Abstracts

Introduction

The Vulnerable Elders Survey-13 (VES-13) is a tool developed in the United States for screening individuals aged 65 and older living in the community and Medicare users ¹1. Saliba D, Elliott M, Rubenstein LZ, Solomon DH, Young RT, Kamberg CJ, et al. The Vulnerable Elders Survey: a tool for identifying vulnerable older people in the community. J Am Geriatr Soc 2001; 49:1691-9.. The purpose of the tool is to identify elderly people with an increased risk of death or functional decline: the so-called vulnerable elders ¹. The tool considers 13 items covering age, self-rated health, functional capacity and physical condition. Using the VES-13 and adopting a score greater than or equal to three as a vulnerability cutpoint, the authors detected vulnerability in 32% of older people and observed that the risk of death or functional decline was four times greater in this group than in groups with a score of less than three ¹.

A study involving prostate cancer patients conducted in Chicago (United States) assessed the psychometric properties of the VES-13 ²2. Mohile SG, Bylow K, Dale W, Dignam J, Petrylac DP, Stadler WM, et al. 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 2007; 109:802-10.. Test-retest reliability was assessed using Pearson's correlation coefficient (r) and criterion validity was tested by comparing VES-13 results with those obtained using a standardized Comprehensive Geriatric Assessment (CGA), which is considered the gold standard. With respect to reliability, the analysis showed a strong correlation (r = 0.92) between the results of the first questionnaire and those of a new questionnaire conducted 30 days later. Separate analyses were undertaken of self-rated health (r = 0.52), physical capacity (r = 0.60) and functional capacity (0.70). With regard to the validity of the VES-13, sensitivity and specificity was 72.7% and 85.7%, respectively, while the positive predictive value was 88.9%, in comparison to the gold standard (CGA) which had an area under the curve (AUC) of 0.90 (95%CI: 0.80-0.99). The tool also proved to be a good predictor of changes in activities of daily living (ADL) and instrumental activities of daily living (IADL), and of comorbidities ².

The VES-13 has been used in the United States and Europe to assess elderly patients with different types of cancer ^3,4,5,63. Rodin MB, Mohile SG. A practical approach to geriatric assessment in oncology. J Clin Oncol 2007; 25:1936-44.. On the whole, the results of these studies confirm the validity and feasibility of using the VES-13 as a tool for identifying vulnerable older people with cancer. However, no studies were found that assessed the construct validity of the VES-13.

In light of the above, we decided to create a cross-cultural adaption of the VES-13 for use with the Brazilian population targeting older people with cancer. In the first stage of the process, an analysis of the referential meaning and general meaning of the items detected strong semantic equivalence, while the pre-test of the synthesised version showed a high rate of understanding and acceptance of the questions among elderly patients ⁷7. Luz LL, Santiago LM, Silva JFS, Mattos IE. Primeira etapa da adaptação transcultural do instrumento The Vulnerable Elders Survey (VES-13) para o Português. Cad Saúde Pública 2013; 29:621-8..

The aim of this study was to assess the psychometric properties of the Brazilian version of the VES-13 in a sample of elderly cancer patients.

Material and methods

Assuming α = 0.05 (two-sided test), 80% power and a prevalence of 0.50, the estimated sample size needed for this analysis was 201 individuals. A sample of 272 cancer patients aged 60 years and over and receiving care at an ambulatory cancer centre in the city of Campo Grande, Mato Grosso State, Brazil, were interviewed between March and April 2012.

Trained interviewers conducted a questionnaire containing the Brazilian version of the VES-13 and other similar questions designed to assess construct validity. The first 108 individuals were interviewed again by the same interviewer after a period of seven to 15 days to measure test-retest reliability.

The study was approved by the Ethics Research Committee of the Sergio Arouca National School of Public Health at the Oswaldo Cruz Foundation (ENSP/Fiocruz) and all participants signed an informed consent form.

The VES-13 is made up of 13 items covering age (one question), self-rated health (one question), physical capacity (six questions) and functional capacity (five questions) and takes an average of five minutes to complete. With regard to item 1 (age), one point is assigned to individuals aged between 75 and 84 years and three points to those aged 85 years and over. Item 2 (self-rated health) has five possible answers (poor, fair, good, very good, and excellent), and one point is assigned for the options poor or fair. Items 3 to 8 (physical capacity) refer to the degree of difficulty in performing six activities and has five possible answers (none, a little, some, a lot of difficulty, or unable to do). One point is given for the answers "a lot of difficulty" or "unable to do the activity", with a maximum limit of two points. Items 9 to 13 (functional capacity) asked about difficulties in performing five activities of daily living due to health. Possible answers are "yes", "no" and "don't do". No points are given for these answers. If the answer is affirmative the individual is asked if he/she needs help to perform the activity (possible answers yes or no). If the answer is "don't do", he/she is asked if this is due to his/her health (possible answers yes/no). An affirmative answer to these latter two questions is assigned four points. The total score ranges between zero and 10 points. The vulnerability cutpoint is a score greater than or equal to three ¹1. Saliba D, Elliott M, Rubenstein LZ, Solomon DH, Young RT, Kamberg CJ, et al. The Vulnerable Elders Survey: a tool for identifying vulnerable older people in the community. J Am Geriatr Soc 2001; 49:1691-9..

During the first stage of the cross-cultural adaptation of the VES-13 for use with the Brazilian population we analysed the conceptual equivalence of the questionnaire and carried out a pre-test of the synthesised version. We found that the concept vulnerable elder had the same meaning in the culture of origin as in the Brazilian culture and that the items which made up the instrument presented similarities in both cultures. The analysis of general and differential meanings showed strong semantic equivalence and pre-test results displayed a good level of understanding and acceptance among the target population ⁷7. Luz LL, Santiago LM, Silva JFS, Mattos IE. Primeira etapa da adaptação transcultural do instrumento The Vulnerable Elders Survey (VES-13) para o Português. Cad Saúde Pública 2013; 29:621-8..

Test-retest reliability was measured using the Spearman correlation coefficient to analyse the correlation between age and self-rated health scores, between the scores for items 3 to 8 and items 9 to 13, and between the total VES-13 scores. Intraclass correlation coefficients with their respective 95% confidence intervals (95%CI) were calculated and the response options for items 1 to 13 were then dichotomized in accordance with the scoring criteria established for these items. The test-retest reliability of the answers to the questions of each item given by the respondents at the two points in time was also assessed using the simple percent agreement and Cohen's kappa coefficient. The kappa coefficient is generally thought to be a more robust measure than the simple percent agreement since it takes into account the agreement occurring by chance. We used the Landis & Koch classification ⁸8. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977; 33:159-74. to evaluate degree of agreement as follows: absent (< 0.10), weak (0.10-0.20), fair (0.21-0.40), moderate (0.41-0.60), substantial (0.61-0.80), or nearly perfect agreement (0.81-1.00).

Principal component analysis was undertaken to evaluate construct validity and explore the dimensions of the tool using Pearson's correlation coefficients. It is important to note that this may lead to an underestimation of the magnitude of correlation coefficients and the possibility that some items end up in the wrong components ⁹9. Costello AB, Osborne JW. Best practices in exploratory factor analysis: four recommendations for getting the most from your analysis. Practical Assessment, Research & Evaluation 2005; 10:1-9.. The adequacy of the correlation matrix was measured using the Bartlett statistical test and the Kaiser-Meyer-Olkin (KMO) measure. Varimax rotation was used to identify the factors which best represented the items contained in the tool. The appropriate number of factors was determined based on the proportion of overall variance explained by each variable.

Construct validity was also measured by testing convergent and discriminant validity using indicators similar to the items contained in the VES-13 taken from tools routinely used to assess elderly patients in health care settings. For the analysis of convergent validity, we considered the strength and direction of the relationship of each item with its alternative indicator, while for discriminant validity, we examined the relationship between VES-13 items and corresponding alternative indicators and the relationship between the functions (physical capacity and functional capacity) and corresponding alternative measurements.

Results

The mean and median age of the participants was 71.6 (± 0.4) and 71 years, respectively. The youngest participant was aged 60 years, while the oldest member of the sample was aged 97. The predominant age group was 60 to 74 years (n = 176, or 67.4% of the sample). The majority of the sample were male (n = 157, or 61.6% of the sample). The prevalence of vulnerability based on a cutpoint score of greater than or equal to three was 50.4% (n = 137). No statistically significant differences were found in relation to gender, age group and vulnerability condition between the overall sample and sub-sample of older people re-interviewed to measure test-retest reliability.

The Spearman correlation coefficient (ρ) for total VES-13 scores in the test-retest was 0.67 (p < 0.001). The scores for the age item (ρ = 0.98; p < 0.001), self-rated health item (ρ = 0.52; p < 0.0001), physical capacity items 3 to 8 (ρ = 0.66; p < 0.001), and functional capacity items 9 to 13 (ρ = 0.48; p < 0.001) showed a strong to moderate statistically significant correlation, demonstrating reasonable test-retest reliability.

Average scores for the VES-13 varied between 0.42 (age) and 1.37 (functional capacity) in the test, and between 0.41 (age) and 1.33 (functional capacity) in the retest. Estimated average scores and intraclass correlation coefficients are shown in Table 1. All the intraclass correlation coefficients were greater than 0.60.

Simple agreement between the answers given to the VES-13 questions in the two interviews ranged between 0.76 and 0.97. The Landis & Koch classification based on the kappa coefficients ⁸8. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977; 33:159-74. showed the following degrees of agreement: nearly perfect agreement (one item), substantial agreement (three items), moderate (four items), and fair agreement (four items). These results are shown in Table 2.

The results of the Bartlett's sphericity test (p < 0.001) and a KMO index of 0.91 confirmed sample adequacy for principal component analysis. Three factors were retained which explained 72.6% of overall sample variance.

Table 3 presents the results of the analysis after the Varimax rotation. The first component included five functional capacity variables related to activities of daily living, each with positive loads (≥ 0.943). In the second component, five of the six physical capacity items presented high positive loads (≥ 0.667) and the self-rated item presented a moderate negative load. With respect to the third component, age was the only variable to present a high positive load. Low loads were observed for three of the physical capacity variables, two of which were negative (reaching or extending arms above shoulder level, and writing, or handling and grasping small objects) and one positive (walking a quarter of mile).

Table 4 shows that the VES-13 has good convergent validity. As a whole, there was a strong and statistically significant correlation between the VES-13 items and alternative indicators. Only two items (managing money and doing light housework) presented values below 0.20 and were also statistically significant. The correlations between the scores of the physical capacity and functional capacity items of the VES-13 and their respective alternative indicators demonstrate convergent validity of the Brazilian version of the tool (Table 5).

The analysis of the correlations between the alternative indicators and VES-13 items also indicate reasonable discriminant validity. The analysis of alternative indicators for physical capacity showed a strong correlation between the alternative indicators picking up an object from the floor and performing moderate physical activities and the bathing item (functional capacity) of the VES-13. The alternative indicator for the functional capacity item "Do you have problems in your daily life due to difficulty in walking?" showed a strong correlation with the VES-13 functional capacity items "difficulty walking 400 meters" and "difficulty performing heavy housework". The alternative indicator "Can you tidy your house?" presented a strong correlation with the VES-13 physical capacity item "difficulty in performing heavy housework" (Table 5).

Discussion

This study is a continuation of the cross-cultural adaptation of the VES-13 for use with the Brazilian population. Adopting a universalist approach ¹⁰10. Herdman M, Fox-Rushby J, Badia X. A model of equivalence in the cultural adaptation of HRQol instruments: the universalist approach. Qual Life Res 1998; 7:323-35. it tests the psychometric properties of the Brazilian version with a sample of cancer patients aged 60 years and over.

Mohile et al. ²2. Mohile SG, Bylow K, Dale W, Dignam J, Petrylac DP, Stadler WM, et al. 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 2007; 109:802-10. assessed the test-retest reliability of the VES-13 by conducting two interviews 30 days apart with a convenience sample of 50 individuals aged 70 years and over and using Pearson correlation coefficient to compare the scores of the first and second interviews. The value of r was 0.92, which is higher than the value observed by the present study using the Spearman correlation coefficient. However, a comparison of specific items showed that values were practically the same in both studies, with the exception of the functional capacity items, whose values were higher in the study conducted by Mohile et al. ². However, these authors did not calculate intraclass correlation coefficients, simple percent agreement or kappa coefficients, and therefore a direct comparison between the studies' findings is not possible.

The analysis of intraclass correlation coefficients and kappa coefficients demonstrated the relative stability of the VES-13 items and categories. In test-retest studies, retested individuals may remember answers given in the first interview, which means that coefficients may be overestimated. Furthermore, real changes in the health of individuals may occur during the interval between the two interviews. An interval of 7 to 15 days was therefore chosen to minimize these problems.

The VES-13 is a tool for identifying vulnerable older people that considers age, self-rated health, physical capacity and functional capacity. Principal component analysis, used to assess construct validity, grouped functional capacity items and physical capacity items into separate components. Although some of these items loaded onto the other component, these loadings were comparatively low, in line with the original proposal of the tool regarding functional capacity and physical capacity. A study conducted in Japan showed that a loss of function in the elderly was preceded by a reduction in mobility ¹¹11. Kono K, Katsumata Y, Arai A, Tamashiro H. Functional status and active life expectancy among senior citizens in a small town in Japan. Arch Gerontol Geriatr 2004; 38:153-66.. Results related to self-rated health were as expected, showing negative loadings on the physical capacity and functional capacity components, grouped into the former component. The separation of age into a third component highlights the importance of giving greater emphasis to the contribution of this variable in the instrument, as suggested by a study conducted in Ireland ¹²12. McGee HM, O'Hanlon A, Barker M, Hickey A, Montgomery A, Conroy R, et al. Vulnerable older people in the community: relationship between the Vulnerable Elders Survey and health service use. J Am Geriatr Soc 2008; 56:8-15..

On the whole, the Brazilian version of the VES-13 presented good results in relation to convergent validity as shown by the positive and statistically significant correlations between the tool's items and alternative indicators. However, the correlation coefficient values of two functional capacity items ("because of your health or a physical condition, do you have any difficulty managing money" and "because of your health or a physical condition, do you have any difficulty doing light housework") were low, albeit statistically significant. A possible explanation for this weaker correlation may be the fact that the alternative indicators are related to only a part of the respective construct and are therefore not fully equivalent, which could have also resulted in incorrect responses from the interviewees. These two questions are more detailed in the VES-13 and even use examples. However, these indicators were maintained since more comprehensive alternatives were not available.

With respect to discriminant validity, physical capacity and functional capacity items were grouped into separate components. Furthermore, the correlation between these VES-13 items and the alternative indicators was generally stronger and more statistically significant, with only two physical capacity items and one functional capacity item showing a weaker correlation with the alternative indicators. This may be due to the complex interrelations between these two categories. Walking and mobility in general are related to activities of daily living which are evaluated within the functional capacity category and therefore physical difficulties may overlap into functional difficulties ¹¹11. Kono K, Katsumata Y, Arai A, Tamashiro H. Functional status and active life expectancy among senior citizens in a small town in Japan. Arch Gerontol Geriatr 2004; 38:153-66..

It was not possible to compare our findings since we were unable to find similar studies that analysed the validity of VES-13 in the Brazilian context.

Although we sought to select adequate alternative indicators from different assessment tools of older people used in Brazil, it is possible that certain indicators were not sufficiently similar to its respective item. However, based on the results obtained, we believe that the Brazilian version of the VES-13 demonstrates good convergent and discriminant validity.

Studies of the psychometric properties of the VES-13 are limited to the analysis of predictive validity for adverse health outcomes or criterion validity in relation to an arbitrary gold standard (CGA) ^{1,2,5,13,14,15}1. Saliba D, Elliott M, Rubenstein LZ, Solomon DH, Young RT, Kamberg CJ, et al. The Vulnerable Elders Survey: a tool for identifying vulnerable older people in the community. J Am Geriatr Soc 2001; 49:1691-9.. It is therefore important to advance the assessment of the Brazilian version of this tool by investigating predictive validity with respect to outcomes such as loss of function, toxicity and death among elderly cancer patients.

Another limitation is the fact that the individuals assessed by this study were part of a convenience sample of patients receiving care at an ambulatory cancer centre in Campo Grande. However, the main objective of adapting the VES-13 was to use the tool with this target group.

The VES-13 is a short tool originally based on a self-reporting questionnaire ¹1. Saliba D, Elliott M, Rubenstein LZ, Solomon DH, Young RT, Kamberg CJ, et al. The Vulnerable Elders Survey: a tool for identifying vulnerable older people in the community. J Am Geriatr Soc 2001; 49:1691-9.. However, considering the context of cancer care and the low level of schooling among the target group, it was decided that the questionnaire would be filled in by an interviewer ⁷7. Luz LL, Santiago LM, Silva JFS, Mattos IE. Primeira etapa da adaptação transcultural do instrumento The Vulnerable Elders Survey (VES-13) para o Português. Cad Saúde Pública 2013; 29:621-8.. Despite this, the Brazilian version of the VES-13 retains its original characteristics as an easily applicable and low-cost tool that can be used in different levels of care ^2,5,132. Mohile SG, Bylow K, Dale W, Dignam J, Petrylac DP, Stadler WM, et al. 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 2007; 109:802-10..

The analyses conducted by this study indicate that the psychometric properties of the cross-cultural adaptation of the VES-13 are consistent and adequate for use with the Brazilian population. However, we emphasise the need for further research to assess predictive validity and criterion validity in comparison with a consensus-based gold standard.

Contributors

L. L. Luz participated in all stages of this study including study design, data collection, statistical analysis, data quality control, and the drafting and revision of this manuscript. L. M. Santiago and I. E. Mattos collaborated with data collection, statistical analysis, and the drafting and revision of this manuscript. J. F. S. Silva contributed towards data collection, data quality control, and the drafting and revision of this manuscript.

Acknowledgments

This study received financial support from Fiocruz/CNPq's Technological Development in Public Health Programme (PDTSP) and the Fundação Carlos Chagas' Unified Health System Research Programme (PPSUS).

References

Publication Dates

History