Frailty status: prevalence and its risk factors
From this multicenter cross-sectional study among outpatient subjects aged 60 years or above with a mean age of 72.4 years, we found that the prevalence of frailty according to frailty assessment using FI-40 items was 25.2%. This prevalence is comparable with the finding from a study by Tan et al., which found that the prevalence of frailty among Singaporean subjects aged 65 years or above with mean age of 76.6 (SD 6.5) years was 27% [
18]. We must bear in mind that the prevalence of frailty depends on the tools used to assess frailty status and the setting (e.g., community-based, hospitalized, or institutionalized). Thus, different tools and different settings may lead to a different result in frailty prevalence. In interpreting the result about the prevalence of frailty in this study in subjects who had received geriatric service in a referral hospital, the prevalence might be different than other Indonesian elderly populations.
Age is one of the risk factors of frailty identified in this study. In this study, we used 70 years as the cut-off to categorize subjects based on age. This cut-off was chosen because the life expectancy of Indonesian elderly is 70.1 years [
19]. We would like to compare those who lived beyond the average life expectancy and younger subjects. The risk of frailty was 2.7 times greater among those aged 70 years or above. This result is similar to reports in other studies conducted in Turkey and Brazil, which identified age as a factor significantly associated with frailty [
20,
21]. Phenotype of aging manifested due to a discrepancy between stressors and protective mechanisms, as well as failure to compensate, which results in defects. The manifestation of aging phenotype will exacerbate frailty, increase susceptibility to disease, and failure to thrive [
22].
Being malnourished or being at risk of malnutrition was identified as a risk factor of frailty in this study. This result is in line with the findings from previous studies by Wei et al. and Eyigor et al. [
20,
23] Nutrition has been proven to be an important factor for the development of frailty. Higher protein intake was associated with lower frailty incidence [
24]. Nutritional states are also associated with frailty state transition and it is suggested to monitor changes in nutritional status to prevent worsening of the frailty state [
20].
In our study, the risk of frailty was 2.9 times greater among elderly with dependent functional status. Similarly, the previous study reported that ADL showed inverted association with frailty. Besides overweight, obesity, and increased comorbidities, physical inactivity is another known risk factor of frailty [
25].
Frailty transition, its prognostic factors, and prognostic scoring system
This cohort study proved that frailty is a dynamic process, even in a relatively short (i.e., 12 months) follow-up period. Our findings were comparable to those with a longer follow-up period [
6,
8]. This probably means that the observation of frailty state transition in a short period of follow-up (i.e., 12 months) may reflect frailty state transition in the next following years. The proportion of subjects who experienced frailty state worsening was higher than those who had constant frailty state and frailty state improvement. The percentage of subjects who experienced frailty state worsening was smaller than that in Wei et al. and Trevisan et al. (27.2% vs 46.8% vs 32.7%) [
9,
23]. Furthermore, frailty states’ worsening in our study was slightly higher than Alencar’s, which had a similar follow-up period (27.2% vs 24.2%) [
7]. Most of our subjects had a constant frailty state (58%). However, frailty state improvement in our study was lower than Faria et al.’s (14.8% vs 45%) [
26]. Nevertheless, it showed that there is a possibility for individuals to transition into lesser frailty states, although the probability is low. Thus, effort on frailty state improvement in prefrail and frail older persons should not be neglected.
Unlike previous studies in a community setting [
6‐
9], the definition of frailty state in this prospective cohort study was performed using FI-40, which enabled the collection of new data on the frailty state transition and its prognostic factors. FI-40 is the most suitable instrument to evaluate frailty state in the subjects, especially in a hospital setting. Furthermore, it is the only instrument that covers all of the frailty factors and uses a continuous scoring system [
27].
Our study found that the older age group (age ≥ 70 years) had increased risk of frailty state worsening in 12 months. This finding was in line with previous longitudinal studies with a longer follow-up period in European and East Asian populations, which found that older age was associated with the transition into worse frailty state [
9,
28]. A previous study in a large Italian population (
n = 2925) with a mean age of 74.4 ± 7.3 years identified older age as one of the factors associated with frailty state worsening over a mean of the 4.4-year cohort (OR 1.11, 95% CI 1.10–1.12,
p < 0.001 for frailty state worsening) [
9]. In addition, a cohort of 2 years on community-dwelling older Chinese (mean age of 73.5 ± 4.7 years for males and 73.8 ± 5.0 for females) reported that increasing age was one of the factors significantly associated with frailty status exacerbation [
28]. Therefore, our finding supports the existing evidence that older age, in this case being 70 years or older, increases the risk of frailty status worsening within a follow-up period as short as 12 months.
The correlation between frailty and HR-QoL in older adults has been extensively studied [
5,
6,
29,
30]. Prior study of QoL and frailty revealed that prefrail and frail subjects presented with worse physical and cognitive HR-QoL compared with fit subjects [
4]. A previous longitudinal study by Gobbens et al. has shown the ability of frailty assessment in predicting the presence of adverse outcomes including poor QoL in 1–2 years [
31]. However, the possibility of QoL in affecting frailty state transition of older adults has scarcely been explored. In our study, the prevalence of subjects with negative QoL (i.e., fair and poor QoL) was 64.2% based on general health domain in the SF-12 questionnaire and those who had a higher risk of experiencing frailty state worsening in 12 months compared with subjects with better QoL (RR 2.5; 95% CI 1.1–5.9). The finding from our prospective observation suggested that there was a clear time relationship between the QoL and frailty state worsening, in which negative QoL occurred before frailty state worsening. A longitudinal study of psychological well-being and frailty found that better psychological well-being based on the CASP-19 QoL questionnaire was protective against being prefrail (RR 0.79; 95% CI 0.71–0.89) and being frail (RR 0.62; 95% CI 0.52–0.74) at 4-year follow-up [
32]. The confidence interval in our study was wide compared to those studies with larger number of subjects, which could be explained by the smaller sample size in our study.
Our study also emphasized the essential role of functional mobility in predicting the frailty state transition. Our study found that subjects with slower usual gait speed (gait speed < 0.8 m/s based on 15-ft walking test) had an increased risk of experiencing frailty state worsening in 12 months. There are two perspectives of assessing frailty: frailty as deficit accumulation and frailty as a phenotype. Our study used the FI-40 questionnaire to assess frailty status as deficits accumulation. The FI-40 did not incorporate the gait speed component in the measurement. Thus, to complement the data on frailty, we also measured subjects’ gait speed, which represents physical frailty components. The results of this study clearly reaffirm that slow gait speed is closely related to frailty, especially the worsening frailty state transition. Walking requires the coordination of various organ systems and consumes energy, thus decreased organ function and increased energy consumption for walking may be reflected through slowing gait speed. Slowing gait speed can be caused by the presence of comorbidities (e.g., cardiovascular disease and musculoskeletal problems) and frailty [
33]. Gait speed can be used to assess multiple organ systems simply and comprehensively [
34]. Numerous prior studies have reported the association between gait speed and frailty. A study by Jung et al. in rural community-dwelling Korean elderly population reported that slower gait speed was associated with worse frailty status [
35]. Furthermore, a cross-sectional study in Malaysia by Badrasawi et al. reported that slower rapid pace gait speed (> 5.2 s) which reflected poor physical function was one of the significant risk factors of frailty [
36]. It was argued that lower rapid pace gait speed had high correlation with slowness because gait speed was one of the criteria that define frailty, besides the four other criteria in Fried criteria [
36]. While the study in the Malaysian population had identified lower rapid gait speed as one of the risk factors of frailty, our study identified lower usual gait speed (gait speed < 0.8 m/s based on 15-ft walking test) as predictors of frailty state worsening in 12 months.
Similar to our study, the previous study by Fallah et al. found that mobility was significantly associated with frailty state transition in 18 months [
37]. It is interesting to notice the different cut-off values for gait speed between our studies. In the present study, subjects were categorized as having lower gait speed when 15-ft walking test results were less than 0.8 m/s, while in their study gait speed less than 0.6 m/s based on a 20-ft walking test was considered slow. The follow-up periods in the studies were both relatively short (12 and 18 months). We can infer from both studies that gait speed can be used to predict frailty status transition in the relatively near future. Another longitudinal study over a 12-month cohort in Malaysian older adults reported that walking was the only type of physical activity according to the PASE questionnaire that was significantly associated with frailty state transition [
38].
To our knowledge, the present study is the first multicenter cross-sectional and longitudinal study on frailty in an outpatient setting. Conducting this study in an outpatient setting enabled frailty state evaluation to be performed comprehensively using FI-40 as the instrument, which had seldom been done in previous community-setting studies [
6‐
9]. Furthermore, we developed a scoring system to identify older patients who have a higher risk of frailty state worsening in 12 months, which can be easily applied by healthcare professionals. Our prognostic study was the first one to develop a prediction model for frailty transition in Indonesia. We applied mathematical models to predict the phenomenon of frailty transition within 12 months. This simple scoring system consisted of three identified prognostic factors (i.e., functional mobility, QoL, and age), which can be assessed in a short duration of time by asking older patients to complete the 15-ft walking test, interviewing them using the general health domain in the SF-12 questionnaire and recording their age. It should be noted that although the association between gait speed and frailty has been clearly established, gait speed alone should not be used independently to predict frailty transition. Besides the emphasis on gait speed, our study also emphasizes the importance of maintaining good QoL and functional mobility in older adults, especially the prefrail and frail ones. Thus, our prognostic study has combined gait speed, age, and QoL to develop a prognostic model that can be used to predict frailty transition in the 12-month period. Because frailty has been known to be associated with poor health-related outcomes, information about the risk of being frail in the future is important for elderly patients and their families as well as health professionals caring for elderly patients. The development of such prognostic model is important to help physicians and their elderly patients in making a well-informed clinical decision related to high-risk procedures, invasive treatment, or diagnostic tests in particular. Our prognostic model has been proven to have good internal validity, however, future study is needed to assess the external validation of this prognostic model in other Indonesian populations.
However, there were some limitations in our study. Firstly, results from the cross-sectional study could not be used to establish causal relationships between frailty and the independent variables. Secondly, the number of drop out subject in prospective cohort study was 5.2%. Thirdly, because the prospective cohort study took place in RSCM, which is the National Referral Hospital, it resulted in the selection of subjects living in a metropolitan area. Thus, the results may not be generalized to other Indonesian elderly living in rural areas.