- Email: [email protected]
Patient frailty predicts worse perioperative outcomes and higher cost after radical cystectomy
Surgical Oncology 32 (2020) 8–13
Contents lists available at ScienceDirect
Surgical Oncology journal homepage: http://www.elsevier.com/locate/suronc
Patient frailty predicts worse perioperative outcomes and higher cost after radical cystectomy worse radical cystectomy outcomes in frails Carlotta Palumbo a, b, *, Sophie Knipper a, c, Angela Pecoraro a, d, Giuseppe Rosiello a, e, Stefano Luzzago a, f, Marina Deuker a, g, Zhe Tian a, Shahrokh F. Shariat h, i, j, k, l, Claudio Simeone b, Alberto Briganti e, Fred Saad a, Alfredo Berruti m, Alessandro Antonelli b, Pierre I. Karakiewicz a a
Cancer Prognostics and Health Outcomes Unit, University of Montreal Health Center, Montreal, Quebec, Canada Urology Unit, ASST Spedali Civili of Brescia. Department of Medical and Surgical Specialties, Radiological Science and Public Health, University of Brescia, Italy Martini Klinik, University Medical Center Hamburg-Eppendorf, Hamburg, Germany d Department of Urology, San Luigi Gonzaga Hospital, University of Turin, Orbassano, Turin, Italy e Division of Experimental Oncology/Unit of Urology, Urological Research Institute (URI), IRCCS San Raffaele Scientific Institute, Milan, Italy, Vita-Salute San Raffaele University, Milan, Italy f Department of Urology, European Institute of Oncology, Milan, Italy g Department of Urology, University Hospital Frankfurt, Frankfurt, Germany h Department of Urology, Comprehensive Cancer Center, Medical University of Vienna, Vienna, Austria i Department of Urology, Weill Cornell Medical College, New York, NY, USA j Department of Urology, University of Texas Southwestern, Dallas, TX, USA k Department of Urology, Second Faculty of Medicine, Charles University, Prague, Czech Republic l Institute for Urology and Reproductive Health, I.M. Sechenov First Moscow State Medical University, Moscow, Russia m Medical Oncology Unit, ASST Spedali Civili of Brescia. Department of Medical and Surgical Specialties, Radiological Science and Public Health, University of Brescia, Italy b c
A R T I C L E I N F O
A B S T R A C T
Keywords: Radical cystectomy Frailty Complications
Background: Relatively few studies investigated the importance of frailty in radical cystectomy (RC) patients. We tested the ability of frailty, using the Johns Hopkins Adjusted Clinical Groups indicator, to predict early peri operative outcomes after RC. Methods: RC patients were identified within the National Inpatient Sample database (2000–2015). The effect of frailty, age and Charlson Comorbidity Index were tested in five separate multivariable models predicting: (1) complications, (2) failure to rescue (FTR), (3) in-hospital mortality, (4) length of stay (LOS) and (5) total hospital charges (THCs). All models were weighted and adjusted for clustering, as well as all available patient and hospital characteristics. Results: Of 23,967 RC patients, 5833 (24.3%) were frail, 7721 (32.2%) were aged �75 years and 2832 (11.8%) had CCI �2. Frailty, age �75 years and CCI �2 were non-overlapping in 86.3% of the cohort. Any two or three of these features were recorded in 12.4 and 1.3%, respectively. Frailty was an independent predictor of all five examined endpoints and the magnitude of its association was stronger or at least equal than that of age �75 years and CCI �2. Conclusion: Frailty, advanced age and comorbidities represent non-overlapping patients’ characteristics. Of those, frailty represents the most consistent and strongest predictor of early adverse outcomes after RC. Ideally, all three indicators should be considered in retrospective, as well as prospective analyses. Pre-surgical recog nition of frail patients should be ideally incorporate in clinical practice in order to address these patients to multimodal pre-habilitation programs that may potentially improve the perioperative prognosis.
* Corresponding author. Urology Unit, ASST Spedali Civili of Brescia. Department of Medical and Surgical Specialties, Radiological Science and Public Health, University of Brescia, Italy, Piazzale Spedali Civili 1, 25123, Brescia, Italy. E-mail address: [email protected] (C. Palumbo). https://doi.org/10.1016/j.suronc.2019.10.014 Received 23 August 2019; Received in revised form 11 October 2019; Accepted 24 October 2019 Available online 25 October 2019 0960-7404/© 2019 Published by Elsevier Ltd.
C. Palumbo et al.
Surgical Oncology 32 (2020) 8–13
1. Introduction
Comorbidity Index (CCI) was defined according to the Deyo adaptation for ICD-9 codes-based databases [26] (Supplementary Table 4) and categorized as CCI 0–1 vs. CCI �2. It is of note that frailty defining di agnoses were distinct from comorbidity diagnoses, with the exception of dementia, for which the ICD-9 codes 290.x that defined dementia in frailty were also part of the definition of dementia in CCI. For purpose of multivariable adjustment, we relied on patient and hospital characteristics that are made available by the NIS. These included gender, ethnicity (Caucasian, African American and Others), year of surgery, income and insurance status (private insurance, Medi care, Medicaid, and other [self-pay]), hospital region (Northeast, Mid west, South, West), hospital teaching vs. non-teaching status and hospital annual volume (low, medium and high), which represents the number of RCs performed at each participating institution during each study calendar year.
Radical cystectomy (RC) is associated with elevated rates of com plications and non-negligible perioperative mortality [1–5]. Age and comorbidities represent established predictors of these outcomes [6,7]. Moreover, advanced age and elevated comorbidities also predicted longer length of stay and higher hospitalization costs [6]. Relative to these established and well-documented predictors of worse periopera tive outcomes after RC, only recently frailty emerged as a prognostic indicator of increasing risk of adverse health events and mortality after surgery [8,9]. Relatively few studies investigated its importance in the context of RC literature [10–19]. To address this limitation, we tested the effect of frailty alongside age and comorbidities on complications, failure to rescue (FTR), in-hospital mortality, length of stay (LOS) and total hospital charges (THCs) within a large cohort of patients treated with RC, identified within the National Inpatient Sample (NIS) database (2000–2015).
2.4. Statistical analyses
2. Materials and methods
The three predictors of interest were frailty (non-frail vs. frail), age (<75 vs. �75 years) and CCI (CCI 0–1 vs. CCI �2). First, medians and interquartile ranges, as well as frequencies and proportions were re ported for continuous and categorical variables, respectively. Second, we examined the proportions and the estimated annual percentage changes (EAPCs) of RC rates according to frailty, age �75 years and CCI �2. Third, five separate endpoints were addressed in multivariable an alyses: (1) overall complications, (2) FTR, (3) in-hospital mortality, (4) LOS and (5) THCs. Specifically, overall complications, FTR and inhospital mortality were addressed in three separate multivariable lo gistic regression models. Multivariable Poisson regression models focused on LOS. Multivariable linear regression models focused on THCs. All multivariable models relied on weighting using a Generalized Estimating Equation (GEE) function to provide more accurate national estimates based on NIS-provided weights [20]. All models were then adjusted for clustering at hospital level, as well as for age, CCI, frailty, gender, year of surgery, ethnicity, insurance status, teaching status, hospital volume, region, income, lymph node dissection and urinary diversion. Within each multivariable model all the three predictors of interest were considered simultaneously along with all the covariates without either forward or backward variable selection. All statistical tests were two-sided with a level of significance set at p < 0.05. All analyses were performed using the R software environment for statistical computing and graphics (version 3.4.1; http://www. r-project. org/).
2.1. Source of data and study population We relied on the National Inpatient Sample (NIS) database (2000–2015) [20] that is composed of longitudinal hospital inpatient databases from the Healthcare Cost and Utilization Project family and includes 20% of United States inpatient hospitalizations. Institutional ethical board had reviewed the research project and it meets re quirements for protection of human subjects. We focused on patients aged�18 years with primary diagnosis of non-metastatic carcinoma of the urinary bladder (International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] codes 188.x), treated with RC (primary procedure ICD-9-CM codes 57.6, 57.7� or 688). Secondary procedure codes were used to identify lymph node dissection (ICD-9-CM codes 40.3 or 40.5), as well as use of ileal conduit or continent ortho topic urinary diversion (ICD-9-CM codes 56.51 and 57.87, respectively). 2.2. Outcomes of interest We focused on four early perioperative outcomes, namely overall complications, failure to rescue (FTR), in-hospital mortality and length of stay (LOS), as well as on total hospital charges (THCs). According to previous established methodology, overall complications were defined using secondary ICD-9-CM diagnostic codes and represented the sum of intraoperative and all postoperative complications (cardiac, respiratory, vascular, operative wound, genitourinary, gastrointestinal, infectious, blood transfusions, miscellaneous medical and miscellaneous surgical, Supplementary Table 1). FTR was defined as mortality after a major complication [21] (Supplementary Table 2). In-hospital mortality was defined from discharge disposition. Length of stay was calculated by subtracting the admission date from the discharge date. Finally, inflation-adjusted THCs were defined according to NIS methodology [20].
3. Results 3.1. Patients’ characteristics and temporal trends Between 2000 and 2015, we identified 23,967 non-metastatic car cinoma of the urinary bladder patients treated with RC (Table 1). Of these, 5833 (24.3%) patients were frail, 7721 (32.2%) were aged �75 years and 2832 (11.8%) had CCI �2. Presence of frailty, age �75 years and CCI �2 represented unique and non-overlapping features in 86.3% of the cohort. Conversely, combination of frailty and either �75 years or CCI �2 was recorded in 12.4% of patients. Finally, presence of all the three features was recorded in 1.3% of the cohort (Fig. 1). Relative to non-frail, frail patients were older (patients aged �75 years 36.8% vs. 30.7%). Minimally-invasive surgery, which included both laparoscopic and robot-assisted procedures, was performed in 9.2% of frail vs. 8.3% of non-frail. Ileal conduit was performed in 72.3% of frail vs. 67.2% of non-frail. Time trends (Fig. 2) revealed an increase of frail patients (EAPC þ3.1%, p < 0.001) and, to a lesser extent, of those with CCI �2 (EAPC þ1.6%, p ¼ 0.003). Conversely, a decrease of patients �75 years (EAPC -0.7%, p < 0.001) was recorded.
2.3. Patients and hospital characteristics Frailty was defined according to the Johns Hopkins Adjusted Clinical Groups (ACG) frailty-defining diagnoses indicator [22], an instrument designed and validated for use in health administrative data [23], based on 10 clusters of frailty-defining diagnoses that comprise the Johns Hopkins ACG frailty-defining diagnosis indicator [22]. It allows to ac count for cognitive, functional and social impairments, since it encom passes dementia, vision impairment, malnutrition, urinary and fecal incontinence, difficulty in walking, falls and social support needs. Frailty was defined using patients’ ICD-9 codes available in the NIS, as previ ously reported [23–25] (Supplementary Table 3), and categorized as frail vs. non-frail. Age was dichotomized as <75 vs. �75 years. Charlson 9
C. Palumbo et al.
Surgical Oncology 32 (2020) 8–13
1.32, p < 0.001), age �75 years (RR 1.06, p < 0.001) and CCI �2 (RR 1.12, p < 0.001) achieved independent predictor status. In multivariable linear regression models predicting THCs, only frail status (þ8003.3, p < 0.001), and CCI �2 (þ3910.7, p < 0.001) achieved independent predictor status.
Table 1 Descriptive characteristic of 23,967 non-metastatic carcinoma of the urinary bladder patients treated with radical cystectomy, identified within the Nation wide Inpatient Sample database between 2000 and 2015. Variable Age, years Age group Frailty Charlson Comorbidity Idex Gender Race Approach Type of diversion Lymph-node dissection Insurance
Region
Income
Annual hospital volume Teaching status Bed size
Number (%) Median (IQR) <75 years �75 years Non-frail Frail 0–1 �2 Female Male Caucasian African-American Other Open Minimally-invasive Ileal conduit Orthotopic diversion Other Performed Not-performed Medicare Medicaid Private Other Midwest Northeast South West First Quartile Second Quartile Third Quartile Fourth Quartile Low Medium High Teaching Non-Teaching Small Medium Large
70 (62–77) 16,246 (67.8) 7721 (32.2) 18,134 (75.7) 5833 (24.3) 21,135 (88.2) 2832 (11.8) 5014 (20.9) 18,953 (79.1) 17,193 (71.7) 997 (4.2) 5777 (24.1) 21,925 (91.5) 2042 (8.5) 16,404 (68.4) 1651 (6.9) 5912 (24.7) 2832 (11.8) 7010 (29.2) 14,887 (62.1) 1011 (4.2) 7089 (29.6) 980 (4.1) 6312 (26.3) 4746 (19.8) 8551 (35.7) 4358 (18.2) 4933 (20.6) 6067 (25.3) 6145 (25.6) 6822 (28.5) 7465 (31.1) 8122 (33.9) 8380 (35) 17,050 (71.1) 6917 (28.9) 2474 (10.3) 4380 (18.3) 17,113 (71.4)
4. Discussion We hypothesized that frailty represents an independent predictor of adverse outcomes in RC patients. To test this hypothesis, we tested frailty along with age and comorbidities on these endpoints, namely overall complication, FTR in-hospital mortality, LOS and THCs. Our analyses resulted in several noteworthy findings. First, frail patients represented 24.3% of our cohort. This rate is consistent with previous NSQIP-based studies [10–16,18], where rates of frail patients ranged from 22.0% [16] to 26.9% [13]. Relative to non-frail, frail patients were older (30.7 vs. 36.8% in age category �75 years). Nonetheless, the rates of frailty among patients younger than 75 years were not negligible. These observations indicate that frailty may affect also younger patients. In consequence, frailty represents an age independent characteristics. Second, frailty, age �75 years and CCI �2 represented unique and non-overlapping patient characteristics in 86.3% of the population. Inconsequence, frailty requires independent quantification and its presence or absence cannot be derived from either age or comorbidities. To the best of our knowledge, no previous investigations examined this concept in a RC population. Third, time trend analyses demonstrated significant increasing rates of frailty in RC patients over time (EAPC þ3.1%, p < 0.001). Also CCI �2 increased but to a lower rate (EAPC þ1.6%, p ¼ 0.003). Finally, the proportion of patients aged �75 years decreased. Frailty is gaining in importance among RC patients, in a rate that exceeds those of both higher comorbidity index and more advanced age. In consequence, it could be argued that frailty should be given at least the same importance of age and elevated number of comorbidities. To the best of our knowledge, this aspect of frailty has not been previously reported. Fourth, frailty achieved independent predictor status as a unique risk variable, even after adjustment for advanced age and comorbidities in the analyses of the five endpoints. This observation further militate in favor of considering frailty as a unique risk variable. Moreover, on average the magnitude of increase of the risk of all five adverse peri operative outcomes was greater or equal for frailty than for age or co morbidity. For example, when considering the overall complications, the magnitude of risk increase of frailty exceeds that of age and was similar to that of comorbidity. Additionally, concerning both LOS and THCs, the magnitude of risk increase of frailty exceeds that of both age and comorbidity. Conversely, in FTR and mortality, frailty was weaker relative to age, but similar to comorbidity. Taken together, presence of frailty appears to be a stronger indicator of three out of five examined endpoints, with the exception of FTR and mortality, where patients aged �75 years were at higher risk than frail patients. In consequence, the importance of frailty is at least similar, if not greater, to that of age and comorbidity. Therefore, consideration of frailty should be invariably included in analyses of RC adverse perioperative outcomes and possibly in similar analyses that are based on older patients. Previous studies [12,14–19] showed that frailty was independently associated with worse outcomes in different context from the current study (30- and 90-day events vs. readmission setting vs. in-hospital events). Nonetheless, these analyses concur in the importance of frailty in predicting adverse outcomes after RC and jointly indicate that frailty should be included alongside with age and CCI as important in dicator of adverse clinical outcomes at both initial hospital stay, as well as at readmission. Moreover, to the best of our knowledge, we are the first to report an association between frailty and FTR in cystectomy patients, since none of the previous studies investigated this topic. This observation is important, since the concept of FTR may be particularly
AbbreviationsIQR ¼ interquartile range.
3.2. Crude rates of outcomes of interest Crude rates of the five outcomes stratified according to frailty, age and CCI categories are summarized in Table 2. Frail patients exhibited higher rates of complications (67.9 vs. 55.8%, p < 0.001), FTR (2.2 vs. 1.2%, p < 0.001) and in-hospital mortality (2.4 vs. 1.5%, p < 0.001), as well as longer LOS (p < 0.001) and higher THCs (p < 0.001). Patients aged �75 years exhibited higher rates of complications, FTR and inhospital mortality, as well as longer LOS and higher THCs were recor ded (all p < 0.001). Patients with CCI �2 exhibited higher rates of complications, FTR and in-hospital mortality increased, as well as longer LOS and higher THCs were reported (all p < 0.001). 3.3. Multivariable analyses After adjustment for all covariates (Table 3), in multivariable logistic regression models predicting overall complications, frailty (odds ratio [OR] 1.54, p < 0.001), age �75 years (OR 1.16, p ¼ 0.003) and CCI �2 (OR 1.54, p < 0.001) achieved independent predictor status. In multi variable logistic regression models predicting FTR, frail status (OR 1.64, p < 0.001), age �75 years (OR 1.98, p < 0.001) and CCI �2 (OR 1.76, p < 0.001) achieved independent predictor status. In multivariable lo gistic regression models predicting in-hospital mortality, frail status (OR 1.45, p ¼ 0.001), age �75 years (OR 2.21, p < 0.001) and CCI �2 (OR 1.69, p < 0.001) achieved independent predictor status. In multivariable Poisson regression models predicting LOS, frail status (relative risk [RR] 10
C. Palumbo et al.
Surgical Oncology 32 (2020) 8–13
Fig. 1. Venn diagram representing distribution of frailty, age 75 years or more and Charlson Comorbidity Index score of 2 or greater, among a cohort of 23,967 patients treated with radical cystectomy for non-metastatic carcinoma of the urinary bladder, identified within the Nationwide Inpatient Sample database between 2000 and 2015.
Fig. 2. Annual rates over time of radical cystectomy for non-metastatic bladder cancer in patients who were respectively frail, aged 75 years or more and with Charlson Comorbidity Index score of 2 or greater, identified within the Nationwide Inpatient Sample database between 2000 and 2015. 11
C. Palumbo et al.
Surgical Oncology 32 (2020) 8–13
Table 2 Crude perioperative outcomes and total hospital charges recorded in 23,967 radical cystectomy patients that are stratified according to frailty status (non-frail vs. frail), age categories <75 vs. �75 years and Charlson Comorbidity Index score categories (CCI 0–1 vs. CCI �2), identified within the Nationwide Inpatient Sample database between 2000 and 2015. Frailty Frail Non-frail Age �75 years <75 years CCI CCI �2 CCI 0-1
Overall complicationsa
Failure to rescuea
In-hospital mortalitya
Length of stayb
Total hospital chargeb
3958 (67.9) 10,112 (55.8)
129 (2.2) 224 (1.2)
141 (2.4) 281 (1.5)
8 (6–10) 9 (7–14)
29,698.9 (21,178.7–43,602.0) 24,538.3 (17,620.5–34,471.2)
4811 (62.3) 9259 (57.0)
202 (2.6) 151 (0.9)
250 (3.2) 172 (1.1)
8 (6–12) 8 (6–11)
26,134.3 (18,195.5–37,983.8) 25,568.7 (18,434.6–35,966.7)
1934 (68.3) 12,136 (57.4)
74 (2.6) 279 (1.3)
85 (3.0) 337 (1.6)
9 (7–13) 8 (6–11)
28,285.8 (20,075.1–40,063.0) 25,451.6 (18,189.4–36066.0)
AbbreviationsCCI ¼ Charlson Comorbidity Index. All comparisons were statistically significant with p < 0.001. a Only number (%) of patients who experienced overall complications, failure to rescue or in-hospital mortality are reported. b Median and interquartile range are reported. Table 3 Separate multivariable regression models, predicting overall complications, failure to rescue, in-hospital mortality, length of stay and total hospital charges in patients treated with radical cystectomy for urothelial carcinoma of the bladder within the National Inpatient Sample database (2000–2015). All models were weighted and adjusted for clustering, as well as age, Charlson comorbidity index, modified frailty index, gender, year of surgery, ethnicity, insurance status, teaching status, hospital volume, region, hospital bed-size, income, lymph node dissection, type or urinary diversion. Frail (vs. non-frail) Overall complicationsa Failure to rescuea In-hospital mortalitya Length of stayb Total hospital chargesc
Age �75 (vs. age <75)
CCI �2 (vs. CCI 0–1)
OR (95% CI)
p
OR (95% CI)
p
OR (95% CI)
p
1.54 (1.44–1.65) 1.64 (1.3–2.07) 1.45 (1.17–1.8) 1.32 (1.28–1.35) 8003.6 (6849.1–9158.2)
<0.001 <0.001 0.001 <0.001 <0.001
1.16 (1.09–1.23) 1.98 (1.55–2.53) 2.21 (1.76–2.76) 1.06 (1.04–1.09) 111.2 ( 854.41–1076.7)
<0.001 <0.001 <0.001 <0.001 0.8
1.54 (1.41–1.68) 1.76 (1.35–2.3) 1.69 (1.31–2.18) 1.12 (1.08–1.16) 3910.7 (2726.9–5094.5)
<0.001 <0.001 <0.001 <0.001 <0.001
AbbreviationsOR ¼ odds ratio, CI ¼ confidence interval; CCI¼Charlson Comorbidity Index. a Multivariable logistic regression models. b Multivariable Poisson regression model, relative risk is reported. c Multivariable linear regression model.
interesting in the frail population because of the definition itself of frailty, which is a decreased ability to restore homeostasis following insults such as complications. It is of note, that previous NSQIP database [10–16] defined frailty according to the Canadian Study of Health and Aging Frailty Index. Although this frailty index is a multidimensional score, it relies at most on comorbidity domains and, therefore, may not entirely capture the different effect of comorbidity and frailty. Indeed, frailty differs from comorbidity, since frailty is a state of health described by a reduced physical reserve and increased vulnerability to stressors [27]. Conversely, the Johns Hopkins ACG frailty-defining diagnosis indicator-based definition of frailty may allow to capture the different effect of comorbidity and frailty. It accounts for cognitive, functional and social impairments, without overlapping with CCI. Taken together, our study represents one of the first attempt to test the ability of the Johns Hopkins ACG frailty-defining diagnosis to predict perioperative outcomes after cystectomy, simultaneously accounting for age and comorbidities. Our analyses showed that frail patients accoun ted for almost 25% of the RC population and did not overlap with age and comorbidity. In consequence, neither age or comorbidity can be used as a surrogate of frailty. Moreover, the proportion of frail patients increased over time. Additionally, frailty was an independent predictor of all five examined endpoints, even after adjustment that included also age and comorbidity. Finally, frailty was the strongest predictor of three out of five examined endpoints. In consequence, frailty should be uni versally assessed and its consideration should be equal to that of wellestablished predictors of outcomes, such as advanced age and pres ence of multiple comorbidities. Multimodal prehabilitation programs, including exercise, nutrition and psychological interventions, could potentially improve the perioperative outcomes of these patients [28].
Nonetheless, ideally these programs should last up to 8 weeks. However, the underlying disease may not allow an eight-week delay prior to cystectomy. In consequence, both frailty and comorbidity might not be modifiable, but rather alongside with age these characteristics are static, yet independent predictors of adverse short-term outcomes. A careful discussion about goals of care could help patients have realistic expec tations and make better informed decisions before the surgery. Despite the strengths of this study, important limitations need to be acknowledged. First, this is a retrospective study based on the NIS database, which is an administrative dataset. Within the NIS, both diagnosis and procedures are derived from administrative codes for the conditions or procedures according to the ICD-9-CM classification. Second, within the NIS database complications and mortality are limited to in-hospital events. In consequence, delayed complications, as well as the readmission rate, could not be examined. Therefore, information on performance status, ASA score, lookback period of CCI assessment, as well as laboratory values, are not available within the NIS database. Third, frailty may be underestimated due to undercoding of frailtydefining diagnoses, which may not be recognized and coded by administrative personnel. Therefore, it may not completely reflect pa tients’ cognition and functional status. Additionally, the use of a binary definition of frailty does not allow evaluation of the effect of different degrees of frailty. Similarly, complications were assessed using sec ondary ICD-9-CM diagnosis codes. Therefore, we were unable to report and score complications in a standardized fashion, such as the ClavienDindo classification. Fourth, the cost analysis was based on hospitalrelated charges that may not be reflective at all of other health-related costs, such as readmission and post-discharge complications rates. Finally, since the NIS database did not provide tumor characteristics, such as stage and grade, we could not adjust our analyses for these 12
C. Palumbo et al.
Surgical Oncology 32 (2020) 8–13
variables.
[8] T.N. Robinson, B. Eiseman, J.I. Wallace, S.D. Church, K.K. McFann, S.M. Pfister, et al., Redefining geriatric preoperative assessment using frailty, disability and comorbidity, Ann. Surg. 250 (2009) 449–455, https://doi.org/10.1097/ SLA.0b013e3181b45598. [9] M.A. Makary, D.L. Segev, P.J. Pronovost, D. Syin, K. Bandeen-Roche, P. Patel, et al., Frailty as a predictor of surgical outcomes in older patients, J. Am. Coll. Surg. 210 (2010) 901–908, https://doi.org/10.1016/j.jamcollsurg.2010.01.028. [10] J.A. Pearl, D. Patil, C.P. Filson, S. Arya, M. Alemozaffar, V.A. Master, et al., Patient frailty and discharge disposition following radical cystectomy, Clin. Genitourin. Cancer 15 (2017) e615–e621, https://doi.org/10.1016/j.clgc.2016.12.013. [11] A.M. Suskind, C. Jin, M.R. Cooperberg, E. Finlayson, W.J. Boscardin, S. Sen, et al., Preoperative frailty is associated with discharge to skilled or assisted living facilities after urologic procedures of varying complexity, Urology 97 (2016) 25–32, https://doi.org/10.1016/j.urology.2016.03.073. [12] M.R. Chappidi, M. Kates, H.D. Patel, J.J. Tosoian, D.R. Kaye, N.A. Sopko, et al., Frailty as a marker of adverse outcomes in patients with bladder cancer undergoing radical cystectomy, Urol. Oncol. Semin. Orig. Investig. 34 (2016), https://doi.org/ 10.1016/j.urolonc.2015.12.010, 256.e1-256.e6. [13] B.L. Taylor, L. Xia, T.J. Guzzo, D.S. Scherr, J.C. Hu, Frailty and greater health care resource utilization following major urologic oncology surgery, Eur. Urol. Oncol. 2 (2019) 21–27, https://doi.org/10.1016/j.euo.2018.06.005. [14] D. Lascano, J.S. Pak, M. Kates, J.B. Finkelstein, M. Silva, E. Hagen, et al., Validation of a frailty index in patients undergoing curative surgery for urologic malignancy and comparison with other risk stratification tools, Urol. Oncol. Semin. Orig. Investig. 33 (2015), https://doi.org/10.1016/j.urolonc.2015.06.002, 426.e1-426. e12. [15] X. Meng, B. Press, A. Renson, J.S. Wysock, S.S. Taneja, W.C. Huang, et al., Discriminative ability of commonly used indexes to predict adverse outcomes after radical cystectomy: comparison of demographic data, American society of anesthesiologists, modified Charlson comorbidity index, and modified frailty index, Clin. Genitourin. Cancer 16 (2018) e843–e850, https://doi.org/10.1016/j. clgc.2018.02.009. [16] N.J. Sathianathen, S. Jarosek, N. Lawrentschuk, D. Bolton, B.R. Konety, A simplified frailty index to predict outcomes after radical cystectomy, Eur Urol Focus (2018), https://doi.org/10.1016/j.euf.2017.12.011. S2405456918300026. [17] S.L. Woldu, O. Sanli, T.N. Clinton, Y. Lotan, Validating the predictors of outcomes after radical cystectomy for bladder cancer: predicting Radical Cystectomy Outcomes, Cancer 125 (2019) 223–231, https://doi.org/10.1002/cncr.31799. [18] M.L. Burg, T.G. Clifford, S.T. Bazargani, M. Lin-Brande, G. Miranda, J. Cai, et al., Frailty as a predictor of complications after radical cystectomy: a prospective study of various preoperative assessments, Urol. Oncol. Semin. Orig. Investig. 37 (2019) 40–47, https://doi.org/10.1016/j.urolonc.2018.10.002. [19] J. Michel, A.N. Goel, V. Golla, A.T. Lenis, D.C. Johnson, K. Chamie, et al., Predicting short-term outcomes after radical cystectomy based on frailty, Urology (2019), https://doi.org/10.1016/j.urology.2019.04.057. S0090429519306107. [20] HCUP-US NIS overview, n.d, https://www.hcup-us.ahrq.gov/nisoverview.jsp. accessed March 21, 2019. [21] Q.-D. Trinh, M. Bianchi, J. Hansen, Z. Tian, F. Abdollah, S.F. Shariat, et al., Inhospital mortality and failure to rescue after cytoreductive nephrectomy, Eur. Urol. 63 (2013) 1107–1114, https://doi.org/10.1016/j.eururo.2012.08.069. [22] The Johns Hopkins ACG® system excerpt from version 11.0, Tech. Ref. Guide 36 (2015). [23] D.H. Kim, S. Schneeweiss, Measuring frailty using claims data for pharmacoepidemiologic studies of mortality in older adults: evidence and recommendations, Pharmacoepidemiol. Drug Saf. 23 (2014) 891–901, https://doi. org/10.1002/pds.3674. [24] D.I. McIsaac, G.L. Bryson, C. van Walraven, Association of frailty and 1-year postoperative mortality following major elective noncardiac surgery: a populationbased cohort study, JAMA Surg. 151 (2016) 538–545, https://doi.org/10.1001/ jamasurg.2015.5085. [25] C. Palumbo, S. Knipper, C. Dzyuba-Negrean, A. Pecoraro, G. Rosiello, Z. Tian, et al., Complication rates, failure to rescue and in-hospital mortality after cytoreductive nephrectomy in the older patients, J. Geriatr. Oncol. (2019), https://doi.org/ 10.1016/j.jgo.2019.06.005. S1879406819301961. [26] R.A. Deyo, D.C. Cherkin, M.A. Ciol, Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases, J. Clin. Epidemiol. 45 (1992) 613–619. [27] A. Clegg, J. Young, S. Iliffe, M.O. Rikkert, K. Rockwood, Frailty in elderly people, Lancet. Lond. Engl. 381 (2013) 752–762, https://doi.org/10.1016/S0140-6736 (12)62167-9. [28] M.L. Alvarez-Nebreda, N. Bentov, R.D. Urman, S. Setia, J.C.-S. Huang, K. Pfeifer, et al., Recommendations for preoperative management of frailty from the society for perioperative assessment and quality improvement (SPAQI), J. Clin. Anesth. 47 (2018) 33–42, https://doi.org/10.1016/j.jclinane.2018.02.011.
5. Conclusions Frailty, advanced age and comorbidities represent non-overlapping patients’ characteristics. Of those, frailty represents the most consis tent and strongest predictor of early adverse outcomes after RC. Ideally, all three indicators should be considered in retrospective, as well as prospective analyses. Pre-surgical recognition of frail patients should be ideally incorporate in clinical practice in order to address these patients to multimodal pre-habilitation programs that may potentially improve the perioperative prognosis. Formatting of funding sources This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. Declaration of competing interest All the authors declare no potential conflicts of interest to disclose. Acknowledgment None. Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi. org/10.1016/j.suronc.2019.10.014. References [1] S.P. Kim, S.A. Boorjian, N.D. Shah, R.J. Karnes, C.J. Weight, J.P. Moriarty, et al., Contemporary trends of in-hospital complications and mortality for radical cystectomy, BJU Int. 110 (2012) 1163–1168, https://doi.org/10.1111/j.1464410X.2012.10990.x. [2] S. Zamboni, F. Soria, R. Mathieu, E. Xylinas, M. Abufaraj, D. D’Andrea, et al., Differences in trends in the use of robot-assisted and open radical cystectomy and changes over time in peri-operative outcomes among selected centres in North America and Europe: an international multicentre collaboration: time changes of use of robotic radical cystectomy, BJU Int. (2019), https://doi.org/10.1111/ bju.14791. [3] M. Moschini, G. Simone, A. Stenzl, I.S. Gill, J. Catto, Critical review of outcomes from radical cystectomy: can complications from radical cystectomy Be reduced by surgical volume and robotic surgery? Eur Urol Focus 2 (2016) 19–29, https://doi. org/10.1016/j.euf.2016.03.001. [4] H. Yu, N.D. Hevelone, S.R. Lipsitz, K.J. Kowalczyk, P.L. Nguyen, T.K. Choueiri, et al., Comparative analysis of outcomes and costs following open radical cystectomy versus robot-assisted laparoscopic radical cystectomy: results from the US Nationwide Inpatient Sample, Eur. Urol. 61 (2012) 1239–1244, https://doi.org/ 10.1016/j.eururo.2012.03.032. [5] A. Antonelli, S. Belotti, L. Cristinelli, V. De Luca, C. Simeone, Comparison of perioperative morbidity of radical cystectomy with neobladder versus ileal conduit: a matched pair analysis of 170 patients, Clin. Genitourin. Cancer 14 (2016) 244–248, https://doi.org/10.1016/j.clgc.2015.07.011. [6] E. Mazzone, F. Preisser, S. Nazzani, Z. Tian, E. Zaffuto, A. Gallina, et al., The effect of age and comorbidities on early postoperative complications after radical cystectomy: a contemporary population-based analysis, J. Geriatr. Oncol. (2019), https://doi.org/10.1016/j.jgo.2019.04.011. S1879406818305046. [7] V. Novotny, S. Zastrow, R. Koch, M.P. Wirth, Radical cystectomy in patients over 70 years of age: impact of comorbidity on perioperative morbidity and mortality, World J. Urol. 30 (2012) 769–776, https://doi.org/10.1007/s00345-011-0782-0.
13
Contents lists available at ScienceDirect
Surgical Oncology journal homepage: http://www.elsevier.com/locate/suronc
Patient frailty predicts worse perioperative outcomes and higher cost after radical cystectomy worse radical cystectomy outcomes in frails Carlotta Palumbo a, b, *, Sophie Knipper a, c, Angela Pecoraro a, d, Giuseppe Rosiello a, e, Stefano Luzzago a, f, Marina Deuker a, g, Zhe Tian a, Shahrokh F. Shariat h, i, j, k, l, Claudio Simeone b, Alberto Briganti e, Fred Saad a, Alfredo Berruti m, Alessandro Antonelli b, Pierre I. Karakiewicz a a
Cancer Prognostics and Health Outcomes Unit, University of Montreal Health Center, Montreal, Quebec, Canada Urology Unit, ASST Spedali Civili of Brescia. Department of Medical and Surgical Specialties, Radiological Science and Public Health, University of Brescia, Italy Martini Klinik, University Medical Center Hamburg-Eppendorf, Hamburg, Germany d Department of Urology, San Luigi Gonzaga Hospital, University of Turin, Orbassano, Turin, Italy e Division of Experimental Oncology/Unit of Urology, Urological Research Institute (URI), IRCCS San Raffaele Scientific Institute, Milan, Italy, Vita-Salute San Raffaele University, Milan, Italy f Department of Urology, European Institute of Oncology, Milan, Italy g Department of Urology, University Hospital Frankfurt, Frankfurt, Germany h Department of Urology, Comprehensive Cancer Center, Medical University of Vienna, Vienna, Austria i Department of Urology, Weill Cornell Medical College, New York, NY, USA j Department of Urology, University of Texas Southwestern, Dallas, TX, USA k Department of Urology, Second Faculty of Medicine, Charles University, Prague, Czech Republic l Institute for Urology and Reproductive Health, I.M. Sechenov First Moscow State Medical University, Moscow, Russia m Medical Oncology Unit, ASST Spedali Civili of Brescia. Department of Medical and Surgical Specialties, Radiological Science and Public Health, University of Brescia, Italy b c
A R T I C L E I N F O
A B S T R A C T
Keywords: Radical cystectomy Frailty Complications
Background: Relatively few studies investigated the importance of frailty in radical cystectomy (RC) patients. We tested the ability of frailty, using the Johns Hopkins Adjusted Clinical Groups indicator, to predict early peri operative outcomes after RC. Methods: RC patients were identified within the National Inpatient Sample database (2000–2015). The effect of frailty, age and Charlson Comorbidity Index were tested in five separate multivariable models predicting: (1) complications, (2) failure to rescue (FTR), (3) in-hospital mortality, (4) length of stay (LOS) and (5) total hospital charges (THCs). All models were weighted and adjusted for clustering, as well as all available patient and hospital characteristics. Results: Of 23,967 RC patients, 5833 (24.3%) were frail, 7721 (32.2%) were aged �75 years and 2832 (11.8%) had CCI �2. Frailty, age �75 years and CCI �2 were non-overlapping in 86.3% of the cohort. Any two or three of these features were recorded in 12.4 and 1.3%, respectively. Frailty was an independent predictor of all five examined endpoints and the magnitude of its association was stronger or at least equal than that of age �75 years and CCI �2. Conclusion: Frailty, advanced age and comorbidities represent non-overlapping patients’ characteristics. Of those, frailty represents the most consistent and strongest predictor of early adverse outcomes after RC. Ideally, all three indicators should be considered in retrospective, as well as prospective analyses. Pre-surgical recog nition of frail patients should be ideally incorporate in clinical practice in order to address these patients to multimodal pre-habilitation programs that may potentially improve the perioperative prognosis.
* Corresponding author. Urology Unit, ASST Spedali Civili of Brescia. Department of Medical and Surgical Specialties, Radiological Science and Public Health, University of Brescia, Italy, Piazzale Spedali Civili 1, 25123, Brescia, Italy. E-mail address: [email protected] (C. Palumbo). https://doi.org/10.1016/j.suronc.2019.10.014 Received 23 August 2019; Received in revised form 11 October 2019; Accepted 24 October 2019 Available online 25 October 2019 0960-7404/© 2019 Published by Elsevier Ltd.
C. Palumbo et al.
Surgical Oncology 32 (2020) 8–13
1. Introduction
Comorbidity Index (CCI) was defined according to the Deyo adaptation for ICD-9 codes-based databases [26] (Supplementary Table 4) and categorized as CCI 0–1 vs. CCI �2. It is of note that frailty defining di agnoses were distinct from comorbidity diagnoses, with the exception of dementia, for which the ICD-9 codes 290.x that defined dementia in frailty were also part of the definition of dementia in CCI. For purpose of multivariable adjustment, we relied on patient and hospital characteristics that are made available by the NIS. These included gender, ethnicity (Caucasian, African American and Others), year of surgery, income and insurance status (private insurance, Medi care, Medicaid, and other [self-pay]), hospital region (Northeast, Mid west, South, West), hospital teaching vs. non-teaching status and hospital annual volume (low, medium and high), which represents the number of RCs performed at each participating institution during each study calendar year.
Radical cystectomy (RC) is associated with elevated rates of com plications and non-negligible perioperative mortality [1–5]. Age and comorbidities represent established predictors of these outcomes [6,7]. Moreover, advanced age and elevated comorbidities also predicted longer length of stay and higher hospitalization costs [6]. Relative to these established and well-documented predictors of worse periopera tive outcomes after RC, only recently frailty emerged as a prognostic indicator of increasing risk of adverse health events and mortality after surgery [8,9]. Relatively few studies investigated its importance in the context of RC literature [10–19]. To address this limitation, we tested the effect of frailty alongside age and comorbidities on complications, failure to rescue (FTR), in-hospital mortality, length of stay (LOS) and total hospital charges (THCs) within a large cohort of patients treated with RC, identified within the National Inpatient Sample (NIS) database (2000–2015).
2.4. Statistical analyses
2. Materials and methods
The three predictors of interest were frailty (non-frail vs. frail), age (<75 vs. �75 years) and CCI (CCI 0–1 vs. CCI �2). First, medians and interquartile ranges, as well as frequencies and proportions were re ported for continuous and categorical variables, respectively. Second, we examined the proportions and the estimated annual percentage changes (EAPCs) of RC rates according to frailty, age �75 years and CCI �2. Third, five separate endpoints were addressed in multivariable an alyses: (1) overall complications, (2) FTR, (3) in-hospital mortality, (4) LOS and (5) THCs. Specifically, overall complications, FTR and inhospital mortality were addressed in three separate multivariable lo gistic regression models. Multivariable Poisson regression models focused on LOS. Multivariable linear regression models focused on THCs. All multivariable models relied on weighting using a Generalized Estimating Equation (GEE) function to provide more accurate national estimates based on NIS-provided weights [20]. All models were then adjusted for clustering at hospital level, as well as for age, CCI, frailty, gender, year of surgery, ethnicity, insurance status, teaching status, hospital volume, region, income, lymph node dissection and urinary diversion. Within each multivariable model all the three predictors of interest were considered simultaneously along with all the covariates without either forward or backward variable selection. All statistical tests were two-sided with a level of significance set at p < 0.05. All analyses were performed using the R software environment for statistical computing and graphics (version 3.4.1; http://www. r-project. org/).
2.1. Source of data and study population We relied on the National Inpatient Sample (NIS) database (2000–2015) [20] that is composed of longitudinal hospital inpatient databases from the Healthcare Cost and Utilization Project family and includes 20% of United States inpatient hospitalizations. Institutional ethical board had reviewed the research project and it meets re quirements for protection of human subjects. We focused on patients aged�18 years with primary diagnosis of non-metastatic carcinoma of the urinary bladder (International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM] codes 188.x), treated with RC (primary procedure ICD-9-CM codes 57.6, 57.7� or 688). Secondary procedure codes were used to identify lymph node dissection (ICD-9-CM codes 40.3 or 40.5), as well as use of ileal conduit or continent ortho topic urinary diversion (ICD-9-CM codes 56.51 and 57.87, respectively). 2.2. Outcomes of interest We focused on four early perioperative outcomes, namely overall complications, failure to rescue (FTR), in-hospital mortality and length of stay (LOS), as well as on total hospital charges (THCs). According to previous established methodology, overall complications were defined using secondary ICD-9-CM diagnostic codes and represented the sum of intraoperative and all postoperative complications (cardiac, respiratory, vascular, operative wound, genitourinary, gastrointestinal, infectious, blood transfusions, miscellaneous medical and miscellaneous surgical, Supplementary Table 1). FTR was defined as mortality after a major complication [21] (Supplementary Table 2). In-hospital mortality was defined from discharge disposition. Length of stay was calculated by subtracting the admission date from the discharge date. Finally, inflation-adjusted THCs were defined according to NIS methodology [20].
3. Results 3.1. Patients’ characteristics and temporal trends Between 2000 and 2015, we identified 23,967 non-metastatic car cinoma of the urinary bladder patients treated with RC (Table 1). Of these, 5833 (24.3%) patients were frail, 7721 (32.2%) were aged �75 years and 2832 (11.8%) had CCI �2. Presence of frailty, age �75 years and CCI �2 represented unique and non-overlapping features in 86.3% of the cohort. Conversely, combination of frailty and either �75 years or CCI �2 was recorded in 12.4% of patients. Finally, presence of all the three features was recorded in 1.3% of the cohort (Fig. 1). Relative to non-frail, frail patients were older (patients aged �75 years 36.8% vs. 30.7%). Minimally-invasive surgery, which included both laparoscopic and robot-assisted procedures, was performed in 9.2% of frail vs. 8.3% of non-frail. Ileal conduit was performed in 72.3% of frail vs. 67.2% of non-frail. Time trends (Fig. 2) revealed an increase of frail patients (EAPC þ3.1%, p < 0.001) and, to a lesser extent, of those with CCI �2 (EAPC þ1.6%, p ¼ 0.003). Conversely, a decrease of patients �75 years (EAPC -0.7%, p < 0.001) was recorded.
2.3. Patients and hospital characteristics Frailty was defined according to the Johns Hopkins Adjusted Clinical Groups (ACG) frailty-defining diagnoses indicator [22], an instrument designed and validated for use in health administrative data [23], based on 10 clusters of frailty-defining diagnoses that comprise the Johns Hopkins ACG frailty-defining diagnosis indicator [22]. It allows to ac count for cognitive, functional and social impairments, since it encom passes dementia, vision impairment, malnutrition, urinary and fecal incontinence, difficulty in walking, falls and social support needs. Frailty was defined using patients’ ICD-9 codes available in the NIS, as previ ously reported [23–25] (Supplementary Table 3), and categorized as frail vs. non-frail. Age was dichotomized as <75 vs. �75 years. Charlson 9
C. Palumbo et al.
Surgical Oncology 32 (2020) 8–13
1.32, p < 0.001), age �75 years (RR 1.06, p < 0.001) and CCI �2 (RR 1.12, p < 0.001) achieved independent predictor status. In multivariable linear regression models predicting THCs, only frail status (þ8003.3, p < 0.001), and CCI �2 (þ3910.7, p < 0.001) achieved independent predictor status.
Table 1 Descriptive characteristic of 23,967 non-metastatic carcinoma of the urinary bladder patients treated with radical cystectomy, identified within the Nation wide Inpatient Sample database between 2000 and 2015. Variable Age, years Age group Frailty Charlson Comorbidity Idex Gender Race Approach Type of diversion Lymph-node dissection Insurance
Region
Income
Annual hospital volume Teaching status Bed size
Number (%) Median (IQR) <75 years �75 years Non-frail Frail 0–1 �2 Female Male Caucasian African-American Other Open Minimally-invasive Ileal conduit Orthotopic diversion Other Performed Not-performed Medicare Medicaid Private Other Midwest Northeast South West First Quartile Second Quartile Third Quartile Fourth Quartile Low Medium High Teaching Non-Teaching Small Medium Large
70 (62–77) 16,246 (67.8) 7721 (32.2) 18,134 (75.7) 5833 (24.3) 21,135 (88.2) 2832 (11.8) 5014 (20.9) 18,953 (79.1) 17,193 (71.7) 997 (4.2) 5777 (24.1) 21,925 (91.5) 2042 (8.5) 16,404 (68.4) 1651 (6.9) 5912 (24.7) 2832 (11.8) 7010 (29.2) 14,887 (62.1) 1011 (4.2) 7089 (29.6) 980 (4.1) 6312 (26.3) 4746 (19.8) 8551 (35.7) 4358 (18.2) 4933 (20.6) 6067 (25.3) 6145 (25.6) 6822 (28.5) 7465 (31.1) 8122 (33.9) 8380 (35) 17,050 (71.1) 6917 (28.9) 2474 (10.3) 4380 (18.3) 17,113 (71.4)
4. Discussion We hypothesized that frailty represents an independent predictor of adverse outcomes in RC patients. To test this hypothesis, we tested frailty along with age and comorbidities on these endpoints, namely overall complication, FTR in-hospital mortality, LOS and THCs. Our analyses resulted in several noteworthy findings. First, frail patients represented 24.3% of our cohort. This rate is consistent with previous NSQIP-based studies [10–16,18], where rates of frail patients ranged from 22.0% [16] to 26.9% [13]. Relative to non-frail, frail patients were older (30.7 vs. 36.8% in age category �75 years). Nonetheless, the rates of frailty among patients younger than 75 years were not negligible. These observations indicate that frailty may affect also younger patients. In consequence, frailty represents an age independent characteristics. Second, frailty, age �75 years and CCI �2 represented unique and non-overlapping patient characteristics in 86.3% of the population. Inconsequence, frailty requires independent quantification and its presence or absence cannot be derived from either age or comorbidities. To the best of our knowledge, no previous investigations examined this concept in a RC population. Third, time trend analyses demonstrated significant increasing rates of frailty in RC patients over time (EAPC þ3.1%, p < 0.001). Also CCI �2 increased but to a lower rate (EAPC þ1.6%, p ¼ 0.003). Finally, the proportion of patients aged �75 years decreased. Frailty is gaining in importance among RC patients, in a rate that exceeds those of both higher comorbidity index and more advanced age. In consequence, it could be argued that frailty should be given at least the same importance of age and elevated number of comorbidities. To the best of our knowledge, this aspect of frailty has not been previously reported. Fourth, frailty achieved independent predictor status as a unique risk variable, even after adjustment for advanced age and comorbidities in the analyses of the five endpoints. This observation further militate in favor of considering frailty as a unique risk variable. Moreover, on average the magnitude of increase of the risk of all five adverse peri operative outcomes was greater or equal for frailty than for age or co morbidity. For example, when considering the overall complications, the magnitude of risk increase of frailty exceeds that of age and was similar to that of comorbidity. Additionally, concerning both LOS and THCs, the magnitude of risk increase of frailty exceeds that of both age and comorbidity. Conversely, in FTR and mortality, frailty was weaker relative to age, but similar to comorbidity. Taken together, presence of frailty appears to be a stronger indicator of three out of five examined endpoints, with the exception of FTR and mortality, where patients aged �75 years were at higher risk than frail patients. In consequence, the importance of frailty is at least similar, if not greater, to that of age and comorbidity. Therefore, consideration of frailty should be invariably included in analyses of RC adverse perioperative outcomes and possibly in similar analyses that are based on older patients. Previous studies [12,14–19] showed that frailty was independently associated with worse outcomes in different context from the current study (30- and 90-day events vs. readmission setting vs. in-hospital events). Nonetheless, these analyses concur in the importance of frailty in predicting adverse outcomes after RC and jointly indicate that frailty should be included alongside with age and CCI as important in dicator of adverse clinical outcomes at both initial hospital stay, as well as at readmission. Moreover, to the best of our knowledge, we are the first to report an association between frailty and FTR in cystectomy patients, since none of the previous studies investigated this topic. This observation is important, since the concept of FTR may be particularly
AbbreviationsIQR ¼ interquartile range.
3.2. Crude rates of outcomes of interest Crude rates of the five outcomes stratified according to frailty, age and CCI categories are summarized in Table 2. Frail patients exhibited higher rates of complications (67.9 vs. 55.8%, p < 0.001), FTR (2.2 vs. 1.2%, p < 0.001) and in-hospital mortality (2.4 vs. 1.5%, p < 0.001), as well as longer LOS (p < 0.001) and higher THCs (p < 0.001). Patients aged �75 years exhibited higher rates of complications, FTR and inhospital mortality, as well as longer LOS and higher THCs were recor ded (all p < 0.001). Patients with CCI �2 exhibited higher rates of complications, FTR and in-hospital mortality increased, as well as longer LOS and higher THCs were reported (all p < 0.001). 3.3. Multivariable analyses After adjustment for all covariates (Table 3), in multivariable logistic regression models predicting overall complications, frailty (odds ratio [OR] 1.54, p < 0.001), age �75 years (OR 1.16, p ¼ 0.003) and CCI �2 (OR 1.54, p < 0.001) achieved independent predictor status. In multi variable logistic regression models predicting FTR, frail status (OR 1.64, p < 0.001), age �75 years (OR 1.98, p < 0.001) and CCI �2 (OR 1.76, p < 0.001) achieved independent predictor status. In multivariable lo gistic regression models predicting in-hospital mortality, frail status (OR 1.45, p ¼ 0.001), age �75 years (OR 2.21, p < 0.001) and CCI �2 (OR 1.69, p < 0.001) achieved independent predictor status. In multivariable Poisson regression models predicting LOS, frail status (relative risk [RR] 10
C. Palumbo et al.
Surgical Oncology 32 (2020) 8–13
Fig. 1. Venn diagram representing distribution of frailty, age 75 years or more and Charlson Comorbidity Index score of 2 or greater, among a cohort of 23,967 patients treated with radical cystectomy for non-metastatic carcinoma of the urinary bladder, identified within the Nationwide Inpatient Sample database between 2000 and 2015.
Fig. 2. Annual rates over time of radical cystectomy for non-metastatic bladder cancer in patients who were respectively frail, aged 75 years or more and with Charlson Comorbidity Index score of 2 or greater, identified within the Nationwide Inpatient Sample database between 2000 and 2015. 11
C. Palumbo et al.
Surgical Oncology 32 (2020) 8–13
Table 2 Crude perioperative outcomes and total hospital charges recorded in 23,967 radical cystectomy patients that are stratified according to frailty status (non-frail vs. frail), age categories <75 vs. �75 years and Charlson Comorbidity Index score categories (CCI 0–1 vs. CCI �2), identified within the Nationwide Inpatient Sample database between 2000 and 2015. Frailty Frail Non-frail Age �75 years <75 years CCI CCI �2 CCI 0-1
Overall complicationsa
Failure to rescuea
In-hospital mortalitya
Length of stayb
Total hospital chargeb
3958 (67.9) 10,112 (55.8)
129 (2.2) 224 (1.2)
141 (2.4) 281 (1.5)
8 (6–10) 9 (7–14)
29,698.9 (21,178.7–43,602.0) 24,538.3 (17,620.5–34,471.2)
4811 (62.3) 9259 (57.0)
202 (2.6) 151 (0.9)
250 (3.2) 172 (1.1)
8 (6–12) 8 (6–11)
26,134.3 (18,195.5–37,983.8) 25,568.7 (18,434.6–35,966.7)
1934 (68.3) 12,136 (57.4)
74 (2.6) 279 (1.3)
85 (3.0) 337 (1.6)
9 (7–13) 8 (6–11)
28,285.8 (20,075.1–40,063.0) 25,451.6 (18,189.4–36066.0)
AbbreviationsCCI ¼ Charlson Comorbidity Index. All comparisons were statistically significant with p < 0.001. a Only number (%) of patients who experienced overall complications, failure to rescue or in-hospital mortality are reported. b Median and interquartile range are reported. Table 3 Separate multivariable regression models, predicting overall complications, failure to rescue, in-hospital mortality, length of stay and total hospital charges in patients treated with radical cystectomy for urothelial carcinoma of the bladder within the National Inpatient Sample database (2000–2015). All models were weighted and adjusted for clustering, as well as age, Charlson comorbidity index, modified frailty index, gender, year of surgery, ethnicity, insurance status, teaching status, hospital volume, region, hospital bed-size, income, lymph node dissection, type or urinary diversion. Frail (vs. non-frail) Overall complicationsa Failure to rescuea In-hospital mortalitya Length of stayb Total hospital chargesc
Age �75 (vs. age <75)
CCI �2 (vs. CCI 0–1)
OR (95% CI)
p
OR (95% CI)
p
OR (95% CI)
p
1.54 (1.44–1.65) 1.64 (1.3–2.07) 1.45 (1.17–1.8) 1.32 (1.28–1.35) 8003.6 (6849.1–9158.2)
<0.001 <0.001 0.001 <0.001 <0.001
1.16 (1.09–1.23) 1.98 (1.55–2.53) 2.21 (1.76–2.76) 1.06 (1.04–1.09) 111.2 ( 854.41–1076.7)
<0.001 <0.001 <0.001 <0.001 0.8
1.54 (1.41–1.68) 1.76 (1.35–2.3) 1.69 (1.31–2.18) 1.12 (1.08–1.16) 3910.7 (2726.9–5094.5)
<0.001 <0.001 <0.001 <0.001 <0.001
AbbreviationsOR ¼ odds ratio, CI ¼ confidence interval; CCI¼Charlson Comorbidity Index. a Multivariable logistic regression models. b Multivariable Poisson regression model, relative risk is reported. c Multivariable linear regression model.
interesting in the frail population because of the definition itself of frailty, which is a decreased ability to restore homeostasis following insults such as complications. It is of note, that previous NSQIP database [10–16] defined frailty according to the Canadian Study of Health and Aging Frailty Index. Although this frailty index is a multidimensional score, it relies at most on comorbidity domains and, therefore, may not entirely capture the different effect of comorbidity and frailty. Indeed, frailty differs from comorbidity, since frailty is a state of health described by a reduced physical reserve and increased vulnerability to stressors [27]. Conversely, the Johns Hopkins ACG frailty-defining diagnosis indicator-based definition of frailty may allow to capture the different effect of comorbidity and frailty. It accounts for cognitive, functional and social impairments, without overlapping with CCI. Taken together, our study represents one of the first attempt to test the ability of the Johns Hopkins ACG frailty-defining diagnosis to predict perioperative outcomes after cystectomy, simultaneously accounting for age and comorbidities. Our analyses showed that frail patients accoun ted for almost 25% of the RC population and did not overlap with age and comorbidity. In consequence, neither age or comorbidity can be used as a surrogate of frailty. Moreover, the proportion of frail patients increased over time. Additionally, frailty was an independent predictor of all five examined endpoints, even after adjustment that included also age and comorbidity. Finally, frailty was the strongest predictor of three out of five examined endpoints. In consequence, frailty should be uni versally assessed and its consideration should be equal to that of wellestablished predictors of outcomes, such as advanced age and pres ence of multiple comorbidities. Multimodal prehabilitation programs, including exercise, nutrition and psychological interventions, could potentially improve the perioperative outcomes of these patients [28].
Nonetheless, ideally these programs should last up to 8 weeks. However, the underlying disease may not allow an eight-week delay prior to cystectomy. In consequence, both frailty and comorbidity might not be modifiable, but rather alongside with age these characteristics are static, yet independent predictors of adverse short-term outcomes. A careful discussion about goals of care could help patients have realistic expec tations and make better informed decisions before the surgery. Despite the strengths of this study, important limitations need to be acknowledged. First, this is a retrospective study based on the NIS database, which is an administrative dataset. Within the NIS, both diagnosis and procedures are derived from administrative codes for the conditions or procedures according to the ICD-9-CM classification. Second, within the NIS database complications and mortality are limited to in-hospital events. In consequence, delayed complications, as well as the readmission rate, could not be examined. Therefore, information on performance status, ASA score, lookback period of CCI assessment, as well as laboratory values, are not available within the NIS database. Third, frailty may be underestimated due to undercoding of frailtydefining diagnoses, which may not be recognized and coded by administrative personnel. Therefore, it may not completely reflect pa tients’ cognition and functional status. Additionally, the use of a binary definition of frailty does not allow evaluation of the effect of different degrees of frailty. Similarly, complications were assessed using sec ondary ICD-9-CM diagnosis codes. Therefore, we were unable to report and score complications in a standardized fashion, such as the ClavienDindo classification. Fourth, the cost analysis was based on hospitalrelated charges that may not be reflective at all of other health-related costs, such as readmission and post-discharge complications rates. Finally, since the NIS database did not provide tumor characteristics, such as stage and grade, we could not adjust our analyses for these 12
C. Palumbo et al.
Surgical Oncology 32 (2020) 8–13
variables.
[8] T.N. Robinson, B. Eiseman, J.I. Wallace, S.D. Church, K.K. McFann, S.M. Pfister, et al., Redefining geriatric preoperative assessment using frailty, disability and comorbidity, Ann. Surg. 250 (2009) 449–455, https://doi.org/10.1097/ SLA.0b013e3181b45598. [9] M.A. Makary, D.L. Segev, P.J. Pronovost, D. Syin, K. Bandeen-Roche, P. Patel, et al., Frailty as a predictor of surgical outcomes in older patients, J. Am. Coll. Surg. 210 (2010) 901–908, https://doi.org/10.1016/j.jamcollsurg.2010.01.028. [10] J.A. Pearl, D. Patil, C.P. Filson, S. Arya, M. Alemozaffar, V.A. Master, et al., Patient frailty and discharge disposition following radical cystectomy, Clin. Genitourin. Cancer 15 (2017) e615–e621, https://doi.org/10.1016/j.clgc.2016.12.013. [11] A.M. Suskind, C. Jin, M.R. Cooperberg, E. Finlayson, W.J. Boscardin, S. Sen, et al., Preoperative frailty is associated with discharge to skilled or assisted living facilities after urologic procedures of varying complexity, Urology 97 (2016) 25–32, https://doi.org/10.1016/j.urology.2016.03.073. [12] M.R. Chappidi, M. Kates, H.D. Patel, J.J. Tosoian, D.R. Kaye, N.A. Sopko, et al., Frailty as a marker of adverse outcomes in patients with bladder cancer undergoing radical cystectomy, Urol. Oncol. Semin. Orig. Investig. 34 (2016), https://doi.org/ 10.1016/j.urolonc.2015.12.010, 256.e1-256.e6. [13] B.L. Taylor, L. Xia, T.J. Guzzo, D.S. Scherr, J.C. Hu, Frailty and greater health care resource utilization following major urologic oncology surgery, Eur. Urol. Oncol. 2 (2019) 21–27, https://doi.org/10.1016/j.euo.2018.06.005. [14] D. Lascano, J.S. Pak, M. Kates, J.B. Finkelstein, M. Silva, E. Hagen, et al., Validation of a frailty index in patients undergoing curative surgery for urologic malignancy and comparison with other risk stratification tools, Urol. Oncol. Semin. Orig. Investig. 33 (2015), https://doi.org/10.1016/j.urolonc.2015.06.002, 426.e1-426. e12. [15] X. Meng, B. Press, A. Renson, J.S. Wysock, S.S. Taneja, W.C. Huang, et al., Discriminative ability of commonly used indexes to predict adverse outcomes after radical cystectomy: comparison of demographic data, American society of anesthesiologists, modified Charlson comorbidity index, and modified frailty index, Clin. Genitourin. Cancer 16 (2018) e843–e850, https://doi.org/10.1016/j. clgc.2018.02.009. [16] N.J. Sathianathen, S. Jarosek, N. Lawrentschuk, D. Bolton, B.R. Konety, A simplified frailty index to predict outcomes after radical cystectomy, Eur Urol Focus (2018), https://doi.org/10.1016/j.euf.2017.12.011. S2405456918300026. [17] S.L. Woldu, O. Sanli, T.N. Clinton, Y. Lotan, Validating the predictors of outcomes after radical cystectomy for bladder cancer: predicting Radical Cystectomy Outcomes, Cancer 125 (2019) 223–231, https://doi.org/10.1002/cncr.31799. [18] M.L. Burg, T.G. Clifford, S.T. Bazargani, M. Lin-Brande, G. Miranda, J. Cai, et al., Frailty as a predictor of complications after radical cystectomy: a prospective study of various preoperative assessments, Urol. Oncol. Semin. Orig. Investig. 37 (2019) 40–47, https://doi.org/10.1016/j.urolonc.2018.10.002. [19] J. Michel, A.N. Goel, V. Golla, A.T. Lenis, D.C. Johnson, K. Chamie, et al., Predicting short-term outcomes after radical cystectomy based on frailty, Urology (2019), https://doi.org/10.1016/j.urology.2019.04.057. S0090429519306107. [20] HCUP-US NIS overview, n.d, https://www.hcup-us.ahrq.gov/nisoverview.jsp. accessed March 21, 2019. [21] Q.-D. Trinh, M. Bianchi, J. Hansen, Z. Tian, F. Abdollah, S.F. Shariat, et al., Inhospital mortality and failure to rescue after cytoreductive nephrectomy, Eur. Urol. 63 (2013) 1107–1114, https://doi.org/10.1016/j.eururo.2012.08.069. [22] The Johns Hopkins ACG® system excerpt from version 11.0, Tech. Ref. Guide 36 (2015). [23] D.H. Kim, S. Schneeweiss, Measuring frailty using claims data for pharmacoepidemiologic studies of mortality in older adults: evidence and recommendations, Pharmacoepidemiol. Drug Saf. 23 (2014) 891–901, https://doi. org/10.1002/pds.3674. [24] D.I. McIsaac, G.L. Bryson, C. van Walraven, Association of frailty and 1-year postoperative mortality following major elective noncardiac surgery: a populationbased cohort study, JAMA Surg. 151 (2016) 538–545, https://doi.org/10.1001/ jamasurg.2015.5085. [25] C. Palumbo, S. Knipper, C. Dzyuba-Negrean, A. Pecoraro, G. Rosiello, Z. Tian, et al., Complication rates, failure to rescue and in-hospital mortality after cytoreductive nephrectomy in the older patients, J. Geriatr. Oncol. (2019), https://doi.org/ 10.1016/j.jgo.2019.06.005. S1879406819301961. [26] R.A. Deyo, D.C. Cherkin, M.A. Ciol, Adapting a clinical comorbidity index for use with ICD-9-CM administrative databases, J. Clin. Epidemiol. 45 (1992) 613–619. [27] A. Clegg, J. Young, S. Iliffe, M.O. Rikkert, K. Rockwood, Frailty in elderly people, Lancet. Lond. Engl. 381 (2013) 752–762, https://doi.org/10.1016/S0140-6736 (12)62167-9. [28] M.L. Alvarez-Nebreda, N. Bentov, R.D. Urman, S. Setia, J.C.-S. Huang, K. Pfeifer, et al., Recommendations for preoperative management of frailty from the society for perioperative assessment and quality improvement (SPAQI), J. Clin. Anesth. 47 (2018) 33–42, https://doi.org/10.1016/j.jclinane.2018.02.011.
5. Conclusions Frailty, advanced age and comorbidities represent non-overlapping patients’ characteristics. Of those, frailty represents the most consis tent and strongest predictor of early adverse outcomes after RC. Ideally, all three indicators should be considered in retrospective, as well as prospective analyses. Pre-surgical recognition of frail patients should be ideally incorporate in clinical practice in order to address these patients to multimodal pre-habilitation programs that may potentially improve the perioperative prognosis. Formatting of funding sources This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors. Declaration of competing interest All the authors declare no potential conflicts of interest to disclose. Acknowledgment None. Appendix A. Supplementary data Supplementary data to this article can be found online at https://doi. org/10.1016/j.suronc.2019.10.014. References [1] S.P. Kim, S.A. Boorjian, N.D. Shah, R.J. Karnes, C.J. Weight, J.P. Moriarty, et al., Contemporary trends of in-hospital complications and mortality for radical cystectomy, BJU Int. 110 (2012) 1163–1168, https://doi.org/10.1111/j.1464410X.2012.10990.x. [2] S. Zamboni, F. Soria, R. Mathieu, E. Xylinas, M. Abufaraj, D. D’Andrea, et al., Differences in trends in the use of robot-assisted and open radical cystectomy and changes over time in peri-operative outcomes among selected centres in North America and Europe: an international multicentre collaboration: time changes of use of robotic radical cystectomy, BJU Int. (2019), https://doi.org/10.1111/ bju.14791. [3] M. Moschini, G. Simone, A. Stenzl, I.S. Gill, J. Catto, Critical review of outcomes from radical cystectomy: can complications from radical cystectomy Be reduced by surgical volume and robotic surgery? Eur Urol Focus 2 (2016) 19–29, https://doi. org/10.1016/j.euf.2016.03.001. [4] H. Yu, N.D. Hevelone, S.R. Lipsitz, K.J. Kowalczyk, P.L. Nguyen, T.K. Choueiri, et al., Comparative analysis of outcomes and costs following open radical cystectomy versus robot-assisted laparoscopic radical cystectomy: results from the US Nationwide Inpatient Sample, Eur. Urol. 61 (2012) 1239–1244, https://doi.org/ 10.1016/j.eururo.2012.03.032. [5] A. Antonelli, S. Belotti, L. Cristinelli, V. De Luca, C. Simeone, Comparison of perioperative morbidity of radical cystectomy with neobladder versus ileal conduit: a matched pair analysis of 170 patients, Clin. Genitourin. Cancer 14 (2016) 244–248, https://doi.org/10.1016/j.clgc.2015.07.011. [6] E. Mazzone, F. Preisser, S. Nazzani, Z. Tian, E. Zaffuto, A. Gallina, et al., The effect of age and comorbidities on early postoperative complications after radical cystectomy: a contemporary population-based analysis, J. Geriatr. Oncol. (2019), https://doi.org/10.1016/j.jgo.2019.04.011. S1879406818305046. [7] V. Novotny, S. Zastrow, R. Koch, M.P. Wirth, Radical cystectomy in patients over 70 years of age: impact of comorbidity on perioperative morbidity and mortality, World J. Urol. 30 (2012) 769–776, https://doi.org/10.1007/s00345-011-0782-0.
13