Prospectively Collected Functional vs. Self-reported Frailty Measures and Postoperative Outcomes after Radical Cystectomy in Older Adults with Bladder Cancer

Abstract

Introduction and Objectives:

The optimal tools to measure frailty in patients with bladder cancer are poorly defined. Our objective was to compare baseline physical function with self-reported frailty assessments, and to evaluate associations between functional vs. self-reported frailty assessments and postoperative outcomes.

Methods:

Adults ≥65 years old with bladder cancer anticipating radical cystectomy (RC) were preoperatively assessed with the Short Physical Performance and Battery Assessment (SPPB): a physical assessment evaluating balance, gait speed, and lower extremity strength; and the Care and Aging Resilience Evaluation-Geriatric Assessment: a self-reported form which generates a validated frailty index (CARE-FI). Associations between the two frailty measures, postoperative length of stay (LOS), and 90-day grade ≥3 Clavien Dindo complications were assessed.

Results:

53 participants were analyzed. Overall, 3/53 were frail per the CARE-FI, while 1/53 demonstrated poor performance per SPPB criteria. CARE-FI and SPPB frailty categories were strongly associated on an ordinal scale (Kendall’s τ-b = 0.88, p = 0.02), agreement between categorical classifications was limited (weighted κ = 0.02). Patients classified as frail or pre-frail by CARE-FI were more likely to experience extended length of stay (≥7 days; p = 0.008, chi-square) and had a longer continuous LOS (p = 0.0013, Wilcoxon rank-sum) and were also more likely to experience a grade ≥3 complication (p = 0.04, chi-square). The SPPB was not significantly associated with either variable per chi squared analysis (p>0.05 for both).

Conclusions:

Functional and self-reported frailty assessments provide different information characterizing individual vulnerability profiles; however, only the CARE-FI was statistically associated with increased LOS and postoperative complications following RC.

Graphical Abstract

1. Introduction

Bladder cancer is the 9^th most common cancer in both males and females, and primarily affects older populations with a high burden of frailty, malnutrition, sarcopenia, and medical complexity.^1,2,3,4 The average age of diagnosis for men is 69 and 73 for women.⁵ Muscle-invasive bladder cancer (MIBC) comprises 20% of new diagnoses, for which the current standard of treatment is radical cystectomy (RC) following neoadjuvant systemic therapy.^1,6 This procedure carries a high morbidity and mortality rate, with 90-day postoperative complication rates between 25-50%, and readmission in up to 30% of cases.^7,8 Given the advanced age and medical complexity of the at-risk population and the morbidity associated with therapy, MIBC is associated with significant treatment burden and potential for impact on quality of life. However, we have limited tools to predict those at greatest risk for adverse events associated with treatment, which is crucial information needed to prepare patients with an understanding of their expected recovery and informed shared decision-making and informed consent.

Frailty is a well-recognized risk factor for adverse outcomes in bladder cancer, yet frailty is not often incorporated formally into risk assessment for patient treatment. Frailty is a multimodal diagnosis of physiological decline, which predisposes patients to physiological stressors through increased vulnerability and morbidity.^9,10 A diagnosis of frailty alone is associated with increased mortality.¹⁰ As patients age, their risk for frailty and its adverse effects on health-related outcomes increases.¹¹ Patients with cancer are at an increased risk for developing frailty due to both the burden of the cancer and the potential complications and risks of treatments, which increase their risk of mortality and decreased quality of life.^10,12 Although evaluation for frailty is supported widely as an evidence-based, guideline-concordant practice in bladder cancer treatment,^13,14,15 it is not widely adopted in practice¹⁶ and few prospective studies have assessed objective measures of frailty as a predictor of postoperative outcomes following radical cystectomy.¹⁷

One of the barriers to adoption of frailty evaluation into practice is that the optimal tools to measure frailty are not well-defined.⁹ Two commonly utilized frailty assessment tools are the Cancer and Aging Resilience Evaluation Geriatric Assessment (CARE-GA), and the Short Performance Physical and Battery Testing (SPPB). The CARE-GA is a validated, self-reported form in which respondents rate their functional status, health-related quality of life, and other psychosocial factors in order to identify areas of vulnerability. ¹⁸ Integrating a geriatric assessment has been shown to be feasible in gastrointestinal oncology patients, but is not routinely used in other oncology practices.¹⁹ The SPPB is a widely used, efficient, and objective evaluation of physical function that is associated with all-cause mortality.²⁰ The SPPB consists of three domains: balance, gait, and strength, which are used as representations of physical function.^20,21 We collected these assessments as part of a larger prospective observational cohort study evaluating post-radical cystectomy recovery trajectories (the University of Washington RC-Bounceback Study). Our objective was to evaluate associations between baseline functional (SPPB) vs. self-reported (CARE-GA) frailty assessment as well as the association of these results with postoperative outcomes. We hypothesized that self-reported CARE-GA frailty, which integrates multiple domains of vulnerability, would outperform physical function as a measure of frailty with respect to predicting post-RC outcomes.

2. Methods

2.1. Ethics Approval Statement

This prospective observational cohort study was approved by the University of Washington Institutional Review Board (STUDY00017395). The study was conducted in accordance with the Declaration of Helsinki. Written informed consent was obtained from all participants prior to enrollment.

2.2. Study design

Participants were identified during pre-operative surgical clearance appointments and were prospectively enrolled from June 2023 to February 2025. Inclusion criteria were adults 65 years or older indicated for open or robotic cystectomy for bladder cancer at our institution. 89 patients were initially screened from the UW-Bounceback cohort. Patients must have completed both the SPPB and CARE-GA within four weeks of their operation to be included in the current analysis, those who did not meet this timeframe were excluded from analysis.

During the preoperative appointment, participants completed the CARE-GA form and a SPPB was recorded and scored in person by the study team. The responses of the CARE-GA were used to calculate a CARE-frailty index (FI) score.²² The frailty index categorized patients as frail (>0.35), pre-frail (0.2-0.35), or robust (0-0.2). SPPB performance was scored as poor (0-5 points), moderate (7-9 points), or good (10-12 points).

2.3. Outcomes of Interest

Postoperative outcomes included length of stay (LOS) and 90-day Clavien-Dindo grade ≥3 complications. Length of hospital stay was recorded through medical records following discharge. 90-day postoperative complications were abstracted from the electronic medical record including regionally linked records to capture out of system adverse events and characterized according to the Clavien-Dindo classification system.²³ Demographics, Eastern Cooperative Oncology Group Performance Status, Charlson Comorbidity Index (CCI) exclusive of bladder cancer diagnosis, and clinical characteristics of the lesion were also collected.

2.4. Statistical Methods

Ordinal associations between preoperative SPPB and CARE-FI classifications were assessed using Kendall’s tau-b, while agreement between categorical classifications was evaluated using weighted Cohen’s kappa with linear weights. Associations between the frailty status measures and postoperative outcomes were analyzed using chi-square tests for categorical outcomes and the Wilcoxon rank-sum test for continuous outcomes. LOS was examined as both a continuous variable and a binary outcome, with extended LOS defined as ≥7 days. Due to small numbers of patients classified as CARE-FI frail or with poor SPPB performance, frailty categories were dichotomized for outcome analyses (frail/pre-frail vs. robust for CARE-FI and poor/moderate vs. good for SPPB). Statistical significance was defined as p < 0.05. All analyses were completed in JMP.

3. Results

3.1. Cohort Demographics

The final cohort included 53 older adults who underwent radical cystectomy (Table A). The median age was 73 years (IQR 68-77). Most participants were male (N=41, 77%) and 43 patients had a CCI of 3 or greater (81%). Fifteen participants had received prior neoadjuvant chemotherapy (28%). Forty-four patients had undergone laparoscopic or robotic cystectomy, and nine patients underwent open cystectomy. 21/53 (40%) of the patients were cachectic by the Fearon criteria, by self-reported weight loss history.²⁴ By CARE-FI, 37 patients were characterized as robust (70%), 13 were pre-frail (24%), and 3 were frail (6%). CARE-FI scoring stratified by CARE-GA domains is demonstrated in Figure A. Per the SPPB scoring at baseline, 31 patients demonstrated good performance (58%), 21 demonstrated moderate performance (40%), and 1 demonstrated poor performance (2%). Combined CARE-FI and SPPB scoring is shown in Table B. Spider plots in Figure B display median domain scores for CARE-GA (Panels A-B) and SPPB components (Panels C-D), stratified by both CARE-FI and SPPB classification methods.

Table A.

Demographics and Clinical Characteristics for 53 older adults who had a radical cystectomy

Demographic	N (%)
Age, year (median, IQR)	73 (68-77)
Male	41 (77%)
Race
Caucasian	52 (98%)
Asian	1 (2%)
Ethnicity
Non-Hispanic	53 (100%)
Charleson Comorbidity Index
≥3	43 (81%)
Eastern Cooperative
Oncology Group (ECOG)
Performance Status
0	38 (72%)
1	13 (24%)
2	2 (4%)
Cachexia	21 (40%)
Smoking Status
Never	30 (57%)
Former	19 (36%)
Current	4 (7%)
Clinical Tumor Stage
Tis-Ta	3 (6%)
T1	11 (20%)
T2	28 (53%)
T3	9 (17%)
T4	2 (4%)
Histologic Subtype
Urothelial	46 (87%)
Squamous	5 (10%)
Adenocarcinoma	1 (2%)
Micropapillary	1 (2%)

Figure A.

HRQOL = Health-Related Quality of Life. Spider plots displaying normalized domain scores for individual study patients stratified by CARE-FI frailty category. Scores are normalized from 0 (deficit absent) to 1 (deficit present), with higher values indicating greater vulnerability across CARE-FI domains.

Table B.

Combined CARE and SPPB Scoring Per Patient

CARE-FI	SPPB
	Poor (n = 1)	Moderate (n = 21)	Good (n = 31)
Frail (n=3)	0 (0%)	0 (0%)	3 (5.7%)
Pre-Frail (n=13)	1 (1.9%)	6 (11.3%)	6 (11.3%)
Robust (n=37)	0 (0%)	15 (28.3%)	22 (41.5%)

Figure B.

CARE-FI = Cancer and Aging Resilience Evaluation Frailty Index; SPPB = Short Physical Performance Battery; HRQOL = Health-Related Quality of Life. Spider plots displaying median normalized domain scores across CARE-FI domains (A, B) and SPPB components (C, D), stratified by CARE-FI frailty status and SPPB performance category. Scores are normalized from 0 (deficit absent) to 1 (deficit present), with higher values indicating greater vulnerability.

3.2. Statistical Analysis

CARE-FI and SPPB frailty categories were strongly associated on an ordinal scale (Kendall’s T-b = 0.88, p = 0.02). In contrast, agreement between categorical classifications was limited (weighted κ = 0.02), suggesting differences in classification thresholds between the instruments. Patients classified as frail or pre-frail by CARE-FI were more likely to experience extended length of stay (Figure C; p = 0.008). In addition, these patients were more likely to experience a Clavien grade ≥3 complication (p = 0.04). In contrast, SPPB frailty categories were not significantly associated with either LOS or Clavien ≥3 complications.

Figure C.

Length of stay in patients according to CARE-GA-assigned frailty status

3.3. Postoperative Outcomes

Within 90 days of surgery, 22 patients (42%) sustained Grade 3 or greater postoperative complications, and 20 patients (38%) were readmitted within 90 days of discharge. Goal discharge was within 5 days of the operation, per the predetermined Enhanced Recovery After Surgery protocol. Median LOS was 6 days (IQR 5-8). Of the 19 patients with a LOS longer than 6 days, four were recommended further management with either home health or physical and occupational therapy, and two were discharged to a skilled nursing facility prior to returning to home.

4. Discussion

In this study, we prospectively compared a validated and widely used physical function assessment (SPPB) to a validated self-reported geriatric assessment (CARE-GA) in older adults who underwent radical cystectomy. We quantified the associations between these measures of frailty and postoperative LOS as well as 90-day severe complications. The key finding of these analyses was the low correlation between these two assessments of vulnerability: a multi-domain assessment of self-reported frailty, the CARE-GA frailty index, and an objective assessment of physical function, the SPPB. This study highlights a difference between self-reported frailty and measured physical function. We observed that while SPPB-assessed frailty is not associated with early postoperative outcomes including LOS, frailty, as measured by the multi-domain CARE-GA was statistically significantly associated with LOS following radical cystectomy. Additionally, only patients classified as frail or pre-frail by the CARE-GA were more likely to experience a 90 day-Clavien 3+ complication. This finding is consistent with the findings of a recent meta-analysis comparing robotic versus open radical cystectomy, which reported that a pre-operative diagnosis of frailty was associated with increased risk of postoperative complications²⁵.

The CARE-GA frailty index was significantly associated with increasing LOS following RC. By identifying risk for frailty pre-operatively, the CARE-GA could potentially be used to better estimate hospital course duration following RC which has implications for preoperative patient counseling and expectation-setting. The CARE-GA has most commonly been utilized as a predictor of frailty in oncologic patients, specifically gastrointestinal malignancies, but to our knowledge, has yet to be studied in urologic oncology settings.²² Previous studies have identified prolonged LOS with older age and increased CCI but no standardized predictor for LOS following RC has been established.²⁶ This study suggests the multidomain CARE-GA, which is a self-report of vulnerability vs. robustness incorporating all domains of a comprehensive geriatric assessment in a single, patient-reported self-assessment that is completed in less than ten minutes by the majority of older adults, more thoroughly evaluates frailty domains than physical function alone. This tool has potential for ease of implementation in routine preoperative assessment. This tool may promote not only assessment of individual risk profiles, but also identify drivers of frailty which can inform personalized prehabilitation intervention recommendations.²⁷ The identification and quantification of an individuals’ vulnerability profile may facilitate personalized preoperative optimization including preoperative conditioning, nutritional services, or increased attention to the development of social support.

Interestingly, the SPPB, a multidomain physical function assessment that evaluates lower extremity strength, balance, and gait speed, was not significantly associated with either LOS or increased risk for Grade 3 or greater complications after RC in this prospective observational cohort of older adults undergoing RC. Although physical function is widely assumed to be a large component of frailty,²⁸ in this preliminary cohort, it was not independently associated with postoperative recovery trajectory. This exemplifies a need to consider each patient in their entirety when evaluating surgical candidacy. Physicality is often used as a surrogate marker for what is considered health, which may be exclusive of patients who require RC. The SPPB is a validated tool endorsed by the National Comprehensive Cancer Network (NCCN) in bladder cancer evaluation of older adults¹³, and has been utilized in assessing sarcopenia in patients with MIBC undergoing cisplatin based chemotherapy and RC.²⁹ The SPPB, if used as routine geriatric screening, has the potential to rapidly identify specific physical function vulnerabilities that can identify specific prehabilitation needs for patients prior to undergoing RC. However, this study suggests it should not be used as a standalone predictor for postoperative outcomes following RC.

Additionally, frailty screening tools have a role in non-operative setting, particularly in patients undergoing systemic cancer treatment. The American Society of Clinical Oncology (ASCO), currently recommends geriatric assessments as a screening tool in these patient populations, for early identification and treatment of baseline deficits.³⁰ For patients undergoing treatment with enfortumab vedotin (EV) and pembrolizumab, which is currently being explored as a perioperative treatment in combination with RC³¹, frailty assessments serve a role. As continued evidence emerges, frailty assessments such as the CARE-GA have the potential to be utilized to further risk stratify patients into operative versus non-operative categories. Future trials must be performed to further assess this application of frailty tools.

4.1. Limitations

There are other frailty measures such as the modified frailty index (MFI) that has been suggested to predict increased Clavien complications in bladder cancer patients undergoing RC, as seen in a retrospective study.³² However, while the MFI has been validated for predicting outcomes in patients undergoing prostatectomy and nephroureterectomy, it has not been demonstrated to be predictive of risk for complications following RC.¹⁷ The CCI is also widely utilized as a preoperative assessment which higher scores are associated with more postoperative complications variable in some studies but not others.^{33,34,35,36,37,38,39,40,41,42,43,44} However, as seen in this study, patients who are diagnosed with bladder cancer innately have an elevated CCI, due to its data points largely relying on age and known comorbid medical conditions, as bladder cancer is typically a cancer of older adults. Further pre-operative evaluation is needed to more adequately predict complications in patients with various comorbidities.

We acknowledge that the modest sample size here is a significant potential limitation to this study. Furthermore, the small sample size ultimately leads to fewer individuals in the frail and poor function categories, which limits the discriminatory power of the tools. Additionally, only a small proportion, 28%, of patients received neoadjuvant chemotherapy prior to surgery. However, the UW RC-Bounceback study is actively accruing patients with the objective of providing detailed predictors of clinical outcomes, recovery trajectories and patient outcomes after RC, and has recently been expanded to multiple academic sites.^45,46 All operations in the current cohort were performed at a high-volume tertiary care center, the University of Washington, which impact the generalizability of the results presented here. Furthermore, Grade III complications or greater were used as a cutoff point for this study, however a recent meta-analysis showed most complications following RC tend to be grade I-II.⁴⁷

4.2. Future Applications

The findings of this prospective observational study highlight the fact that physical function is a component of frailty, which is a multidomain assessment of physiologic decline. Thus, while understanding physical function of patients prior to major surgical interventions is of interest, it cannot replace a more comprehensive assessment. Functional assessments and self-reported multi-domain frailty assessments provide different information when characterizing vulnerability profiles, as demonstrated by the results of this study. The overall evaluation of frailty using an efficient, multidomain, self-reported assessment such as the CARE-GA as a risk stratification tool may help to identify patients at risk for protracted recovery courses after RC, as well as to identify personalized vulnerability profiles that may inform the development of individualized treatment plans for patients undergoing RC. Subsequent steps for this study include further analysis with updated data collection for the UW RC-Bounceback study at additional academic sites, including Emory University and University of California San Diego. Future applications of these preoperative frailty assessments may include prediction of long-term recovery trajectories as well as oncologic outcomes, in addition to patient-reported outcomes including postoperative health-related quality of life.

5. Conclusion

In this prospective cohort of older adults with bladder cancer undergoing radical cystectomy, the CARE-GA frailty index was associated with postoperative LOS and 90 day-Clavien 3+ complications. As such, this tool offers the potential to efficiently assess patients in clinic across multiple domains classically captured in a comprehensive geriatric assessment and may inform estimation of early postoperative recovery trajectories. Furthermore, this tool has the potential to inform shared-decision making, detailed and personalized informed consent discussions and may facilitate evaluation of personalized vulnerability profiles which could inform individualized prehabilitation interventions targeting improvement of candidacy for therapy, as well as optimization of outcomes. Conversely, here we observed physical function assessments alone did not correlate significantly with self-reported multidomain assessment of frailty nor early postoperative outcomes. Our results speak to a potential association, however given the limited sample size, further studies are needed to adequately assess the optimal tools to determine frailty as a predictor of postoperative complications.

Highlights:

• Optimal tools to determine frailty are poorly defined in bladder cancer patients

• Physical assessments are not correlated with self-reported assessments of frailty

• Increased length of hospital stay may be anticipated with geriatric assessments

• Physical assessments alone are not predictive of postoperative outcomes