Author + information
- Received July 19, 2014
- Revision received November 19, 2014
- Accepted December 16, 2014
- Published online March 17, 2015.
- Giuseppe Sergi, MD, PhD∗,
- Nicola Veronese, MD∗∗ (, )
- Luigi Fontana, MD, PhD†,‡,§,
- Marina De Rui, MD∗,
- Francesco Bolzetta, MD∗,
- Sabina Zambon, MD‖,¶,
- Maria-Chiara Corti, MD#,
- Giovannella Baggio, MD∗∗,
- Elena Debora Toffanello, MD, PhD∗,
- Gaetano Crepaldi, MD, PhD‖,
- Egle Perissinotto, ScD†† and
- Enzo Manzato, MD, PhD∗,‖
- ∗Department of Medicine DIMED, Geriatrics Division, University of Padua, Padua, Italy
- †Department of Clinical and Experimental Sciences, Brescia University Medical School, Brescia, Italy
- ‡Division of Geriatrics and Nutritional Science and Center for Human Nutrition, Washington University School of Medicine, St. Louis, Missouri
- §CEINGE Biotecnologie Avanzate, Naples, Italy
- ‖National Research Council, Aging Branch, Institute of Neuroscience, Padua, Italy
- ¶Department of Medical and Surgical Sciences, University of Padua, Padua, Italy
- #Azienda Unità Locale Socio Sanitaria 16, Padua, Italy
- ∗∗Division of Internal Medicine, Azienda Ospedaliera, Padua, Italy
- ††Department of Cardiac, Thoracic and Vascular Sciences, Unit of Biostatistics, Epidemiology and Public Health, University of Padua, Padua, Italy
- ↵∗Reprint requests and correspondence:
Dr. Nicola Veronese, Department of Medicine–DIMED, Geriatrics Division, University of Padua, Via Giustiniani, 2 35128 Padua, Italy.
Background Frailty is an important risk factor for cardiovascular disease (CVD), but the impact of early, potentially reversible stages of frailty on CVD risk is unknown.
Objectives This study sought to ascertain whether pre-frailty can predict the onset of CVD in a cohort of community-dwelling, not disabled, elderly people.
Methods A sample of 1,567 participants age 65 to 96 years without frailty or disability at baseline was followed for 4.4 years. Pre-frailty was defined as the presence of 1 or 2 modified Fried criteria (unintentional weight loss, low physical activity level, weakness, exhaustion, and slow gait speed), and incident CVD as onset of coronary artery diseases, heart failure, stroke, peripheral artery disease, or CVD-related mortality.
Results During follow-up, 551 participants developed CVD. Compared with participants who did not become frail, those with 1 modified Fried criterion (p = 0.03) and those with 2 criteria (p = 0.001) had a significantly higher risk of CVD, even after adjusting for several potential confounders (traditional risk factors for CVD, inflammatory markers, and hemoglobin A1c levels). Low energy expenditure (p = 0.03), exhaustion (p = 0.01), and slow gait speed (p = 0.03) were significantly associated with the onset of CVD, whereas unintentional weight loss and weakness were not.
Conclusions Our findings suggest that pre-frailty, which is potentially reversible, is independently associated with a higher risk of older adults developing CVD. Among the physical domains of pre-frailty, low gait speed seems to be the best predictor of future CVD.
Frailty is a geriatric syndrome reflecting a state of reduced physiological reserve and increased vulnerability to poor resolution of homeostasis after a stressor event that occurs in 25% to 50% of elderly patients with cardiovascular disease (CVD) (1,2). The frailty phenotype is also a strong independent predictor of disability, hospital and nursing home admission, poor surgical outcomes, and mortality (3–6). Data from observational studies show a significant correlation between frailty and CVD morbidity and mortality in the elderly (7–9), and assessing these patients for frailty has been found instrumental in refining their prognosis and defining optimal care and treatment for elderly individuals (10).
Several metabolic and hormonal factors, some of which have also been implicated in the pathogenesis of CVD (e.g., chronic low-grade inflammation and insulin resistance) (1,2), are responsible for the gradual biological and functional decline that marks the transition from a state of robustness to pre-frailty and, ultimately, to frailty and its complications (11). It has been hypothesized that the early detection of a pre-frail status may provide a window of opportunity for more aggressive preventive or therapeutic interventions that might contain disability, hospitalization, and mortality (12,13). Few and contrasting data are available, however, on the possible association between pre-frailty and CVD risk in elderly men and women without CVD or disability (3,5).
The aim of this prospective study is to investigate the impact of pre-frailty, defined using the 5 modified Fried criteria (3) of: 1) unintentional weight loss; 2) low physical activity level; 3) weakness; 4) exhaustion; and 5) slow gait speed, on the risk of developing CVD in a cohort of community-dwelling elderly individuals with no disabilities, CVD, cancer, or dementia at baseline. We adopted a modified Fried scale to ascertain pre-frailty because this is the most often used frailty index (3) and it has consistently been shown to predict disability and mortality in large cohorts of elderly patients with and without CVD (4–9).
This study was based on an analysis of 4.4 ± 1.2 years of follow-up data obtained in the Progetto Veneto Anziani (Pro.V.A.) population-based prospective cohort study conducted in 2 separate geographic areas near the city of Padua in the Veneto Region of Northern Italy. The study enrolled 3,099 age- and sex-stratified community-dwelling individuals age ≥65 years (1,854 women and 1,245 men) from 1995 to 1997. Participants came from a random sample of white men and women using a multistage stratified method described elsewhere (14). After excluding subjects with a diagnosis of CVD at baseline (n = 694), subjects with frailty (n = 200) or any disability in activities of daily living (n = 499), and subjects who declined to attend follow-up visits (n = 139), the final sample consisted of 1,567 Italian subjects.
At baseline and follow-up examinations, each participant underwent a detailed structured interview, clinical evaluation, and blood chemistry and performance-based tests to document an extensive range of demographic, biological, and medical characteristics, with the support of standardized algorithms (14). Cognitive function was assessed with the 30-item Mini-Mental State Examination and depression with the Geriatric Depression Scale (GDS) (15,16). Face-to-face interviews and clinical assessments were performed by trained physicians and nurses. The present study was approved by the Human Studies Committee of Padua University and by the Veneto Region’s Local Health Units No. 15 and No. 18. All subjects gave their informed consent.
Body weight was measured on a calibrated balance scale. Height was determined using a stadiometer to the nearest 0.5 cm. Waist circumference was obtained by using a cloth tape, with the waist defined as the mid-point between the highest point of the iliac crest and the lowest part of the costal margin in the mid-axillary line. Body mass index (BMI) was calculated as: weight (in kilograms)/height (in meters) squared. Blood pressure (BP) was measured with a mercury sphygmomanometer (Erkameter 300, Erka, Bad Tölz, Germany) in both arms after the participant had been resting quietly for >5 min in a seated position. For the ankle-brachial index (ABI) test, systolic BP was measured with BP cuffs on the right brachial artery and both posterior tibial arteries. ABI was calculated as the ratio of the average ankle systolic to arm systolic BP, taking the lowest value for reference; a value <0.9 was used for the diagnosis of peripheral artery disease (17).
Physical performance was assessed as follows:
• Short Physical Performance Battery, derived from 3 objective physical function tests (18). Each test was scored from 0 (inability to complete the test) to 4 (highest level of performance). The scores from all 3 tests were summarized in a composite score of 0 to 12 with higher scores reflecting better physical function.
1. Tandem test: Participants were asked to maintain their balance, progressing from having their feet side-by-side to a semi-tandem and then full-tandem position.
2. Gait speed: The best performance achieved in 2 walks at usual pace along a 4-m corridor was recorded in meters per second. Participants were allowed to use canes or walkers.
3. Chair stands time: Participants were asked to stand up and sit down 5 times as quickly as possible, with their hands folded across their chest. The time taken to complete the test was recorded in seconds.
• Handgrip strength: Handgrip strength was measured (in kilograms) using a JAMAR hand-held dynamometer (BK-7498, Fred Sammons, Inc., Burr Ridge, Illinois). The best result obtained at 3 attempts on the dominant side was used for our analyses.
Fried et al. (3) defined frailty using 5 measurable items. In our study, weakness and slow gait speed were respectively defined using the best handgrip strength value and the best timed walk over 4 m at usual pace, stratified by sex and BMI cutoffs, as originally proposed by Fried et al. (3). Fried’s work had yet to be published when our study was designed (1995 to 1997), so the other 3 frailty items were defined slightly differently. In our study, a low energy expenditure was defined as weekly physical activity <383 kcal/week in males and <270 kcal/week in females, calculated from the sum of all the leisure-time activities performed during an ideal week in the previous month; unintentional weight loss was defined as a self-reported unintentional weight loss >5 kg over the past year; and exhaustion was ascertained by asking the question on the GDS scale, “Do you feel full of energy?” Participants were considered exhausted if they gave a negative answer and also had a GDS score ≥10, as previously reported (19). A modified version of the Fried index was consequently used in our study.
Participants were classified as frail if they met 3 or more of the 5 modified Fried criteria, as pre-frail if they met 1 or 2, or as nonfrail if they met none of the criteria. Subjects unable to perform the handgrip strength or walking tests were considered as having weakness or slow gait speed, respectively.
Coronary heart disease was defined as a history of revascularization, hospitalization for myocardial infarction (MI), electrocardiographic evidence of MI, or self-reported history of MI or angina, accompanied by the use of antianginal medication. Cerebrovascular disease was defined as a documented (e.g., from medical charts) history of stroke, transient ischemic attack, or carotid endarterectomy, confirmed by an in-depth neurological medical examination. Heart failure (HF) was defined on the basis of the participants’ history, physical examination, chest radiography, and use of medication. An HF event was confirmed if, in addition to a physician’s diagnosis, there were documented HF signs and symptoms; supporting findings (e.g., pulmonary edema on radiograph); or HF treatments were being used, including diuretics, digitalis, angiotensin-converting enzyme inhibitors, or beta-blockers. Both at baseline and follow-up, the physician performing the physical examination diagnosed any CVD using all the previously listed measures. Another cardiologist confirmed any CVD diagnoses using a standardized algorithm considering all the medical information collected on each participant. Any disagreement was resolved by consensus (14). In the follow-up assessment, CVD mortality (codes from 390–459 according to International Classification of Diseases-9th Revision, 2002) was also included. A copy of the official death certificate of all deceased participants was obtained.
Blood samples were obtained after an overnight fast for biochemical tests, which were performed at the city hospital’s central laboratory using standard, quality-control procedures. Glycated hemoglobin (HbA1c) was measured using high-performance liquid chromatography. Serum 25-hydroxyvitamin D was measured by radioimmunoassay (RIA kit, DiaSorin, Stillwater, Minnesota). The estimated glomerular filtration rate was calculated using the Modification of Diet in Renal Disease formula. Total cholesterol, triglycerides, and high-density lipoprotein cholesterol were calculated using the enzymatic method. Low-density lipoprotein cholesterol was calculated with the Friedewald equation, unless triglycerides were higher than 400 mg/dl. The erythrocyte sedimentation rate was measured using the Westergren method and sodium citrate-anticoagulated blood samples.
All measurements obtained at baseline and follow-up were used in the data analysis. Participants’ characteristics were summarized using mean ± SD for continuous variables and counts and percentages for categorical variables. For continuous variables, normal distributions were tested using the Shapiro-Wilk test. Age- and sex-adjusted p values for trends were calculated as follows: for continuous variables the differences between the means of the covariates by number of frailty criteria were analyzed using a general linear model; logistic regression was applied for categorical variables.
The age-standardized incidence of CVD was calculated as the number of new cases of CVD per 1,000 person-years during follow-up, standardized according to the age structure of the Italian population in 1991, and compared using logistic binary regression. Cox proportional hazards models were used to assess associations between participants meeting 0, 1, or 2 frailty criteria and those with incident CVD. The time of first recorded event was considered as the “time for event.” Known factors associated with frailty and/or CVD were considered for inclusion in the analysis. To explore whether a variable should be included as a predictor in the final survival model, the log-rank test of equality across strata was performed for all categorical variables, and Cox univariate proportional hazards regression for all continuous variables. The predictors included in the final model were all variables reaching a p < 0.20 in the univariate analyses (Online Table 1). Collinearity among covariates was quantified using the variance inflation factor among the covariates. Waist circumference and BMI had a high collinearity (variance inflation factor = 4.58), but waist circumference was preferred because in older people it seems to be the best predictor of CVD onset (20). Hazard ratios (HRs) and 95% confidence intervals (CIs) were used to compare CVD rates by number of frailty criteria taking those with 0 criteria for reference. HRs were also calculated for specific frailty criteria using the same covariates as in the final model and taking for reference participants without a given condition. All analyses were performed using SPSS 21.0 for Windows (SPSS Inc., Chicago, Illinois). All statistical tests were two-tailed and statistical significance was assumed for a p value <0.05.
The sample consisted of 1,567 elderly subjects (mean age 73.6 ± 6.7 years; 61% were women) with a mean BMI of 27.69 ± 4.40 kg/m2, with no CVD, frailty, or disability at baseline. Table 1 shows the baseline characteristics by pre-frailty criteria according to the number of frailty criteria met. Pre-frail participants were older and more frequently women (p for trend <0.0001 for both variables). After adjusting for age and sex, the variables of BMI, waist circumference, and GDS score positively correlated with our modified Fried frailty score (Table 1). In contrast, current smoking and systolic BP were negatively associated with the modified Fried frailty score (Table 1). No significant differences in medical conditions were found across the groups, with the exception of chronic obstructive pulmonary disease, which was significantly higher in the pre-frail participants (p for trend <0.0001). Compared with nonfrail participants, those classified as pre-frail according to our modified Fried index had higher white blood cell counts, erythrocyte sedimentation rate, HbA1c, uric acid, and high-density lipoprotein cholesterol levels, but lower ABI, low-density lipoprotein cholesterol, 25-hydroxyvitamin D levels, and estimated glomerular filtration rate (Table 1).
Over a period of 4.4 years, 551 CV events were recorded (84 participants died of CVD; 27 developed severe angina; 36 had acute MI, 249 had HF, and 8 had stroke; and 147 developed peripheral artery disease). As shown in Table 2, the age-adjusted incidence of CVD in the sample as a whole was 75 events per 1,000 person-years (95% CI: 64 to 86), with a significant trend for the incidence of CVD higher in those meeting 2 frailty criteria (p for trend <0.0001).
Compared with those who did not become frail, participants meeting 1 criterion were at higher risk of CVD (HR: 1.25; 95% CI: 1.05 to 1.64), whereas those meeting 2 criteria were at 80% higher risk of experiencing a new CV event (HR: 1.79; 95% CI: 1.27 to 2.52) (Table 2). The higher risk of CV events during follow-up was confirmed by the cumulative incidence curve, adjusted for the same covariates as those used in the final Cox model (Figure 1). The other covariates significantly associated with the onset of new CV events during follow-up were age (HR: 1.05; 95% CI: 1.03 to 1.07; p < 0.0001), GDS (HR: 1.02; 95% CI: 1.002 to 1.05; p = 0.045), and atrial fibrillation (HR: 1.55; 95% CI: 1.03 to 2.91; p = 0.03). The use of low-dose aspirin was associated with a significant reduction in the risk of new CV events (HR: 0.78; 95% CI: 0.52 to 0.96; p = 0.03) (Online Table 2). As shown in Figure 2, the frailty parameters proving the best predictors of new CVD events were low energy expenditure (HR: 1.70; 95% CI: 1.07 to 3.50; p = 0.03), exhaustion (HR: 1.53; 95% CI: 1.09 to 2.14; p = 0.01), and slow gait speed (HR: 1.28; 95% CI: 1.03 to 1.71; p = 0.03), whereas unintentional weight loss and weakness were not associated with any significantly higher risk of new cardiovascular events (p = 0.51 and 0.19, respectively).
In our large cohort of community-dwelling elderly men and women, pre-frailty significantly predicted incident CVD. A 1-U increase in the modified Fried score, signifying multisystem decline in physiological reserves and increased vulnerability to stressors, was associated with a 25% increased CVD risk. These findings were consistent in both sexes and independent of known clinical and metabolic risk factors. In particular, pre-frailty was a significant predictor of incident HF, and more than half of the CVD events were HF-related. These findings are consistent with data from several cross-sectional studies and one longitudinal study showing that HF is the most frequent type of CVD in frail and pre-frail elderly individuals (7,21,22).
The prevalence of pre-frailty in our cohort was approximately 40% and, as expected, higher in women and older individuals (23,24). Given that this population initially had no CVD or disabilities, this prevalence is extremely high, underscoring the potential importance of assessing pre-frailty in apparently healthy elderly men and women. Pre-frailty has been associated with a 4-fold higher risk to become frail over a 4-year follow-up (3), so it may be that this study’s worse outcomes were related to pre-frail adults transitioning to frail phenotypes during follow-up. It has been suggested that pre-frailty is a reversible clinical condition if treated in the early stages (3), meaning that such interventions as cardiac rehabilitation, physical exercise, vitamin D supplementation, and reducing unnecessary medication may be able to arrest frailty or delay the transition from pre-frailty to frailty, and thereby contain the increase in CVD risk (Central Illustration) (25–28).
Pre-frailty and frailty have been described as biological syndromes resulting from the dysregulation of multiple metabolic pathways (1,11). Consistently, we found pre-frailty significantly associated with measures of central adiposity, long-term glucose control, and inflammation. The ABI was significantly lower in pre-frail patients too, strongly suggesting early atherosclerotic disease. Progressively adjusting for these correlates did not attenuate the association between pre-frailty and incident CVD, however, implying that responsibility for the association between pre-frailty and CVD lies in as-of-yet unknown factors that need to be identified. Pre-frailty may also represent the sum of several negative conditions and diseases that patients have accumulated over a lifetime, making them more sensitive to adverse health outcomes, including CVD. Interestingly, the prevalence of clinical conditions, such as depression, chronic obstructive pulmonary disease, and hypovitaminosis D (thought to play an independent role in the pathogenesis of CVD [29–31]), was significantly higher in the participants who became pre-frail. In particular, depression was an important predictor of incident CVD in our sample, probably because depression and frailty share the same metabolic changes (i.e., more inflammation and hypercortisolemia, and an increased adrenergic tone) that have key roles in the pathogenesis of sarcopenia and CVD (32). In our study, the risk of developing a new CV event increased by 2% for each point on the GDS, reinforcing the importance of diagnosing and treating depression in elderly men and women at high risk of CVD (33,34).
Another important finding of our study was the relative importance of certain frailty criteria in predicting CVD events in pre-frail men and women. Low energy expenditure, exhaustion, and slow gait speed strongly predicted incident CVD, whereas unintentional weight loss and handgrip strength did not. Our data are consistent with other studies highlighting the importance of walking speed as the best predictor of frailty in patients with prevalent CVD (35). In contrast with other studies on middle-aged adults and disabled women (36,37), we failed to find any significant association between handgrip strength and the onset of new CVD. This would support the conviction that the prognostic importance of physical performance tests, and frailty criteria in general, depends on the demographic characteristics of the population under study and on the outcome being investigated. In our population, a low gait speed may be better able than a weak handgrip to reflect systemic subclinical CV abnormalities and lesions, such as an abnormal ABI, left ventricular hypertrophy, carotid intima-media thickening, and silent carotid plaques (38,39). Consistent with this idea, it has been reported that slow gait speed is a powerful predictor of hospitalization and post-operative morbidity and mortality in patients with CVD (40,41).
First, because no imaging was done to identify subclinical cardiovascular damage, we cannot rule out the possibility of participants having a decline in heart function already at baseline. It is well known that frail subjects could have an underlying quiescent CVD, such as coronary artery disease (42). In our sample, however, the presence of CVD was carefully assessed by highly trained cardiologists (even though no strict definitions of CVD and HF were used). A second limitation of our study lies in that weight loss was self-reported and physical performance level was investigated using a nonstructural test. Finally, although the interviews were conducted with the support of relatives, we cannot exclude the presence of bias.
We demonstrated a significant association between pre-frailty and the risk of incident CVD (irrespective of any classical cardiometabolic risk factors). This suggests that targeting pre-frailty as a potentially reversible risk factor for CVD in the elderly could have significant implications. Among the physical domains of pre-frailty, low gait speed seems to be the single best predictor of future CVD. More studies are warranted to examine the effects of action to treat pre-frailty or early stages of frailty for the purposes of preventing CVD.
COMPETENCY IN PATIENT CARE: Screening elderly people without overt disability for “pre-frailty” could identify individuals more likely to harbor cardiovascular disease.
TRANSLATIONAL OUTLOOK: Clinical trials of interventions targeting the various domains of frailty are needed to ascertain whether treatment can prevent clinical cardiovascular events.
The authors acknowledge all interviewers, nurses, and physicians who participated in the study.
For supplemental tables, please see the online version of this article.
This study was funded by the Fondazione Cassa di Risparmio di Padua e Rovigo, the University of Padua, the Azienda Unità Locale Socio Sanitaria 15 and 18 of the Veneto Region, the Intramural Research Program of the National Institute on Aging, and the Veneto Region Research Project 104/02 (Dr. Corti). All authors have reported that they have no relationships relevant to the contents of this paper to disclose.
Drs. Sergi and Veronese contributed equally to this paper.
- Abbreviations and Acronyms
- ankle-brachial index
- body mass index
- blood pressure
- confidence interval
- cardiovascular disease
- Geriatric Depression Scale
- heart failure
- hazard ratio
- myocardial infarction
- Received July 19, 2014.
- Revision received November 19, 2014.
- Accepted December 16, 2014.
- American College of Cardiology Foundation
- Fried L.P.,
- Tangen C.M.,
- Walston J.,
- et al.
- Singh M.,
- Rihal C.S.,
- Lennon R.J.,
- et al.
- Afilalo J.,
- Alexander K.P.,
- Mack M.J.,
- et al.
- Corti M.C.,
- Guralnik J.M.,
- Sartori L.,
- et al.
- Yesavage J.A.,
- Brink T.L.,
- Rose T.L.,
- et al.
- Melamed M.L.,
- Muntner P.,
- Michos E.D.,
- et al.
- Guralnik J.M.,
- Simonsick E.M.,
- Ferrucci L.,
- et al.
- Newman A.B.,
- Gottdiener J.S.,
- Mcburnie M.A.,
- et al.
- Bouillon K.,
- Batty G.D.,
- Hamer M.,
- et al.
- Rockwood K.
- Singh M.,
- Stewart R.,
- White H.
- Gharacholou S.M.,
- Reid K.J.,
- Arnold S.V.,
- et al.
- Wang L.,
- Song Y.,
- Manson J.E.,
- et al.
- Calverley P.M.,
- Scott S.
- Girod J.P.,
- Brotman D.J.
- Hamer M.,
- Kivimaki M.,
- Lahiri A.,
- et al.
- Elbaz A.,
- Ripert M.,
- Tavernier B.,
- et al.
- Chaudhry S.I.,
- McAvay G.,
- Chen S.,
- et al.
- Afilalo J.,
- Eisenberg M.J.,
- Morin J.F.,
- et al.