This was a cross-sectional study. The data collection occurred in all PHC units in the municipality of Jatai (Brazil) from July to December 2018. The sample size for this study was calculated considering the total elderly population (≥60 years old) in Jatai (n = 10,853), a prevalence of 50%, a margin of error of 5%, and a confidence level of 95%, similarly as described in previous studies [1,14]. Consequently, the estimated sample size consisted of 370 individuals. Finally, 10% was added to compensate for possible missing information in the questionnaires. Therefore, the final sample size calculation comprised 407 older adults. Each participating PHC unit included the same number of individuals [15].
Regarding the eligibility criteria, we included adults aged 60 years or older. In addition, exclusion criteria consisted of being assisted in PHC for less than 12 months, residing in the rural area and having cognitive impairment [defined as a score ≤5 in the 10-point cognitive screener (10-CS)]. These exclusion criteria were chosen because we considered it is important to have been adequately assisted at a PHC center for a certain period of time to assure the collection of reliable information; significant cognitive impairment causes problems on self-reported data; and rural and urban areas have different characteristics, and approximately 85% of elderly adults live in urban areas [16–18].
Patients received information about this study, and the ones who agreed to participate and met the eligibility criteria signed the informed consent form.
Trained researchers contacted the patients while they were waiting for medical consultation, vaccination, wound care, or other health services in the PHC unit [1,14]. Interviews were carried out in a standardized manner using a questionnaire based on previous elderly cohorts [19,20]. Regarding sociodemographic data, we collected information on age, sex, level of schooling (years of formal education), ethnicity (white, black, and others), and living arrangements (living alone or not). Regarding the health history, we collected information on self-reported hypertension, depression, and number of medications. In addition, cognitive impairment was evaluated by the 10-CS, a validated tool for Brazilian older adults [21].
Health vulnerability was measured using the VES–13 version that was validated to Brazilian Portuguese [2,22]. The VES–13 has 13 items with specific scores: age (1 point for 75–84 years old; 3 points for ≥85 years old), self-rated health status evaluation (1 point for bad or regular), physical activity limitations [1 point for too much difficulty (or unable) in performing each item, with a maximum score of 2 points. The items were bending, crouching or kneeling; lifting or carrying objects weighing approximately 5 kilograms; raising or extending arms above shoulder level; handling and holding small objects; walking 400 meters; doing heavy house work such as scrubbing floors or cleaning windows], limitations for basic and instrumental activities of daily life (4 point for difficulty in performing one or more item, with a maximum score of 4 points. The items were buying personal items; dealing with money; walking through the room; performing light housework; taking a shower or bath). Consequently, the possible VES–13 scores vary from 0 to 10, with higher scores associated with more health vulnerability. Older adults with scores of 3 or more are classically classified as vulnerable [6]. However, we used the VES–13 as a continuous variable (from 0 to 10) in this study.
The estimated limited life expectancy in 10 years was based on Suemoto index. This index predicts the mortality risk in 10 years for community-dwelling older adults. It uses information on age, sex, self-reported diabetes, heart disease, lung disease, cancer, smoking status, alcohol use, body mass index, physical activity, difficulty with bathing or showering, difficulty with walking several blocks, today’s date correctly reported, and self-reported health status [4]. This index reports a predicted mortality in 10 years that varies from 0 to 100%, based on the effect size of each of these variables and mortality in the test cohort. We used the online calculator for the Suemoto index available at e-prognosis (https://eprognosis.ucsf.edu/suemoto.php). We considered that LLE was present when the predicted 10-year mortality risk was 50% or more. The cutoff of 50% were previously used in another study related to cancer screening in older adults [12].
The Statistical analysis was performed. The numerical variables were presented as mean and standard deviations (SD), while categorical variables were expressed as absolute numbers and percentages. The dependent variable was LLE (predicted mortality risk ≥50% calculated by the Suemoto index). The independent variable was the VES–13 score (continuous variable). Missing data for these variables were not imputed. We compared participants according to the presence of LLE using chi-square test for categorical variables and unpaired t-test or Mann-Whitney test for numerical ones. To test the association between limited life expectancy and health vulnerability measured by the VES–13, we used logistic regression analysis adjusted for age, sex, ethnicity, and education. We calculated the area under the Receiver Operating Characteristic (AUROC) curve to determine the discrimination of the VES–13 in identifying individuals with LLE. Additionally, we determined the best VES–13 cutoff with the best accuracy to identify these individuals using the Youden index [23]. We used the Stata 15 software (College Station, TX: StataCorp LLC) for the statistical analysis. The alpha level was set at the 5% level.