Incidence of post-operative delirium increases as severity of frailty increases

Abstract

Objective

The surgical population is ageing and often frail. Frailty increases the risk for poor post-operative outcomes such as delirium, which carries significant morbidity, mortality and cost. Frailty is often measured in a binary manner, limiting pre-operative counselling. The goal of this study was to determine the relationship between categorical frailty severity level and post-operative delirium.

Methods

We performed an analysis of a retrospective cohort of older adults from 12 January 2018 to 3 January 2020 admitted to a tertiary medical center for elective surgery. All participants underwent frailty screening prior to inpatient elective surgery with at least two post-operative delirium assessments. Planned ICU admissions were excluded. Procedures were risk-stratified by the Operative Stress Score (OSS). Categorical frailty severity level (Not Frail, Mild, Moderate, and Severe Frailty) was measured using the Edmonton Frail Scale. Delirium was determined using the 4 A’s Test and Confusion Assessment Method-Intensive Care Unit.

Results

In sum, 324 patients were included. The overall post-operative delirium incidence was 4.6% (15 individuals), which increased significantly as the categorical frailty severity level increased (2% not frail, 6% mild frailty, 23% moderate frailty; P < 0.001) corresponding to increasing odds of delirium (OR 2.57 [0.62, 10.66] mild vs. not frail; OR 12.10 [3.57, 40.99] moderate vs. not frail).

Conclusions

Incidence of post-operative delirium increases as categorical frailty severity level increases. This suggests that frailty severity should be considered when counselling older adults about their risk for post-operative delirium prior to surgery.

Key Points

• Frailty increases the risk for delirium, which carries significant morbidity, mortality and cost.

• Delirium incidence increases as frailty severity increases.

• Understanding frailty severity may help us better counsel older adults about post-operative delirium risk.

Introduction

The rapid growth in the population of older adults in the USA means that almost half of all inpatient surgeries are now performed on adults over 65 years of age [1] A particular concern is that a significant proportion of these older adults has frailty, a condition associated with a high risk for poor post-operative outcomes such as complications including delirium, ICU admission, longer hospital length of stay, loss of independence, 30-day readmission and 30-day mortality [2–6]. To mitigate these problems, the American College of Surgeons in partnership with the John A. Hartford Foundation, created the Geriatric Surgical Verification Program and published best practice guidelines for geriatric surgical management, the foundation of which is risk stratification with a frailty screen [7]. In 2018, our urban tertiary medical center created a Geriatric Surgical Pathway (GSP), instituted preoperative frailty screening with the Edmonton Frail Scale (EFS), and implemented best practice standards with the goal of preventing the poor post-operative outcomes frail older adults face [3, 8]. This program has been evaluated to show reduction in post-operative complications and institutionalisation, as well as initial positive findings for return on investment [3, 4].

However, one of the outcomes of interest is post-operative delirium, which is independently associated with significant morbidity and mortality, and can be difficult to treat once it occurs [9]. To this effort, we implemented twice daily delirium screening on the surgical wards for all older adults as part of our GSP. This allowed for a recent analysis examining the association between frailty and post-operative delirium. This study found that frailty determined by the EFS in a binary manner was associated with a 4.86-fold increase in odds of post-operative delirium [5].

However, as we move toward precision medicine, we aim to better understand how frailty severity is associated with the incidence of post-operative delirium. We hypothesised that pre-operative frailty severity level will directly correlate with higher incidence of post-operative delirium. With a better understanding of the complex relationships between the severity of frailty and incident post-operative delirium, we may be able to more accurately identify patients who are at higher risk for post-operative delirium, providing more information to make better informed surgical decisions.

Materials and methods

As part of our GSP, geriatric patients scheduled for elective inpatient surgery undergo routine frailty screening with the EFS in an outpatient pre-operative clinic prior to admission for surgery [3]. We performed a secondary data analysis of an established cohort of 324 patients 65 years or older who underwent this pre-operative EFS assessment prior to undergoing inpatient elective surgery at a tertiary academic medical center from December 2018 to March 2020 [5]. These were patients admitted on the day of their elective surgery and not inpatients awaiting surgery. To be included in the analysis, patients had to have a length of stay of at least one day and at least two inpatient delirium screens. Patients with a planned post-operative ICU stay were excluded. Our primary goal was to determine the association between severity of pre-operative frailty determined by the EFS and incident post-operative delirium.

Individuals' demographic and clinical characteristics, including frailty and delirium scores, were collected through the EPIC electronic medical record (EMR) [3]. Patients were risk stratified for post-operative delirium using individual (demographics and frailty severity) and surgery (primary anaesthesia type and operative stress) specific characteristics. This project was reviewed and approved by the Johns Hopkins University Institutional Review Board.

Frailty

Severity of pre-operative frailty was determined using the EFS, a 10-item frailty screen [3, 8]. The EFS was chosen as the screen of choice for geriatric surgery at our institution because many of the individual components of the EFS (cognitive impairment, functional dependence, social support, polypharmacy, weight loss, depression, incontinence, mobility) were found to be useful in guiding the pre-operative multidisciplinary team in determining what inpatient interventions are needed for a particular individual to prevent poor outcomes during their stay [3, 4, 8]. For example, a patient with functional or mobility limitations may be counselled earlier about the potential need for post-operative rehabilitation. Or a patient with weight loss may be recommended for a nutrition consultation. The screen is performed by nurses and medical assistance in an outpatient pre-operative clinic which is then recorded in the EPIC EMR. This screening takes approximately 10 minutes to complete and is scored from 1 to 17 with patients classified into the following frailty levels: (1) not frail (< 6 points); (2) mild frailty (6–7 points); moderate frailty (8–9 points); or severe frailty (≥ 10 points) [3, 8].

Operative Stress Score

Procedures performed (Supplementary Table 1) were risk-stratified for surgical heterogeneity by their Operative Stress Score (OSS) to assess for potential effect modification of pre-operative frailty on incidence of post-operative delirium due to differences in the stress of the surgery performed [10–12]. The OSS was recently developed by Shinall et al. and assigns a score ranging from 1 to 5 reflecting the perceived physiologic stress to a person from a given procedure by experts in the field. A score of 1 reflects a minimally stressful procedure (e.g. cystourethroscopy), whereas a score of 5 reflects procedures inducing the highest amount of physiologic stress (e.g. liver transplant) [10]. Previous studies have shown an association between increasing OSS and increasing post-operative mortality [10, 12]. OSS was categorised into the following stress levels: mild stress (OSS of 1–2); moderate stress (OSS of 3); or severe stress (OSS of 4–5). Because this is a relatively new assessment measure, 20 individuals underwent a procedure that did not have a corresponding OSS score available for risk stratification [10–12].

Outcomes

The primary outcome of interest was the incidence of post-operative delirium, defined as any inpatient positive delirium assessment following surgery. For the purpose of this study, a single post-operative positive delirium assessment was considered an occurrence, given delirium is classically a waxing and waning alteration in attention and mentation. Delirium was measured either via the 4A’s Test (4AT) or the Confusion Assessment Method for the Intensive Care Unit (CAM-ICU) if a patient had a need-based upgrade to the ICU. The 4AT is a four-question delirium screen assessing alertness, orientation, attention and fluctuation. It is scored from 0 to 12 where a score ≥ 4 points is consistent with delirium. The CAM-ICU assesses four domains and is considered positive if a patient has fluctuating cognition, inattention and either disorganised thinking or an altered level of consciousness [13, 14]. Post-operative delirium was considered positive if there was any positive score from either a 4AT or CAM-ICU assessment any time through post-operative day three or discharge if sooner. These screens are performed routinely by nursing staff.

Statistical analysis

Demographic and clinical characteristics of individuals were summarised and compared across pre-operative frailty levels (not frail, mild frailty, moderate frailty and severe frailty) using analysis of variance (ANOVA) for continuous measures and Chi-squared or Fisher's exact tests for categorical variables. The proportion of individuals developing post-operative delirium was compared across frailty levels using Score tests for trend. Unadjusted logistic regression was used to evaluate the association between frailty levels and post-operative delirium. To determine the potential impact of confounding on these effect estimates, multivariable analyses which included adjustment for age, sex, race, primary anaesthesia type and OSS were performed. The reference group for these analyses were individuals with no frailty. We repeated these same analyses with EFS score as a continuous measure. Results of regression models are presented as odds ratios (OR) with corresponding 95% confidence intervals (CI). In addition, Hosmer and Lemeshow's goodness of fit tests were conducted for each model. We also performed exploratory analysis of any potential effect modification on the relationship between pre-operative frailty with incident post-operative delirium by the operative stress of the procedure performed. The proportion of individuals developing post-operative delirium across each frailty level was compared within each OSS category using Score tests for trend. Sensitivity analyses limited to those individuals who had frailty screening within 120 and within 90 days of surgical admission were conducted. A P value <0.05 was considered statistically significant. All analyses were conducted using STATA 14.0 (StataCorp, College Station, TX).

Results

A total of 324 elective surgical patients met the inclusion criteria. The average age of individuals was 73 years (SD = 6.3), 40% were female and 74% were White (Table 1). General anaesthesia was used in 80% of surgeries and most surgeries performed were of mild to moderate stress level based on the OSS (41% of procedures had an OSS of 1–2 and 48% had an OSS of 3). Frailty screens were performed at a median of 55 days prior to surgical admission (25th and 75th percentiles 26 and 125 days, respectively). Most individuals (77%) were classified as not frail, 15% had mild frailty and 8% had moderate frailty. No individual had severe pre-operative frailty. Those with higher frailty severity levels were older (P = 0.003) and less often underwent general anaesthesia (P < 0.001). There was no significant difference in sex, race, or OSS between frailty severity levels.

The overall incidence of post-operative delirium was 4.6% (15 individuals) (Table 2). There was a significantly higher rate of post-operative delirium incidence with increasing frailty severity level (2.4% for not frail, 6% for mild frailty and 23.1% for moderate frailty).

Table 3 presents the logistic regression analyses evaluating the association of frailty severity levels with the odds of post-operative delirium. Compared to their not frail counterparts, individuals with mild frailty had a non-significant 2.57 [95% CI: 0.62, 10.66] higher odds of post-operative delirium, whereas individuals with moderate frailty had a statistically significant 12.10 [95% CI: 3.57, 40.99] higher odds of post-operative delirium. The inclusion of age, sex or race did not materially impact the magnitude or interpretation of the association between frailty severity level and odds of post-operative delirium.

To more fully explore the relationship between post-operative delirium and frailty, we also evaluated the rate of post-operative delirium by the raw EFS score. A one-point increase in the EFS score was associated with a statistically significant 51% [21, 87%] higher odds of developing post-operative delirium. This association did not change appreciably when including age, sex or race. Figure 1 presents the predicted probabilities of post-operative delirium by EFS score, which increases across the range of EFS scores.

Findings were similar for sensitivity analyses limited to those individuals who had frailty screening within 120 and 90 days of surgical admission (Supplementary Tables 2 and 3).

Exploratory analysis was completed to assess the potential for effect modification of the relationship between pre-operative frailty with incident post-operative delirium by the operative stress of the procedure performed (Table 4). For individuals who underwent procedures of mild stress level (OSS 1–2), there was no significant relationship between increasing pre-operative frailty severity level and incidence of post-operative delirium (delirium incidence of 2.1% for not frail, 5.9% for mild frailty and 8.3% for moderate frailty; P = 0.192 for trend). However, there was a significant relationship between increasing pre-operative frailty severity level and post-operative delirium incidence for those who underwent procedures of moderate stress level (OSS 3) (delirium incidence of 0.9% for not frail, 4.2% for mild frailty and 35.7% for moderate frailty; P < 0.001 for trend). There was insufficient number to assess the effect for surgeries of severe stress level (OSS 4–5).

Discussion

This study demonstrates an association of higher incidence of post-operative delirium with increasing frailty severity, measured by both categorical and continuous variables. In our cohort, we demonstrate that the proportion of individuals with incident post-operative delirium increases as an individual’s pre-operative frailty severity category increased ranging from 2% of not frail individuals to 23% of individuals with moderate frailty. Better understanding the interrelationship between frailty severity, operative stress and incident delirium will help guide precision pre-operative risk evaluations, identification of patient needs while admitted for surgery, and development of quality improvement pathways to prevent incident delirium.

On exploratory analysis, we saw that this relationship between increasing incidence of post-operative delirium with increasing categorical frailty severity level was only significant for surgeries performed that were of moderate stress level. We had insufficient surgeries performed of severe stress levels to determine if this trend would continue for higher stress procedures.

These findings suggest that pre-operative categorical frailty severity level, rather than a binary categorisation of frail versus not frail, may be more helpful when counselling patients about their unique risk for developing delirium after surgery. In addition, based on exploratory analysis, this may be most important to consider for surgeries of at least moderate stress level.

Two recent metanalyses from 2020 and 2021 focused on the effect of frailty on post-operative outcomes following surgery including delirium [6, 15]. Aucoin and colleagues examined the association of several frail scales to post-operative outcomes, 12 of which addressed delirium [6]. Zhang and colleagues specifically looked at studies examining the association between frailty screens and delirium, and twenty-one studies were in surgical populations [15]. Between the two there were several studies showing a relationship between pre-operative frailty and post-operative delirium; however, only a few studies examined this association used frailty as a categorical or continuous variable [16–24]. These studies showed that, in general, frailty identified by the Fried Frailty Scale, FRAIL Scale, Clinical Frailty Scale and the Sinai Abbreviated Geriatric Evaluation was associated with significantly higher risk for delirium as severity of frailty increased [16, 19–24]. Our findings show that the EFS is also useful for determining risk for delirium across categorical levels of frailty severity.

There are several limitations to this study. First, this study was conducted at a single institution, thus results may not be generalisable to all geriatric patients. In addition, our cohort consisted of older adults undergoing inpatient elective surgery which limits external validity to emergency surgery populations. More rapid frailty screens are needed to embed in an emergency surgery setting in order to understand the risk for delirium associated with frailty following emergency surgery. Finally, there was a minimal number of incident post-operative delirium cases (15), which resulted in limited power to detect differences in post-operative delirium between groups and high levels of uncertainty for estimates comparing odds of post-operative delirium across groups. Low incidence of post-operative delirium in our study may reflect the underlying demographics of our cohort with a relatively young average age (73 years), the exclusion of planned ICU admissions, or effect of previous institutional surgical program quality improvements through Enhanced Recovery after Surgery and our GSP [3, 4]. Further research in larger cohorts on the interrelationship between frailty severity, operative stress and incident delirium is needed to further refine our understanding. In addition, we did not measure or assess the impact of all potential contributors to post-operative delirium beyond demographics, frailty severity, primary anaesthesia type and OSS. Further studies are needed to broaden our understanding of the risk factors for post-operative delirium including the administration of deliriogenic medications, time to mobilisation, oral intake and more, which may have further care implications. It is also important to determine the relationship post-operative delirium has on secondary outcomes important to the healthcare system such as the length of stay, discharge disposition and readmissions.

There are several strengths to this study. First, at our tertiary medical center, we have high rates of frailty screening with the EFS and robust delirium screening allowing for a granular assessment of the association of post-operative delirium with frailty severity beyond the typical binary cutoff. Most literature focuses on post-operative delirium risk stratification based on a binary cut-off, and we demonstrate that categorical frailty severity level or score itself offers more information to a provider when counselling a patient on their unique risk for post-operative delirium [6, 15, 25–27]. Second, we include an assessment of the stress induced by the procedures performed via the OSS, which is often overlooked in the surgical literature [10–12]. Although limited, we see a trend toward increasing post-operative delirium with increasing operative stress in those with moderate frailty, which should be verified in a larger study.

Conclusion

The incidence of post-operative delirium was higher with increasing frailty severity levels. This suggests that frailty level, and perhaps the operative stress of the procedure to be performed for individuals with moderate frailty, may offer additional information when counselling older adults about their risk for post-operative delirium prior to surgery.

Declaration of Conflicts of Interest

None.

Declaration of Sources of Funding

This work was supported by grant number T32AG066576 (to A.E.), grant numbers R01AG076525, R01AG057725, R01AG057667 (to E.O.) and grant number UL1TR003098 (to N.Y.W.) from the National Institute on Aging and the National Center for Advancing Translational Sciences of the National Institutes of Health.

References