Multidimensional frailty and quality of life: data from the English Longitudinal Study of Ageing

Literature regarding frailty and quality of life (QoL) is limited to a few cross-sectional studies. Therefore, using data from the English Longitudinal Study on Ageing (ELSA), we aimed to assess associations between multidimensional frailty, determined by MPI (Multidimensional Prognostic Index), with mortality and good QoL expectancy (GQoLE) indicators, over 10 years of follow-up. Overall, multidimensional frailty was strongly associated with poorer QoL and higher mortality risk. It is thus of importance to screen early for the presence of multidimensional frailty in order to try to improve QoL in older adults.

Frailty is associated with a decline in multiple physiological systems with a resultant elevated vulnerability to stressors [1]. It is widely known that frailty increases the risk of multiple adverse health outcomes in older people, such as disability, falls, hospitalization, institutionalization and, consequently, mortality [2]. Frailty is highly prevalent among older people. For example, a recent study using comprehensive geriatric assessment (CGA) to identify multidimensional frailty found that frailty may affect more than a quarter of older people, with a significant difference across settings [3].

While specific tools for frailty have been developed and studied, there is also evidence that elements of the standard CGA, often collected during geriatric clinical practice, can be used to assess frailty. In this regard, the MPI is commonly used for clinical decision making that captures different aspects of frailty, some of which have been extensively used to study frailty [4, 5]. Although the MPI was originally developed and validated in hospitalized older people [6], it is widely applied in other settings and conditions such as ambulatory clinical settings [7] or community-based studies that have included > 54,000 subjects, to date [8‐11].

There is a large body of literature that has consistently found MPI to be associated with several negative outcomes. However, research regarding the impact of higher MPI values, which can be considered a proxy measure of multidimensional frailty, on QoL is still unknown. Previous research has reported that frailty, mainly identified using the physical frailty phenotype, is associated with poor QoL [12]. However, these studies utilized small sample sizes and were of a cross-sectional design, consequently limiting the representativeness of findings [12]. Cross-sectional studies, often failed to include more severe ill patients, introducing a possible “informative censoring” [13]. Moreover, QoL is an important outcome and increasing research has shown that it should be included in epidemiological investigation [14], since one objective of modern geriatric medicine is to implement interventions for increasing living years in good QoL and not only for increasing life expectancy, LE [15].

To date, no study has explored the association between multidimensional frailty and QoL utilizing a longitudinal design. Therefore, the aim of the present study was to investigate associations between multidimensional frailty, assessed by MPI, with mortality and GQoLE indicators, in a large representative sample of older English adults, over 10 years of follow-up.

Of the 9432 participants who took part in wave 2 (baseline) of the ELSA study, 3186 were excluded because they were younger than 60 years, and 2 had missing data in relation to MPI items. Therefore, our analytic study population included 6244 subjects (Fig. 1, not weighted data).

Table 1 shows the data according to survival status during follow-up. The 1392 participants who died during the follow-up period of 10 years had significantly higher MPI values than their counterparts (4852) at the baseline evaluation (mean MPI: 0.39 ± 0.21 vs. 0.25 ± 0.17; p < 0.0001). Participants died during the follow-up period were significantly older, more frequently male, less educated, and more frequently current smokers than their counterparts (Table 1).

After adjusting for sex, age, education, marital status, and smoking status, compared to people in the low-risk group, people in the moderate (HR = 4.27; 95% CI 3.55–5.14) and severe-risk groups (HR = 10.3; 95% CI 7.88–13.5; p < 0.0001) were more likely to die (Table 2). Similar results were evident modeling MPI as a continuous variable.

As shown in Table 2 and in Fig. 2, MPI was overall accurate in predicting mortality. The Harrell’s C-index for MPI as increase in 0.10 points was 80.8 (95% CI 79.6–81.9). Similarly, the ROC curves at 3, 6, and 9 years show that the AUCs were 70.4, 75.7, and 79.8, respectively.

Finally, as reported in Fig. 3 and in Supplementary Table 2, LE as described by WHO in abridged life tables for 2005 (27) among women aged 60–64 years was 23.6 years, and they can expect to live 11.8 years with good QoL (i.e., approximately 50% of their remaining LE). Moreover, considering MPI groups, women aged 60–64 years in the MPI low-risk group could expect to live 14.9 years with good QoL, those in the moderate group 10.2 and those in the severe-risk group 5.7 years (accounting for 63%, 43% and 24% of their LEs, respectively). Men aged 60–64 years could expect to live a further 20.7 years, of which 11.5 years would be lived with good QoL (50% of their LE). The corresponding GQoLE among men aged 60–64 years in the low MPI risk group was 12.1 years, in the moderate MPI risk group 8.4 and 2.8 years in the severe MPI risk group (accounting for 58%, 41%, and 13% of their remaining life, respectively). Moreover, a trend toward loss of GQoLE was observed with increasing age and more evident in men than in women.

In the present study, including 6244 community-dwelling older subjects living in the UK, we found that higher MPI values at baseline, indicating the presence of multidimensional frailty, were associated with a significant higher risk of mortality over 10 years of follow-up. The accuracy of MPI in predicting mortality was overall good. To the best of our knowledge, this is the first study in a large cohort of older people showing that multidimensional frailty, as assessed by higher MPI values, is associated with worse QoL expectancies during a long follow-up.

Previous studies have reported a significant association between MPI and negative outcomes (in particular mortality) in hospitalized older patients affected by several independent medical conditions [30‐35]. However, more recent studies have shown that slightly modified versions of the MPI can predict negative outcomes in community-dwelling older people. For example, in the Swedish National Study on Aging and Care in Kungsholmen (SNACK), higher MPI values at baseline were associated with a higher mortality risk and higher number of days spent in hospital [11]. Similarly, in an Italian population higher MPI values were associated with a higher risk of mortality, with a good accuracy [36]. Finally, in the InCHIANTI study, MPI had a good accuracy in predicting mortality over 15 years of follow-up [17]. Moreover, in other community-dwelling studies, MPI was associated with a higher risk of depression [8], cardiovascular diseases [37], and falls [9] and it is widely known that all these conditions can decrease QoL in older people. Taken together, these findings support the notion that MPI can be used in community-based settings with an accuracy like that found in hospital-based studies and in other settings including long-term care facilities and ambulatory units, since an advantage of the MPI is its plasticity [3, 10].

The current research made another important step forward in better understanding the importance of a CGA-based tool in older people, since higher MPI values at baseline not only predicted higher mortality risk, but also poorer QoL. This finding supports a recent study in approximately 500,000 participants that found considering frailty may change LE estimates by up to 7.2 years [38]. Multidimensional frailty present at the baseline was associated with a dramatical reduction of GQoLE. For example, men and women between 60 and 64 years have approximately 20 years of expected survival, but only half in good QoL. Specifically, participants in MPI 3 (i.e., frail) have approximately 10 years less in good QoL than their counterparts in MPI 1 (i.e., robust) indicating that frailty should be diagnosed early and treated to improving QoL in older people.

It may be suggested that MPI in community-dwelling older people may help in identifying areas which deserve more or less aggressive approaches based on the impaired domains [3]. For example, preventive suggestions (such as vaccinations, healthy diet and increasing physical activity) could be suggested to people in MPI 1 low-risk group in order to avoid the transition to more advanced stages of frailty and consequently to poorer QoL during follow-up. Moreover, in participants already frail, it is important to recommend other interventions, such as deprescribing, resolving social issues or optimizing functional status [3]. The literature suggests that some interventions are effective in improving QoL in frail older people. Multidisciplinary treatment and exercise programs seem to be able to improve QoL in older people affected by frailty, even if the research is still limited to a few studies and with not univocal results [39].

Our findings should be interpreted within the studies limitations. First, the study was of an observational nature. Second, MPI was derived from available data and not according to the original tool: since different approaches for detecting frailty can lead to different findings, other studies are needed to compare the ability of this version of MPI, with the original version. Finally, we did not assess the changes of MPI during the follow-up period.

In conclusion, multidimensional frailty was associated with a significant higher risk of mortality in a large cohort of older English adults, with a good accuracy, confirming that this tool can be used in the community for detecting multidimensional frailty. Moreover, multidimensional frailty was also associated with a relevant reduction in years lived in good QoL, suggesting that the multidimensional approach should be used in older people to allow for the early identification of those that may be at risk of not only a reduced LE but also poor QoL.