Secondary Logo

Journal Logo

Featured Articles: Original Clinical Research Report

Functional Outcomes of Frail Patients After Cardiac Surgery: An Observational Study

Nakano, Mitsunori MD*; Nomura, Yohei MD*; Suffredini, Giancarlo MD; Bush, Brian MD; Tian, Jing MS; Yamaguchi, Atsushi MD, PhD*; Walston, Jeremy MD, PhD§; Hasan, Rani MD, MHS; Mandal, Kaushik MD; Schena, Stefano MD, PhD#; Hogue, Charles W. MD**; Brown, Charles H. IV MD, MHS

Author Information
doi: 10.1213/ANE.0000000000004786

Abstract

KEY POINTS

  • Question: Are functional outcomes worse among frail patients undergoing cardiac surgery compared with nonfrail patients?
  • Findings: In adjusted models, frail patients had greater functional decline at 1 month after cardiac surgery.
  • Meaning: Frailty may identify patients at risk of functional decline at 1 month after cardiac surgery, and therefore perioperative strategies to optimize frail cardiac surgery patients are needed.

Frailty is a geriatric syndrome characterized by lack of resilience to stressors.1 The prevalence of frailty has been estimated to be 10%–60% in patients with cardiovascular disease,2 and in multiple studies, frailty has been associated with major morbidity/mortality and increased length of stay after cardiac surgery.3,4 However, as noted by 2 recent systematic reviews,5,6 there are few studies that have determined whether frailty identifies patients at risk for worse functional outcomes or nonhome discharge destination. There is also little information on whether delirium interacts with frailty to increase susceptibility to functional decline, but such an interaction would identify the most vulnerable older adults at discharge. For older adults undergoing surgery, preservation of functional status and avoiding institutionalization are key goals, while for health systems, minimizing length of stay is becoming increasingly important.

Although there are many ways to measure frailty, the Fried frailty scale has been widely used and validated in both community-dwelling and hospitalized patients.7 Much of the popularity of the Fried scale may be traced to its basis in a biologic phenotype of sarcopenia and low energy expenditure. Patients categorized as frail are thought to be at high risk for functional decline after the stress of cardiac surgery, but this association has not been well documented. However, such information is critically important for risk stratification, surgical decision-making, and perioperative planning. Importantly, given the biologic basis of frailty, such information could provide a compelling rationale for the development and implementation of prehabilitation programs for frail older adults undergoing cardiac surgery that are targeted at the underlying biology of sarcopenia and low energy expenditure.8

In this nested observational study, we hypothesized that baseline frailty would be associated with functional decline at 1 month (primary outcome), decline in absolute Instrumental Activities of Daily Living (IADL) score at 1 month, increased discharge to a nonhome location, and longer duration of hospitalization, and that delirium would modify these associations.

METHODS

Study procedures were approved by the institutional review board at Johns Hopkins School of Medicine ([email protected]). Written informed consent was obtained from patients. The data in this analysis were collected during 2 trials, each of which was submitted for registration on www.clinicaltrials.gov before enrollment (NCT00981474; principal investigator, C.W.H.; submitted September 21, 2009 and NCT02587039; principal investigator, C.H.B.; submitted July 9, 2014). This manuscript adheres to the applicable STrengthening the Reporting of OBservational studies in Epidemiology (STROBE) guidelines.

Study Design and Patients

This was an observational study nested in 2 separate single-site trials at an academic center. Each trial was conducted by the same team with identical measurement of frailty and functional outcomes. The primary question (association of frailty with 1-month functional decline) was prespecified before patient enrollment for patients in this analysis. Data from the 2 trials were combined for this study.

The primary aim of the first trial was to determine whether targeted mean arterial pressure during cardiopulmonary bypass based on cerebral autoregulation monitoring reduced a composite outcome (stroke, ischemic lesions, cognitive change) compared with standard blood pressure management (registration NCT00981474). Patients ≥55 years old were enrolled from August 2014 to May 2016. Inclusion criteria were coronary artery bypass (CAB) and/or valve surgery that required cardiopulmonary bypass and who were at high risk for neurologic complications9 as determined by a Johns Hopkins risk score that generally excluded patients in the lowest quartile of risk. Exclusion criteria were baseline delirium, hepatic dysfunction, contraindications to magnetic resonance imaging (MRI), dialysis, non-English speaking, inability to follow-up, and emergency surgery. Eighty-three patients were included from this trial.

The primary aim of the second trial was to determine whether remote ischemic preconditioning reduced postoperative delirium (trial registration NCT02587039). Patients were enrolled from July 2014 to December 2015. Inclusion criteria were age ≥65 years old and undergoing CAB and/or valve operation that required cardiopulmonary bypass. Exclusion criteria were baseline delirium, Mini-Mental State Examination score <23, inability to speak/read English, severe hearing impairment, inability to tolerate a tourniquet, hemoglobinopathy, or intraoperative ketamine. Fifty patients were included from this trial.

Data from the 2 studies were merged based on common definitions of data elements. Data on the combined group of patients have been published separately, in a parallel manuscript examining the association of frailty and postoperative delirium/cognition.10 Data from the first trial have been published in manuscripts examining the associations of delirium and cognition11 and the effects of an intervention to optimize mean arterial pressure.12

Preoperative Assessment

Baseline frailty assessments were performed using the validated scale from Fried et al7 evaluating 5 domains: (1) shrinking, defined as unintentional weight loss of ≥10 pounds in the last year; (2) weakness, determined by grip strength, adjusted for gender and body mass index (BMI); (3) exhaustion, determined by 2 questions from the modified 10-item Center for Epidemiological Studies-Depression scale13; (4) low physical activity, determined by the modified Minnesota Leisure Time Activities Questionnaire14; and (5) slowed walking speed, as measured at normal pace over 15 feet. Each of the 5 domains yielded a score of 0 or 1 based on cutoffs previously described. Twelve patients who refused to walk were scored as “1” for gait speed. Frail patients were defined as total scores 3–5.7

Outcome Assessment

At baseline and 1-month follow-up, patients completed the Older Americans Resources and Services IADL questionnaire.15 Baseline and discharge location were obtained from the medical record; discharge location was categorized as discharge to home, including to hotel or with family, or discharge to a nonhome location, including subacute or acute rehabilitation. Duration of hospitalization and incidence of postoperative complications (atrial fibrillation, new dialysis, new intra-aortic balloon pump, inotropic drugs>24 hours, mechanical ventilation>48 hours, sepsis, stroke) was obtained from the medical record. Delirium was assessed on 3 of the first 4 days using the Confusion Assessment Method16 and Confusion Assessment Method for the intensive care unit (ICU)17 (intubated patients). For days on which patients were not assessed in person, a validated chart review was used (sensitivity 74%, specificity 83%).18

Perioperative Management

In the operating room, general anesthesia was induced with fentanyl, midazolam, and/or propofol and was maintained with isoflurane and a nondepolarizing muscle relaxant. In the ICU, patients were generally sedated with propofol until readiness for tracheal extubation, with a protocolized goal of extubation within 6 hours. For patients in whom extubation was anticipated to be delayed (>48–72 hours), fentanyl and/or midazolam infusions were considered. Mean arterial pressure targets in the ICU were 65–90 mm Hg depending on the procedure and surgeon. Inotropes were weaned based on estimates of adequate perfusion, including physical examination, hemodynamic variables, and laboratory values. Early mobilization was emphasized, with a goal of ambulating by postoperative day 2.

Statistical Analysis

The primary exposure was frailty (defined as a binary variable [frail versus not-frail]), as assessed by the Fried criteria.7 The primary outcome was functional decline (defined as IADL decline of ≥2 points). A cutoff of 2 points for functional decline was chosen based on surveys of geriatricians indicating clinically significance19 and based on similar definitions in other studies of functional decline after cardiac surgery.20 Secondary outcomes were absolute decline in IADL score, discharge location (home versus nonhome), and duration of hospitalization. For all analyses, a P value <.05 was considered significant. However, all analyses using secondary outcomes should be considered exploratory and hypothesis generating. No patients with baseline frailty assessments died in the hospital. Baseline patient characteristics were compared using Student t tests, Wilcoxon rank-sum tests, and Fisher exact tests to assess potentially confounding variables.

For analyses examining the association of frailty and postoperative outcomes, the primary outcome was functional decline (IADL decline ≥2 points). The associations of frailty and both functional decline and discharge location were examined using logistic regression. The association of frailty and absolute decline in IADL score was examined using linear regression with nonparametric bias-corrected bootstrapped confidence intervals (CIs), a method chosen because the distribution of the outcome was not normal, with a relatively small sample size. Generalized linear models with a Poisson distribution were used to examine the association of frailty and duration of hospitalization. The fit of the Poisson distribution was assessed using a comparison of residual deviance with degrees of freedom. Variables for which to adjust were considered a priori before analysis (based on review of the literature and potential associations with frailty and functional status) and included age, gender, race, and logistic European Score for Cardiac Operative Risk Evaluation (EuroSCORE). Of note, the logistic EuroSCORE includes several variables, which were thought to differ by frailty status, such as age, comorbidities, and surgical characteristics. The association of individual components of the frailty score with each outcome was examined in unadjusted and adjusted regression models, with each individual component incorporated into separate models.

As a post hoc analysis, to minimize selection bias in our observational study, we used an inverse probabilities of treatment weight (IPTW) propensity score (PS) method.21 PSs were calculated by generating a logistic regression model to predict the probability of each patient with observed frailty status at baseline based on the following variables, which were chosen based on review of the literature for potential association with frailty and/or functional decline: age, gender, race, education, prior stroke, hypertension, heart failure, vascular disease, chronic obstructive pulmonary disease, tobacco use, diabetes melliotus, logistic EuroSCORE, surgery, cardiopulmonary bypass, delirium, and baseline IADL score. IPTW was created by using 1/PS. A range of PS distributions was considered. To address the extreme values of the IPTW, truncation was performed using percentile cutpoints—with weights >95th percentile and <5th percentile set equal to the 95th and fifth percentiles, respectively. Finally, weighted models to estimate the effect of frailty on outcomes were performed with truncated IPTW as the weights. As a sensitivity analysis to examine potential overfitting of the logistic regression model, we generated separate PS using only variables associated with frailty in univariate analysis (P < .20). Nine variables were identified: age, gender, race, education, hypertension, heart failure, diabetics, logistic EuroSCORE, and baseline IADL. These PSs were used for separate weighted IPTW models, with similar methodology as described above.

We also examined the modifying effect of delirium by testing the significance of an interaction term in adjusted regression models that examined the association of frailty with functional decline, change in absolute IADL score, discharge location, and duration of hospitalization.

A potential mediating effect of postoperative complications was examined using path analysis in MPlus (Muthén & Muthén, Los Angeles, CA). Here, the “total” association (“effect”) of frailty with outcomes is decomposed into a direct effect on outcomes, independent of complications, and an “indirect” effect in which frailty is associated with complications, which in turn are associated with outcomes. To evaluate the indirect effect more fully, the components of the indirect effect were examined individually, that is, the association of the exposure (frailty) with the mediator (complications) and the associations of the mediator (complications) with each outcome. A probit link was specified in the models with binary mediators or outcomes. This link conceptualizes binary data as a dichotomized version of an underlying standard normal random variable. Path coefficients in this framework are interpreted in the same way as linear regression coefficients, either literally (for continuous outcomes) or with respect to the normal variable underlying the binary dichotomization of it. Mediation was adjudicated based on the size and significance of the indirect effect and the significance of the exposure/complication and complication/outcome relationships. In interpreting the indirect effect, we also examined whether there were exposure–mediator interactions, by examining the interaction of frailty and complications in separate models with each outcome.

The sample size for this nested cohort study was determined by the number of patients with available frailty and outcome assessments. We determined that a sample size of 133 patients would provide a power of 0.76 to detect a difference in an IADL decline of ≥2 at a significance level of .05, assuming a frailty prevalence of 33%, and a decline in IADL of 25% in nonfrail patients and 50% in frail patients.20

RESULTS

Patient Characteristics

Table 1. - Patient and Surgical Characteristics
Nonfrail
N = 89
Frail
N = 44
P
Age (y), mean (SD) 71.3 (7.1) 73.5 (8.1) .117
Male, n (%) 69 (78) 28 (64) .090
Race, n (%)
 Caucasian 75 (84) 35 (80) .195
 African American 8 (9) 8 (18)
 Other 6 (7) 1 (2)
Education, n (%) .181
 High school or below 31 (36) 21 (48)
 Above high school 56 (64) 23 (52)
Comorbidities, n (%)
 Prior stroke 12 (14) 7 (16) .707
 Hypertension 84 (94) 38 (86) .114
 Congestive heart failure 13 (15) 12 (27) .079
 Peripheral vascular disease 11 (12) 9 (20) .219
 COPD 5(6) 2 (5) .739
 Tobacco (prior) 45 (63) 24 (63) .946
 Diabetes 30 (34) 22 (50) .070
Logistic EuroSCORE, median (IQR) 3.35 (2.17–6.66) 6.29 (3.29–9.47) .003
Surgery, n (%) .873
 CAB 26 (48) 18 (41)
 CAB + valve 7 (13) 8 (18)
 Valve 19 (35) 17 (39)
Other 2 (4) 1 (2)
Cardiopulmonary bypass duration (min), median (IQR) 117 (84–154) 128 (80–154) .669
Incident delirium 37 (42%) 19 (48%) .56
Abbreviations: CAB, coronary artery bypass; COPD, chronic obstructive pulmonary disease; EuroSCORE, European Score for Cardiac Operative Risk Evaluation; IQR, interquartile range; SD, standard deviation.

Table 2. - Postoperative Complications
Nonfrail
N = 89
Frail
N = 44
P
Atrial fibrillation, n (%) 17 (19.1) 13 (29.6) .18a
Incident dialysis, n (%) 0 0 N/A
New intra-aortic balloon pump, n (%) 4 (4.5) 5 (11.4) .16b
Mechanical ventilation >48 h, n (%) 1 (1.1) 3 (6.8) .11b
Multiple inotropic drugs >24 h, n (%) 1 (1.1) 2 (4.6) .25b
Single inotropic drug >24 h, n (%) 7 (7.9) 6 (13.6) .29a
Sepsis, n (%) 0 1 (2.3) .33b
Stroke, n (%) 2 (2.3) 4 (9.1) .09b
Compositec, n (%) 27 (30.3) 23 (52.3) .014a
Abbreviation: N/A, not applicable.
aχ2 test.
bFisher exact test.
cComposite was defined as any complication listed in Table 2.

Data were available on 133 patients with baseline frailty measurement. A flow diagram is shown in Supplemental Digital Content, Appendix A, http://links.lww.com/AA/D68. Compared to patients in the trial examining remote ischemic preconditioning, patients in the trial examining cerebral autoregulation were younger (70.7 ± 8.0 years old vs 74.2 ± 6.1 years old), were more likely to be men (81% vs 60%), and had similar median logistic EuroSCORE (4.51 [interquartile range {IQR}, 2.27–9.26] vs 4.22 [IQR, 2.71–7.68]). The overall prevalence of frailty at baseline was 33% (44 of 133). Characteristics of patients by frailty status are presented in Table 1. Frail patients had significantly higher logistic EuroSCORE compared with nonfrail patients. Frail patients had a higher incidence of a composite score of postoperative complications compared to nonfrail patients (52% [23 of 44] vs 30% [27 of 89]; P = .01), as shown in Table 2, and this difference was significant in a model adjusted for age, logistic EuroSCORE, and bypass time (odds ratio [OR], 2.38 [95% CI, 1.11–5.11]; P = .027).

Functional Decline in IADL at 1 Month (Primary Outcome)

Of the 133 patients in the study sample, IADL data were missing in 6 patients at baseline and in an additional 3 patients at 1 month after discharge. Thus, 1-month IADL data were available in 98% (124 of 127) of patients with baseline IADL data. At baseline, the median IADL score (of 14 points total) was 14 (IQR, 13–14) in frail patients and 14 (IQR, 14–14) in nonfrail patients. A total of 91% of patients had a baseline IADL score of 13 or 14. At the 1-month follow-up, functional decline in IADL score (≥2-point decline; primary outcome) occurred in 29% (36 of 124) of patients overall. Among frail patients, functional decline occurred in 44% (17 of 39), while in nonfrail patients, functional decline occurred in 22% (19 of 85; Figure 1A). The odds of functional decline in IADL score from baseline to 1 month were higher for frail patients than for nonfrail patients in a model adjusted for age, gender, race, and logistic EuroSCORE (OR = 2.41 [95% CI, 1.03–5.63]; P = .04) and in a PS-adjusted model (OR = 2.32 [95% CI, 1.27–4.32]; P = .006). In a sensitivity analysis that included the baseline IADL score as a covariate in addition to age, gender, race, and logistic EuroSCORE, the inferences were similar (OR = 2.45 [95% CI, 1.03–5.81]; P = .04).

Figure 1.
Figure 1.:
A, Functional decline (defined as decrease in Instrumental Activities of Daily Living score ≥2), (B) discharge to nonhome location, and (C) number of days in hospital by frailty status. P values are unadjusted comparisons.
Table 3. - Odds of Functional Decline in IADLs, Change in IADL Score, Odds of Discharge to a Nonhome Location, and Increasing Number of Days in Hospital for Frail Patients Compared With Nonfrail Patients
Unadjusted Adjusteda Propensity Score Adjustedb
Estimate (95% CI) P Estimate (95% CI) P Estimate (95% CI) P
Odds ratio of functional decline (≥2) in IADL score from baseline to 1 moc
 Nonfrail Ref Ref Ref
 Frail 2.7 (1.2–6.1) .02 2.4 (1.03–5.6) .04 2.3 (1.3–4.3) .006
Change in IADL score from baseline to 1 mo
 Nonfrail Ref Ref Ref
 Frail −1.5 (−2.7 to −0.4) .01 −1.5 (−2.8 to −0.3) .02 −1.0 (−2.0 to −0.07) .04
Odds ratio of discharge to new nonhome locationd
 Nonfrail Ref Ref Ref
 Frail 4.2 (1.9–9.3) .0005 3.3 (1.4–7.7) .007 2.5 (1.4–4.5) .002
Increasing number of days in hospitale
 Nonfrail Ref Ref Ref
 Frail 1.4 (1.3–1.6) <.0001 1.4 (1.2–1.5) <.0001 1.3 (1.1–1.5) <.001
Abbreviations: CI, confidence interval; IADL, Instrumental Activities of Daily Living.
aAdjusted by age, gender, race, and logistic European Score for Cardiac Operative Risk Evaluation.
bWe used an inverse probabilities of treatment weight propensity score method. Propensity scores were calculated by generating a logistic regression model to predict the probability of each patient with observed frailty status at baseline based on the following variables, which were chosen based on review of the literature for potential association with frailty and/or functional decline: age, gender, race, education, prior stroke, hypertension, congestive heart failure, peripheral vascular disease, chronic obstructive pulmonary disease, tobacco use, diabetes mellitus, logistic European Score for Cardiac Operative Risk Evaluation, surgery, cardiopulmonary bypass, delirium, and baseline IADL score. Inverse probabilities of treatment weight were created by using 1/propensity scores. A range of propensity score distributions was considered. To address the extreme values of the inverse probabilities of treatment weight, truncation was performed using the percentile cutpoints—all weights with value above the 95th percentile were set equal to the 95th percentile and all weights with value below the fifth percentile were set equal to the fifth percentile. Finally, weighted models to estimate the effect of frailty on outcomes were performed with truncated inverse probabilities of treatment weight as the weights.
cEstimate refers to the odds ratio of functional decline in IADL score in frail patients compared with nonfrail patients.
dEstimate refers to the odds ratio of discharge to a new nonhome location in frail patients compared with nonfrail patients.
eEstimate refers to the estimated increasing number of days in the hospital for frail patients compared with nonfrail patients.

The change in IADL score from baseline to 1 month (n = 124) in frail patients was a median change (ie, decline) of −1 (IQR, −3 to 0) while in nonfrail patients was a median change of 0 (IQR, −1 to 0; P = .018). The decline in absolute IADL score (secondary outcome) in 124 patients with baseline and 1-month data was greater in frail than in nonfrail patients in a model adjusted for age, gender, race, and logistic EuroSCORE (−1.48 [95% CI, −2.77 to −0.30]; P = .019) and in a PS-adjusted model (−1.04 [95% CI, −2.0 to −0.07]; P = .035; Table 3). In a sensitivity analysis using 1-month IADL as the outcome with adjustment for baseline IADL score, age, gender, race, and logistic EuroSCORE, frail patients had lower 1-month IADL scores compared with nonfrail patients (−1.53 [95% CI, −2.57 to −0.49]; P = .004).

Discharge Location (Secondary Exploratory Outcome)

Discharge location was a secondary exploratory outcome. The overall incidence of discharge to a new nonhome location was 28% (37 of 133). The incidence of nonhome discharge in frail patients was 48% (21 of 44) and in nonfrail patients was 18% (16 of 89; Figure 1B). As shown in Table 3, the odds of being discharged to a nonhome location were 3.25-fold higher (95% CI, 1.37–7.69; P = .007) for frail patients than for nonfrail patients, adjusted for age, gender, race, and logistic EuroSCORE, with similar inferences in a PS-adjusted model.

Duration of Hospitalization (Secondary Exploratory Outcome)

Duration of hospitalization was a secondary exploratory outcome. The median number of days in the hospital after surgery for frail patients was 9 (IQR, 7.0–11.5), while for nonfrail patients it was 7 (IQR, 6–8; P < .001; Figure 1C). As shown in Table 3, the duration of hospitalization in frail patients was estimated to be 1.35 days longer (95% CI, 1.19–1.52; P < .0001) for frail patients than for nonfrail patients, adjusted for age, gender, race, and logistic EuroSCORE, and 1.29 days longer (95% CI, 1.12–1.50; P < .001) in a PS-adjusted model.

Individual Components of the Frailty Score (Secondary Exploratory Analyses)

Figure 2.
Figure 2.:
A, Functional decline (defined as decrease in Instrumental Activities of Daily Living score ≥2) and (B) discharge to nonhome location according to individual components of the frailty phenotype. “Positive” refers to exceeding the cutoff threshold to be considered frail for that particular criterion. P values are unadjusted comparisons.

Functional decline and discharge to a nonhome location based on each individual component of the frailty phenotype are shown in Figure 2, with slow gait speed being significantly associated with both of these outcomes. In adjusted models in which each item on the frailty score was assessed in separate models, slow gait speed was associated with functional decline (OR = 2.72 [95% CI, 1.14–6.49]; P = .03), absolute IADL decline (β coefficient = −2.12 [95% CI, −3.14 to −1.09]; P < .001), discharge to a nonhome location (OR = 4.23 [95% CI, 1.72–10.38]; P = .002), and number of days in the hospital β(β coefficient = 1.42 [95% CI, 1.25–1.61]; P < .001). Additionally, weight loss β(β coefficient = 1.19 [95% CI, 1.04–1.36]; P = .01), weakness β(β coefficient = 1.28 [95% CI, 1.12–1.46]; P < .001), and exhaustion β(β coefficient = 1.15 [95% CI, 1.02–1.30]; P = .02) were all individually associated with increasing number of days in the hospital in separate models. All other associations between individual components of the frailty phenotype and functional outcomes were not significant.

Delirium as a Potentially Modifying Factor of the Association Between Frailty and Outcomes

Delirium assessments were available on 128 patients (4 missing due to staff availability and 1 missing due to excessive sedation). The association of frailty and 1-month absolute decline in IADL score was significantly different by delirium status (P interaction = .03). For patients with delirium, frail patients had a greater decline in 1-month IADL score compared to nonfrail patients (−2.2 [95% CI, −4.0 to −0.4]; P = .02), while for patients without delirium, the difference in functional decline for frail compared to nonfrail patients was less (−0.12 [95% CI, −1.0 to 0.78]; P = .79). Delirium did not modify the association of frailty with 1-month functional decline, duration of hospitalization, or discharge location.

Postoperative Complications as Mediating Factors

A potential mediating effect of postoperative complications was examined, with the hypothesis that complications would mediate the association of frailty with functional status and hospital outcomes. In these models, the total effect is interpreted as the effect of frailty on the propensity of each individual functional outcome, without considering mediation. The size and significance of the indirect effect reflect the amount of mediation. Complications did not mediate the association between frailty and functional IADL decline (total effect = 0.53, standard error [SE] = 0.26, P = .02; indirect effect = −0.05, SE = 0.09, P = .60) or between frailty and discharge location (total effect = 0.69, SE = 0.27, P = .01; indirect effect = 0.14, SE = 0.10, P = .16). However, complications partially mediated the association of frailty and hospital duration ≥10 days (total effect = 0.77, SE = 0.25, P = .002; indirect effect = 0.30, SE = 0.15, P = .043). In this model, the association of frailty and complications was 0.54 (SE = 0.24; P = .026), and the association of complications and hospital duration ≥10 days was 0.56 (SE = 0.12; P < .001). There was no interaction (P = .45) between frailty and complications in a model with hospital duration ≥10 days as the outcome.

DISCUSSION

In this study, we found that frail patients were at increased risk of functional decline at 1 month after surgery. In exploratory analyses, we found that frail patients were at increased risk for discharge to a new nonhome location and increased duration of hospitalization compared to nonfrail patients. Delirium modified the association of frailty and change in IADL score at 1 month. Postoperative complications partially mediated the association of frailty and duration of hospitalization ≥10 days.

Frailty has been conceptualized as an inability to withstand stress, and multiple studies have found that frailty is a predictor of morbidity and mortality following cardiac surgery.3 The results of our study were concordant, with more complications after surgery occurring in the frail. In particular, frailty has been an important focus in the aortic valve replacement population, with the goal of stratifying patients into appropriate candidates for surgical or transcatheter aortic valve replacement or even nonoperative medical therapy. Yet there is a paucity of data describing the relationship between frailty and function after conventional cardiac surgery, even though preservation of functional status is a key patient-centered goal.5

In open surgery, the association between frailty and functional decline has not been consistent,22,23 with 2 systematic reviews finding few well-done studies and advocating for more research focusing on functional outcomes.5,6 One recent study found that frailty was associated with worse functional trajectories after cardiac surgery, but the frailty measure was based on an index of comorbidities.24 A second well-done multicenter study reported that frailty, as assessed by a number of measures, was associated with a composite of 1-year mortality and disability.25 In this study, the Essential Frailty Toolset (a composite measure of chair stands, cognition, anemia, and albumin) performed the best for prediction of death and disability at 1 year in comparison to 6 other frailty scales. However, intermediate functional status that may be important for recovery was not reported, nor was the contribution of complications or delirium. It is noteworthy that delirium modified the association of frailty and 1-month functional change in our study and highlights the negative synergy between these 2 geriatric syndromes. With regard to discharge location, Lee et al26 found in >3800 cardiac surgical patients that frailty was associated with new institutional discharge, but the frailty prevalence was only 4%, and the frailty definition was unique to this study. In the transcatheter aortic procedure, frailty appears to be associated with functional decline at 6 months or later.27,28

A recent systematic review6 highlighted an important role for single-component frailty instruments, and in our study, it is noteworthy that individual domains of the Fried frailty assessment differed with respect to outcomes. Specifically, slowed walking speed was the only domain that was consistently associated with both decline in IADLs and discharge to a nonhome location. These results support the measurement of at least gait speed for risk stratification in busy preoperative clinics and potentially support change in gait speed as a surrogate outcome for prehabilitation programs. These results may be particularly informative given the myriad of frailty scales that can be used.29

Identification of modifiable risk factors for functional decline may allow for targeted strategies based on the underlying biology of frailty. For instance, frail patients (especially those with delirium or at high risk for delirium) might benefit from enhanced physical therapy, early consideration of discharge planning, or even from prehabilitation programs before surgery.30 In contrast, frailty assessments that are based on an index of various comorbidities may not be as amenable to targeted interventions. In this study, delirium did modify the association of frailty with change in IADL score and thus appears to interact with frailty. However, frailty was not associated with delirium as in other studies, in large part since prefrailty was not considered as a separate exposure.10 Although frail patients were at increased risk for postoperative complications, complications did not mediate the relationship between frailty and either decline in IADL or discharge location, although complications mediated the relationship with prolonged hospitalization.

There are several strengths to this study. Frailty was assessed using the well-validated Fried scale by an experienced team. Functional outcomes were measured prospectively and reflected important patient-related outcomes. Analyses were adjusted for important confounding variables and included a PS analysis. There are several limitations. Although the frailty and outcome assessments were similar, these results reflect a combination of 2 studies, with different enrollment criteria and goals. The Fried frailty scale is thought to reflect a biologic phenotype of frailty but can be time consuming to implement. Other measures of frailty, such as the frailty index or many single-item measures, were not examined. Thus, it is unclear whether these results would be similar using other conceptualizations of frailty, especially since different frailty scales measure different underlying biology. The primary functional outcomes were assessed at 1 month, but functional status at later time points (ie, 3–12 months) may better reflect recovery. These longer-term outcomes would be particularly important to parse out what aspects of frailty are driven by cardiac comorbidities in comparison to traits such as sarcopenia. There is a possibility of residual confounding given baseline differences in comorbidity by frailty status, even though we adjusted for several variables that were determined a priori. Finally, these results reflect patients only at 1 academic medical center with important exclusion criteria that may limit generalizability.

In conclusion, frail patients undergoing cardiac surgery were at higher risk of functional decline at 1 month after surgery. Further research is needed to determine how to incorporate frailty assessment into perioperative management to optimize functional status early after cardiac surgery.

ACKNOWLEDGMENTS

The authors are grateful for the support of the Johns Hopkins Department of Anesthesiology & Critical Care Medicine Clinical Research Core (Michelle Parish, RN, Elizabeth White, RN, Mirinda Anderson, RN) and Laura Max, PA, for study conduct and regulatory support.

DISCLOSURES

Name: Mitsunori Nakano, MD.

Contribution: This author participated in the design of the study, revised the manuscript for critical intellectual content, and approved the final version.

Conflicts of Interest: None.

Name: Yohei Nomura, MD.

Contribution: This author participated in data acquisition, revised the manuscript for critical intellectual content, and approved the final version.

Conflicts of Interest: None.

Name: Giancarlo Suffredini, MD.

Contribution: This author participated in the design of the study, revised the manuscript for critical intellectual content, and approved the final version.

Conflicts of Interest: None.

Name: Brian Bush, MD.

Contribution: This author participated in the design of the study, revised the manuscript for critical intellectual content, and approved the final version.

Conflicts of Interest: None.

Name: Jing Tian, MS.

Contribution: This author participated in data acquisition, conducted data analysis, revised the manuscript for critical intellectual content, and approved the final version.

Conflicts of Interest: None.

Name: Atsushi Yamaguchi, MD, PhD.

Contribution: This author participated in the design of the study, revised the manuscript for critical intellectual content, and approved the final version.

Conflicts of Interest: None.

Name: Jeremy Walston, MD, PhD.

Contribution: This author participated in the design of the study, revised the manuscript for critical intellectual content, and approved the final version.

Conflicts of Interest: None.

Name: Rani Hasan, MD, MHS.

Contribution: This author participated in the design of the study, revised the manuscript for critical intellectual content, and approved the final version.

Conflicts of Interest: None.

Name: Kaushik Mandal, MD.

Contribution: This author participated in the design of the study, revised the manuscript for critical intellectual content, and approved the final version.

Conflicts of Interest: None.

Name: Stefano Schena, MD, PhD.

Contribution: This author participated in the design of the study, revised the manuscript for critical intellectual content, and approved the final version.

Conflicts of Interest: None.

Name: Charles W. Hogue, MD.

Contribution: This author participated in the design of the study, revised the manuscript for critical intellectual content, and approved the final version.

Conflicts of Interest: C. W. Hogue is a consultant and provides lectures for Medtronic/Covidien, Inc (Boulder, CO). He is a consultant to Merck, Inc (Kenilworth, NJ).

Name: Charles H. Brown IV, MD, MHS.

Contribution: This author participated in data acquisition, drafted and revised the manuscript for critical intellectual content, and approved the final version.

Conflicts of Interest: C. H. Brown IV has consulted for and received grant funding from Medtronic.

This manuscript was handled by: Robert Whittington, MD.

FOOTNOTES

    REFERENCES

    1. Walston J, Hadley EC, Ferrucci L, et al. Research agenda for frailty in older adults: toward a better understanding of physiology and etiology: summary from the American Geriatrics Society/National Institute on Aging Research Conference on Frailty in Older Adults. J Am Geriatr Soc. 2006;54:991–1001.
    2. Graham A, Brown CH 4th.. Frailty, aging, and cardiovascular surgery. Anesth Analg. 2017;124:1053–1060.
    3. Sündermann S, Dademasch A, Rastan A, et al. One-year follow-up of patients undergoing elective cardiac surgery assessed with the comprehensive assessment of frailty test and its simplified form. Interact Cardiovasc Thorac Surg. 2011;13:119–123.
    4. Sündermann SH, Dademasch A, Seifert B, et al. Frailty is a predictor of short- and mid-term mortality after elective cardiac surgery independently of age. Interact Cardiovasc Thorac Surg. 2014;18:580–585.
    5. Lin HS, Watts JN, Peel NM, Hubbard RE. Frailty and post-operative outcomes in older surgical patients: a systematic review. BMC Geriatr. 2016;16:157.
    6. Kim DH, Kim CA, Placide S, Lipsitz LA, Marcantonio ER. Preoperative frailty assessment and outcomes at 6 months or later in older adults undergoing cardiac surgical procedures: a systematic review. Ann Intern Med. 2016;165:650–660.
    7. Fried LP, Tangen CM, Walston J, et al.; Cardiovascular Health Study Collaborative Research Group. Frailty in older adults: evidence for a phenotype. J Gerontol A Biol Sci Med Sci. 2001;56:M146–M156.
    8. Stammers AN, Kehler DS, Afilalo J, et al. Protocol for the PREHAB study-Pre-operative Rehabilitation for reduction of Hospitalization After coronary Bypass and valvular surgery: a randomised controlled trial. BMJ Open. 2015;5:e007250.
    9. McKhann GM, Grega MA, Borowicz LM Jr, et al. Encephalopathy and stroke after coronary artery bypass grafting: incidence, consequences, and prediction. Arch Neurol. 2002;59:1422–1428.
    10. Nomura Y, Nakano M, Bush B, et al. Observational study examining the association of baseline frailty and postcardiac surgery delirium and cognitive change. Anesth Analg. 2019;129:507–514.
    11. Brown CH 4th, Probert J, Healy R, et al. Cognitive decline after delirium in patients undergoing cardiac surgery. Anesthesiology. 2018;129:406–416.
    12. Brown CH, Neufeld KJ, Tian J, et al. Effect of targeting mean arterial pressure during cardiopulmonary bypass by monitoring cerebral autoregulation on postsurgical delirium among older patients: a nested randomized clinical trial. JAMA Surg. 2019 May 22 [Epub ahead of print].
    13. Orme JG, Reis J, Herz EJ. Factorial and discriminant validity of the Center for Epidemiological Studies Depression (CES-D) scale. J Clin Psychol. 1986;42:28–33.
    14. Taylor HL, Jacobs DR Jr, Schucker B, Knudsen J, Leon AS, Debacker G. A questionnaire for the assessment of leisure time physical activities. J Chronic Dis. 1978;31:741–755.
    15. Pfeiffer E. Multidimensional Functional Assessment: The OARS Methodology. A Manual. 1978.2nd ed. Durham, NC: Duke University Center for the Study of Aging and Human Development.
    16. Inouye SK, van Dyck CH, Alessi CA, Balkin S, Siegal AP, Horwitz RI. Clarifying confusion: the confusion assessment method. A new method for detection of delirium. Ann Intern Med. 1990;113:941–948.
    17. Ely EW, Margolin R, Francis J, et al. Evaluation of delirium in critically ill patients: validation of the Confusion Assessment Method for the Intensive Care Unit (CAM-ICU). Crit Care Med. 2001;29:1370–1379.
    18. Inouye SK, Leo-Summers L, Zhang Y, Bogardus ST Jr, Leslie DL, Agostini JV. A chart-based method for identification of delirium: validation compared with interviewer ratings using the confusion assessment method. J Am Geriatr Soc. 2005;53:312–318.
    19. Abdulaziz KE, Brehaut J, Taljaard M, et al. National survey of family physicians to define functional decline in elderly patients with minor trauma. BMC Fam Pract. 2016;17:117.
    20. Rudolph JL, Inouye SK, Jones RN, et al. Delirium: an independent predictor of functional decline after cardiac surgery. J Am Geriatr Soc. 2010;58:643–649.
    21. Kurth T, Walker AM, Glynn RJ, et al. Results of multivariable logistic regression, propensity matching, propensity adjustment, and propensity-based weighting under conditions of nonuniform effect. Am J Epidemiol. 2006;163:262–270.
    22. Partridge JS, Fuller M, Harari D, Taylor PR, Martin FC, Dhesi JK. Frailty and poor functional status are common in arterial vascular surgical patients and affect postoperative outcomes. Int J Surg. 2015;18:57–63.
    23. Rønning B, Wyller TB, Jordhøy MS, et al. Frailty indicators and functional status in older patients after colorectal cancer surgery. J Geriatr Oncol. 2014;5:26–32.
    24. Kim DH, Afilalo J, Shi SM, et al. Evaluation of changes in functional status in the year after aortic valve replacement. JAMA Intern Med. 2019;179:383–391.
    25. Afilalo J, Lauck S, Kim DH, et al. Frailty in older adults undergoing aortic valve replacement: the FRAILTY-AVR Study. J Am Coll Cardiol. 2017;70:689–700.
    26. Lee DH, Buth KJ, Martin BJ, Yip AM, Hirsch GM. Frail patients are at increased risk for mortality and prolonged institutional care after cardiac surgery. Circulation. 2010;121:973–978.
    27. Schoenenberger AW, Stortecky S, Neumann S, et al. Predictors of functional decline in elderly patients undergoing transcatheter aortic valve implantation (TAVI). Eur Heart J. 2013;34:684–692.
    28. Green P, Arnold SV, Cohen DJ, et al. Relation of frailty to outcomes after transcatheter aortic valve replacement (from the PARTNER trial). Am J Cardiol. 2015;116:264–269.
    29. Smith CR. Frailty is to predictive as Jello is to wall. J Thorac Cardiovasc Surg. 2018;156:177.
    30. Arora RC, Brown CH 4th, Sanjanwala RM, McKelvie R. “NEW” prehabilitation: a 3-way approach to improve postoperative survival and health-related quality of life in cardiac surgery patients. Can J Cardiol. 2018;34:839–849.

    Supplemental Digital Content

    Copyright © 2020 International Anesthesia Research Society