Frailty and cardiovascular outcomes in the National Health and Aging Trends Study.

AIMS: Physical frailty is a commonly encountered geriatric syndrome among older adults without coronary heart disease (CHD). The impact of frailty on the incidence of long-term cardiovascular outcomes is not known.We aimed to evaluate the long-term association of frailty, measured by the Fried frailty phenotype, with all-cause-mortality and MACE among older adults without a history of CHD at baseline in the National Health and Aging Trends Study. METHODS AND
RESULTS: We used the National Health and Aging Trends Study, a prospective cohort study linked to a Medicare sample. Participants with a prior history of CHD were excluded. Frailty was measured during the baseline visit using the Fried physical frailty phenotype. Cardiovascular outcomes were assessed during a 6-year follow-up.Of the 4656 study participants, 3259 (70%) had no history of CHD 1 year prior to their baseline visit. Compared to those without frailty, subjects with frailty were older (mean age 82.1 vs. 75.1 years, P < 0.001), more likely to be female (68.3% vs. 54.9%, P < 0.001), and belong to an ethnic minority. The prevalence of hypertension, falls, disability, anxiety/depression, and multimorbidity was much higher in the frail and pre-frail than the non-frail participants. In a Cox time-to-event multivariable model and during 6-year follow-up, the incidences of death and of each individual cardiovascular outcomes were all significantly higher in the frail than in the non-frail patients including major adverse cardiovascular event (MACE) [hazard ratio (HR) 1.77, 95% confidence interval (CI) 1.53, 2.06], death (HR 2.70, 95% CI 2.16, 3.38), acute myocardial infarction (HR 1.95, 95% CI 1.31, 2.90), stroke (HR 1.71, 95% CI 1.34, 2.17), peripheral vascular disease (HR 1.80, 95% CI 1.44, 2.27), and coronary artery disease (HR 1.35, 95% CI 1.11, 1.65).
CONCLUSION: In patients without CHD, frailty is a risk factor for the development of MACEs. Efforts to identify frailty in patients without CHD and interventions to limit or reverse frailty status are needed and, if successful, may limit subsequent adverse cardiovascular events.

See page 3866 for the editorial comment on this article (

Introduction

One challenge in the clinical management of the rapidly expanding older adult population in the USA is the increased prevalence of frailty, an important geriatric syndrome. This is particularly relevant in the cardiovascular sphere as the prevalence and incidence of cardiovascular disease are also markedly increased with age. The physical frailty phenotype is a clinical state in which there is increased vulnerability to stressors due to diminished reserves across multiple physiological systems, resulting in functional decline, increased mortality, and a higher likelihood of complications from disease and from therapeutic interventions.,

In prior studies that examined the influence of frailty on cardiovascular outcomes, the assessment of frailty was performed in study populations at high cardiovascular risk, including those with acute coronary syndromes, peripheral vascular disease, and valvular heart disease. For example, Farooqi et al. have shown that frailty can provide an incremental prognostic value in addition to traditional cardiovascular risk assessment, but this meta-analysis included a mix of studies with and without cardiovascular disease. However, among patients without known coronary heart disease (CHD), the long-term association between frailty and major adverse cardiovascular events (MACE) remains largely unknown. In this study, we aimed to evaluate the long-term association of frailty, measured by the Fried frailty phenotype, with all-cause-mortality and MACE among older adults without a history of CHD using the National Health and Aging Trends Study (NHATS).

Methods

The source and study population

We examined the 2011 NHATS baseline cohort. NHATS is a prospective cohort study funded by the National Institute on Aging (U01AG032947) that studies functioning in later life. The source population for this study is derived from a sample of Medicare beneficiaries aged 65 years and older, a nationally representative cohort of older patients in the community. These older adults were interviewed in 2011 during their baseline visit and annual re-interviews were performed for each participant to document changes, trends, and dynamics in later life functioning. Detailed information on geriatric risks, including frailty, physical and cognitive capacity, activities of daily living (ADL), and the social, physical, and technological environments were collected. African Americans and patients from older ages were oversampled from the Medicare enrolment file. For each participant, the NHATS repository is linked to Medicare data that were available prior to the 2011 baseline visit.

The study population included adults ≥65 years of age enrolled during the 2011 NHATS baseline visit who also had linked Medicare data available for analysis prior to their baseline visit. For each participant, CHD was identified 12 months prior to the 2011 NHATS baseline visit using International Classification of Diseases-9th Revision 410–414, 410.0–410.9, 410.00–410.02, 410.10–410.12, 410,20–410,22, 410.30–410.32, 410.40–410.42, 410.50–410.52, 410.60–410.62, 410.70–410.72, 410.80–410.82, 410.90–41.92, and 4292.

Frailty assessment

Frailty in each older patient in the NHATS-CMS study was assessed using the five domains of the Fried physical frailty phenotype: exhaustion, low physical activity, weakness, slowness, and shrinking (i.e. unintentional weight loss). If three or more, out of the five criteria, were present, the individual was categorized as frail and those with one or two of the five were categorized as ‘pre-frail’. Out of the total study population (n = 3259), 16% (n = 527) were categorized as ‘frail’ and 47% (n = 1535) were categorized as ‘non-frail’. Detailed definitions of meeting each criterion were previously published. For missing frailty data, a multiple imputation methods was adapted, and it was similar to previously published work that used the imputed frailty dataset out of 10 replicas. The estimates from running separate models on the 10 replicates were pooled together to obtain the final estimates. The pooling of the estimates was performed in such a way that appropriately accounted for the uncertainty in the missing frailty data imputation (see Statistical analysis section).

Cardiovascular outcomes

A MACE was defined as death from any cause, acute myocardial infarction, any subsequent CHD, stroke, or peripheral vascular disease, whichever came first. To address competing risks of death, MACE2 was defined as acute myocardial infarction, any subsequent CHD, stroke, or peripheral vascular disease, whichever came first, excluding all-cause mortality. Secondary cardiovascular endpoints included each of these individual components identified in the Center of Medicare and Medicaid Services database during the 6-year follow-up. Furthermore, any primary hospital admission for subsequent CHD was identified and reported separately.

Geriatric risks

For each patient, specific geriatric risks were assessed during the NHATS follow-up visits. These included measures of functioning [ADL and instrumental ADL (IADL), and functional limitations], cognitive function (any form of cognitive impairment, dementia/Alzheimer’s disease), disability, and mobility disability. For each older participant, the Katz scale was performed to assess independence in (i) self-care (ADL: bathing, dressing, eating, toileting); (ii) household activities (IADL: doing laundry, preparing meals, shopping for groceries and for personal items, medication management, handling bills and banking); and (iii) mobility (getting around inside, going outside, getting out of bed). Screening for cognitive dysfunction was performed to assess functions related to memory, orientation, and executive function. For patients with severe cognitive impairment, a proxy interview was conducted, and the proxy was asked about the function of the participant. Dementia status was ascertained using the following instruments: (i) a physician report indicating that the participant has dementia or Alzheimer’s disease; (ii) a scoring indicating a probable dementia administered to proxies; and (iii) results from cognitive tests that evaluate memory, orientation, and executive function. Disability was measured using the American Community Survey Disability Questions. Outcomes related to mobility, self-care, and household activities were performed independently for each participant during follow-up visits. Loss of independence was defined as patients reporting never or rarely going outside or the use of devices to go outside.

Demographic characteristics, medical conditions, and healthcare utilization

Each older adult enrolled in the study was asked whether their physician had ever told them they had any of the following medical conditions: high blood pressure, diabetes mellitus, stroke, any cardiac disorder, arthritis, lung or bone disease, and cognitive impairment or dementia. Hospitalization within the past 12 months and baseline assessment on self-care, mobility, and household activities were collected.

Statistical analysis

Participants with a history of CHD and stroke were excluded. During the 2011 baseline NHATS visit, participants were categorized into three distinct groups: no frailty, pre-frailty, and frailty as assessed by the Fried physical frailty phenotype. Demographics, smoking status, comorbidity, hospitalizations, emergency department visits, falls, self-care, mobility, household activities, depression, anxiety, and cognitive impairment at baseline were reported for the frail and the non-frail. Frequencies and percentages were calculated for categorical variables and mean ± standard deviation for continuous variables. Data on self-care, mobility, and household activities are presented as cumulative proportions at 6 years for the frail vs. the non-frail group (Table ).

Proportional hazard models were used to assess the association between frailty and cardiovascular outcomes among older adults at 6-year follow-up. Patients were censored if they developed the cardiovascular outcomes of interest or if they were lost to follow-up. To address confounding by age, demographics, and other risk factors, we performed three additional multivariable Cox models. Model 2 adjusted for age and sex; Model 3 adjusted for age, sex, race/ethnicity, body mass index (BMI), and smoking status; and Model 4 adjusted for age, sex, race/ethnicity, BMI, smoking status, diabetes, hypertension, number of comorbid diseases, and dependency status (as a surrogate measure for composite functional status). To explore sensitivity of findings to dementia status, we performed a sensitivity analysis by excluding those with probable or definitive dementia (Supplementary material online, ). The assumption for Cox proportional hazard models was checked by plotting the Schoenfeld residuals against survival time for each primary and secondary cardiovascular outcome by frailty group. As sensitivity analysis, we fitted stratified Cox models that allowed the form of the underlying baseline hazard function to vary across age categories and between sexes (i.e. violation of the proportion hazard assumption) (see Supplementary material online, and ). Kaplan–Meier curves were constructed to evaluate the association of frailty status at baseline with MACE and each individual cardiovascular outcome. Log-rank statistic was calculated for each curve. To test for interaction between frailty, as categorical variable, with each individual cardiovascular risk factors, likelihood ratio tests were performed to compare models with and without the interaction term. We have tested interactions of frailty with cardiovascular disease risk factors (i.e. BMI, smoking, diabetes, and hypertension) in Model 4 with each individual MACE outcome (Supplementary material online, ). To facilitate interpretation, the hazard ratios of frailty and pre-frailty were presented separately by the level of each significant modifier in the supplementary material. For missing data on frailty, we adopted a two-step approach. First, if a test (grip or walking test) was not done because of health/safety concerns, a value of zero was assigned to indicate worst performance. Second, for remaining missing values, we employed multiple imputation (10 replicates) using chained equations (see details in Bandeen-Roche et al.). A separate model was fitted using each imputed dataset, and the parameter estimates (i.e. regression coefficients and standard errors) obtained from each model were then combined into one set of inferential statistics via the STATA ‘mi estimate’ command that accounted for the uncertainty in the imputed values.

All tests are two-sided, and the statistically significant level is set at P < 0.05. Data analyses were conducted using SAS (v.9.4; SAS Institute Inc, Cary, NC, USA) and STATA version 15 MP (Stata Corp., College Station, TX, USA). The Johns Hopkins Medicine Institutional Review Board approved this study.

Results

Of the 4656 patients enrolled in the 2011 NHATS baseline visit, the mean age was 75 years and 60% of the study population was ≥75 years of age. Female participants constituted 61% of the cohort and the majority enrolled was non-Hispanic Whites. On average, the majority was overweight, and more than half of the cohort smoked at least one cigarette per day. The majority of this older population had multiple chronic conditions and 15% of the cohort had four or more chronic comorbidities. The most prevalent medical conditions were hypertension, arthritis, and osteoporosis. Approximately 21% of the study population was living with diabetes mellitus, and 6.1% had dementia at baseline.

Of the 3259 patients who had no history of CHD or stroke prior to their baseline NHATS visits, 1535 (47%) patients were pre-frail and 527 (16%) patients had physical frailty according to the Fried frailty phenotype. Of the total study population, 478 (15%) had missing frailty data at baseline and these estimates were imputed (see Methods section). Patients who were frail were older, more likely to be women and belong to an ethnic minority as compared to non-frail patients. Frail patients had higher prevalence of hypertension, diabetes mellitus, history of prior cardiovascular risk factors, dementia, lung disease, and arthritis than non-frail patients. The overall number of chronic comorbid conditions was also higher among frail patients with approximately one in three patients reported having four or more chronic medical conditions (Table ). Frail patients were more likely to be admitted to the hospital and had more emergency department visits in 12 months prior to their baseline NHATS visits, than did non-frail patients. When evaluating measures of disability at baseline, including self-care, mobility disability, and household activities disability, patients with frailty were more likely to report significant impairment, than did non-frail patients. The overall disability level (i.e. having difficulties requiring help) among the frail group was as high as 76.8%, but only 10.0% reported having difficulties requiring help in the non-frail group. Frail patients also had high cognitive impairment at baseline and ∼40% had probable dementia at baseline (Table ).

The age-adjusted incidence of cardiovascular outcomes at the 6-year follow-up is presented in Table . Frail patients developed more cardiovascular outcomes than did the pre-frail and non-frail groups over the 6-year follow-up, including a MACE, death, acute myocardial infarction, stroke, peripheral vascular disease, or any coronary artery disease (Figure ). In an unadjusted Cox proportional hazards model, frailty and pre-frailty were associated with MACE and with each individual component of cardiovascular outcomes: all-cause death, acute myocardial infarction, peripheral vascular disease, and any subsequent coronary artery disease, as compared to non-frail patients. After adjusting for age, sex, race/ethnicity, census division, residence and income, BMI, traditional cardiovascular risk factors, dependency, and the number of concomitant chronic medical conditions, frailty remains highly associated with MACE, death, and peripheral vascular disease at the 6-year follow-up in the NHATS study (Table ). In a sensitivity analysis excluding those patients with definitive or probable dementia (n = 2832), both frailty and pre-frailty were associated with MACE and with each individual cardiovascular outcome during follow-up when compared to the non-frail group (Supplementary material online, ). In a stratified Cox model that allowed the form of the underlying baseline hazard function to vary across age categories and between sexes, the results largely remained the same (Supplementary material online, and ). Modified association of frail by smoking status on all-cause mortality, acute myocardial infarction, and coronary artery disease as outcomes are presented in Supplementary material online, and that of frailty on peripheral vascular disease by hypertension is presented in Supplementary material online, .

(A) Kaplan–Meier survival curve illustrating major adverse cardiovascular event (MACE)-free over 6-year follow-up by frailty status at baseline in the NHATS-CMS study among patients without a history of coronary heart disease (log-rank P < 0.001). MACE was defined as a composite of all-cause mortality, acute myocardial infarction, stroke, peripheral vascular disease, and subsequent coronary disease. (B) Kaplan–Meier survival curve illustrating MACE2-free over 6-year follow-up by frailty status at baseline in the NHATS-CMS study among patients without a history of coronary heart disease (log-rank P < 0.001). MACE2 was defined as a composite of acute myocardial infarction, stroke, peripheral vascular disease, and subsequent coronary disease. (C) Kaplan–Meier survival curve illustrating the survival over 6-year follow-up by frailty status at baseline in the NHATS-CMS study among patients without a history of coronary heart disease (log-rank P < 0.001). (D) Kaplan–Meier survival curve illustrating acute myocardial infarction-free survival over 6-year follow-up by frailty status at baseline in the NHATS-CMS study among patients without a history of coronary heart disease (log-rank P = 0.003). (E) Kaplan–Meier survival curve illustrating stroke-free survival over 6-year follow-up by frailty status at baseline in the NHATS-CMS study among patients without a history of coronary heart disease (log-rank P < 0.001). (F) Kaplan–Meier survival curve illustrating peripheral vascular disease-free survival over 6-year follow-up by frailty status at baseline in the NHATS-CMS study among patients without a history of coronary heart disease (log-rank P < 0.001).

Discussion

We examined the association of physical frailty phenotype with cardiovascular outcomes among older adults in the NHATS without prior CHD during the 6-year follow-up. The major findings of this study are as follows: (i) participants without CHD at baseline who exhibit pre-frailty or physical frailty, as measured by the Fried frailty phenotype, had a high prevalence of multiple chronic conditions, baseline disability, mobility disability, and cognitive dysfunction as compared to non-frail CHD participants; (ii) participants with baseline pre-frailty and physical frailty also had higher rates of healthcare utilization with more emergency department visits, admissions to the inpatient service, and longer hospital lengths of stay; and (iii) as compared to non-frail subjects, pre-frail and frail older patients had a higher risk of developing MACE, including mortality during the 6-year follow-up, even after adjusting for demographic characteristics, traditional cardiovascular risk factors, and multimorbidity at baseline ().

In this cohort of older adults free of CHD at baseline, we estimated that the prevalence of frailty is ∼16%, which is significantly lower than the prevalence of frailty among patients with preexisting cardiovascular disease., Consistent with our estimates, pooled analysis from 46 studies that enrolled participants with frailty showed that 1 in 6 community-dwelling older adults lives with frailty. While the older patients in our study were free of known cardiac disease at baseline, many older adults with frailty frequently have coexisting cardiovascular risk factors. A bidirectional association between frailty and multimorbidity exists, in which the coexistence of these two geriatric syndromes will lead to the progressive worsening of both. In a systematic review and meta-analysis of 48 observational studies, 70% of frail older adults examined also had multimorbidity with two or more coexisting conditions. In our study, 60% of older patients reported multimorbidity, defined as two or more coexisting chronic medical conditions. Hypertension was the most commonly encountered cardiovascular risk factor and 1 in 5 patients had diabetes mellitus. The burden of these cardiovascular risk factors is clearly higher in the frail, than the non-frailty cohort. Vetrano et al. reported the important observation that the vast majority of older adults with frailty are also multimorbid, but very few older adults with multimorbidity are also frail. The authors hypothesize that multimorbidity plays an important deterministic role in the development of frailty syndrome. Frailty and multimorbidity, including hypertension, diabetes mellitus, and other traditional cardiovascular risk factors, can potentially share common pathophysiology mechanisms that put older adults at risk for the development of cardiovascular disease including, inflammation, coagulopathy, and metabolic dysregulation.

In a retrospective cohort study, middle aged participants from the Civil Service departments in London were examined based on their cardiovascular risk at baseline. Those with four different cardiovascular disease risk scores (Framingham cardiovascular disease, Framingham CHD, Framingham stroke, and Systematic Coronary Risk Evaluation) were associated with an elevated risk of frailty, measured using the physical frailty phenotype. Data from the British Regional Heart Study also showed that older adults frailty in older age was associated with a number of cardiovascular risk factors. Taken together with the results of our study, this highlights the bidirectional association between frailty and cardiovascular disease mentioned previously. In a cross-sectional study, Fernandes et al. investigated the association between frailty, measured by the physical frailty phenotype, and cardiovascular risk measured by the Framingham risk score. The investigators found that frailty and pre-frailty were associated with increased cardiovascular risk. Frailty and cardiovascular disease risk were measured at the same time. Veronese et al. evaluated the prognostic value of a multidimensional prognostic index, an instrument grounded in comprehensive geriatric assessment, and self-reported cardiovascular outcomes over 8 years of follow-up. The multidimensional index predicted the onset of cardiovascular disease in community dwellers affected by, or at risk for, osteoarthritis. Our study complements these findings by measuring frailty and pre-frailty, using the Fried physical frailty phenotype, and incidences of the outcomes were ascertained in the CMS database during 6 years of follow-up. The cumulative knowledge continues to highlight the importance of frailty as a risk factor for cardiovascular disease and trigger the need for integration of frailty assessment in the cardiovascular profile of older adults., Similar to our findings, Marinus et al. reported a higher prevalence of frailty in older female than male participants. This propensity to frailty in older female patients may have differential power of prediction when compared to older patients at risk for cardiovascular disease. Newman and colleagues examined participants enrolled in the Cardiovascular Health Study and reported that the physical frailty phenotype at baseline was strongly associated with imaging markers of subclinical atherosclerosis including carotid stenosis, impaired ankle-brachial index, and other electro- and echocardiographic variables. Progression of these subclinical atherosclerotic cardiovascular conditions to overt clinical events is likely driven by the pathophysiologic mechanisms present in frail older adults including higher oxidative stress; elevated circulating inflammatory biomarkers including C-reactive protein, and neutrophils, white cell counts, and interleukin-6, and measures of coagulopathy, including D-dimer and fibrinogen. Prior research has shown that even among patients who meet only one or two of the Fried criteria, also referred to as ‘pre-frail’, there is a higher risk of developing cardiovascular disease after adjustment for traditional risk factors, inflammatory markers, and glycated haemoglobin during a follow-up period of 4.4 years. In the NHATS study, we adjusted for baseline demographic variables, BMI, multimorbidity, and other traditional risk factors for cardiovascular disease. Over 6 years of follow-up in the NHATS study, we found that older patients with pre-frailty and frailty exhibit higher incidences of all-cause mortality and MACE, as a composite mainly driven by death, stroke, and peripheral vascular disease. In a meta-analysis that included cross-sectional and prospective cohort studies, Veronese and colleagues reported that frailty was associated with an approximate three-fold increased risk of cardiovascular disease when compared to robust patients. To complement these findings, our study population enables the evaluation of frailty in older adults as predictor of MACE in patients with or without previous cardiovascular disease. Efforts to study geriatric syndromes during acute cardiovascular illnesses are well recognized by the cardiovascular community at large,, but the ability to integrate the assessment of frailty in the care for older patients at risk for cardiovascular disease is limited because of the lack of efficacious therapies to prevent or reverse the development of physical frailty. However, several initiatives are underway to test the influence of physical activity programs, nutritional interventions, cognitive training, and a combination of these to prevent or reverse frailty in older adults. Because cardiovascular disease remains the most common cause of mortality in older adults, efforts to establish the efficacy and safety of such interventions in cardiovascular practice, similar to other therapies targeting traditional cardiovascular risk factors, are needed. The 2020 ESC Guidelines for the management of acute coronary syndromes in patients presenting without persistent ST-segment elevation acknowledged the importance of integrating frailty in the management and balancing the risks of each individual treatment (i.e. medical therapy plus invasive strategy) against the risk of harm from not offering therapy given perceived risk of complications. Our work underscores the urgent need to establish robust clinical trial data to inform management for older patients with frailty at risk for cardiovascular disease.

This study has potential limitations. First, MACEs were diagnosed using data obtained from the Medicare claims database from hospital and outpatient encounters after the 2011 baseline NHATS visit. While this method of studying cardiovascular outcomes is widely used in health services research, the severity and degree of acute myocardial infarction, stroke, vascular disease, and cardiac-specific mortality could not be ascertained. Despite this limitation, this large study is novel because it is the first to evaluate the temporal relationship between frailty and incident cardiovascular disease during 6 years of follow-up among patients without a history of CHD in the Medicare database. Second, it is plausible that patients with frailty have undiagnosed cardiovascular disease, which in turn led to a higher incidence of cardiovascular events during follow-up. Third, physical frailty was measured at the baseline visit as a binary variable with impairment of three or more domains in the Fried criteria. However, frailty can be a reversible and dynamic physiologic process and may change over time. When evaluating the association between physical frailty and cardiovascular outcomes, significant confounding exists, and caution is needed when interpreting the unadjusted estimates presented in Table . To provide a comprehensive assessment of the influence of frailty on cardiovascular outcomes, a multivariable Cox regression model, stratified by Cox modelling techniques conditioning on age, and other sensitivity analyses were provided to mitigate the influence of confounding by indication.

Conclusion

In the NHATS study, we found that pre-frailty and physical frailty phenotype are associated with a significant risk for mortality and the development of MACE during 6 years of follow-up, even after controlling for traditional cardiovascular risk factors. Efforts to integrate frailty assessment as part of primary cardiovascular prevention programs in older adults at risk for cardiovascular disease are essential in daily clinical cardiovascular practice. Testing the efficacy and safety of physical activity programs, nutritional interventions, and cognitive training to prevent or reverse physical frailty in patients at risk for the development of cardiovascular disease is needed as the US older adult population expands rapidly in the coming decades.

Supplementary material

Supplementary material is available at European Heart Journal online.

Click here for additional data file.

Table 1

Characteristics of the study population of patients without a history of coronary heart disease enrolled in the National Health and Aging Trends Study by physical frailty phenotype

Characteristics	Total (n = 3259)	No frailty (n = 1197)	Pre-frailty (n = 1535)	Frailtya (n = 527)	P-value
Age, years, mean	77.6	75.1	78.0	82.1	<0.001
Age, years, %
65–69	18.8	25.7	16.4	9.0
70–74	20.7	26.6	19.3	11.6
75–79	20.0	21.1	21.0	14.6	<0.001
80–84	19.6	16.3	22.2	19.8
85–89	12.6	6.7	13.4	23.9
≥90	8.2	3.5	7.8	20.2
Sex, %
Female	60.7	54.9	62.6	68.3	<0.001
Male	39.3	45.1	37.4	31.7
Race, %
Non-Hispanic white	72.0	76.9	72.3	60.3
Non-Hispanic black	21.2	17.4	20.9	31.2	<0.001
Hispanic	4.2	3.3	4.0	6.5
Others	2.6	2.4	2.8	2.0
BMI, kg/m2, mean	27.1	26.9	27.5	26.5	0.001
Smoking status, %  Smoke at least 1 cigarette/day	49.2	50.9	49.7	44.2	0.044
Comorbidities, %
Arthritis	54.6	41.4	57.7	75.5	<0.001
Diabetes mellitus	21.2	16.5	21.5	30.8	<0.001
Hypertension	63.8	56.6	66.8	71.3	<0.001
Lung disease	13.7	8.1	15.5	20.9	<0.001
Osteoporosis	21.3	15.9	21.8	32.0	<0.001
Dementia	6.1	1.0	5.0	20.9	<0.001
No. chronic diseases, %
0–1	35.4	52.3	30.7	10.6
2–3	49.4	41.8	54.5	51.7	<0.001
≥4	15.2	5.9	14.8	37.7
Cancer, %	26.6	24.4	28.4	26.6	0.071
Hospital stay past 12 months, %	16.7	8.9	17.2	32.8	<0.001
Any fall past month, %	30.1	18.6	32.2	50.2	<0.001
Disability, %
No difficulty	74.6	94.0	73.7	33.2
Difficulty but no help	11.9	4.5	15.0	20.1	<0.001
Help	13.4	1.5	11.3	46.7
Mobility disability, %
No difficulty	67.6	90.2	65.9	21.2
Difficulty but no help	17.5	8.8	22.2	23.5	<0.001
Help	14.9	1.0	11.9	55.3
Household activities disability, %
No difficulty	62.2	86.4	58.9	16.7
Difficulty but no help	12.3	8.0	16.1	11.1	<0.001
Help	25.5	5.6	25.0	72.2
Overall disability level, %
No difficulty	51.0	77.7	44.4	9.4
Difficulty but no help	20.3	15.4	26.4	13.6	<0.001
Help	28.7	6.8	29.2	77.0
Depression, % (PHQ2 score ≥3)	13.8	4.8	14.1	33.8	<0.001
Anxiety, %
GAD2 score ≥3	11.3	4.5	11.7	26.1	<0.001
No. ED visits, %
0	76.0	84.1	75.3	59.8	<0.001
1	15.9	12.7	17.0	19.7
≥2	8.1	3.2	7.7	20.6
No. hospitalizations, %					<0.001
0	88.5	94.4	88.4	75.4
1	8.9	4.8	9.5	16.4
≥2	2.6	0.8	2.1	8.2
Total LOS in hospital, days, mean	1.00	0.31	0.93	2.74	<0.001
No. physician visits, mean	7.24	5.99	7.65	8.89	<0.001
No. ADL impairment, %					<0.001
0	63.4	88.3	59.4	18.4
1–2	21.8	10.5	28.9	27.1
≥3	14.8	1.3	11.7	54.5
No. IADL impairments, %
0	63.5	87.0	61.0	17.6
1–2	20.7	11.6	25.7	26.4	<0.001
≥3	15.8	1.4	13.3	56.1
Cognitive impairment, %	8.6	2.6	8.0	26.8	<0.001
AD8 dementia, %	5.8	0.6	3.6	23.8	<0.001
Dementia (probable), %	13.1	3.5	11.3	40.0	<0.001

AD8, AD8 dementia screening interview; ADL, activities of daily living; BMI, body mass index; ED, emergency department; GAD2, generalized anxiety disorder 2-item; IADL, instrumental activities of daily living; LOS, length of stay; PHQ2, patient health questionnaire-2.

Frailty was assessed by the physical frailty phenotype paradigm that is grounded in five criteria: exhaustion, low physical activity, weakness, slowness, and shrinking (www.nhats.org).

Table 2

The age-adjusted incidence of major adverse cardiovascular events by physical frailty phenotype among older adults without history of coronary heart disease outcomes in the National Health and Aging Trends Study during the 6-year follow-up

Outcome	Total (n = 3259)	No frailty (n = 1197)	Pre-frailty (n = 1535)	Frailty a (n = 527)
MACE1, %	64.7	49.6	68.6	87.5
MACE2, %	56.3	45.0	60.1	70.8
Death, %	28.6	14.3	29.5	58.6
AMI, %	9.0	6.4	9.7	12.8
Stroke, %	24.2	18.9	25.3	32.7
PVD, %	27.7	18.0	30.6	41.5
CAD, %	37.0	30.4	39.3	45.2

MACE1: a composite of acute myocardial infarction, stroke, peripheral vascular disease, coronary artery disease, and all-cause mortality; MACE2: a composite of acute myocardial infarction, stroke, peripheral vascular disease, coronary artery disease.

AMI, acute myocardial infarction; CAD, coronary artery disease; MACE, major adverse cardiovascular event; PVD, peripheral vascular disease.

Frailty and pre-frailty were assessed by the physical frailty phenotype paradigm that is grounded in five criteria: exhaustion, low physical activity, weakness, slowness, and shrinking (www.nhats.org).

Table 3

Proportional hazards regression model evaluating the influence of physical frailty status on 6-year cardiovascular outcomes among older adults without a history of coronary heart disease in the National Health and Aging Trends Study

	MACE 1	MACE 2	Death	AMI	Stroke	PVD	CAD
	HR (95% CI)	HR (95% CI)	HR (95% CI)	HR (95% CI)	HR (95% CI)	HR (95% CI)	HR (95% CI)
Model 1a
Pre-frailty	1.48 (1.33, 1.64)	1.42 (1.28, 1.59)	1.79 (1.49, 2.15)	1.51 (1.14, 2.01)	1.32 (1.12, 1.57)	1.66 (1.41, 1.97)	1.33 (1.16, 1.53)
Frailty	2.34 (2.06, 2.67)	2.09 (1.81, 2.40)	3.70 (3.03, 4.50)	2.35 (1.65, 3.33)	2.00 (1.62, 2.46)	2.53 (2.07, 3.09)	1.80 (1.51, 2.15)
Model 2b
Pre-frailty	1.47 (1.32, 1.63)	1.42 (1.27, 1.58)	1.75 (1.46, 2.11)	1.46 (1.10, 1.95)	1.36 (1.15, 1.61)	1.62 (1.37, 1.91)	1.32 (1.16, 1.52)
Frailty	2.28 (1.99, 2.61)	2.05 (1.77, 2.37)	3.62 (2.95, 4.44)	2.35 (1.63, 3.38)	2.11 (1.69, 2.62)	2.32 (1.88, 2.85)	1.75 (1.46, 2.10)
Model 3c
Pre-frailty	1.43 (1.28, 1.58)	1.36 (1.22, 1.52)	1.78 (1.48, 2.14)	1.42 (1.06, 1.90)	1.32 (1.11, 1.56)	1.56 (1.31, 1.85)	1.27 (1.11, 1.46)
Frailty	2.15 (1.88, 2.47)	1.91 (1.65, 2.21)	3.61 (2.93, 4.43)	2.20 (1.52, 3.18)	1.98 (1.58, 2.47)	2.12 (1.72, 2.62)	1.65 (1.37, 1.98)
Model 4d
Pre-frailty	1.34 (1.21, 1.49)	1.29 (1.15, 1.44)	1.64 (1.36, 1.98)	1.36 (1.01, 1.82)	1.25 (1.05, 1.49)	1.49 (1.25, 1.77)	1.17 (1.02, 1.35)
Frailty	1.77 (1.53, 2.06)	1.59 (1.35, 1.87)	2.70 (2.16, 3.38)	1.95 (1.31, 2.90)	1.71 (1.34, 2.17)	1.80 (1.44, 2.27)	1.35 (1.11, 1.65)

MACE1: a composite of acute myocardial infarction, stroke, peripheral vascular disease, coronary artery disease, and all-cause mortality; MACE2: a composite of acute myocardial infarction, stroke, peripheral vascular disease, coronary artery disease.

AMI, acute myocardial infarction; CAD, coronary artery disease; MACE, major adverse cardiovascular event; PVD, peripheral vascular disease.

Model 1 was adjusted for age.

Model 2 was adjusted for age, sex, race/ethnicity, census division, residence, and income.

Model 3 was adjusted for age, sex, race/ethnicity, census division, residence, income, body mass index, smoking status, diabetes, and hypertension.

Model 4 was adjusted for age, sex, race/ethnicity, census division, residence, income, body mass index, smoking status, diabetes, hypertension, dependency, and number of chronic diseases.