Frail Patients Are at Increased Risk for Mortality and Prolonged Institutional Care After Cardiac Surgery
Background— Frailty is an emerging concept in medicine yet to be explored as a risk factor in cardiac surgery. Where elderly patients are increasingly referred for cardiac surgery, the prevalence of a frail group among these is also on the rise. We assessed frailty as a risk factor for adverse outcomes after cardiac surgery.
Methods and Results— Functional measures of frailty and clinical data were collected prospectively for all cardiac surgery patients at a single center. Frailty was defined as any impairment in activities of daily living (Katz index), ambulation, or a documented history of dementia. Of 3826 patients, 157 (4.1%) were frail. Frail patients were older, were more likely to be female, and had risk factors for adverse surgical outcomes. By logistic regression, frailty was an independent predictor of in-hospital mortality (odds ratio 1.8, 95% CI 1.1 to 3.0), as well as institutional discharge (odds ratio 6.3, 95% CI 4.2 to 9.4). Frailty was an independent predictor of reduced midterm survival (hazard ratio 1.5, 95% CI 1.1 to 2.2).
Conclusions— Frailty is a risk for postoperative complications and an independent predictor of in-hospital mortality, institutional discharge, and reduced midterm survival. Frailty screening improves risk assessment in cardiac surgery patients and may identify a subgroup of patients who may benefit from innovative processes of care.
Received December 5, 2008; accepted December 14, 2009.
Advances in modern medicine have led to people living longer and healthier lives. The fastest growing demographic in North America is that of octogenarians. As a result, the number of people of advanced age presenting for cardiac surgery is increasing.1 However, elderly patients generally present with a larger burden of disease and higher potential perioperative morbidity and mortality despite advanced pharmacotherapy and aggressive surgical management. In surgical populations, age is an independent risk factor for postoperative complications, including mortality and major adverse clinical events.2–5 Although age has been demonstrated to be a risk factor for both mortality and major morbidity in cardiac surgical outcomes, it has also been amply demonstrated in several studies that excellent outcomes can be achieved in selected populations of elderly patients.6–9
Clinical Perspective on p 978
Chronological age does not always reflect biological age, and elderly people have a range of biological status that varies from robust to frail.10,11 Frailty is an emerging concept in clinical medicine, most extensively investigated in community-dwelling geriatric populations, in which it has been demonstrated that frail patients are predisposed to falls, hospitalization, institutionalization, and mortality.12–14 Frailty has been less thoroughly investigated as a risk factor for patients undergoing procedural interventions.15 It has not been investigated as a risk factor for cardiac surgical intervention. We proposed that preoperative assessment of patients for frailty would result in more accurate risk assessment for cardiac surgery and may identify a challenging subset of patients who would benefit from novel processes of care.
In an effort to determine the degree of frailty in our patient population, in June 2004, we initiated data collection concerning the Katz index of activities of daily living (ADL), an internationally validated measure of dependency in elderly patients.16 In addition, we collected data on independence in ambulation and the presence of dementia. The goal of this study was to determine the impact of these measures of frailty on mortality and the need for prolonged postoperative institutional care.
We identified all patients undergoing cardiac surgery at the Queen Elizabeth II Health Sciences Centre in Halifax, Nova Scotia, Canada, between June 2004 and December 2007. The Queen Elizabeth II Health Sciences Centre is the sole cardiac surgical center serving the entire province of Nova Scotia as well as parts of the surrounding provinces. For patients undergoing more than 1 cardiac surgical procedure during the study period, only the index procedure was considered.
The study protocol was approved by the local institutional research ethics board.
Data Collection and Variable Selection
The Maritime Heart Center Cardiac Surgery Registry is a detailed clinical database that has prospectively captured pre-, intra-, and postoperative information on all of the cardiac surgery patients at the Queen Elizabeth II Health Sciences Centre since March 1995, with ongoing auditing to ensure data accuracy. Collection of preoperative measures of frailty within the registry began in June 2004. Trained nurse practitioners or clinical associates administered questions to patients and family members concerning the Katz index of ADL (independence in feeding, bathing, dressing, transferring, toileting, and urinary continence), as well as independence in ambulation (no walking aid or assist required). Additionally, clear evidence of a previous diagnosis of dementia by a specialist physician was sought from the patient’s record. Patients with any deficiency in the Katz index of ADL, in ambulation, or with a previous diagnosis of dementia were defined as frail for the purposes of this study.
Other preoperative characteristics included age at time of surgery, sex, diabetes, chronic obstructive pulmonary disease (COPD, defined as pharmacological therapy for the treatment of chronic pulmonary compromise, or forced expiratory volume in 1 second <75% of predicted value), congestive heart failure (CHF, defined as at least 3 of presence of dyspnea, rales thought to represent pulmonary congestion, peripheral edema, cardiomegaly on chest x-ray, or chest x-ray compatible with interstitial edema), preoperative renal failure (RF, defined as serum creatinine >176 μmol/L), cerebrovascular disease (CVD, defined as any transient ischemic attack, reversible ischemic neurological deficit, cerebrovascular accident or stroke, history of cerebrovascular surgery, or any carotid disease, including asymptomatic carotid disease), peripheral vascular disease (defined as history of aneurysm and/or occlusive vascular disease with or without previous extracardiac vascular surgery), left ventricular ejection fraction <40%, urgency of surgery (elective [stable at home], in-house [requiring hospitalization until the time of surgery], urgent [requiring surgery within 24 hours to minimize further clinical deterioration], or emergent [no delay in surgery]), body mass index (BMI, kg/m2), complexity of procedure (isolated coronary artery bypass grafting versus other cardiac surgery), and reoperation (repeat cardiac surgery).
For analysis of midterm survival, clinical data from the Maritime Heart Center Cardiac Surgery Registry were linked to the administrative Vital Statistics database for the province. This database is created by the Department of Health and housed by the Population Health Research Unit in the Department of Community Health and Epidemiology at Dalhousie University. Links with administrative data were available for all of the cardiac surgery patients in this study who resided in Nova Scotia at the time of surgery and were eligible for provincial medical insurance.
Process of Surgical Care
The great majority of cases were performed on pump (<1.4% of coronary artery bypass grafting cases during the study period were off-pump coronary artery bypass grafting). In the majority of cases, cardiopulmonary bypass with modest hypothermia (32°C) was used with a crystalloid prime. Cannulation was routinely performed by distal ascending aorta/proximal arch and right atrium. Synthesis oxygenators were used. Antegrade, cold blood, 4:1 ratio cardioplegia was used in cases without significant aortic insufficiency or in the case of aortic valve surgery by direct osteal delivery. Retrograde cardioplegia was added as clinically indicated (poor left ventricular function, aortic insufficiency).
Primary outcomes in this study were in-hospital mortality, midterm all-cause mortality (including in-hospital and after discharge deaths), and discharge to an institution (community hospital, rehabilitation or restorative care facility, or skilled nursing facility, among patients discharged alive). Secondary in-hospital outcomes included transfusion, low cardiac output syndrome, sepsis, pneumonia, permanent stroke, delirium, prolonged ventilation (≥24 hours), postoperative RF (serum creatinine >176 μmol/L, not present preoperatively), and prolonged postoperative stay (length of stay, ≥10 days).
Preoperative characteristics and postoperative in-hospital outcomes in frail versus non-frail patients were compared univariately using χ2 tests or Fisher exact tests for categorical variables and t tests or Wilcoxon rank sum tests for continuous variables. To satisfy the assumption of linearity inherent in logistic regression and proportional hazards modeling, age and BMI were transformed using a restricted cubic spline function with 3 knots placed at the 10th, 50th, and 90th percentiles of each distribution.17
Clinically relevant preoperative variables with χ2 P<0.20 were included in the fully adjusted multivariable models. Colinearity was assessed through correlation matrices as Pearson r≥0.3; only 1 variable for each correlated pair was retained based on clinical importance.
The association of frailty with in-hospital mortality and institutional discharge was examined by logistic regression models fully adjusted for relevant prognostic variables. Predictive accuracy of each model was assessed by the receiver operating characteristic curve.18 A bootstrap procedure was performed on 200 subsamples, and the 95% CI of the receiver operating characteristic was obtained from the 2.5th and 97.5th percentiles of the bootstrap distribution.
The impact of frailty on all-cause mortality was examined by a fully adjusted Cox proportional hazards model. To assess the proportional hazards assumption, a time-dependent covariate was created for each predictor as a function of survival time, and tests of proportionality were applied. For variables that did not satisfy the assumption of proportional hazards, time-dependent covariates were included in the models. Statistical analysis was performed using SAS software version 9.2 (SAS, Cary, North Carolina).
The authors had full access to the data and take full responsibility for its integrity. All of the authors have read and agree to the article as written.
Between June 2004 and December 2007, a total of 3826 patients had cardiac surgery at the Queen Elizabeth II Health Sciences Centre. Of these 3826 patients, 64 (1.7%) demonstrated a deficiency in the Katz index of ADL, 124 patients (3.2%) had some degree of dependence in ambulation, and 22 patients (0.6%) carried a previous diagnosis of dementia. The frail group comprised 157 (4.1%) patients having at least 1 of these deficits. Among the frail patients, 106 (67.5%) had a deficit in only 1 of these categories, 49 (31.2%) in 2 categories, and 2 (1.3%) in all 3 categories, In-hospital deaths in each frail category were 15 (14.2%), 7 (14.3%), and 1 (50%), respectively.
The baseline preoperative characteristics are presented in Table 1. Frail patients were older than nonfrail patients with a median age of 71 years (interquartile range 61 to 78) versus 66 years (interquartile range 57 to 74; P=0.0001), although the age range was similar in both groups. Frail patients were more likely to be female and presented with greater comorbidity burden, including higher rates of diabetes, COPD, CHF, RF, and CVD. Furthermore, frail patients presented with greater acuity and underwent more complex operative procedures than the nonfrail patients (all P<0.05, Table 1). However, the 2 groups did not differ significantly in other characteristics, as shown in Table 1.
Age and BMI were used as continuous variables in analysis. To satisfy the assumption of linearity in logistic regression and proportional hazards modeling, a restricted cubic spline transformation was applied to age, with knots at 49, 66, and 80 years, corresponding to the 10th, 50th, and 90th percentiles of the distribution. Similarly, a restricted cubic spline transformation was applied to BMI, with knots at 22.7, 28.1, and 35.7 kg/m2, corresponding to the 10th, 50th, and 90th percentiles of the distribution.
A total of 3254 cases (85%) were linked with the Vital Statistics administrative database maintained by the province, with follow-up through March 2008. Median follow-up time was 1.8 years (interquartile range, 0.9 to 2.8 years). Cases that did not link with the Vital Statistics database were 555 patients not residing in Nova Scotia at the time of surgery and 17 who resided in the province but were ineligible for provincial medical insurance. Compared with cases that linked, cases that did not link were similar with regard to frailty, age, sex, comorbidities, and urgency of surgery. However, the cases that did not link were more likely to be complex cases or require reoperation. This reflects the referral pattern for out-of-province patients in our practice.
Unadjusted In-Hospital Outcomes
Mortality, discharge to an institution, transfusion, low cardiac output syndrome, sepsis, pneumonia, delirium, prolonged ventilation, postoperative RF, and prolonged length of stay were more prevalent among frail than nonfrail patients (all P<0.05, Table 2).
In unadjusted analysis, frailty was associated with increased in-hospital mortality (nonfrail n=164 [4.5%], frail n=23 [14.7%]; P<0.0001, Table 2). Other preoperative factors associated univariately with increased mortality included age, female sex, CHF, COPD, RF, CVD, peripheral vascular disease, BMI, urgency of surgery, complex procedure, and reoperation (all P<0.01, data not shown). By logistic regression analysis, frailty was identified as an independent risk factor for in-hospital mortality (odds ratio [OR] 1.8, 95% CI 1.1 to 3.0; Table 3).
Frailty was predictive of discharge to an institution for prolonged care in univariate analyses (nonfrail n=316 [9.0%], frail n=65 [48.5%]; P<0.0001; Table 2). Other preoperative factors having a univariate association with prolonged institutional care were age, female sex, diabetes, COPD, CHF, RF, CVD, peripheral vascular disease, BMI, urgency of surgery, and complex procedure (all P<0.02, data not shown). In logistic regression, frailty was an independent predictor of institutional discharge (OR 6.3, 95% CI 4.2 to 9.4; Table 4).
By univariate analysis, frailty was associated with reduced midterm survival (nonfrail n=330 deaths [10.6%], frail n=41 deaths [29.5%]; P<0.0001). Other preoperative factors associated univariately with reduced midterm survival were age, female sex, diabetes, COPD, CHF, RF, CVD, peripheral vascular disease, BMI, urgency of surgery, complex procedure, and reoperation (all P<0.01, data not shown). After adjusting for other relevant risk factors by Cox proportional hazards modeling, frailty was identified as an independent risk factor for reduced midterm survival (hazard ratio 1.5, 95% CI 1.1 to 2.2, Table 5). Adjusted survival at 2 years was 84% (95% CI 79% to 91%) in frail and 89% (95% CI 85% to 93%) in nonfrail patients (Figure).
The interaction of frailty and age was examined and was not a significant predictor of any of the 3 primary outcomes. Therefore, this interaction term was not included in the final models.
Ours is the first study, to our knowledge, to demonstrate frailty as a risk factor for adverse clinical outcomes after cardiac surgery, including mortality and prolonged institutional care. Analyzing a consecutive cohort of all of the patients undergoing cardiac surgery in a single institution, we found that frailty markedly increased the risk for these outcomes. Furthermore, the effect of frailty was independent of age. Several aspects of this study merit comment.
There is no generally agreed-on definition of frailty because it is an emerging concept and has been operationally defined with a variety of scales. It is agreed that frailty is a biological state or syndrome of decreased resistance to stressors that results from deterioration in multiple physiological systems.12,19,20 Frailty includes a constellation of clinical attributes, including loss of skeletal muscle mass, low activity levels, and poor endurance.12 It has been demonstrated that frailty and the onset of dependence in ADLs are strongly associated.13 This informed our decision to use the Katz index of ADL16 as part of our frailty measure, a widely accepted measure of overall dependency in elderly people.15 We also included information concerning independence in ambulation and a history of dementia, both measures having been associated with adverse clinical outcomes in a number of studies.21–23
The majority of frailty studies have used a geriatric patient cohort, whether community dwelling or institutionalized, rather than patients undergoing surgical interventions. These studies have demonstrated that frailty is predictive of adverse health outcomes, including falls, hospitalization, institutionalization, and mortality.12–14 A study by Dasgupta et al evaluated the outcomes of patients undergoing noncardiac surgical procedures and concluded that frailty was independently associated with postoperative complications, increased length of hospitalization, and inability to be discharged home.15 However, no studies take measures of frailty into account in patients undergoing cardiac surgery. We hypothesized that frailty is associated with an increased risk of mortality and prolonged institutional care in patients undergoing cardiac surgery.
There were significant differences in the preoperative characteristics between the frail and nonfrail populations. Overall, the frail patients were sicker (as demonstrated by an increased burden of disease), older, and, interestingly, more likely to undergo complex procedures. This latter observation may reflect a referral bias in that clinicians withhold referring frail elderly patients for coronary artery bypass grafting unless they have an insurmountable burden of disease precluding medical management.
The impact of frailty on mortality is perhaps expected in that by definition frail patients have diminished physiological reserve and capacity to maintain homeostasis. Furthermore, at baseline the frail patients had a decreased ability to mobilize and ambulate, predisposing them to postoperative pneumonia, reintubation, and urinary tract infections related to prolonged catheterization. All of these factors increase the risk of a protracted recovery and prolonged postoperative institutional care.
In this study, we have demonstrated the utility of easily applied measures of frailty in predicting both mortality and prolonged institutional care. Importantly, we considered even 1 impairment in any of these measures as evidence of frailty. Despite this low threshold, our operational definition of frailty had a marked impact on the outcomes of interest. Thus, our definition of frailty may provide a relatively insensitive measure of frailty, unable to distinguish more subtle risks for these adverse outcomes. Other more comprehensive measures of frailty, although more time-consuming and thus costly to administer, may be more appropriate for measuring preoperative frailty. Ideally, in addition to the measures we used, frailty assessment would include an assessment of nutrition, depression, social support, and cognitive impairment. Other authors have used an instrument that measures these parameters and demonstrated a similar relationship between frailty and outcome in noncardiac surgery patients.14,24 Thus, more sensitive measures of preoperative frailty may allow us to better discern more subtle risk for adverse outcomes.
Analysis of longitudinal data demonstrated that frail patients experienced increased in-hospital mortality, increased rates of institutional discharge, and reduced midterm survival. These data will help to inform frail patients and their clinicians about the advisability of surgical intervention. Our findings raise the question of whether it is justifiable to perform cardiac surgical operations on frail patients in the first instance. Our own data indicate that even in our center, where cases are vetted in a peer-reviewed, multidisciplinary “cath conference,” patients who qualify as frail are being operated on. Thus, it is important for cardiac caregivers to engage frail patients in fully informed consent. Recent studies indicate that the majority of surgical consent discussions fail to meet criteria for fully informed consent.25,26 It may be that decision aids designed to convey risks and benefits of cardiac surgical intervention in frail patients, particularly with regard to increased rates of prolonged institutional care, hospital mortality, and reduced long-term survival, would afford a more fully informed, shared decision-making process. It is equally important that the providers of cardiac surgical care work to mitigate the risks faced by frail patients through modification of processes of care. Possibilities include programs to increase mobility in frail patients in a safe, supervised manner, and to address nutritional deficiencies prior to intervention.
Several limitations should be noted. First, this is a single-center, retrospective study. The relatively recent implementation of measures of frailty in our database limits our power to discriminate more fully about other major adverse outcomes. Furthermore, prolonged institutional care, although a reasonable surrogate for potential failure to return to a fully independent life, is not conclusive in this regard. Information concerning the late functional status of patients, beyond discharge from the secondary institution, would more fully inform us about the potential for eventual recovery of independence in these patients.
In summary, we have demonstrated that preoperative functional assessment of frailty increases our ability to predict patients at greater risk for mortality, prolonged institutional care, and reduced midterm survival. It identifies a subset of patients who need to be more fully informed about risks, thus improving informed, shared decision making about the advisability of surgical intervention. It may also identify a subset of patients who could benefit from altered processes of care designed to offset the burden of frailty.
Sources of Funding
This study was supported in part by a Canadian Institutes for Health Research Team Grant to the Canadian Cardiovascular Outcomes Research Team.
Rosamond W, Flegal K, Furie K, Go A, Greenlund K, Haase N, Hailpern SM, Ho M, Howard V, Kissela B, Kittner S, Lloyd-Jones D, McDermott M, Meigs J, Moy C, Nichol G, O'Donnell C, Roger V, Sorlie P, Steinberger J, Thom T, Wilson M, Hong Y, Am Heart Association Statistics Committee and Stroke Statistics Subcommittee. Heart disease and stroke statistics:2008 update: a report from the Am Heart Association Statistics Committee and Stroke Statistics Subcommittee. Circulation. 2008; 117: e25–e146.
Arozullah AM, Khuri SF, Henderson WG, Daley J. Participants in the National Veterans Affairs Surgical Quality Improvement Program: development and validation of a multifactorial risk index for predicting postoperative pneumonia after major noncardiac surgery. Ann Intern Med. 2001; 135: 847–857.
Alibhai SM, Leach M, Tomlinson G, Krahn MD, Fleshner N, Holowaty E, Naglie G. 30-day mortality and major complications after radical prostatectomy: influence of age and comorbidity. J Natl Cancer Inst. 2005; 97: 1525–1532.
Baskett R, Buth K, Ghali W, Norris C, Maas T, Maitland A, Ross D, Forgie R, Hirsch G. Outcomes in octogenarians undergoing coronary artery bypass grafting. CMAJ. 2005; 172: 1183–1186.
Huber CH, Goeber V, Berdat P, Carrel T, Eckstein F. Benefits of cardiac surgery in octogenarians: a postoperative quality of life assessment. Eur J Cardiothorac Surg. 2007; 31: 1099–1105.
Fried LP, Tangen CM, Walston J, Newman AB, Hirsch C, Gottdiener J, Seeman T, Tracy R, Kop WJ, Burke G, McBurnie MA, 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.
Efron B. Nonparametric estimates of standard error: the jackknife, the bootstrap and other methods. Biometrika. 1981; 68: 589–599.
Campbell AJ, Buchner DM. Unstable disability and the fluctuations of frailty. Age Ageing. 1997; 26: 315–318.
De Rooij SEJA, Govers AC, Korevaar JC, Giesbers AW, Levi ML, de Jonge E. Cognitive, functional, and quality-of-life outcomes of patients aged 80 and older who survived at least 1 year after planned or unplanned surgery or medical intensive care treatment. J Am Geriatr Soc. 2008; 56: 816–822.
Lang PO, Heitz D, Hédelin G, Dramé M, Jovenin N, Ankri J, Somme D, Novella JL, Gauvain JB, Couturier P, Voisin T, De Wazière B, Gonthier R, Jeandel C, Jolly D, Saint-Jean O, Blanchard F. Early markers of prolonged hospital stays in older people: a prospective, multicenter study of 908 inpatients in French acute hospitals. J Am Geriatr Soc. 2006; 54: 1031–1039.
Rolfson DB, Majumdar SR, Tsuyuki RT, Tahir A, Rockwood K. Validity and reliability of the Edmonton Frail Scale. Age Ageing. 2006; 35: 526–529.
Brezis M, Israel S, Weinstein-Birenshtock A, Pogoda P, Sharon A, Tauber R. Quality of informed consent for invasive procedures. Int J Qual Health Care. 2008; 20: 352–357.
Frailty is an emerging concept in clinical medicine, most extensively investigated in community dwelling geriatric populations where it has been demonstrated that frail patients are predisposed to falls, hospitalization, institutionalization, and mortality. Frailty has been less thoroughly investigated as a risk factor for patients undergoing procedural interventions. Its role as a risk factor for cardiac surgical intervention has not been investigated previously. We have prospectively examined our cardiac patient population for frailty as measured by The Katz Index of Activities of Daily Living, an internationally validated measure of dependency in elderly patients, as well as for deficits in independent ambulation and for documented history of dementia. Patients having any defect in these measures were defined as frail. We demonstrated that frailty was an independent predictor of in-hospital mortality, reduced medium term survival, as well as discharge to institutional care rather than home. Although age was also a predictor of these outcomes, after adjusting for age, frailty remained an independent predictor of these outcomes. Our data have implications for frail patients who have cardiac disease amenable to surgical repair. As a result of our work, both patients and surgeons can be better informed about potential adverse outcomes before arriving at a decision to go ahead with cardiac surgical intervention. Additionally, frail patients would potentially benefit from altered approaches to care that could mitigate the risks these patients face, for example, in improving mobilization and nutrition prior to a planned intervention.
Earlier versions of this article were presented at the Canadian Cardiovascular Congress 2008, Toronto, Ontario, Canada, October 25–29, 2008, and at the American Heart Association Scientific Sessions 2008, New Orleans, Louisiana, November 8–12, 2008.