Baseline assessments

Licensed nurses recorded demographic data, including age, sex, number of prescribed medications and medical history in face-to-face interviews. Participants were asked about their history regarding the following diagnoses: stroke, Parkinson's disease, hypertension, heart disease, diabetes mellitus and osteoporosis. We measured participants’ height and weight and calculated their body mass index (BMI). Global cognitive function was assessed using the MMSE,7 with a cut-off point of 23/24.10 Depressive symptoms were measured using the 15-item Geriatric Depression Scale (GDS).11 The cut-off score of ≥6 has a sensitivity of 82% and a specificity of 75% with a structured clinical interview for depression.12

Operationalisation of the physical frailty phenotype

We considered the physical frailty phenotype to be characterised by limitations in three or more of the following five conditions based on those used in Fried et al's2 original studies: slowness, weakness, exhaustion, low activity and weight loss. Participants who had none of these components were considered to be robust; those with one or two components were considered to be prefrail.

A majority of previous prospective cohort studies seem to agree with the use of walking speeds for health predictors in ageing.13 Walking speed was measured in seconds using a stopwatch. Participants were asked to walk on a flat and straight surface at a comfortable walking speed. Two markers were used to indicate the start and end of a 2.4 m walk path, with a 2 m section to be traversed before passing the start marker, such that participants were walking at a comfortable pace by the time they reached the timed path. Participants were asked to continue walking for an additional 2 m past the end of the path to ensure a consistent walking pace while on the timed path. Slowness was established according to a predetermined cut-off (<1.0 m/s).6 Together with slowness, low handgrip strength is considered an important indicator of health outcome such as fractures,14 disability15 and death.16 Weakness was defined using maximum grip strength. Grip strength was measured in kilograms using a Smedley-type handheld dynamometer (GRIP-D; Takei Ltd, Niigata, Japan). In addition, weakness was established according to a sex-specific cut-off (<26 kg for men and <18 kg for women).17 Exhaustion was considered to be present if the participant responded with ‘yes’ to the following questions taken from the Kihon-Checklist, a self-reported comprehensive health checklist developed by the Japanese Ministry of Health, Labour and Welfare18: “In the last two weeks, have you felt tired for no reason?” We evaluated the role of physical activity by asking the following questions about time spent engaged in sports and exercise: (1) “Do you engage in moderate levels of physical exercise or sports aimed at health?” and (2) “Do you engage in low levels of physical exercise aimed at health?” Participants who answered “no” to both of these questions were classified as low activity.6 Weight loss was assessed by a response of “yes” to the question, “Have you lost 2 kg or more in the past six months?”18

Outcomes

Participants were followed monthly for incident certification of need of care according to the LTCI system during the 2 years after the baseline assessment. Japan implemented a mandatory social LTCI system on 1 April 2000.8 ,9 Every Japanese citizen aged 65 and older is eligible for benefits (institutional and community-based services, but not cash) in cases of physical and/or mental disability. The computer-aided standardised needs-assessment system used by the mandatory social LTCI system categorises people into seven levels of needs.9 To determine an individual's level of nursing care need, a trained local government official visits that individual's home and administers a questionnaire on current physical and mental status (73 items in 7 dimensions; eg, paralysis and limitation of joint movement, movement and balance, complex movement, conditions requiring special assistance, activities of daily living/instrumental activities of daily living, communication and cognition, behavioural problems) and use of medical procedures (12 items). The results of this questionnaire are then entered into the computer to calculate the applicant's standardised scores for the seven dimensions of physical and mental status and the estimated time for nine categories of care (grooming/bathing, eating, toileting, transferring, eating, assistance with instrumental activities of daily living, behavioural problems, rehabilitation and medical services), after which a care needs level based on the total estimated time taken for care is assigned. After this, the Nursing Care Needs Certification Board, which comprises physicians, nurses and other experts in health and social services, reviews and confirms the care needs level.9 We defined onset of disability as the point at which a participant was certified as needing care according to the LTCI classification.

Statistical analyses

Student's t test and Pearson's χ² test were used to test differences in baseline characteristics between participants with incidence of disability during the 2 years after baseline assessment and those without.

We calculated the cumulative incidence of disability during follow-up according to baseline frailty status (frail, prefrail and robust) and corresponding to each frailty component (slowness, weakness, exhaustion, low activity and weight loss) with Kaplan-Meier curves. Intergroup differences were estimated by the log-rank test.

Cox proportional hazards regression models were used to analyse the associations between frailty phenotype and disability risk. The first model (model 1) was adjusted for age and sex. We then used a multiple adjustment model adjusted for age, sex, BMI, MMSE, number of prescribed medications, hypertension, heart disease, diabetes mellitus, osteoporosis and GDS (model 2). These covariates were included as categorical (age, sex and diagnoses) and continuous variables (BMI, MMSE, number of prescribed medications and GDS). We estimated adjusted HRs for incidence of disability and their 95% CIs.

Stratified analyses were performed to examine the relationship between frailty and disability risk in different subgroups defined by sex, age (74/75 years old), cognitive function (MMSE score 23/24) and depressive symptoms (GDS score 5/6).12 Adjusted HRs for incidence of disability and their 95% CIs were also estimated in the stratified analyses.

All analyses were conducted using IBM SPSS Statistics 19.0 (IBM Japan Tokyo). The level of statistical significance was set at p<0.05.

Clinical and policy implications

Many different ways to assess physical frailty were reported in previous studies from around the world,5 with the majority of cohort studies conducted in Western countries.19 Thus, it might be inappropriate to extend the results of these studies to Asian countries. Indeed, the European Working Group on Sarcopenia in Older People20 and the Asian Working Group for Sarcopenia (AWGS)17 have different diagnostic cut-offs for the frailty phenotype. Thus, assessing frailty phenotype in an Asian population would develop a more comprehensive definition of the concept and lead to better-designed studies on its effect on the risk of disability among community-dwelling older adults in Asian countries. In this prospective cohort study of community-dwelling older adults, individuals with frail or prefrail phenotype at baseline had an increased risk of disability incidence during the 2 years after baseline assessment. These results support findings from previous cohort studies with large samples.2 ,3 Regarding the components of frailty, slowness, weakness and weight loss were more strongly associated with incident disability than the other components. The associations between frailty and the incidence of disability remained across various strata defined by sex, age, cognitive function and depressive symptoms. Specifically, participants with both a frail phenotype and lower cognitive function (MMSE scores <24) had the highest disability incidence rate (30.3%) during the 2 years after baseline assessment (figure 3). Thus, physical frailty and lower cognitive function could have additive effects on the risk for disability incidence.

The results of this prospective study showed that participants with the slowness component (defined as having a walking speed slower than 1.0 m/s) had more than a twofold higher risk of disability. However, there is no consensus regarding the cut-off point for walking speed as an indicator of slowness.21–23 Although additional studies are necessary to determine the optimal cut-off values, slowness defined as a walking speed slower than 1.0 m/s could be useful as a component of frailty for predicting disability and preventing functional decline among community-dwelling older adults who are relatively well functioning. In this study, weakness was also determined using modified cut-off values of handgrip strength for Asian populations suggested in a consensus report from AWGS. The AWGS recommends using <26 kg for men and <18 kg for women as the cut-off values for handgrip strength among community-dwelling older adults in Asia.17 Our findings indicated that low handgrip strength, as suggested by AWGS, was independently associated with the incidence of disability after adjustment for potential covariates; thus, these modified cut-off values would be appropriate for diagnosing frailty in Asian populations.

Two components of frailty, exhaustion and weight loss, assessed using items in the Kihon-Checklist, identified prevalence rates similar to those reported in a previous cohort study sampling more than 5000 community-dwelling older adults.2 Although physical activity was assessed by two simple questions according to participation in sports or physical exercises in this study, we should recognise that many other kinds of activities such as domestic tasks and gardening could play important roles with regard to physical activity in older people. We thus need to consider these points, despite the fact that the total prevalence of frailty in our study was very similar to that reported in the Cardiovascular Health Study.2 There seems to be a general consensus on the essential components of physical frailty phenotype, and the present study also indicated these impacts on incident disability in the Japanese older samples. Frailty is due to an accumulation of deficits in areas including physical and cognitive impairment and psychosocial risk factors.24

Although frailty has generally been regarded as an important concept and several multidimensional instruments have been developed to measure frailty in its totality, there is still considerable variety in how the concept is defined and measured. Specifically, there appear to be two major approaches to defining and measuring frailty: namely, regarding it as either a multifactorial construct (comprising social, psychological and physical aspects) or a mainly physical one. Thus far, the social and psychological dimensions of frailty have not been sufficiently verified.25 For instance, there is insufficient evidence regarding the operational definition of cognitive frailty and the validity of its measurements.26 A recent conference defined cognitive frailty as a clinical entity characterised by cognitive impairment related to physical causes with potential reversibility,27 making it a useful target for the secondary prevention of cognitive problems in older people.27 Indeed, considering physical frailty and cognitive impairment as a single complex phenotype may be central to the prevention of dementia and its subtypes, although this should be confirmed with secondary preventive trials on cognitively frail older individuals.28 In addition, according to the integral conceptual model of frailty, where frailty is affected by physical, psychological and social factors, life-course determinants such as sociodemographic characteristics, lifestyle, life events and environment-related factors can directly influence frailty as well as the onset of diseases that lead to frailty.29 Therefore, further studies on frailty that focus on physical as well as cognitive and psychosocial domains will be needed.

Another contentious point in defining and measuring frailty is whether to include performance-based measurements, especially for the identification of physical frailty.5 Frailty questionnaires appear to be a highly feasible method for obtaining data from large samples and for assessing participants in a busy clinical practice setting; however, slowness and weakness as assessed by performance-based methods were more strongly associated with incident disability in our study. Our findings indicate that combining questionnaires and performance-based assessments could be an effective method to identify older adults with frailty phenotype as a way to predict risk for disability incidence. Furthermore, assessments of walking speed and handgrip strength are very simple and easy to implement in community settings, and are good predictors for health outcomes.30 A notable point of our findings is that older adults with physical frailty and lower cognitive function (MMSE scores <24) concurrently represented the highest percentage, more than 30%, of incident disability in stratified analyses. Previous studies have indicated the association between physical frailty and cognitive impairment among non-demented community-dwelling older adults31–33 and shown that cognitive decline leads to higher risks of poor health.34 Our findings also suggest that physical frailty and lower cognitive function have additive effects on disability incidence.

Strengths and limitations

A major strength of this study is the application of a monthly follow-up of disability using a mandatory social LTCI in Japan. Since most frailty models were developed in white populations, different cut-offs for frailty should be considered when examining different populations.4 Although few prospective cohort studies regarding frailty phenotype and disability have been reported in Asia, this study included a large-scale prospective sample of community-dwelling Japanese older adults and the application of a comprehensive measure of physical frailty including questionnaires as well as physical performance measurements.

Nevertheless, several limitations should be considered. In the multivariate analyses, although some diagnoses, such as hypertension, heart disease, diabetes mellitus and osteoporosis, were included, several potential clinical confounders, such as haematological diseases including anaemia, oncological diseases and eye diseases causing severe visual impairment, were not included. In addition, these clinical conditions were based on self-report. We should therefore consider these issues carefully in interpreting the results. This study involved community-dwelling older people who were relatively well functioning and able to participate in the assessments at the community centre on their own. Therefore, this is likely to lead to an underestimation of the actual incidence of disability. In addition, our follow-up period was shorter than that in previous studies.2 ,3 ,35 ,36 Another limitation is that the causes of the incidence of disability were not determined. The major causes of incident disability certification by the LTCI include poststroke, dementia and severe stage of frailty. Moreover, anybody aged 65 and older (and anyone aged 40 to 64 with an ageing-related disability) is eligible for LTCI.37 Thus, future studies examining causes of disability incidence and the longitudinal relationships between frailty and disability using longer follow-up data would be helpful for the development of preventive strategies for disability.