors for delirium and postoperative cognitive dysfunction (POCD) include preoperative cognitive status and medical comorbidities which are sensitive but not always specific for risk in older surgical patients. Physical frailty describes the impact of medical comorbidities on a patient’s overall state of health, patients who are frail exhibit an increased vulnerability to stress.^4,5 Frailty assessment scores could augment perioperative provider’s understanding of an older surgical patient’s risk because it includes features not captured in the American Society of Anesthesiology (ASA) physical status such as tiredness, reduced ability to walk distances or climb stairs, and unintentional weight loss.^4–6

Studies of community dwelling older adults have shown that physical frailty can be associated with increased incidence of cognitive impairment and delirium.^7–11 Recently, a study of cardiac surgery patients demonstrated that frailty was strongly associated with postoperative delirium, and marginally with longer term cognitive dysfunction.¹² Physical frailty and cognitive impairment clearly overlap, though they are not synonymous. Patients with cognitive frailty (frailty and concurrent cognitive impairment) have increased white matter hyperintensities when compared to patients without cognitive frailty.¹³ Worse performance on executive and working memory neuropsychological tests has been found in geriatric outpatient patients with greater frailty.¹⁴ However, no one has demonstrated whether frailty is associated with delirium and longer term postoperative cognitive dysfunction in noncardiac surgery.

To address whether preoperative physical frailty is associated with delirium and postoperative cognitive dysfunction in older major noncardiac surgery patients, we performed a secondary analysis of a prospective cohort study in a group who had frailty screening with the FRAIL Scale, delirium screening, and in-depth cognitive testing before and after surgery. We hypothesized that frailty would be related to cognitive decline 3 months after surgery and to postoperative delirium. We examine baseline cognition and cognitive decline by composite score and also by cognitive domain, acknowledging prior to work which emphasizes a relationship between frailty and executive function. Better insight into the relationship between preoperative frailty and postoperative cognitive complications may elucidate important areas for future study and interventions.

METHODS

Protocol

The study is a secondary analysis of a prospective cohort (Optimizing Postoperative Cognitive Dysfunction in the Elderly- PRESERVE, ClinicalTrials.Gov: NCT02650687, The Icahn School of Medicine, New York, NY). The study was approved by the Mount Sinai Institutional Review Board and all subjects provided written informed consent. The cohort consisted of English-speaking patients over the age of 65 years having elective “major” noncardiac surgery with general anesthesia; major was defined as a procedure requiring at least a 2-day period of postoperative inpatient care. Patients with a diagnosis of dementia (either listed in the medical record or reported by the patient) were excluded, however we did not exclude patients with mild cognitive impairment, subjective cognitive complaints, nor was there a minimum cognitive performance required to participate. We excluded patients with recent cardiac or intracranial procedures, inability to consent for themselves, or undergoing emergency surgery. Potentially eligible patients were identified daily through the computerized scheduling system at the Icahn School of Medicine at Mount Sinai. The study team worked with surgeons to obtain assent to approach eligible patients who were then contacted over the phone and screened for interest and inclusion/exclusion criteria. Candidates who consented to the study received a home visit for preoperative cognitive testing performed by trained research coordinators at least 24 hours and up to 30 days before surgery.

The primary purpose of the parent PRESERVE study was to 1) look at cognitive dysfunction after surgery by domain and 2) examine the relationship between POCD and functional impairment.

Frailty screening (FRAIL Score)

All patients were assessed for frailty using the FRAIL Scale by trained research coordinators. This scale was developed by the Geriatric Advisory Panel of the International Academy of Nutrition and Aging as a screening tool that could be administered by many types of health professionals or caregivers and has been validated in a diverse group of patients.^15,16 The FRAIL Scale consists of five domains: Fatigue, Resistance, Ambulation, Illnesses, Loss of weight. “Fatigue” for the scale is defined as subjective feeling of tiredness over past 4 weeks “most or all” of the time. “Resistance” is scored positive if the patient has difficulty climbing 10 steps without an assist device or stopping to rest. “Ambulation” is scored positive if the patient has difficulty walking 2 blocks independently. The presence of “Illnesses” is scored positive if the patient has 5 or more co-morbidities: hypertension, diabetes, cancer, asthma, chronic lung disease, heart attack, congestive heart failure, angina, arthritis, stroke, and kidney disease. “Loss of Weight” is scored positive if the patient has lost more than 5% of bodyweight over past year. A score of 1 is assigned to each positive domain. A score of 0 is categorized as robust, 1–2 as prefrail, and ≥ 3 as frail. The sensitivity of the FRAIL scale compares well to other frailty scores and has been associated with mortality and complications.^17,18

Cognitive testing

All patients were assessed with a full neuropsychological battery prior to surgery at the same time the FRAIL scale was administered (within 30 days but at least 24 hours prior) and again at 3 months after surgery. We used the California Verbal Learning Test (CVLT-II) and the Uniform Dataset Battery (UDS) from the Alzheimer’s Disease Research Center.¹⁹ UDS tests include: Trail Making Test, subtests from the Wechsler Adult Intelligence Scale (WAIS-R Digit Symbol and WAIS-III Digit Span), Logical Memory Story A, Immediate and Delayed Recall, Animal and Vegetable verbal fluency, Boston Naming Test, and the Mini-Mental Status Exam (MMSE). Executive function was tested with the Trail Making Test, subtests from the Wechsler Adult Intelligence Scale (WAIS-R Digit Symbol and WAIS-III Digit Span), and the Boston Naming Test. Memory and language was tested with category fluency (Animal and Vegetable verbal fluency) and logical memory (Immediate and Delayed Recall). Testing was performed by trained personnel in a quiet environment, often the patient’s home.

Delirium Assessment

All patients were assessed with the Confusion Assessment Method validated for the intensive care unit detection of delirium (CAM-ICU) by trained research personnel twice each day throughout their hospitalization.²⁰ The CAM-ICU was chosen due to its sensitivity and specificity in identifying delirium in both verbal and nonverbal patients²⁰ and its prior use in the post-anesthesia recovery unit.²¹ The CAM-ICU consists of four criteria: acute fluctuating mentation (I), inattention (II), disorganized thinking (III) and altered level of consciousness (IV). Patients that meet criteria I, II and either III or IV are diagnosed with delirium and labeled as CAM-ICU positive.²⁰ Altered level of consciousness was determined by the Richmond Agitation and Sedation Scale.²⁰ The same research personnel administered the CAM-ICU to each patient in the presurgical and postsurgical time points. Separate research personnel reviewed the scores to determine and adjudicate the presence of delirium. All data was double entered into the database to ensure reliability.

Anesthesia and surgical events

Anesthesiologists received minimal directions regarding the choice of general anesthetic, as previous literature has not found a difference between different anesthetic techniques (e.g. total intravenous anesthesia and gas anesthetics) and cognitive outcomes. The protocol requested that the clinical anesthesia team avoid the use of midazolam, nitrous oxide, ketamine, and etomidate when possible as the parent study investigated the effects of processed EEG and these drugs are not accounted for in the device’s proprietary algorithm. We recorded major intraoperative variables such as blood pressure trends, anesthetics and analgesic administered, and surgical and anesthesia duration. The majority of patients were extubated at the end of surgery and transferred to the post-anesthesia recovery unit or intensive care unit.

Analysis

Using parametric (Chi – square, Fisher’s exact) or nonparametric (Wilcoxon rank sum, Kruskal Wallis) statistical methods as appropriate we compared the frail, prefrail and robust groups on the following baseline characteristics: demographics characteristics, baseline anesthesia risk (age, gender, the ASA Physical Status) and the Hospital Anxiety and Depression Scale (HADS).¹⁰ Continuous variables are presented as mean ± standard deviation (SD) and categorical variables as count (%). Our threshold for statistical significance was p< 0.05. However, for both postoperative delirium and POCD, we used alpha=0.025 as the significance level to adjust for multiple comparison (two outcomes) and we reported the corresponding 97.5% confidence interval.

Analysis of cognition

For each cognitive test, the individual scores obtained at baseline and 3 months after surgery were normalized to the cohort mean to create z-scores (the difference between the individual score and the baseline average was divided by the baseline standard deviation) as previously reported for cognitive tests in surgical patients.^22,23 Prior to normalization, logarithmic transformation was performed for Trail Making Test and the sign was corrected so that when change scores were created, higher values were always “better.” The average of the normalized baseline z scores across all individual tests was then used to indicate the overall baseline cognitive performance. Group comparison of changes in cognitive score from baseline to 3 months was adjusted for their respective baseline score using least squares regression. Cluster analysis was used to perform grouping of the cognitive tests to create domain scores in the manner presented by Price et al.²⁴ The variable cluster analysis (VARCLUS) procedure in SAS (v 9.4) was implemented to find clusters of cognitive tests that were correlated as possible within the group and not with tests in other clusters. The algorithm was binary and divisive –i.e. at the beginning of the analysis all cognitive tests start in one cluster and splitting continues until a stopping criterion (based on eigenvalues) is reached. Factor analysis generated factor structures similar to the clusters generated by the cluster analysis procedure.

Identification of POCD

We calculated the change in z-score from baseline to 3 months. The effect of learning from repetitive testing (practice effect) was calculated as the average of the domain-specific change scores across all individuals. The difference between an individual’s change score combined with the practice effect was compared to the average score of our population at baseline. Patients with a decline of 1 or more SD were defined as having POCD. We chose to use the standard deviation of the study cohort at baseline to define the threshold for POCD because the group was lower scoring prior to surgery in almost every test relative to published normative data in community-dwelling nonsurgical patients¹⁰ therefore we did not pursue comparison to non-surgical controls (Supplemental Table 1).¹⁹

Analysis of Post-Operative Delirium

A logistic regression was performed to examine delirium and its relationship to frailty adjusted for covariates with p<0.2 and apriori delirium risk factors. We used backward selection, the initial variables in the model included frailty (frail/prefrail, robust), baseline cognitive score, age, education, surgery duration, ASA status, type of surgery, and gender.

Sample Size Justification

The sample size for the parent study focused on the relationship between postoperative cognitive dysfunction and functional impairment. Of the 150 patients who had 3-month follow-up, 68.7% were frail or prefrail and the incidence of POCD and delirium is 12.3% and 22.1%, respectively. For demonstration, we assume that an odds ratio of 2 is clinically meaningful risk for POCD, and 2.5 for delirium between frail and robust patients. Utilizing a two-group continuity corrected X² test with a 0.050 two-sided significance level, we would have 20% power to detect the odds ratio of 2 for POCD and 46% for postoperative delirium. Using a 0.025 two-sided test (corrected for 2 outcomes), the power is 14% and 36%, respectively.

RESULTS

178 patients met inclusion criteria and consented to this study, and 167 underwent surgery. Cause of exclusions prior to surgery included surgery rescheduled at a different institution (n=3) or decision not to proceed with surgery (n=3). Follow up occurred 3 months after surgery, 150 patients were available and completed the 3-month testing. The most common cause of dropout at 3 months were non-related medical issues (n=7), followed by socioeconomic issues (n=5). Five patients were lost to follow up.

Table 1 summarizes the demographics and baseline cognitive test scores of patients and a comparison of those who tested as frail and those who did not. Of the 167 total patients, 31 patients (18.6%) tested as frail, 72 as prefrail (48.1%). The sample had a median age of 70 years with the median age of the frail cohort being similar (71 years). The distribution of race in the cohort was 74.7% White non-Hispanic, 17.5% Black, 7.2% Hispanic, and 0.6% Asian. There was no significant difference of race between patients who were and were not frail. Women comprised 55.1% of the cohort. The frail group had a larger proportion of women (74.2%). Patients who tested as frail showed significantly lower baseline cognitive z-scores in the domain of executive function and on a composite score of all tests.

Table 1-.

Comparison of patients categorized as robust, prefrail, and frail on FRAIL scale screening prior to surgery

	Robust n = 64	Prefrail n=72	Frail n = 31	p- value#
Demographics
Age (years)	70 [67,74.5]	70.5 [67,75]	71 [66,74]	0.74
Education (years)	16.5 [14,19]	16 [13,19]	14 [12,16]	0.008
Race ⱡ				0.12
Asian	0	1 (1.4)	0
Black	12 (19.1)	8 (11.1)	9 (29.0)
Hispanic	3 (4.8)	5 (6.9)	4 (12.9)
White/nonHispanic/Unknown	48 (76.2)	58 (80.6)	18 (58.1)
Gender				0.05
Male	33 (51.6)	34 (47.2)	8 (25.8)
Female	31 (48.4)	38 (52.8)	23 (74.2)
ASA status				0.02
II	29 (45.3)	22 (30.6)	6 (19.4)
III	35 (54.7)	45 (62.5)	24 (77.4)
IV	0	5 (6.9)	1 (3.2)
Surgery category				0.053
Spine	17 (26.6)	36 (50)	18 (58.1)
Thoracic	9 (14.1)	8 (11.1)	2 (6.5)
Urologic	13 (20.3)	12 (16.7)	3 (9.7)
General	25 (39.1)	16 (22.2)	8 (25.8)
Surgery duration (min)	167 [124,205.5]	183.5 [128,249]	204 [144,263]	0.075
Baseline cognition*
MMSE	29 [28,30]	29 [28,30]	28 [27,30]	0.29
CVLT$	0.02 [−0.45,0.65]	0.09 [−0.55,0.68]	−0.24 [−0.89,0.34]	0.12
Memory/Language$	0.09 [−0.4,0.69]	0.02 [−0.48,0.49]	−0.37 [−0.97,0.38]	0.051
Speed/Executive$	0.26 [−0.34,0.66]	0.06 [−0.27,0.44]	−0.23 [−0.75,−0.08]	0.002
All$	0.11 [−0.29,0.62]	0.1 [−0.23,0.48]	−0.17 [−0.49,0.07]	0.004

Continuous variables are presented as median [IQR] and categorical variables as count (%).

^*Raw cognitive scores listed in Supplemental Table 1

^ⱡn = 166

^#Univariate comparisons used Fisher’s exact test, Kruskal-Wallis test, or chi-square test as appropriate.

^$average z scores - the difference between the individual score and the baseline average was divided by the baseline standard deviation

The univariate analysis of cognitive outcomes (at 3 months) and in-hospital delirium are summarized in Table 2. Fourteen percent of the study sample experienced cognitive decline 3 months after surgery on a composite score of all tests. There was no significant difference between the proportion of POCD between patients who tested as frail, prefrail or robust. An exploratory analysis with continuous cognitive outcome at 3 months in a linear regression model showed no significant difference between frailty groups and cognitive outcome (results not shown). Postoperative delirium occurred in 25.3% of all patients. Patients who tested as frail had a significantly higher proportion of postoperative delirium (39.3% vs 29.8% in the prefrail cohort, and 12.7% in the robust cohort, p= 0.016).

Table 2-.

In-hospital postoperative delirium and 3-month cognition in robust, prefrail, and frail patients N=150

	Robust n = 64	Prefrail n=72	Frail n = 31	P value#
Delirium				0.016
No	48 (87.3)	47 (70.2)	17 (60.7)
Yes	7 (12.7)	20 (29.8)	11 (39.3)
MMSE score change	0 [−0.53,0]	0 [−0.53,0.53]	0 [−0.79,0]	0.18
Cognitive decline > 1SD
CVLT	10 (18.2)	9 (13.4)	5 (17.9)	0.74
Memory/Language	8 (14.6)	9 (13.4)	4 (14.3)	0.98
Speed/Executive	6 (10.9)	9 (13.4)	6 (21.4)	0.42
All test composite	6 (10.9)	9 (13.4)	6 (21.4)	0.42

^*Continuous variables are presented median [Q1, Q3] and categorical variables as count [%].

^ƗFor log-transformed change scores, we performed log transformation first, reversed the sign and then took the difference.

^#Univariate comparisons used Kruskal-Wallis test for continuous data, or chi-square test or Fisher’s exact test for binary data, as appropriate. 2-sided alpha=0.025 was used for significance level.

Table 3 shows results of the logistic regression model postoperative delirium. After backward selection, only frailty and baseline cognition remained in the model. The combined cohort of patients who tested as prefrail or frail had an estimated 2.7 times odds of developing delirium compared to the robust cohort (97.5% confidence interval (CI) 1.0, 7.3). Baseline cognitive score was associated with post-operative delirium (OR 0.4, 97.5% CI 0.2, 0.8); a higher baseline composite cognitive score was associated with lower odds of developing delirium.

Table 3.

Logistic Regression Model of Postoperative Delirium adjusted for covariates^*

Parameters	Odds Ratio	97.5% Confidence Intervals
Frail or Prefrail#	2.7	[1.0, 7.3]
Robust #	1.0
Baseline Cognitive Score	0.4	[0.2, 0.7]

^*Original model included age, gender, education, surgical duration, surgical type, ASA status which were eliminated during backward selection (methods section)

^#Reference category is robust

DISCUSSION

In a cohort of older adults having major noncardiac surgery we found that testing prefrail or frail on the FRAIL scale administered prior to surgery significantly increased the odds of delirium even after adjustment for baseline cognition. While the estimated 2.7 odds of delirium are clinically significant, the large confidence interval suggests variation and belies a relatively small sample size (see sample size justification). As expected, baseline cognitive score was significantly associated with postoperative delirium; higher scores were protective. Baseline frailty status, however, was not associated with odds of POCD. Our findings suggest that a very simple form of frailty screening can help identify patients who have increased odds of developing delirium. Our findings also suggest that patients who test as prefrail behave similarly to patients who are categorized as frail and should be considered to have higher odds of developing delirium. This is important because prefrail patients may not be obvious to practitioners without use of a frailty screening tool. Future studies are warranted to further investigate the relationship between delirium and frailty; a larger study may be needed to definitively describe the relationship between preoperative frailty and longer-term cognitive outcomes.

Brown et al found that frailty in cardiac surgery patients was associated with lower cognitive scores at 1 month but not at later follow up.¹² We did not have an early one month time point for comparison, and at 3 months did not find a significant difference in cognitive decline in frail patients. Additional studies are needed to understand whether frail patients are predisposed towards early cognitive dysfunction as they may need additional support to optimize their cognitive and physical recovery. Regarding delirium, we found that there is an independent association with frailty even after adjusting for baseline cognition, which was lower in the frail and prefrail group. Our methodology addresses a limitation of the Brown paper which did not have available preoperative cognitive scores and could not adjust for differences in baseline cognition.

Our findings are concordant with many studies which show that cognitive reserve is protective for delirium. The Successful Aging after Elective Surgery study (SAGES) cohort found that cognitive markers of reserve attenuated the risk associated with postoperative inflammation as measured by CRP.²⁵ Similarly, participation in cognitive activities in later life such as reading and email was associated with lower delirium incidence.²⁶ The association of frailty with delirium, but not longer term cognitive dysfunction, provides further support for the concept that delirium and POCD may be distinct manifestations of perioperative neurocognitive deficits as put forth by Daiello et al.²⁷

Frailty screening has recently been recommended by the American College of Surgeons and as it becomes part of preoperative evaluation, it will be important to understand which frailty screening tools best predict outcomes in surgical patients. The wider geriatric literature has a long history of utilizing multidimensional frailty scales inclusive of physical measures and some have even included a measure of cognition (and also emotional state).²⁸ In fact, the term cognitive frailty refers to the co-occurrence of physical frailty and cognitive impairment. In the perioperative literature, most frailty studies have not simultaneously examined cognition. While we used a frailty scale that does not specifically test for cognition, we did include in-depth measures of cognition. We found a significant difference in baseline cognition and executive function among robust, prefrail and frail groups. This finding is similar to an outpatient geriatric study that showed worse performance on executive and working memory neuropsychological tests in patients with greater frailty.¹⁴ The co-occurrence of frailty and worse performance on cognitive tests and specifically executive function may be an important area of future study in assessing synergistic risks of delirium. A recent large Canadian administrative cohort in patients with the co-occurrence of frailty and dementia demonstrated a synergistic negative effect on survival.²⁹ Our findings suggest that perioperative frailty screening should include a measure of cognition, either as part of the frailty screen, or as a separate test to predict postoperative delirium.

Limitations:

Our study includes extensive preoperative and postoperative cognitive testing and rigorous delirium testing by trained research personnel, however, there are a few important limitations to the applicability of our results. First, this study is a secondary analysis of a previously performed study “Optimizing Postoperative Cognitive Dysfunction in the Elderly- PRESERVE” study. Sample size calculation for the parent study was performed for the primary question of investigating the relationship between POCD and postoperative functional impairment and were not designed to find a relationship between frailty and delirium and frailty and POCD. Based on the incidences of frailty, delirium and POCD, we had only a 14% power to detect an OR of 2 for POCD and 36% power to detect an OR of 2.5 for delirium. However, given that the incidence of frailty and POCD we found are in-line with what is clinically observed, it is likely that physical frailty confers only a modest (if any) increase in odds of longer term postoperative cognitive impairment. Our study excluded patients with dementia; this subgroup may have a unique risk structure for cognitive and physical decline, therefore future studies should seek to include them or focus on them primarily. Our choice of delirium assessment was the CAM-ICU, which is very specific but moderately sensitive in ill but non-intubated patients therefore we may have missed some cases;³⁰ although our delirium incidence was within range of described for older adults having major noncardiac surgery.³¹

Conclusion

We have found that patients who test positive for frailty or prefrailty on the FRAIL scale have increased odds of in-hospital delirium, but not postoperative cognitive dysfunction at 3 months after surgery. Importantly, our findings suggest that a simple frailty screen can identify an important risk factor for delirium. Preoperative screening programs should include a measure of cognition, or a multidimensional frailty scale which includes a measure of cognition. Our study did not include patients with a diagnosis of dementia; we demonstrated that worse “sub-clinical” cognitive function is associated with delirium. This demonstrates that cognitive testing can identify patients at increased odds for delirium who would otherwise go unnoticed. Future studies aimed at understanding the relationship between preoperative cognition, frailty, and postoperative delirium will continue to enhance our understanding of the pathophysiology of cognitive frailty and potential modifiable risk factors with opportunities to improve recovery by applying additional early perioperative support.

Key Points Summary.

Question:

What is the relationship between frailty and postoperative cognitive dysfunction and frailty and postoperative delirium after major noncardiac surgery?

Findings:

Testing as frail or prefrail was associated with marginally increased estimated odds for postoperative delirium but not POCD when compared to the robust group after cardiac surgery.

Meaning:

Non-cardiac surgery patients who tested “frail” or “prefrail” are at increased odds for postoperative delirium.

Glossary of Terms

ASA

American Society of Anesthesiology

CAM-ICU

confusion assessment method for the intensive care unit

CI

confidence interval

FRAIL scale

a simple questionnaire which categorizes patients as robust, prefrail or frail.

HADS

Hospital Anxiety and Depression Scale

POCD

postoperative cognitive dysfunction

PRESERVE study

Optimizing Postoperative Cognitive Dysfunction in the Elderly Study

SAGES study

Successful Aging after Elective Surgery study

SD

standard deviation

UDS

Uniform Dataset Battery

VARCLUS

variable cluster analysis procedure in SAS (v 9.4)