Agreement between six frailty assessment tolos in Colombian seniors: SABE Colombia

Concordancia entre seis herramientas de tamización de fragilidad en personas mayores colombianas: SABE Colombia

Concordância entre seis instrumentos de triagem de fragilidade em pessoas idosas colombianas: SABE Colômbia

Judarcid Guzmán-Sánchez jgumaz@unal.edu.co

Universidad Nacional de Colombia, Colombia

 https://ror.org/059yx9a68

Milena Lima de Moraes mmoraes@unal.edu.co

Universidad Nacional de Colombia, Colombia

 https://ror.org/059yx9a68

Esta obra está bajo una Licencia Creative Commons Atribución-NoComercial 4.0 Internacional.

Frailty has been established as a fundamental pillar within comprehensive geriatric assessments, as it helps identify an individual’s vulnerability (1). It is one of the most useful elements both for research and for defining therapeutic plans, interventions, health planning and resource allocation (2), relegating chronological age to second place as a prognostic predictor.(3). In addition, assessing it is strategically relevant in the field of public health because it offers valuable tools for designing preventive programs and organizing services aimed at promoting health (4). The underlying pathophysiology of frailty is complex and multifactorial, involving various systemic levels and specific components, as detailed in Table 1.

Table 1.Pathophysiological levels of the frailty síndrome

Table 1. Pathophysiological levels of the frailty síndrome
Table 1.Pathophysiological levels of the frailty síndrome

Source: prepared by the authors

Multiple tools have been designed for their operation, which has made it difficult to compare prevalence, clinical trial results and clinical decisions (5). An additional challenge lies in the divergent definitions, different measures and cut-off points used by each instrument. Many of them are not practical for community settings (6). This may generate inaccurate information about who requires specialized care in specific domains and create variations in the clinical interpretation of pathophysiological processes associated with frailty (7).

The prevalence of frailty on a global level varies widely depending on the instrument used to measure it, as well as the region that is assessed (8). It can range from 3.9% to 59.4% for frailty, and from 13.4% to 71.6% for pre- frailty (9). This condition shows a direct association with age, with a marked increase after 75 years of age (10).

Moreover, the overall incidence has been estimated at 43.4 cases of frailty and 150.6 cases of pre-frailty per 1,000 people per year, and it is higher in women (11).

In Colombia, evidence with respect to frailty in the senior population in the community is still limited (12,13). Recently, a study in five cities in the country, based on the FRAIL scale, reported a prevalence of 9.6% frailty and 60.8% pre-frailty.(14). However, when assessed by Fried’s criteria, the prevalence was 17.9% for frailty and 63.3% for pre-frailty (15). To date, no national studies have been published that simultaneously compare multiple validated tools for frailty assessments in a community setting.

The objective of this study was to identify frailty using different assessment tools: FRAIL, Study of Osteoporotic Fractures (SOF), gait speed (GS), Short Physical Performance Battery (SPPB), Modified Frailty Index (mFI) and Clinical Frailty Scale (CFS) and to assess the agreement between the results obtained from frailty assessments in the Colombian population.

Population/Study Design

This study is a cross-sectional analytical study based on secondary data from the SABE Colombia study, which is part of the National System of Population Studies and Surveys for Health. This was the first study exclusively for seniors in Colombia residing in homes (not in institutions), which was conducted between 2014 and 2015. The methods used, statistical sampling, data quality control and response rate have been previously published and are available at https://www.minsalud.gov.co/salud/publica/epidemiologia/Paginas/Estudios-y-encuestas.aspx(16-19).

The SABE Colombia participants were selected following a multistage probability sampling design, in order to achieve a representative sample of Colombian seniors. The total sample size was 23,694 people (18). Persons aged 60 years or older who were able to communicate with the research team, and who provided written informed consent, were included. Those with a total score of less than 13 on the Minimental test were excluded (20). All participants were interviewed, and their responses were recorded on tablets, while printed versions of the survey were applied and then digitized in some circumstances due to insecurity and lack of access to the internet to upload the surveys. All data were collected and managed using a database program designed specifically for SABE Colombia (Synkron, folder synchronization) (16-19).

A proprietary database was created with the variables needed to apply the 6 frailty assessment tools: 2 tools that assess physical frailty - FRAIL (10) and SOF (11); 2 tools that measure physical performance - GS (12) and SPPB (13); 1 tool that assesses cumulative deficit - mFI (14) and the last one, which is a tool based on clinical judgment - CFS (15). Individuals who failed to begin or complete the different objective tests, such as the chair test (the main cause of exclusion in this study) and gait speed, were excluded. A total of 4,704 elderly persons were excluded from analysis with the FRAIL scale because of the absence of data on some of the required variables or because their answers corresponded to the categories “do not know” or “no answer.” 19,650 individuals were excluded from the SOF scale because they did not attempt the chair test, did not complete it, or had inconsistent values. In the case of the Clinical Frailty Scale (CFS), 7,878 participants were excluded because of missing information on one or more of the considered variables. With respect to the Modified Frailty Index (mFI), 309 individuals were excluded for similar reasons.

For the gait speed test, 2,058 records corresponding to participants who did not attempt the exercise, did not complete it or presented inconsistent data were eliminated. Finally, 19,326 people were excluded from the Short Physical Performance Battery (SPPB) because they had not completed one or more of the three tests that comprise it or because of invalid data in those assessments.

SABE Colombia has prior approval from the Institutional Human Ethics Review Committee of the Faculty of Health of Universidad del Valle (minutes No. 09-014 and 011- 015) and from the Bioethics Committee of Universidad de Caldas (code CBCS-021-14). In addition, this study was approved by the Ethics Committee of the Faculty of Medicine of Universidad Nacional (minutes No. 004, B.FM. 1002-CE-0047-23).

Data Analysis

Characterization variables are presented as means and standard deviations, or as a proportions of their percentage. The frequency of frailty was calculated for each tool that was studied. In order to evaluate the degree of correlation and agreement between the results of the tools, they were categorized as dichotomous, into frail and not frail. In addition, when the tool presented the possibility of classification as pre-frailty, this category was grouped as not frail. For the agreement assessment, only people who presented complete data for the application of the 6 tools concomitantly were included. McNemar’s test was used to compare proportions of tool scores and Cohen’s kappa coefficient was used to assess agreement, interpreting kappa values up to 0.20 agreement as insignificant, 0.21-0.40 agreement as medium, 0.41-0.60 agreement as moderate, and >0.60 agreement as substantial. All analyses were performed with Statistical Package for Social Sciences® (SPSS) version 25. Amounts where P < 0.05 were considered statistically significant.

A total of 23,694 data of seniors were analyzed and distributed in different subsamples according to the studied tools. Of these, 2,336 people submitted complete data for the application of the 6 tools concomitantly (Figure 1).

Figure 1. Sample and subsamples of seniors assessed by different frailty assessment tools
Figure 1.Sample and subsamples of seniors assessed by different frailty assessment tools
Source: prepared by authors

The sociodemographic and health characteristics of the frail population identified with the 6 studied frailty assessment tools are presented in Table 2. By any of the measures of frailty that were used, the population was mainly female, from socioeconomic strata 1 and 2, with a low level of education. The most frequent chronic diseases among the frail population were high blood pressure, dyslipidemia and diabetes mellitus.

Table 2.Sociodemographic and health characteristics of the frail population identified according to 6 frailty assessment tolos

Table 2. Sociodemographic and health characteristics of the frail population identified according to 6 frailty assessment tolos
Table 2.Sociodemographic and health characteristics of the frail population identified according to 6 frailty assessment tolos

Source: prepared by the authors

A variable prevalence of frailty was identified according to the tool that was used. A higher prevalence was observed using GS, which is a tool that assesses physical performance. Meanwhile, a lower prevalence was observed using FRAIL, which assesses physical frailty (Figure 2).

Figure 2. Prevalences of frailty (95% confidence interval) assessed by the 6 studied tools
Figure 2.Prevalences of frailty (95% confidence interval) assessed by the 6 studied tools

SOF: Study of osteoporotic fractures; GS: gait speed; SPPB: Short physical performance battery; mFI: modified frailty index; CFS: Clinical frailty scale.

Source: prepared by authors

Two medium agreements were observed (SOF vs GS, κ=0.21 p<0.001; and FRAIL vs CFS κ=0.27 p<0.001), as well as one moderate agreement (SOF vs SPPB, κ=0.54 p<0.001) between the evaluated tools (Figure 3 and Table 3).

Figure 3. Cohen’s kappa coefficient to assess agreement between 6 frailty assessment tools (n = 2,336).
Figure 3.Cohen’s kappa coefficient to assess agreement between 6 frailty assessment tools (n = 2,336).

*All kappa coefficients presented p < 0.001. vs: versus, SOF: Study of osteoporotic fractures; GS: gait speed; SPPB: Short physical performance battery; mFI: modified frailty index; CFS: Clinical frailty scale.

Source: prepared by authors

Table 3. Agreement between 6 frailty assessment tools (n=2,336)
Table 3. Agreement between 6 frailty assessment tools (n=2,336)

McNemar: McNemar test.

Source: prepared by the authors

The main findings of this study were the medium and moderate agreements between tools measuring physical frailty (FRAIL and SOF) with tools assessing physical performance (GS and SPPB), as well as the judgment- based tool (CFS). It is important to highlight that there was also a lack of agreement between tools that have the same focus, such as FRAIL and SOF (which measure physical frailty) and GS and SPPB (which measure physical performance).

SOF presented agreement with the two tools that measure physical performance (SOF vs GS, κ=0.21 p<0.001; SOF and SPPB (κ=0.54 p<0.001). Physical performance is a central aspect of physical frailty, as measured by SOF (21), reason why these results indicate that this tool fulfills (with medium and moderate agreement) the objective of assessing this aspect in Colombian seniors. Frailty, as determined by SOF, is associated with risk of falls, fractures and mortality, and it is a valid tool for the community setting (21,22).

In the overall sample, SOF and SPPB presented the closest prevalence of frailty (42.4% and 42.7%, respectively), while GS had the highest observed prevalence (62.7%). Although GS is associated with physical frailty and adverse outcomes (23), it is likely that it generates an overestimation of frailty because GS is the only objective measurement. On the other hand, the other studied tools have at least 3 evaluation components, which allow analyzing other relevant aspects in identifying this syndrome.

FRAIL also measures physical frailty, but unlike the SOF, it does not have any measure of physical performance among its evaluation components. It is based solely on referral data. This tool presented medium agreement with the CFS (κ=0.27 p<0.001), which is a tool based on clinical judgment. Although they consist of different assessment components, both are tools composed of subjective assessments of functional status (21). In a multicenter study conducted in 5 European countries, medium agreement between FRAIL and CFS in seniors in primary care was also observed (κ=0.40, p<0.05) (24).

It is worth noting the lack of agreement between tools with the same focus, such as FRAIL and SOF (which measure physical frailty), GS and SPPB (which measure physical performance), as well as the lack of agreement between the cumulative deficit tool mFI and all the evaluated tools and between the judgment-based tool CFS and the physical performance tools GS and SPPB. In disagreement with this finding, Oviedo-Briones et al. (24) observed medium agreement between mFI and CFS in seniors in primary care (κ=0.23, p<0.05) (24).

The study by Ocampo-Chaparro et al. (15) evaluated physical frailty using Fried’s phenotype as a tool based on data from SABE Colombia (15) and found a prevalence of 17.9%. This diverges from the prevalences observed in this study, in which, after applying 5 tools, the observed prevalences were higher. The only tool with a lower prevalence than that of Ocampo-Chaparro et al. was FRAIL (9.1%). This can be explained by the fact that FRAIL originally classified its results into not frail, pre- frailty and frail. By dichotomizing its results, pre-frailty persons, representing 87.4% (n=16,611), were categorized as not frail, which may contribute to underestimating the prevalence. The prevalence with FRAIL in the present study is similar to that which Oviedo-Briones et al. found, at 8.7% in primary care (24).

There was a higher prevalence of frailty in the female population across most tools. The average for FRAIL was 63.5%, SOF 63.2%, CFS 61.3%, and SPPB 66.8%, which is consistent with the global literature, identifying female gender as a key factor influencing not only the aging trajectory but also the course of frailty (23). This difference can be explained both by a more constant and progressive decrease in lean body mass and muscle strength in women (24), as well as by a longer life expectancy (15). Only two tools, mFI and GS, showed greater frailty in the male population, which differs from the global literature if we consider that the former tool corresponds to a multidomain scale and the latter to an objective test. This leads us to think that the considerable number of eliminated data explains these differences and that the low female representation in the case of gait speed is due to an inability to perform the test, which would place the majority of older Colombian women in the frail population group.

A higher prevalence of frailty was found in strata 1, 2 and 3 for the different tools evaluated. It has already been shown in systematic reviews that poverty has direct and indirect effects on the level of frailty (25), and this association can be explained by the serious consequences that a poor economic situation has on an individual’s health and well- being and on the way they age in general (26). In addition, completed secondary education and higher education are the most infrequent levels of education among the frail population, which coincides with previous descriptions. They have identified the influence of level of education on social vulnerability, physical frailty, poor health status and lower long-term survival (27,28).

A high coexistence of frailty and chronic diseases was found. High blood pressure was the most frequent, followed by dyslipidemia, arthritis and diabetes mellitus, with a significant coexistence of frailty and mental illness. A meta-analysis showed a prevalence of multimorbidity of 72% in frail individuals and a prevalence of frailty 16% among multimorbid individuals, suggesting a bidirectional association (29).

As a recommendation, frailty is considered a syndrome that deserves systematic detection through the different available tools (1), perhaps requiring recalibration of both the cut-off level and the relative weight assigned to each variable in each scale to avoid producing biased estimates.

There is medium to moderate agreement between the SOF physical frailty tool and the physical performance tools (GS and SPPB), and medium agreement between the FRAIL physical frailty tool and the CFS judgment- based tool. However, there is no agreement between the mFI cumulative deficit tool and any of the studied tools, nor between the physical performance tools and the CFS judgment-based tool, reason why they cannot replace each other.

Conflicts of interest

The authors declare that they have no conflicts of interest.

Funding

The SABE study was financed by a Colciencias fund and the Colombian Ministry of Health and Social Protection (2013, No. 764); this study was funded by resources from Universidad Nacional de Colombia (Resolution 03/2017).

How to reference: Guzmán-Sánchez J, Moraes ML. Agreement between six frailty assessment tolos in Colombian seniors: SABE Colombia. MedUNAB [Internet]. 2025;28(1):36-46. doi: https://doi.org/10.29375/01237047.5149

Author Contributions: JGS participated in conceptualization, data curation, formal analysis, methodology, visualization, writing (original draft) and review/ editing. MLM participated in conceptualization, data curation, statistical analysis, methodology, visualization, writing (original draft) and review/ editing. All authors reviewed and approved the submitted version.

redalyc-journal-id: 719

https://revistas.unab.edu.co/index.php/medunab/article/view/5149 (html)

https://revistas.unab.edu.co/index.php/medunab/article/view/5149/4199 (pdf)

https://revistas.unab.edu.co/index.php/medunab/article/view/5149/4200 (pdf)

https://revistas.unab.edu.co/index.php/medunab/article/view/5149/4271 (xml)