Fax: (773) 834-3538
A pilot study of the vulnerable elders survey-13 compared with the comprehensive geriatric assessment for identifying disability in older patients with prostate cancer who receive androgen ablation†
Article first published online: 11 JAN 2007
Copyright © 2007 American Cancer Society
Volume 109, Issue 4, pages 802–810, 15 February 2007
How to Cite
Mohile, S. G., Bylow, K., Dale, W., Dignam, J., Martin, K., Petrylak, D. P., Stadler, W. M. and Rodin, M. (2007), A pilot study of the vulnerable elders survey-13 compared with the comprehensive geriatric assessment for identifying disability in older patients with prostate cancer who receive androgen ablation. Cancer, 109: 802–810. doi: 10.1002/cncr.22495
Presented in part as a poster at the Annual Meeting of the American Society of Clinical Oncology, Atlanta, Georgia, June 2–6, 2006.
- Issue published online: 2 FEB 2007
- Article first published online: 11 JAN 2007
- Manuscript Accepted: 20 NOV 2006
- Manuscript Revised: 16 NOV 2006
- Manuscript Received: 15 AUG 2006
- American Society of Clinical Oncology Young Investigator Award (to S.G.M.)
- geriatric assessment;
- prostate cancer;
- vulnerable elders;
- functional impairment
Impairments in geriatric domains adversely affect health outcomes of the elderly. The Comprehensive Geriatric Assessment (CGA) is a key component of the treatment approach for older cancer patients, but it is time consuming. In this pilot study, the authors evaluated the validity of a brief, functionally based screening tool, the Vulnerable Elders Survey-13 (VES-13), for identifying older patients with prostate cancer (PCa) with impairment in the oncology clinic setting.
Patients with PCa aged ≥70 years who actively were receiving androgen ablation treatment and who were followed within the clinics at the University of Chicago were eligible. Patients self-completed the VES-13 and CGA instruments and repeated the VES-13 1 month later. Physical performance and cognitive assessments were administered by a research assistant.
Of 50 participating patients, 50% were identified as impaired by the VES-13 (score ≥3). Sixty percent of patients scored as impaired on ≥2 tests within the CGA, exhibiting deficits in multiple domains. The reliability of the VES-13 (Pearson correlation coefficient) was 0.92. The cut-off score of 3 on the VES-13 had 72.7% sensitivity and 85.7% specificity for CGA deficits and was highly predictive for identifying impairment (area under the receiver operating characteristic curve, 0.90). Patients who had mean VES-13 scores ≥3 performed significantly worse on evaluations of activities of daily living (P = .001), physical performance (P = .002), comorbidity (P = .004), and cognitive impairment (P = .003).
Functional and cognitive impairments are highly prevalent among older patients with PCa who receive androgen ablation in oncology clinics. The current results indicated that the brief VES-13 performed nearly as well as a conventional CGA in detecting geriatric impairment in this population. Cancer 2007. © 2007 American Cancer Society.
Prostate cancer is an age-associated disease. Greater than 70% of all patients with prostate cancer are diagnosed in men aged >65 years in the United States.1 The incidence of prostate cancer increases exponentially with age; the probability of developing prostate cancer increases from 2.2% (1 in 45 men) for those aged 40 years to 59 years to 13.7% (1 in 7 men) for those aged ≥60 years.2 Because of more sensitive diagnostic techniques, prostate cancer is being diagnosed more frequently and at earlier stages.3 These statistics portend a substantial increase in the number of men who will be diagnosed with prostate cancer and who will require evaluation for treatment.
The most widely used therapeutic modality in systemic hormone-sensitive prostate cancer (ie, biochemical prostate-specific antigen [PSA]-only recurrence after local therapy or overt metastatic disease) is androgen suppression by orchiectomy or gonadotropin-releasing hormone agonists. Although the timing of treatment initiation for patients with asymptomatic disease is controversial,4 androgen ablation increasingly is employed earlier in the disease course.5, 6 Although older men with prostate cancer have a higher incidence of low-risk disease characteristics, they are more likely to be treated with androgen ablation than other modalities, including watchful waiting.3, 7 Because androgen ablation is continued life-long, many men live with the side effects from androgen ablation for many years.8 These adverse effects include complications of osteoporosis, sarcopenia, declining physical performance, and potential cognitive effects. The prevalence of functional, cognitive, and physical impairments in an at-risk population of older men with prostate cancer undergoing treatment of androgen ablation is not well documented.
Older patients are a heterogeneous group, and the spectrum of impairment can range from those who are independent, to those who are at moderate risk of health deterioration (vulnerable), and those who are at a high risk of functional decline or mortality (frail).9–11 Disability and comorbidity have distinct and possibly synergistic influences on underlying vulnerability and frailty in the elderly. Disability, or functional impairment, is defined as dependency in performing tasks that allow for self care and living in the community and also may include difficulties with physical mobility. Comorbidity, the concurrent presence of ≥2 medical illnesses, usually chronic in nature, is highly prevalent in older individuals. Frailty, which Fried et al defined as a state in which patients are highly vulnerable to adverse health outcomes, is a physiologic state in which there is increased vulnerability to stressors that results from decreased physiologic reserves.12 Clinical criteria proposed by Balducci and Extermann to identify frail elders include age ≥85 years, dependence in ≥1 activities of daily living (ADLs), the presence of ≥3 comorbid conditions, and the presence of ≥1 geriatric syndromes (eg, dementia, incontinence, falls).13 Although definitions still are evolving, vulnerable elders have a greater likelihood of having modifiable risk factors for health deterioration than elders who are frail. Therefore, vulnerable elders may be targeted for interventions to improve cancer-related and overall outcomes.
According to the National Cancer Comprehensive Network guidelines, a multidimensional comprehensive geriatric assessment (CGA) should be a key part of the treatment approach for vulnerable older cancer patients.14 The CGA includes evaluation of comorbidity, functional status, physical performance, cognitive ability, psychological status, medication review, and social support. The benefits of geriatric assessment in older patients may include prolongation of life and prevention of hospitalizations and admissions to long-term care facilities,15–17 prevention of geriatric syndromes,18–20 recognition of cognitive deficit,21 improvement of health status,22 and detection of unsuspected conditions that may affect cancer treatment in >50% of patients aged ≥70 years.23, 24 Despite recent studies that demonstrated the feasibility of CGA in oncology, its adoption as the standard of care has been slow because of a lack of resources, difficulties with interpreting results, and difficulties with implementing targeted interventions in specialty clinic settings, such as urology or oncology.25–27 A short, simple, validated screening procedure that could be adapted to the specialty clinic setting to quickly identify those patients who are at risk for geriatric disability would be valuable. With such screening, impaired patients could be offered referral to specific geriatric programs for interventions, whereas older patients who are not at risk would be spared the more cumbersome CGA. Currently, little is known about the usefulness of brief screening tools in selecting those older cancer patients who would benefit most from the full CGA with targeted interventions.
The Vulnerable Elders Survey-13 (VES-13) is a self-administered survey that consists of 1 item for age and an additional 12 items that assess self-related health, functional capacity, and physical performance.10, 11 In the national sample of elders from the Medicare Current Beneficiary Survey that was used to validate the VES-13, a score ≥3 identified 32% of individuals as vulnerable.10, 11 This identified group had >4 times the risk of death or functional decline over 2 years compared with elders who scored <3. Higher scores predict increasing risk for functional decline and/or death.28 The average time elders took to complete the VES-13 was <5 minutes.29 Although the utility of the VES-13 compared with a more complete geriatric assessment has not been tested previously, the VES-13 has been used in oncology to help with patient selection, risk-stratification, and toxicity evaluation.30 In the current pilot study, we examined the prevalence of geriatric impairment in older patients with prostate cancer who were receiving androgen ablation and evaluated the utility of the VES-13 in identifying impairment compared with the CGA.
MATERIALS AND METHODS
Patient Population and Research Design
The study population included a convenience sample of patients aged ≥70 years who were receiving androgen ablation for histologically confirmed prostate cancer. Patients were required to have systemic prostate cancer, which was defined as having started androgen ablation for a rising PSA level after local therapy (ie, an increased PSA level on ≥2 successive measurements ≥2 weeks apart),31 or asymptomatic, metastatic disease. Other inclusion criteria included adequate command of the English language, ability to give informed consent, no history of prior cytotoxic chemotherapy use, and no other active cancer diagnosis. All patients had to be receiving their primary oncology care at the University of Chicago Genitourinary Oncology clinics. Patients who exhibited severe cognitive impairment, as measured by the Short Portable Mental Status Questionnaire (SPMSQ) (>5 errors),32 and/or who had less than an 8th-grade education, and who did not have a medical proxy for medical decision-making were excluded from study participation.
In this cross-sectional, observational study design, eligible patients were screened with the VES-13 and completed a standardized CGA. The CGA consisted of a compilation of reliable and validated tools that assessed major geriatric domains, including functional status, physical performance, comorbidity, number of medications, cognition, and social support.32–41 It has been demonstrated that impairments in these geriatric domains have a negative impact on health outcomes in the elderly.17, 24, 35, 38, 42–44 Cut-off scores for impairment on the individual assessment tools are associated prospectively with increased risk for subsequent disability or mortality in the community-dwelling elderly population (Table 1).
|Test (Study)||Geriatric domain||No. of questions||Administration [Minutes]||Score range||Cut-off point associated with adverse outcomes*|
|VES-13 (Saliba et al, 200110, 11)||Functionally based screening measure||13||Self-administered ||0–10||≥3|
|ADL (Reuben et al, 199215; Stuck et al, 199321; Katz et al, 196333)||Function||8||Self-administered [5–10]||0–16||≤14|
|IADL (Fried et al, 200412; Stuck et al, 199321; Lawton, 198834)||Function||7||Self-administered [5–10]||0–14||≤12|
|SPPB (Guralnik et al, 1994,40 199535)||Objective evaluation of function/physical performance||3 Separate physical performance tests||Administered by member of research team [10–15]||0–12||<9|
|CALGB (Charlson et al, 198736)†||Comorbidity||18||Self-administered ||0–54||>10|
|No. of medications (Juurlink et al, 200338)||Comorbidity/toxicity potential from drug interactions||1||Self-administered [1–5]||0–∞||≥5|
|RAND MOS Social Support Scale (Ware and Sherbourne, 199237)||Social support/access to medical care and support||5||Self-administered [1–5]||0–5||<4|
|Short Portable Mental Status Questionnaire (Wenger et al, 200329; Pfeiffer, 197532; Stump et al, 200142)||Cognition/risk for dementia||10||Administered by member of research team [10–15]||0–10||>3|
The reliability of the VES-13 in this population was determined by collecting survey results at a first visit and then 1 month later. The validity of the VES-13 in this population was assessed by comparing the VES-13 results with results from a simultaneous CGA. Patients self-administered surveys, and a trained member of the research team administered physical and cognitive performance measures. Any missing responses from the self-administered portion of the research interviews were completed in follow-up interviews over the telephone.
The total score for the VES-13 and scores for each test within the CGA were recorded. In addition, each test was scored dichotomously, indicating impairment or no impairment according to published values. Based on previous research, meeting the cut-off scores for impairment in ≥2 tests within the CGA signifies vulnerability (eg, increased risk for future disability or mortality).20, 21, 27, 28, 35 Impairment on the CGA was defined as meeting the cut-off scores for impairment on ≥2 of 7 individual tests within the CGA. This definition was chosen to select the cancer patients who were impaired on multiple geriatric domains who were most likely to benefit from specific interventions. Impairment on the VES-13 was not included in this definition.
Questions that described patient demographics were included with the first survey. Disease characteristics, including Gleason score (histologic grading of tumor aggressiveness), previous time on androgen ablation in months, and disease status (ie, asymptomatic, rising PSA without overt metastatic disease vs the presence of metastatic disease by symptoms or on imaging studies), were extracted from the medical record.
Data were managed in a specific research file that was stored in an area with access limited to members of the research team. Participation in the study was entirely voluntary, and ethical standards for human participation were followed strictly. The University of Chicago Institutional Review Board approved the procedures followed.
We planned to recruit a sufficient number of participants to provide a suitably precise estimate of the proportion that scored as impaired on the VES-13. A sample of 50 enrolled participants would provide an estimate of this proportion to within ± ≤14%, depending on the prevalence value. Based on previous research in older community-dwelling adults, we estimated that >30% of our sample would score as impaired on the VES-13.11 In a study of the VES-13 administered to elders who had similar socioeconomic and demographic characteristics, >60% scored as impaired.45
To examine patient and disease characteristics, descriptive statistics and summary statistics were employed. Reliability of the VES-13 score was assessed with the Pearson correlation coefficient. To assess the contribution of age alone to the total score, we also examined other subcategories within the VES-13 for reliability.
To determine the most appropriate cut-off point for impairment for this sample, a receiver operating characteristic (ROC) analysis was employed. The ROC evaluated the VES-13 as a screening measure for impairment compared with the CGA (using the definition of impairment on the CGA as deficits on ≥2 individual tests within the battery).21, 46 The area under the ROC curve (AUC) was calculated to reflect the predictive value of the VES-13 for identifying impairment. An AUC of 0.5 represents predictive ability no better than chance, whereas an AUC of 1.0 indicates perfect predictive ability.
The VES-13 score with the most appropriate sensitivity and specificity for identification of impairment was to be used for further analyses. In addition, because a VES-13 score ≥3 is associated prospectively with adverse outcomes in the elderly,11 this score also was to be included in further analyses. Confirmatory sensitivity and specificity analyses and positive and negative predictive values were computed comparing the screening test, the VES-13, with the gold standard for diagnosing geriatric risk factors, the CGA. In addition, trade-offs between sensitivity and specificity for identifying impairment with VES-13 compared with each test within the CGA were examined. Differences in results on the CGA battery by group, ie, impaired versus not impaired on the VES-13 measure, were obtained by using 2-sample t tests. STATA software (version 9.0) was used for all statistical analyses.
Fifty-eight patients consented to participate in the study, and 50 patients returned completed surveys and were included in the final data analysis. There were no obvious differences in age, race, and prostate cancer characteristics between the 8 patients who were excluded and the patients who completed study procedures.
Patient and Disease Characteristics
The patients who participated in this study represented an older age group (ages 70–92 years). The sample population represented in this study was well educated and primarily was married. Greater than 33% of the patients in our sample were African Americans (Table 2).
|Time on ADT, mo|
|Marital status, %|
|Disease status, %|
|Baseline VES-13 scores|
|Score range (of 10 possible)||0–9|
|Score ≥3, %||50|
Although the majority of patients had intermediate or higher grade tumors (≥6) according to Gleason score criteria, 80% had systemic disease according to PSA criteria only. These patients had no evidence of overt metastatic disease by imaging criteria and had minimal or no disease symptoms. There was a wide range of treatment lengths with androgen ablation (range, 3–96 months); however, 80% of men in the sample had been on androgen deprivation for >12 months.
Distribution of VES-13 and CGA Scores
There was a wide range of scores on the VES-13 within the sample population (range, 0–9). The median score was 2.5, and the an interquartile ratio was 4.5. Using scores ≥3 to signify impairment,10, 11 50% of this population scored as impaired. VES-13 scores within each category are depicted in Table 3. Reflecting the older age of this sample, 72% and 24% of patients were ages ≥75 years and ≥85 years, respectively. Forty-two percent of patients scored the maximum of 2 points within the physical disability section, and 30% of patients had difficulty or did not perform ≥1 of the tasks that evaluated function.
|Category||Frequency (%), n = 50||Points accumulated per scoring instructions|
|Fair or poor||34||1|
|Good, very good, or excellent||66||0|
|Physical disability: No. of items marked with a lot of difficulty or unable to do (1 point for each response: maximum, 2 points)*|
|Functional disability: No. of items marked as needing help or not performing task because of health (4 points for ≥1 response)†|
This sample also was impaired using the CGA as the gold standard (Table 4). Sixty percent of patients demonstrated deficits in ≥2 tests within the CGA. The median number of impaired tests was 3 (range, 0–7 impaired tests), and 25% of patients were impaired in ≥4 domains within the CGA.
|Test||Score range||Abnormal score||Percentage impaired||Sensitivity||Specificity||PPV||NPV|
|CGA||NA||Deficits in ≥2 tests||60||72.7||85.7||88.9||66.7|
|ADL||0–16||≤14 (Dependence in any ADL)||24||83.3||60.5||40||92|
|IADL||0–14||≤12 (Dependence in any IADL)||42||76.2||69||64||80|
|Comorbidity score||0–54||>10 or ≥2 Comorbidities that interfere “somewhat” with daily function||34||76.4||63.6||52||84|
|No. of medications||0–∞||>5||46||69.6||66.7||64||72|
|MOS Social Support||0–5||Average score <4||18||33.3||46.3||12||76|
|Short Portable Mental Status Questionnaire||0–10||≥3 Errors (mild cognitive impairment)||24||75||57.9||36||88|
Reliability of the VES-13 Measure
The reliability of the total VES-13 measure was 0.92 using the Pearson correlation coefficient. After excluding age, which was 1 of the criteria in the VES-13, the reliability of subcategories within the VES-13 remained acceptably high (self-rated health, 0.52; physical ability, 0.60; functional ability, 0.70).
Sensitivity and Specificity Analyses
An ROC was constructed to determine the appropriate VES-13 cut-off point for impairment in this sample and was designed to compare the VES-13 with the CGA as the gold standard (Fig. 1). Overall, the VES-13 was highly predictive for identifying impairment compared with the CGA, with an AUC of 0.900 (standard error, 0.05; 95% confidence interval for detecting asymptomatic normals, 0.800–0.995). The VES-13 remained predictive for identifying impairment when excluding patients who scored in the impaired range because of age alone (≥85 years). Except for social support, the VES-13 also was predictive for impairment in specific geriatric domains compared with individual tests within the CGA (Table 4). Finally, 2-sample t tests were used to examine group differences in scores on individual measures within the CGA. Patients who were impaired by the VES-13 screening measure performed significantly worse on all tests with the exception of social support (Table 5).
|Test||Range||Patients with mean VES scores <3||Patients with mean VES scores ≥3||P|
|No. of medications||0–∞||3.7||6.0||.004|
A high proportion of older patients with prostate cancer who are receiving androgen ablation have geriatric impairment that may place them at greater risk for decline or death. Fifty percent of the sample in this study met the definition of vulnerability on the VES-13 (scores ≥3), and 60% of patients scored in the impaired range on ≥2 individual tests within the CGA. High levels of impairment were noted in measures that were designed to evaluate the ability of patients to live in the community independently (Instrumental Activities of Daily Living), the risk of future disability and falls (Short Physical Performance Battery),35, 40 and the risk for dementia (SPMSQ).32, 39, 42 Our results are consistent with 1 other published study by investigators who detected significant disability by administering the CGA to patients with prostate cancer at various stages of disease.47
The prevalence of impairment in this sample was greater than that of the original Medicare population. In the Medicare population, the prevalence of scores ≥3 was 32% versus 50% in our sample.10, 11 This difference may be explained by multiple factors. The sample in our study was older, and increased age is an important independent predictor of functional decline and death, even when considering multiple other risk factors. In addition, the proportion of vulnerable older patients may be higher in urban populations. In a previous study, the VES-13 was administered to 412 older clinic patients of the University of Chicago South Shore Geriatrics multidisciplinary clinic.45 The proportion of patients at that geriatric clinic defined as vulnerable according to the screening tool was double the proportion reported in the original development studies. The prostate cancer sample in our study was similar to that urban geriatric population, because the oncology clinics serve the same catchment area. Finally, patients with prostate cancer may have higher impairment levels associated with their disease or treatment with androgen ablation.
Because of the known differences in our sample's characteristics compared with the original studies, we measured the utility of the brief VES-13 tool to detect impairment compared with a lengthier, more time-consuming CGA. We observed that the VES-13 had high predictive value for identifying impairment compared with the CGA using a cut-off score ≥3 in our sample. Because of the previously defined clinical meaningfulness of scores ≥3,10, 11 we chose this score to signify vulnerability. We also considered other possible cut-off scores: Although a cut-off score ≥1 had the highest sensitivity, we did not choose this score because of its lower specificity and because many patients scored in the impaired range with this score based on age alone. Patients who had cut-off scores between 2 and 4 had testing characteristics that were very similar (Fig. 1); all of these scores were highly sensitive and specific and correctly classified approximately 80% of patients. Given the traditional use of the cut-off score ≥3 and its comparable testing characteristics, we chose it for further analyses, although we acknowledge that other cut-off scores may be equally useful, depending on context. Overall, these results establish the utility and feasibility of using a screening measure to detect geriatric impairment in an older, disease- and treatment-specific cancer population in the specialty clinic setting. Additional studies to clarify further the testing characteristics of the VES-13 using different cut-off scores in other populations are needed.
It is noteworthy that 24% of the patients were scored as impaired on the VES-13 solely because they were aged ≥85 years. Although a full CGA without screening, thus, could be advocated for all oldest-old patients, a screening test like the VES-13 could help identify physical and functional impairments, which may add additional risk for adverse cancer outcomes, thus requiring more immediate intervention. In this pilot study, all patients aged ≥85 years (n = 12) had additional deficits within the VES-13. In this subset, the median score on the VES-13 was 5.5 (range, from 4 to 9 out of a total possible score of 10). Further prospective research will be necessary to determine whether functional and/or physical disability adds prognostic value over age alone within a cancer context. The age ≥85 years criteria in the VES-13 serves as a reminder for oncologists that such patients who are receiving care are at high risk for disability independent of other factors. Such a reminder is an opportunity to improve oncologic care through attention to geriatric issues. Overall, the VES-13 remains predictive for identifying impairment compared with the CGA, even after excluding patients aged ≥85 years with an ROC AUC of 0.751 (standard error, 0.08; 95% confidence interval for detecting asymptomatic normals, 0.600–0.901; results not shown).
The results from this pilot study should be interpreted with caution, because there are limitations in the study design. Defining a gold standard for the detection of impairment in geriatric domains is somewhat arbitrary, because there is a plethora of tools for identifying geriatric deficits. However, our tools are in common use, validated, and well documented in outcomes literature. Our definition of impaired CGA as 2 tests impaired within the battery, rather than 1 test, is conservative and would underestimate the sensitivity of the VES-13 in detecting older prostate cancer patients with lesser impairment. Patients with multiple impairments within the CGA are vulnerable and are at greater risk for future disability, decline, or death.48 In addition, there is the potential for selection bias. Despite our efforts, we may not have captured all eligible patients. The patients who were not captured may have had inherently different characteristics than the patients who were included, because healthier patients may not have been recognized as candidates for the study. This bias would overestimate the levels of impairment in this older male population. Alternatively, a healthy volunteer bias would underestimate the prevalence of frailty in this clinical population. The cross-sectional design and small sample size did not allow for the establishment of causal links between patient and disease characteristics and geriatric impairment. Androgen ablation and its well documented, long-term complications may be associated with the high prevalence of impairment noted in this report.49 The research hypothesis that androgen ablation may cause geriatric impairment should be studied further in prospective study designs.
The current results may not be generalizable to all elderly cancer patients. Because of varied natural histories and treatment patterns, it will be important to identify cancer-specific and cancer treatment-specific measures that identify geriatric impairment. Hurria et al published a feasibility study that evaluated a geriatric assessment measure for elderly patients who were receiving chemotherapy, and that measure is being evaluated currently in a large, prospective trial.26, 27 Their work and our study are examples of how future research may be designed to help select measures for other elderly cancer populations.
In a sample of older prostate cancer patients who were receiving androgen ablation, the VES-13 was predictive for identifying vulnerability compared with the CGA and demonstrated feasibility for use in the specialty clinic setting. The numbers of vulnerable older patients with cancer will increase as the population ages50; therefore, it is imperative that we find a better way to identify older patients who have underlying geriatric disability, so that we can better determine appropriate intervention strategies. Further research will be necessary to examine the relation between underlying geriatric impairment and outcomes of elderly cancer patients.
- 9Epidemiology of cancer and aging. J Oncol Manag. 2005; 14: 47–50..