Value of Flemish Version of the Triage Risk Screening Tool in Predicting Unfavorable Outcomes after Elective Cancer Surgery: A Propensity Score-Matched Retrospective Cohort Study Open Access

Introduction: The Flemish version of the Triage Risk Screening Tool (fTRST), derived from the Triage Risk Screening Tool for assessing risk of readmission to the emergency department, is increasingly used as a simple screening tool in oncology. This study aimed to evaluate the utility of the fTRST in the context of elective surgical treatment for urologic cancer patients. Methods: We included 886 patients who underwent major urologic cancer surgery at our institution between 2020 and 2022 and underwent preoperative screening, including fTRST. We set the fTRST cutoff at 2 and used propensity score matching and multivariate regression analysis to assess how fTRST affected two postoperative outcomes: ambulation failure and delirium. Results: Of the 886 patients, 693 (78%) had an fTRST score <2, and 193 (22%) had an fTRST score ≥2 (high likelihood of frailty). After matching the groups by propensity scores, we compared the outcomes of 131 patients in each group. We found that the group with fTRST ≥2 had significantly higher rates of ambulation failure (15 vs. 11%, p = 0.03) and delirium (16 vs. 11%, p = 0.008) than the group with fTRST <2. Multivariate logistic regression analysis showed that fTRST score ≥2 was an independent risk factor for postoperative ambulation failure (odds ratio [OR] = 4.05, p = 0.02), along with age ≥75 years (OR = 6.62, p = 0.02), preoperative benzodiazepine medications (OR = 5.12, p = 0.01), and receiving radical cystectomy (OR = 9.30, p = 0.02). Similarly, for delirium, fTRST score ≥2 was an independent risk factor (OR = 2.88, p = 0.03), along with preoperative benzodiazepine medications (OR = 4.38, p = 0.002). Conclusion: The fTRST might be a screening tool with great potential for identifying patients at high risk for unfavorable postoperative outcomes in elective urologic cancer surgery.

With the increase in life expectancy in the general population worldwide, providing better cancer care for older patients has become a major concern [1], especially in the field of urology, which has a relatively high proportion of older patients [2]. A comprehensive geriatric assessment (CGA), which is an evaluation of older patients to identify age-related impairments, is recommended to perform for better managing cancer in older patients [3, 4]. CGA provides important prognostic value for mortality, activities of daily living, and dependency [5].

Nevertheless, CGA requires a lot of time and specialized human resources and is not required for every patient [6]. Hence, there is growing interest in the use of less resource-consuming and simpler geriatric screening tools as alternatives to CGA. Screening tools used in oncology include the Vulnerable Elders Survey-13 (VES-13) [7], the Geriatric-8 (G8) [8‒10], the Flemish version of the Triage Risk Screening Tool (fTRST) [10‒13], abbreviated CGA [14], instrumental activities of daily living (IADL) [15], and so on. The fTRST was first reported by Meldon et al. [12] in 2003 as a 6-item Triage Risk Screening Tool (TRST) to assess the risk of readmission to the emergency department, followed by a shorter 5-item Flemish version (fTRST) [11], which has since been increasingly used in the field of oncology [10, 16].

The fTRST covers multiple domains of CGA, such as cognitive function, polypharmacy, falls, and social support (Table 1). The G8 and VES-13 are useful for identifying patients who need more intensive CGA [7‒10]. They can also estimate the risk of chemotherapy-related toxicity and mortality, following the ASCO guidelines for older patients receiving chemotherapy [17]. The fTRST has the advantage of having fewer items (five) and a simpler score calculation than the G8 and VES-13 [11]. However, it is less comprehensive in assessing physical function, as it only asks about history of falls. On the other hand, the fTRST has questions on social support and history of hospitalization, which are not in the G8 and VES-13, and measures cognitive function and polypharmacy. The fTRST is a balanced indicator with a small number of items. Although fTRST is a simple and useful screening tool and has demonstrated high prognostic power for functional decline and overall survival when used in an oncologic population [10], adequate validations of functional prognosis associated with surgical treatment are lacking. Therefore, this study aimed to evaluate the utility of the fTRST when focused on the context of surgical treatment in cancer care.

This observational study included 968 consecutive patients who underwent major urological cancer surgery (radical cystectomy [RC], radical prostatectomy [RP], radical nephrectomy [RN], partial nephrectomy [PN], or radical nephroureterectomy [RNU]) at our department (National Cancer Center Hospital East, Chiba, Japan) between January 2020 and October 2022. All surgeries were performed by the urologic surgical team that included three highly skilled and experienced surgeons. During the study period, 960 patients aged 40 or older undergoing scheduled surgery were asked questions and tested by trained nurses on multiple geriatric screening, including fTRST, as part of hospital protocols at the preoperative outpatient clinic. Of these patients, 74 were excluded from the study because the geriatric screening was not completed at the preoperative clinic due to conflicting hospital visit schedules: as a result, 886 patients were analyzed. All included patients provided their written informed consent for elective surgery. Participation in the study could be waived by opting out, and the study was approved by the National Cancer Center Institutional Review Board (2018–159).

In the preoperative outpatient clinic, trained nurses screened patients for 5 items related to fTRST. These items were cognitive impairment (2 points), living alone or having no available, willing, or able caregiver (1 point), difficulty with walking or transfers or falls in the past 6 months (1 point), hospitalization in the last 3 months (1 point), and polypharmacy; taking five or more medications (1 point). Table 1 shows the questions and the corresponding scores for calculating fTRST. Cognitive function in fTRST was assessed using mini-cognitive assessment instrument (Mini-Cog) [18]. Mini-Cog is a brief screening tool for detecting cognitive impairment in older adults, and it consists of two components: a three-item recall test and a clock-drawing test [18]. The total score ranges from 0 to 5, with lower scores indicating higher likelihood of cognitive impairment [18]. We considered Mini-Cog <3 as an indicator of potential cognitive impairment in the evaluation of fTRST (Table 1). For the “hospitalization in the last 3 months” item, we excluded cases in which the patient was hospitalized for a short period of time for scheduled cancer tests. The fTRST score was calculated by adding up the points of these five items (range 0–6) [13]. Based on early data from the medical oncology setting, a patient with fTRST ≥2 was considered to be at high risk for frailty [11, 13, 16]. We also evaluated the validity of using 1 as the cutoff value for fTRST [10]. We also collected and analyzed the following clinical data: age, sex, body mass index, Eastern Cooperative Oncology Group-performance status (ECOG-PS), Charlson comorbidity index (CCI), existing depression or dementia, a history of alcohol consumption (once a week or more), a history of stroke, preoperative medications, cancer stages (I–IV) based on the 8th edition of the tumor-node-metastasis staging classification for urologic cancers [19], surgical form (RC, RP, RN, PN, or RNU), surgical approaches (open, laparoscopic, or robot-assisted), operative time, estimated blood loss, blood transfusion status, admission to surgical intensive care unit, complications within 30 days after surgery according to the Clavien-Dindo classification (CDC), and length of hospital stay. Previous studies have shown that age, CCI, and ECOG-PS can be divided into two groups: less than 75 versus 75 or more, less than 5 versus 5 or more, and less than 2 versus 2 or more [20‒22]. Similarly, other continuous variables were split into two categories based on their median values.

This study aimed to examine the association between fTRST and several endpoints that indicate postoperative complications. Specifically, we focused on two endpoints: ambulation failure and delirium. We defined these endpoints as follows:

Ambulation failure: a condition in which the patient could not walk independently, with or without a cane, within 2 days after surgery and needed physical therapy intervention. The need for physical therapy was determined by the physicians based on the patient’s physical and medical condition, which was discussed at the daily postoperative conference. Patients who were discharged in a bedridden state and transferred to other hospitals were also classified as having ambulation failure.

Delirium: a state of confusion that was screened for at least twice a day by trained nurses using the Confusion Assessment Method for all postoperative patients. Patients who were suspected of having delirium were evaluated by a consultation-liaison psychiatry (CLP) team using the American Psychiatric Association’s Diagnostic and Statistical Manual of Mental Disorders (DSM-5) [23].

We compared the baseline characteristics of patients according to their fTRST scores (≥2 vs. <2). We presented continuous variables as median and interquartile range (IQR) and categorical variables as numbers and percentages. We used the Mann-Whitney U test for continuous variables and the χ² test or Fisher’s exact probability test for categorical variables to evaluate differences in the distribution between the two groups. We performed zero-cell correction using Haldane-Anscombe correction [24].

We calculated the area under the curve (AUC) of the receiver operating characteristic curve to evaluate the predictive ability of fTRST for each endpoint. We calculated a propensity score based on the covariates of age, sex, ECOG-PS, CCI, cancer type, and stage groups of cancer, and matched subjects in each group 1:1 based on the propensity score. We performed the same analysis (calculating AUC), for the cohort that we matched using propensity score matching (PSM). We also used multivariate logistic regression analyses to estimate whether fTRST was a significant predictor for postoperative unfavorable outcomes; we generated reduced models by eliminating the variable with the highest p value in each iteration using backward elimination. We examined the performance on endpoints when 1 was used as the cutoff for fTRST as well as when 2 was used as the cutoff, and we used DeLong’s test to evaluate differences in AUC at cutoff 1 and AUC at cutoff 2 [25]. We considered p values <0.05 (two-sided) as statistically significant. We used JMP 13 software (SAS Institute Inc., Cary, NC, USA) and statistical software R version 4.2.1 (R Foundation, Vienna, Austria) for all statistical analyses. We used GraphPad Prism 9.0 (GraphPad Software, Inc., San Diego, CA, USA) for figure preparation.

The demographics of 886 patients are shown in Table 2. The median age was 72 years (IQR 67–77 years), and 803 patients (91%) were male. In total, 117 (13%), 533 (60%), 43 (5%), 117 (13%), and 76 patients (9%) underwent RC, RP, RN, PN, and RNU, respectively. The median interval from the screening of fTRST to scheduled surgery was 34 days (IQR 21–60 days). Of the 886 patients, there were 693 (78%) patients with fTRST scores <2 and 193 (22%) patients with fTRST scores ≥2 (high likelihood of frailty). In the cohort before PSM, fTRST scores ≥2 were significantly associated with older age, females, higher ECOG-PS, a higher CCI, higher rates of previously diagnosed dementia, preoperative benzodiazepine use, longer operative time, blood transfusion, the higher admission rate of surgical intensive care unit, and longer hospital stay. In addition, fTRST scores ≥2 were significantly more frequent in cancer stages IV, patients undergoing RC those undergoing open surgeries, and those experiencing CDC 3 or more complications. The fTRST scores ≥2 were less frequent in patients undergoing RP, undergoing robot-assisted surgeries, and those experiencing CDC 1 or fewer complications. Figure 1 illustrates the distribution of patients according to their fTRST scores.

Matching based on the propensity score resulted in 131 matches each in the fTRST scores <2 and fTRST scores ≥2 groups. After matching, there was a significant difference in the rate of admission to the surgical intensive care unit, but no significant difference in other clinical factors (Table 2).

Of the 886 patients, 57 (6%) experienced ambulation failure, and 55 (6%) experienced delirium. Of the 193 patients with fTRST ≥2, 38 (20%) experienced ambulation failure, and 35 (18%) experienced delirium. These rates were significantly higher than those in the group with fTRST <2 (ambulation failure: p < 0.001; delirium: p < 0.001). Among the 262 patients in the post-PSM cohort, 28 (11%) experienced ambulation failure and 28 (11%) experienced delirium. Of the 131 patients with fTRST ≥2, 20 (15%) experienced ambulation failure, and 21 (16%) experienced delirium. These rates were significantly higher than those in the group with fTRST <2 (ambulation failure: p = 0.03; delirium: p = 0.008). Figure 2 shows the receiver operating characteristic curves for the predictive performance of fTRST scores for unfavorable postoperative outcomes. Figure 2a shows the analysis in the pre-PSM cohort (N = 886), and Figure 2b shows the analysis in the post-PSM cohort (N = 262). Regarding ambulation failure, fTRST had good discriminating ability in the pre-PSM cohort (AUC = 0.80, Fig. 2a) and fair discriminating ability in the post-PSM cohort (AUC = 0.71, Fig. 2b). Likewise, regarding delirium, fTRST had fair discriminating ability in both the pre-PSM and post-PSM cohorts (AUC = 0.78, Fig. 2a; AUC = 0.72, Fig. 2b).

We further evaluated clinical factors associated with postoperative unfavorable outcomes in the post-PSM cohort (N = 262) using multivariate logistic regression analysis. In the post-PSM cohort, both ambulation failure and delirium occurred in 28 patients (11%). Table 3 shows that fTRST score ≥2 was an independent risk factor for postoperative ambulation failure (odds ratio [OR] = 4.05, 95% CI = 1.22–13.5, p = 0.02), along with age ≥75 years (OR = 6.62, 95% CI = 1.42–30.9, p = 0.02), preoperative benzodiazepine medications (OR = 5.12, 95% CI = 1.46–18.0, p = 0.01), and receiving RC (OR = 9.30, 95% CI = 1.33–65.1, p = 0.02). Similarly, for delirium, fTRST score ≥2 was an independent risk factor (OR = 2.88, 95% CI = 1.10–7.53, p = 0.03), along with preoperative benzodiazepine medications (OR = 4.38, 95% CI = 1.70–11.3, p = 0.002).

Table 4 shows the sensitivity, specificity, positive predictive value, negative predictive value, and AUC for each outcome in the post-matching cohort, using 1 and 2 as cutoff values for fTRST. Out of 262 cases, 77 (29%) had fTRST <1 and 185 (71%) had fTRST ≥1. The AUCs for ambulation failure and delirium did not differ significantly between the cutoff values of 1 and 2 (ambulation failure: AUC = 0.63 with cutoff 1, AUC = 0.62 with cutoff 2, p = 0.91; delirium: AUC = 0.61 with cutoff 1, AUC = 0.64 with cutoff 2, p = 0.33).

We showed that fTRST scores were significantly and independently associated with postoperative ambulation failure and delirium in a cohort of patients who underwent elective major surgery for urologic cancer. These results suggest the clinical importance and use of screening with fTRST in terms of predicting unfavorable postoperative outcomes in patients who underwent major surgery for urologic cancer. The fTRST was derived from TRST, a triage screening used in emergency settings, and its usefulness as a screening tool in an oncologic population has been evaluated [10], and the current validation shows that fTRST screening is also useful in the perioperative context. The fTRST might be applicable to tailor the perioperative care plan of the elderly who plan to undergo major surgery for cancer.

We used a cutoff value of 2 for fTRST, following previous studies [11, 13, 16]. Initially, fTRST was evaluated in a global geriatric (non-oncologic) population, and the cutoff for having a geriatric risk profile was defined at ≥2 [11, 26]. In a study of 937 cancer patients, Kenis et al. [10] showed that the predictive performance of fTRST for the risk profile of the elderly according to the gold standard was AUC = 0.67 (95% CI = 0.65–0.70) for cutoff 1 and AUC = 0.74 (95% CI = 0.71–0.76), with cutoff 2 being significantly better (p < 0.001). However, for sensitivity, cutoff 1 (91.3%, 95% CI = 88.9–93.3%) was superior to cutoff 2 (91.3%, 95% CI = 88.9–93.3%), and C Kennis et al. [10] concluded that cutoff 1 might be appropriate, considering the importance of sensitivity in screening. As shown in Table 4, there was no significant difference in AUC between cutoff 1 and cutoff 2 for the outcome aimed in our study (ambulation failure: AUC = 0.625 for cutoff 1, AUC = 0.620 for cutoff 2, p = 0.91; delirium: AUC = 0.605 at cutoff 1, AUC = 0.640 at cutoff 2, p = 0.33). Since we assessed the outcome right after surgery, a highly invasive treatment, it is possible that ambulation failure and delirium occurred in a certain number of patients in the fTRST = 1 patient group, and this may have led to the same performance at cutoff 1 as at cutoff 2. A possible implication of this is that patients with fTRST above 1 might require careful perioperative management, regardless of whether the fTRST exceeds 2 or not.

There are several limitations in the present study. First, although the number of patients included is relatively large, this single-center study might have an inherent selection bias due to its retrospective nature. Second, a postoperative follow-up period was too short to evaluate the association between fTRST and life expectancy in the present analysis. Third, the present study only performed one preoperative screening and was unable to assess the inter- and intra-observer agreement of the fTRST score. Fourth, in the present study, the Mini-Cog [18] was used to determine the “cognitive impairment” item, while the mini-mental state examination (MMSE) was used in another study as an evaluation index [10]. Although it has been reported that Mini-Cog is not inferior to MMSE in detecting cognitive impairment [27], the data would have been more robust if a comparison with the judgment made using MMSE had also been made. Finally, we have not compared the performance of fTRST to other screening tools, and we cannot rule out the possibility that other geriatric screening tools might have even greater accuracy for predicting unfavorable postoperative outcomes. We conclude that fTRST is a simple and rapid screening tool with great potential for identifying patients with a geriatric risk profile at high risk for unfavorable postoperative outcomes in the context of scheduled cancer surgery.

The research was conducted ethically in accordance with the World Medical Association Declaration of Helsinki. The study was approved by the National Cancer Center Institutional Review Board (2018–159). All included patients provided their written informed consent for elective surgery. Participation in the study could be waived by opting out.

The authors have no conflicts of interest to declare.

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Shugo Yajima: conceptualization, methodology, visualization, and writing – original draft preparation. Yasukazu Nakanishi: conceptualization, methodology, and writing – original draft. Ryo Andei Ogasawara, Naoki Imasato, Kohei Hirose, Madoka Kataoka, and Sao Katsumura: data curation and writing – review and editing. Hitoshi Masuda: conceptualization, methodology, supervision, and writing – review and editing.

The data that support the findings of this study are not publicly available due to their containing information that could compromise the privacy of research participants but are available from Yasukazu Nakanishi upon reasonable request.