Screening for frailty among older patients with cancer that qualify for abdominal surgery

JO U RN A L OF GE RI A TR IC O N COL O G Y 6 (2 0 1 5 ) 5 2 –5 9

Available online at www.sciencedirect.com

ScienceDirect

Screening for frailty among older patients with cancer that qualify for abdominal surgery Jakub Kenig⁎, Beata Zychiewicz, Urszula Olszewska, Piotr Richter 3rd Department of General Surgery, Jagiellonian University Medical College, Krakow, Poland

AR TIC LE I N FO

ABS TR ACT

Article history:

Objective: The Geriatric Assessment (GA) is an established method for evaluating and optimizing

Received 14 January 2014

diagnostic and treatment plans. However, it requires experience and is time-consuming.

Received in revised form 2 August 2014

Therefore, a variety of screening methods have been developed. The aim of this study was to

Accepted 10 September 2014

compare their accuracy for predicting frailty among older patients with cancer qualified for

Available online 30 September 2014

abdominal surgery based on comparison to the GA.

Keywords:

were prospectively enrolled. The diagnostic performance of eight screening tests was

Geriatric Assessment

evaluated: The Vulnerable Elderly Survey (VES-13), Triage Risk Screening Tool (TRST),

Frailty screening

Geriatric 8 (G8), Groningen Frailty Index (GFI), abbreviated Comprehensive Geriatric

Surgery in the elderly

Assessment (aCGA), Rockwood, Balducci and Fried score.

Material and Methods: One hundred and thirty five consecutive patients ≥ 65 years of age

Results: The prevalence of frailty as diagnosed by the GA was 73%. Screening methods identified frail patients in 40–75.5% of cases. The sensitivity and specificity of these tests in predicting frailty were 52%–97% (Fried score-G8) and 44–100% (G8-Rockwood score), respectively. The positive and negative predictive values were 82–100% (Balducci–Rockwood) and 43–84% (TRST–G8), respectively. Age significantly influenced the predictive value of the screening tests whereas gender and type of cancer did not. Conclusion: At present, there is no universal screening test that adequately identifies frailty in at risk older patients. The results of this study showed that the aCGA and G8 were the best screens for older patients with cancer that qualified for elective abdominal surgery; the G8 had the highest sensitivity and negative predictive value and the aCGA was a good overall assessment tool. © 2014 Elsevier Ltd. All rights reserved.

1. Introduction About one-half of cancer cases and two-thirds of cancer deaths occur in patients 65 years of age or older.1 The progress in medicine, including the extension of life, suggests that the number of older patients with cancer will significantly increase in the coming years. This group of patients is very heterogeneous with regard to co-morbidity and physical reserve. Therefore, the routine format of medical history, physical examination, biochemistry and imaging tests often

does not provide the information needed for optimal and tailored treatment. To help guide treatment decisions a Geriatric Assessment (GA) was introduced. It helps detect patients with decreased physiological reserve arising from cumulative deficits in several physiological systems that in turn can result in diminished resistance to stressors.2,3 The GA objectively assesses the health status of older patients, focusing on somatic, functional and psychosocial domains; it has proven to be of great value for clinicians treating geriatric patients.4

⁎ Corresponding author at: 3rd Department of General Surgery, Pradnicka 35-37, 31-202 Krakow, Poland. Tel.: +48 12 633 19 95; fax: +48 12 633 31 05. E-mail address: [email protected] (J. Kenig).

http://dx.doi.org/10.1016/j.jgo.2014.09.179 1879-4068/© 2014 Elsevier Ltd. All rights reserved.

53

J O U RN A L OF GE RI A TR IC O N CO LOG Y 6 ( 20 1 5 ) 5 2 – 59

Table 1 – Glossary of the different screening tests used in the study. Developed for

Number of items

Range

Cut-off score a

VES-135 TRST6 G87 GFI8 aCGA9

General older population Older patients at ED Oncology patients General older population Oncology patients

13 5 8 15 15

Rockwood10 Balducci11 Fried score3

General older population General older population General older population

4 4 5

0–15 0–6 0–17 0–15 ADL: 3 IADL: 4 GDS: 4 MMS: 4 0–3 0–4 0–5

≥3 ≥1 ≤14 ≥4 ≥1 dependent ≥1 dependent ≥2 ≤6 ≥2 1 ≥3

Test

Vulnerable Elders Survey (VES-13), Triage Risk Screening Tool (TRST), Geriatric 8 (G8), Groningen Frailty Index (GFI), abbreviated Comprehensive Geriatric Assessment (aCGA); Emergency Department (ED). a Cut-off score for a patient to be considered frail.

However, the GA requires experience, it is time-consuming to administer and not necessary in all patients. Therefore, a variety of screening methods have been developed to identify vulnerable patients that require full Geriatric Assessment prior to treatment. The Vulnerable Elders Survey (VES-13),5 Triage Risk Screening Tool (TRST),6 Geriatric 8 (G8),7 Groningen Frailty Index (GFI),8 abbreviated Comprehensive Geriatric Assessment (aCGA),9 Rockwood,10 Balducci,11 and Fried3 screening tests are commonly used. High sensitivity (for the correct identification of frail patients) and high specificity (which limits the number of healthy patients that undergo the GA) would be desirable features for an ideal screening test of at risk frail elderly patients with cancer. This is particularly important in the case of patients eligible for surgery, which can easily destabilize the homeostasis of the body. Therefore, it is important to determine the efficacy of such tests among oncologic patients that qualify for surgery. The aim of this prospective study was to compare the accuracy of the above-mentioned screening methods for predicting frailty among older patients with cancer qualified for abdominal surgery based on comparison to the GA.

2. Material and Methods 2.1. Study Population Between June 2012 and December 2013, 135 consecutive patients 65 years of age or older, with solid abdominal tumors

in need of surgery under general anesthesia, were enrolled in this prospective study. Two patients refused to participate in the study. All patients met the qualifications for surgery at the tertiary referral hospital. The Ethics Committee approved this study and informed consent was obtained from all patients or their caregiver. Patients that were unable to give informed consent and those that had peritoneal carcinomatosis were excluded.

2.2. Screening Tests and Geriatric Assessment Prior to surgery, eight tests that screened for frailty (VES-13, TRST, G8, GFI, aCGA, Rockwood, Balducci, and the Fried) were carried out. A glossary for the different screening tests used in this study is shown in Table 1. All patients had also a GA, used as the reference to compare all other tests. The GA included validated sections such as: activities of daily living (ADL),12 instrumental activities of daily living (IADL),13 the Blessed Orientation-Memory-Concentration (BOMC) Test,14 the Clock Drawing Test (CDT-test),15 Charlson Comorbidity Scale,16 Geriatric Depression Scale,17 Timed Up and Go (TUG),18 as well as a full nutritional assessment (MNA).19 The results and scores of each section were recorded. In addition, each test was scored on a dichotomous scale, based on whether there was or was not impairment in any of the parameters. Trained persons on the geriatric team administered all surveys 1– 5 days before surgery. Based on prior published literature, the detection of deficits in two or more GA domains indicated an increased risk of disability or death20; this was used as the

Table 2 – Glossary of the tests used in the geriatric assessment. Test ADL12 IADL13 TUG18 Charlson Comorbidity Scale16 Geriatric Depression Scale17 BOMC Test14 CDT-test15 MNA full16 a

Functional status Functional status Physical activity Comorbidity Depression Cognitive assessment Cognitive assessment Nutritional assessment

Cut-off score indicating impairment in the domain.

Number of items

Range

Cut-off score a

6 8 1 19 15 6 7 18

0–6 0–8 0–∞ 0–37 0–15 0–28 0–7 0–30

<5 ≤7 ≥ 15 >3 >5 >10 >3 <24

54

JO U RN A L OF GE RI A TR IC O N COL O G Y 6 (2 0 1 5 ) 5 2 –5 9

Table 3 – Baseline characteristics of patients. Factor Number [n] • 65–74 • 75–84 • 85+ Mean age [years ± SD (range)] Type of cancer [n (%)]: • Gastric • Pancreas • Colon • Rectum • Other Performance status (ECOG 0–5) [n (%)]: • Score 0 • Score 1 • Score 2 • Score 3 • Score 4 Comorbidity (CCS 0–37): • No comorbidity • Comorbidity score 1–3 • Comorbidity >3 Laboratory results [mean ± SD]: • Hemoglobin (g/dl) • Creatinine (μmol/l) • Albumin (g/l) Oncological treatment in the past

Entire population

Female

Male

p value

135 71 54 10 75 ± 6.6 (65–92)

71 34 30 7 76 ± 6.5 (65–92)

64 37 24 3 74 ± 6.5 (65–90)

ns ns ns ns ns

11 23 57 31 13

(8.1%) (17%) (42.2%) (23.1%) (9.6%)

5 (7%) 10 (14.1%) 34 (47.9%) 16 (22.5%) 6 (8.5%)

6 (9.3%) 13 (20.3%) 23 (36%) 15 (23.4%) 7 (11%)

ns ns ns ns ns

16 (11.9%) 59 (43.7%) 28 (20.7%) 24 (17.8%) 8 (5.9%)

4 (5.6%) 34 (47.9%) 15 (21.1%) 14 (19.7%) 4 (5.6%)

12 (18.8%) 25 (39.1%) 13 (20.3%) 10 (15.6%) 4 (6.2%)

ns ns ns ns ns

8 (5.9%) 62 (45.9%) 65 (48.2%)

6 (8.4%) 34 (47.9%) 31 (43.7%)

2 (3.1%) 28 (43.8%) 34 (53.1%)

ns ns ns

12.1 ± 2.1 91.3 ± 54.2 37.8 ± 8.2 13 (9.6%)

12.0 ± 2 91.5 ± 68.5 36.7 ± 9.6 7 (9.9%)

12.2 ± 2.2 91.1 ± 32.3 38.9 ± 6.3 6 (9.4%)

ns ns ns ns

Standard deviation (SD); Charlson Comorbidity Scale (CCS); Eastern Cooperative Oncology Group (ECOG); not significant (ns).

cut-off score and our frailty definition. A detailed list of the tests, with range and cut-off score, according to the GA, is shown in Table 2.

2.3. Statistical Analysis Qualitative and quantitative data were used to describe the study results. Quantitative parameters are expressed as the mean value ± standard deviation or median (range) as appropriate. The remaining cases have been summarized as counts and percentages. The data were analyzed using Statistica 10.0 software (StatSoft). The Shapiro–Wilk W and the Kolmogorov– Smirnov tests, with the Killiefors correction, were used to confirm the normality of the distribution of the results. Next, the data were subjected to parametric or non-parametric tests. A Receiver Operating Characteristic (ROC) curve was used to evaluate the diagnostic performance of the screening tools for frailty, with the GA considered the reference test and used for comparison. The area under the curve (AUC) reflected the predictive ability of the screening tests to detect frailty. A statistical comparison of both tests was carried out as described by Hanley and McNeil.21,22 In addition, the sensitivity, specificity and positive/negative predictive values of the above tests were also determined. Further, an internal analysis was performed among subgroups based on: age (patients 65–74 years of age, 75–84 years of age and 85+), gender (male, female) and type of the cancer (gastric, pancreas, colon, rectum, other). Statistical significance was defined as a two-sided p ≤ 0.05.

3. Results The baseline characteristics of the patients are shown in Table 3. The prevalence of frailty as determined by the GA was 73% (99 patients). Among this group 54 patients were female (76%) and 45 (70%) were male (p > 0.05). According to the frailty screening methods, the frailty prevalence was: 56% (VES-13), 47% (TRST), 85% (G8), 49% (GFI), 65% (aCGA), 40% (Rockwood), 75.5% (Balducci), and 40% (Fried). Detailed results of Geriatric Assessment were listed in Table 4. Furthermore, 44.4% of patients (60) had the ECOG PS ≥ 2.

3.1. Comparison of Predicted Frailty Risk Table 5 summarizes the sensitivity, specificity, PPV, NPV and accuracy of each screening method studied for predicting frailty, as compared to the GA, reference test. These results are presented based on cut off values reported in primary publications (Table 1). The sensitivity and specificity of these tests in predicting frailty were 52%–97% (being the lowest for Fried score and the highest for G8) and 44–100% (G8 and Rockwood score), respectively. The positive and negative predictive value were 82–100% (Balducci and Rockwood) and 43–84% (TRST and G8), respectively. The accuracy of these tests were 62–84% (being the lowest for Fried score and the highest for aCGA). ECOG-PS ≥ 2, had the sensitivity, specificity, positive/negative predictive value and accuracy for predicting frailty of 54%, 81%, 88%/61% and 61%, respectively.

55

J O U RN A L OF GE RI A TR IC O N CO LOG Y 6 ( 20 1 5 ) 5 2 – 59

Table 4 – Patient characteristic including elements of reference Geriatric Assessment. Test

Number of items

ADL (cut-off score <5) • Independent • Dependent IADL (cut-off score ≤ 7) • Independent • Dependent TUG (cut-off score ≥15 s) • <15 s • ≥15 s CCS (cut-off score >3) •3 • >3 GDS (cut-off score >5) • Normal • Depressed BOMC test (cut-off score >10) • No impairment • Impaired CDT-test (cut-off score >3) • No impairment • Impaired MNA full (cut-off score <24) • No malnutrition • Malnutrition

104 (77%) 31 (23%) 67 (49.6%) 68 (50.4%)

the screening tests. Significant difference was observed in the case of TRST, GFI, aCGA, and Rockwood. In the case of VES-13 and Fried scores the difference was on the borderline of statistical significance (p = 0.06 and p = 0.07, respectively). The values of G8 were comparatively high in all subgroups. For the VES-13, G8 and Balducci scores, the patients were divided into only two age subgroups, due to the pattern of results making the detailed ROC analysis impossible.

3.2. Comparison of Discrimination

69 (51%) 66 (49%)

The relationship among all scores was statistically significant. The area under the curve, in the ROC plot, is shown in Table 7, with sufficient (Balducci score) to very good (aCGA) predictive value for the screening tests. The AUC under the aCGA curve was significantly higher than the curve (from p = 0.0001 to p < 0.03) for the other screening tests, indicating a better discriminatory ability.

70 (52%) 65 (48%) 89 (66%) 46 (34%) 88 (65%) 47 (35%)

3.3. Calibration of Prediction Scores

87 (64%) 48 (36%)

The fit of the model was good in all cases, as shown by the Hosmer and Lemeshow test (p > 0.05). However, for the aCGA score the calibration was significantly better when it was compared to the other screening tests.

40 (30%) 95 (70%)

Activities of daily living (ADL); Instrumental activities of daily living (IADL); Blessed Orientation-Memory-Concentration (BOMC) Test; Clock Drawing Test (CDT-test); Charlson Comorbidity Scale (CCS); Geriatric Depression Scale (GDS); Timed Up and Go (TUG); full nutritional assessment (MNA).

The values of the scores, including internal analysis within the subgroups: 65–74 years of age, 75–84 years of age and 85+ are shown in Table 6. There was a significantly higher predictive value with regard to the patients' age, for most of

3.4. Gender and Type of Cancer Table 8 summarizes the area under the curve, in the ROC plot, and the results are divided into subgroups based on the type of cancer. There was no statistically significant difference in predictive value for the screening tests among the subgroups (p > 0.05). The highest values for the AUC were observed for the aCGA screening test. There was a similar predictive value for the aCGA and the VES-13 for a patient in the colon cancer subgroup; however, the G8 had a higher (but not statistically

Table 5 – Summary statistics of screening tests for predicting frailty based on geriatric assessment. Test VES-13 TRST

Sensitivity

Specificity

69% 95% CI (58–77) 59% 95% CI (48–68)

81% 95% CI (63–91) 86% 95% CI (70–95) 44% 95% CI (30–64) 86% 95% CI (70–95) 86% 95% CI (70–95)

91% 95% 92% 95% 83% 95% 93% 95% 94% 95%

95% CI (88–100) 50% 95% CI (33–67) 92% 95% CI (83–99) 81% 95% CI (74–87)

95% 82% 95% 94% 95% 88% 95%

G8 GFI aCGA Rockwood Balducci Fried score ECOG PS

95% 64% 95% 84% 95% 54% 95% 84% 95% 52% 95% 54% 95%

CI (89–98) CI (52–72) CI (75–90) CI (43–64) CI (76–91) CI (46–68) CI (46–62)

Positive predictive value CI (81–96) CI (82–97) CI (75–89) CI (83–97) CI (87–99) CI (91–100) CI (73–89) CI (84–99) CI (74–87)

Negative predictive value 52% 95% CI (35–61) 43% 95% CI (32–55) 95% 46% 95% 66% 95% 44% 95% 55% 95% 59% 95% 61% 95%

CI (57–93) CI (34–58) CI (51–79) CI (33–55) CI (37–71) CI (48–70) CI (52–68)

Accuracy 72% 95% 66% 95% 83% 95% 69% 95% 95% 66% 95% 76% 95% 62% 95% 61% 95%

CI (60–80) CI (56–73) CI (68–88) CI (58–77) CI (73–91) CI (59–70) CI (61–86) CI (65–78) CI (52–68)

Vulnerable Elders Survey (VES-13), Triage Risk Screening Tool (TRST), Geriatric 8 (G8), Groningen Frailty Index (GFI), abbreviated Comprehensive Geriatric Assessment (aCGA); Eastern Cooperative Oncology Group Performance Status (ECOG PS); Confidence Interval (CI). a The highest values are marked in red.

56

JO U RN A L OF GE RI A TR IC O N COL O G Y 6 (2 0 1 5 ) 5 2 –5 9

Table 6 – Summary statistics for screening tests within the subgroups 65–74 years, 75–84 years, and 85+ years old. TEST

Sensitivity

Specificity

Positive predictive value

Negative predictive value

Accuracy

p value

57% 91%

91% 0%

92% 89%

51% 0%

68% 82%

p = 0.06

48% 57% 89%

92% 78% 100%

64% 91% 94%

92% 30% 0%

64% 61% 84%

p = 0.02

97% 97%

50% 0%

80% 89%

88% 0%

81% 87%

p = 0.13

50% 70% 82%

93% 67% 100%

92% 92% 100%

52% 33% 33%

66% 70% 83%

p = 0.01

75% 89% 100%

86% 78% 100%

92% 95% 100%

68% 59% 100%

80% 87% 100%

p < 0.001

36% 60% 100%

100% 100% 100%

100% 100% 100%

48% 22% 100%

60% 66% 100%

p = 0.001

80% 94%

53% 25%

78% 91%

57% 33%

71% 87%

p = 0.69

43% 52% 82%

96% 78% 100%

95% 92% 100%

50% 25% 33%

63% 57% 83%

p = 0.07

a

VES-13 • 65–79 • 80+ TRST • 65–74 • 75–84 • 85+ G8 a • 65–79 • 80+ GFI • 65–74 • 75–84 • 85+ aCGA • 65–74 • 75–84 • 85+ Rockwood • 65–74 • 75–84 • 85+ Balducci a • 65–79 • 80+ Fried score • 65–74 • 75–84 • 85+

Vulnerable Elders Survey (VES-13), Triage Risk Screening Tool (TRST), Geriatric 8 (G8), Groningen Frailty Index (GFI), abbreviated Comprehensive Geriatric Assessment (aCGA). a Groups were re-categorized due to lack of cases in each subgroups.

significant) AUC for other cancer subgroups. In addition, the analysis of the area under the curve showed that gender did not influence the predictive value of the screening tests (p > 0.05). However, these results must be interpreted with caution due to the low power of the test in some categories (gastric and other cancer group).

Table 7 – Area under ROC curve for each screening score (area ± SE). Test

AUC

SE

95% CI

VES-13 TRST G8 GFI aCGA Rockwood Balducci Fried

0.75 0.72 0.71 0.74 0.85 0.77 0.67 0.72

0.05 0.05 0.06 0.05 0.04 0.04 0.06 0.05

0.65–0.84 0.6–0.8 0.6–0.82 0.65–0.8 0.8–0.93 0.69–0.84 0.6–0.8 0.63–0.81

AUC — area under the curve (ROC), SE — standard error of area, type I error probability of area; Vulnerable Elders Survey (VES-13), Triage Risk Screening Tool (TRST), Geriatric 8 (G8), Groningen Frailty Index (GFI), abbreviated Comprehensive Geriatric Assessment (aCGA). The highest value is marked bold.

4. Discussion Our findings demonstrated that at present, there is no universal screening test that adequately identifies frail older patients. The aCGA and G8 turned to be the best screens for older patients with cancer that qualified for elective abdominal surgery; the G8 had the highest sensitivity and negative predictive value and the aCGA was a good overall assessment tool. Moreover, age significantly influenced the predictive value of the screening tests whereas gender and type of cancer did not. There is no doubt that the Geriatric Assessment is an established method for evaluating older patients and optimizing preoperative diagnostic and treatment plans.A Cochrane meta-analysis of 22 trials with over 10,000 patients admitted as an emergency, comparing the GA with standard care, revealed an increase in the relative risk for both being alive and discharged to home at 6 and 12 months follow-up for patients that had the GA assessment.4 The GA is routinely recommended in surgical practice; it has a major impact on the detection of unknown geriatric problems and it enhances the decision making process for determining which patients are most qualified for surgical intervention.23–25 The International Society of Geriatric

57

J O U RN A L OF GE RI A TR IC O N CO LOG Y 6 ( 20 1 5 ) 5 2 – 59

Table 8 – Summary statistics for screening tests within the subgroups based on the type of cancer. Test

Gastric AUC

Pancreas AUC

Colon AUC

Rectum AUC

Other AUC

p value

VES-13 TRST G8 GFI aCGA Rockwood Balducci Fried score

0.53 0.55 –a 0.67 0.73 0.58 0.62 0.48

0.68 0.80 0.8 0.74 0.85 0.73 0.52 0.73

0.89 0.77 0.72 0.71 0.89 0.84 0.66 0.80

0.68 0.67 0.70 0.81 0.86 0.72 0.82 0.63

0.41 0.8 0.96 0.75 0.92 0.71 0.32 0.67

0.56 0.55 0.36 0.44 0.49 0.49 0.31 0.49

Vulnerable Elders Survey (VES-13), Triage Risk Screening Tool (TRST), Geriatric 8 (G8), Groningen Frailty Index (GFI), abbreviated Comprehensive Geriatric Assessment (aCGA), area under the curve (AUC). The highest values are marked bold. a Statistical calculations were not possible due to the pattern of results.

Oncology (SIOG) and the National Comprehensive Cancer Network (NCCN) include such recommendations.20 This unique feature can also be seen in this study. The family, internal medicine physicians and cardiologists consulted on all of the enrolled patients 1–14 days prior to surgery and found no medical contraindications to elective surgery. However, the GA requires experience, it is time-consuming to administer and not necessary in all patients. Can we then afford a two-stage assessment procedure (full GA only for selected group of patients), knowing that cancer surgery is a huge burden for the organism of older patient and energy expenditure is often compared to running a marathon? The published meta-analysis showed that the median prevalence of frailty, based on the screening scores, was 49% (range 12– 83%).26 In our study the median was 52.5% (range 40–85%, depending on the screening test). Table 9 summarizes the sensitivity and specificity of screening tests from a recently published meta-analysis and systematic review compared to the results of this study. The prevalence of the frailty syndrome, based on the full GA, was 73%. In the published literature a median of 68% (range 28–94%) of patients were considered frail.24–27 However, abovementioned studies included mostly papers evaluating cancer patients receiving chemotherapy, making the direct comparison difficult. Furthermore, interpreting these data, we have to take into account, that there are various GAs presented in the literature. Although, they use validated methods to investigate

problematic area they differ in both the GA instruments and cut-off values used for this purpose. Most of the studies investigate at least three of the following domains: cognitive function, mood and depression, nutritional status, ADL, IADL, co-morbidity, polypharmacy, mobility and social support. Such differences in the methodology make the comparison of reference GA significantly more complex. In our study, we assess all above listed domains apart from the social support and polypharmacy (although the question regarding the number of drugs used by the patients is included in the nutritional assessment). This particular set of measures to detect frailty is used at our institution and was chosen following the definition formulated by SIOG, which states that, at least, a GA for older patients with cancer should include assessment of functional status, cognition and mood.20 Moreover, most of the published papers are not assessing patients directly considered for surgery; presenting GA before adjuvant treatment, a mix of patients with different malignances (digestive tract, breast and/or hematological cancer) or not reporting the treatment type.27–30 Of note, Kenis C et al. evaluated relevance of a systemic geriatric screening on 1967 patients from 10 hospitals and 52% (897) received surgery. However, the screening results of surgical patients were not presented separately.25 Pope D et al. published the results of an international prospective study with 460 consecutive older cancer patients, who received Pre-operative Assessment of Cancer in the Elderly (PACE),

Table 9 – Summary statistics for screening tests.

Test

VES-13 TRST G8 GFI aCGA Rockwood Balducci Fried score

Published meta-analysis and systematic review24–27 (mostly non-surgical patients)

Present study Sensitivity

Specificity

Sensitivity range

Specificity range

69% 59% 97% 64% 84% 54% 84% 52%

81% 86% 44% 86% 86% 100% 50% 92%

39–88% 91% 77–92% 39–62% 51% 47% 94% 25–37%

62–100% 47% 39–75% 69–87% 97% 88% 50% 86–96%

Vulnerable Elders Survey (VES-13), Triage Risk Screening Tool (TRST), Geriatric 8 (G8), Groningen Frailty Index (GFI), abbreviated Comprehensive Geriatric Assessment (aCGA).

58

JO U RN A L OF GE RI A TR IC O N COL O G Y 6 (2 0 1 5 ) 5 2 –5 9

incorporating a battery of validated instruments in order to provide information about the preoperative functional reserve. Two thirds of the patients had good functional and mental status according to PACE. 61% and 65% of patients with normal performance status and ASA, respectively, had an abnormal outcome on at least one other component of PACE. Similarly, in our study 61.3% of patients with ECOG-PS ≤ 1 were frail based on the GA. However, only 31.7% of patients were with gastrointestinal cancer and the rest had breast (47%) and genito-urinary cancers (15.4%). Furthermore, there was no frailty-screening test used in the study, allowing any comparison.31,32 Kristjansson et al. compared a pre-operative multi-domain Geriatric Assessment to a modified version of the physical phenotype of frailty in a cohort of older adults with colorectal cancer. The agreement between the classifications was poor; CGA-frailty was identified in 43% of patients, while the physical phenotype was identified in only 13% patients.33 Comparing with other previously published reports concerning general cancer patient population, the results of this study are consistent with published reports. As in other studies, frailty-screening methods with the highest sensitivity (G8) lacked specificity. Similarly, those tests with the highest specificity scores (Rockwood) lacked sensitivity. Only the abbreviated Comprehensive Geriatric Assessment score had both an acceptable sensitivity and specificity, at 84% and 86%, respectively. This finding is not surprising, because most of the screening tests (VES-13, TRST, GFI, Rockwood, Balducci, Fried criteria) were developed based on older general populations. Only the G8 and aCGA were designed specifically for geriatric oncology patients.3–11 Both of these tests, in this study showed the highest sensitivity (97% and 84%, respectively) with good specificity in the case of the aCGA. Such a good result in the case of G8 could be also influenced by the origin of this screening tool. It was created based on the MNA. Thus, we observed significant correlation between both variables and the accuracy of MNA in predicting frailty was app. 80%. The specificity of the G8 was rather low at 47%, suggesting that only every second fit person was recognized as fit based on the reference GA assessment. Unlike the study reported by Kellen et al., the sensitivity and the NPV of the aCGA in this population were much higher (84% vs. 51% and 66% vs. 48%, respectively). For the Kellen et al. study, the findings were surprisingly low; the aCGA was specially designed to identify which CGA domains should be further assessed as confirmed in this study.34 One of the most important characteristics of a screening tool is its ability to exclude the possibility of vulnerability, which is equivalent to a negative predictive value. The G8 had the best score for the NPV (84%), whereas for the aCGA tool more than one third of the patients were incorrectly considered fit by the screening score. The aCGA is not the most accurate screening tool with clear cut-offs; it was designed so that the CGA could be used more efficiently. In addition, for the assessment of cognition in the full CGA, the Blessed OMCand CDT-tests were used and not the full MMSE questionnaire; this makes the comparison more complex. The VES-13 and Fried criteria assess mainly functional status and do not precisely identify impairments in other geriatric domains such as mood, nutritional status or cognitive level, although all of these domains are connected to each other

at some level. The TRST was designed for the screening of frailty in the emergency department, which is not relevant to this study population. It has a very low sensitivity identifying only more than one half of patients with good specificity, in contrast to a published study.35 Age has an important influence on the predictive value of screening tests, as shown in Table 5. With a greater age, the predictive value increases significantly. This is likely associated with the number of impaired domains observed in older patients. The higher the overall CGA score, the easier it is for the screening test to detect frailty. Gender and cancer type did not significantly influence the predictive validity of the tests. However, the results of gastric and other cancer group must be interpreted with caution due to the low power. This sample of geriatric patients with cancer showed a high prevalence of vulnerability that was functional and physical as well as cognitive, emotional and social. Therefore, at present, screening tools for older patients cannot replace the GA; this is due to the insufficient discriminative power to select patients for further assessment. They might be helpful in a busy clinical practice and in facilities that do not have trained personal for Geriatric Assessment. However, dedicated screening scores may be used in a given population, to patients that they were designed for, with good efficacy. However, additional prospective, multicenter randomized studies are needed for further evaluation of the most sensitive tools that can be used. The limitations of this study include the fact that the data are from one clinical site and the population sample was not characteristic of the general population. The inclusion criteria were also limited only to patients undergoing abdominal cancer surgery. Furthermore, another limitation is that we did not look at whether frailty screening can predict outcome (postoperative mortality and morbidity), which will be the aim of our ongoing study. Despite these limitations, this is the first reported study to prospectively assess the efficacy of eight screening tests for older patients with cancer that qualified for abdominal surgery.

5. Conclusions At present, there is no universal screening test available for frail older patients. The aCGA and G8 were the best tests in this study for older patients with cancer that qualified for elective abdominal surgery; the G8 had the highest sensitivity and negative predictive value and the aCGA was overall an acceptable tool.

Disclosures and Conflict of Interest Statements The authors have no conflicts of interest to disclose.

Author Contributions Study concept and design: J Kenig Data acquisition: J Kenig, B Zychiewicz, U Olszewska Quality control of data and algorithms: J Kenig, B Zychiewicz, U Olszewska

J O U RN A L OF GE RI A TR IC O N CO LOG Y 6 ( 20 1 5 ) 5 2 – 59

Data analysis and interpretation: J Kenig, B Zychiewicz, U Olszewska, P Richter Statistical analysis: J Kenig, B Zychiewicz, U Olszewska Manuscript preparation and editing: J Kenig, B Zychiewicz, U Olszewska, P Richter Manuscript review: J Kenig, P Richter

REFERENCES

1. Hurria A, Gupta S, Zauderer M, Zuckermna EL, Cohen JJ, Muss H, et al. Developing a cancer-specific geriatric assessment: a feasibility study. Cancer 2005;104(9):1998–2005. 2. Rockwood K, Mitnitski A. Frailty defined by deficit accumulation and geriatric medicine defined by frailty. Clin Geriatr Med 2011;27:17–26. 3. Fried LP, Tangen CM, Walston J, Newman AB, Hirsch C, Gottdiener J, et al. Frailty in older adults: evidence for a phenotype. J Gerontol A Biol Sci Med Sci 2001;56:146–156. 4. Ellis G, Whitehead MA, O'Neill D, Langhorne P, Robinson D. Comprehensive geriatric assessment for older adults admitted to hospital. Cochrane Database Syst Rev 2011;7: CD006211. 5. Saliba S, Elliott M, Rubenstein LA, Solomon DH, Young RT, Kamberg CJ, et al. The Vulnerable Elders Survey (VES-13): a tool for identifying vulnerable elders in the community. JAGS 2001;49:1691–1699. 6. Meldon SW, Mion LC, Palmer RM, Drew BL, Connor JT, Lewicki LJ, et al. A brief risk-stratification tool to predict repeat emergency department visits and hospitalizations in older patients discharged from the emergency department. Acad Emerg Med 2003;10:224–232. 7. Soubeyran P, Bellera C, Goyard J, Heitz D, Cure H, Rousselot H, et al. Validation of the G8 screening tool in geriatric oncology: the ONCODAGE project. J Clin Oncol. 2011;29 (suppl; abstr 9001; 2011 ASCO Annual Meeting). 8. Slaets JP. Vulnerability in the elderly: frailty. Med Clin North Am 2006;90:593–601. 9. Overcash JA, Beckstead J, Moody L, Extermann M, Cobb S. The abbreviated comprehensive geriatric assessment (aCGA) for use in the older cancer patient as a prescreen: scoring and interpretation. Crit Rev Oncol Hematol 2006;59:205–210. 10. Rockwood K, Stadnyk K, MacKnight C, McDowell I, Hebert, D.B. R. A brief clinical instrument to classify frailty in elderly people. Lancet 1999;353:205–206. 11. Balducci L, Beghe C. The application of the principles of geriatrics to the management of the older person with cancer. Crit Rev Oncol Hematol 2000;35(3):147–154. 12. Katz S, Akpom CA. A measure of primary sociobiological function. Int J Health Serv 1976;6:493–507. 13. Lawton MP, Brody EM. Assessment of older people: self-maintaining and instrumental activities of daily living. Gerontologist 1969;9:179–186. 14. Tomlinson BE, Roth M. The association between quantitative measures of dementia and senile change in cerebral grey matter of elderly subjects. Br J Psychiatry 1968;114:797–811. 15. Watson YI, Arfken CL, Birge SL. Clock completion: an objective screening test for dementia. J Am Geriatr Soc 1993;41:1235–1240. 16. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 1987;40:373–383. 17. Brink TL, Yesavage JA, Lum O, Heersema H, Michael Adey BA, Rose TL. Screening test for geriatric depression. Clin Gerontol 1982;1:37–44. 18. Podsiadlo D, Richardson S. The Timed “Up & Go”: a test of basic functional mobility for frail elderly persons. J Am Geriatr Soc 1991;39:142–148.

59

19. Guigoz Y, Vellas B, Garry PJ. Mini Nutritional Assessment: a practical assessment tool for grading the nutritional state of elderly patients. Facts Res Gerontol 1994;2:15–59. 20. Extermann M, Aapro M, Bernabei R, Cohen HJ, Dros J-P, Lichtman S, et al. Use of comprehensive geriatric assessment in older cancer patients: recommendations from the task force on CGA of the International Society of Geriatric Oncology (SIOG). Crit Rev Oncol Hematol 2005;55:241–252. 21. Hanley JA, Hajin-Tilaki KO. Sampling variability of nonparametric estimates of the areas under receiver operating characteristics curves: an update. Acad Radiol 1997;4:49–58. 22. Hanley JA, McNeil BJ. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology 1982;143:29–36. 23. Audisio R, PACE study participants, et al. Shall we operate? Preoperative assessment in elderly cancer patients (PACE) can help. A SIOG surgical task force prospective study. Crit Rev Oncol Hematol 2008;65:156–163. 24. Puts MT, Hardt J, Monette J, Girre V, Springall E, Alibhai SM, et al. Use of geriatric assessment for older adults in the oncology setting: a systematic review. JNCI J Natl Cancer Inst 2012;104(15):1134–1164. 25. Kenis C, Bron D, Libert Y, Decoster L, Van Puyvelde K, Scalliet P, et al. Relevance of a systematic geriatric screening and assessment in older patients with cancer: results of a prospective multicentric study. Ann Oncol 2013;24:1306–1312. 26. Hamaker ME, Jonker JM, de Rooij SE, de Rooij SE, Vos AG, Smorenburg CH, van Munster BC, et al. Frailty screening methods for predicting outcome of a comprehensive geriatric assessment in elderly patients with cancer: a systematic review. Lancet Oncol 2012;13:37–44. 27. Owusu C, Koroukian SM, Schluchter M, Bakaki P, Berger NA. Screening older cancer patients for a Comprehensive Geriatric Assessment: a comparison of three instruments. J Geriatr Oncol 2011;2:121–129. 28. Pottel L, Boterberg T, Pottel H, Goethals L, Van Den Noortgate N, Duprez F, et al. Could the combined test VES-13 + (max-G8) represent an interesting alternative screening tool for identification of potentially vulnerable elderly head and neck cancer patients? J Geriatr Oncol 2011;2(Suppl 1):21. 29. Pottel L, Boterberg T, Pottel H, Goethals L, Van Den Noortgate N, Duprez F, et al. Determination of an adequate screening tool for identification of vulnerable elderly head and neck cancer patients treated with radio(chemo)therapy. J Geriatr Oncol 2012;3:24–32. 30. Bellera CA, Rainfray M, Mathoulin-Pélissier S, Mertens C, Delva F, Fonck M, et al. Screening older cancer patients: first evaluation of the G-8 geriatric screening tool. Ann Oncol 2012;23:2166–2172. 31. Pope D, Ramesh HSJ, Gennari R, Corsini M, Maffezzini M, Hoekstra HJ, et al. Pre-operative assessment of cancer in the elderly (PACE): a comprehensive assessment of underlying characteristics of elderly cancer patients prior to elective surgery. Surg Oncol 2006;15(4):189–197. 32. Audisio RA, Pope D, Ramesh HSJ, Gennari R, van Leeuwen BL, West C, et al. Shall we operate? Preoperative assessment in the elderly cancer patient (PACE) can help. A SIOG surgical task force prospective study, Clin Rev Oncol Hematol 2008;65:156–163. 33. Kristjansson SR, Jordhoy MS, Nesbakken A, Wyller TB. A comparison of two methods to measuring frailty in elderly patients with colorectal cancer. Crit Rev Oncol Hematol 2008;68:30–36. 34. Kellen E, Bulens P, Deckx L, Schouten H, Van Dijk M, Verdonck I, et al. Identifying an accurate pre-screening tool in geriatric oncology. Crit Rev Oncol Hematol 2010;75:243–248. 35. Rojas V, Pedro Pablo M, Herrera ME, Carrasco M, Bartolotti C, Galindo H. Are the diagnostic tools of ECOG, VES-13 and scales of frailty of Balducci and Rockwood useful to investigate the vulnerability in the older people with cancer? J Geriatr Oncol 2013;4(1):85–86.