Frailty in NHANES: Comparing the frailty index and phenotype

https://doi.org/10.1016/j.archger.2015.01.016Get rights and content

Highlights

  • The frailty index and phenotype confirmed previously established characteristics of frailty scales.

  • The frailty index may discriminate levels of frailty better at the lower to middle end of the frailty continuum.

  • Both measures were associated with disability, self-reported health and healthcare utilization.

Abstract

The two most commonly employed frailty measures are the frailty phenotype and the frailty index. We compared them to examine whether they demonstrated common characteristics of frailty scales, and to examine their association with adverse health measures including disability, self-reported health, and healthcare utilization. The study examined adults aged 50+ (n = 4096) from a sequential, cross-sectional sample (2003–2004; 2005–2006), National Health and Nutrition Examination Survey. The frailty phenotype was modified from a previously adapted version and a 46-item frailty index was created following a standard protocol. Both measures demonstrated a right-skewed distribution, higher levels of frailty in women, exponential increase with age and associations with high healthcare utilization and poor self-reported health. More people classified as frail by the modified phenotype had ADL disability (97.8%) compared with the frailty index (56.6%) and similarly for IADL disability (95% vs. 85.6%). The prevalence of frailty was 3.6% using the modified frailty phenotype and 34% using the frailty index. Frailty index scores in those who were classified as robust by the modified phenotype were still significantly associated with poor self-reported health and high healthcare utilization. The frailty index and the modified frailty phenotype each confirmed previously established characteristics of frailty scales. The agreement between frailty and disability was high with each measure, suggesting that frailty is not simply a pre-disability stage. Overall, the frailty index classified more people as frail, and suggested that it may have the ability to discriminate better at the lower to middle end of the frailty continuum.

Introduction

Chronological age – the number of years since birth – differs from biological age – the active rate at which the body is aging (Searle, Mitnitski, Gahbauer, Gill, & Rockwood, 2008). Biological age may be better at describing quality of life, life expectancy and current level of health (Mitnitski, Graham, Mogilner, & Rockwood, 2002). The measurement of frailty could represent an assessment of biological age and thereby is a useful estimate of an individual's health status (Rockwood, 2005a, De Lepeleire et al., 2009). Frailty signifies an increased vulnerability to adverse health outcomes, reflecting an age-associated decline in multiple physiological systems (Fried et al., 2001, Rockwood and Theou, 2012, Rockwood, 2005b, Xue, 2011). The two most commonly used approaches to frailty differ, viewing frailty either as a syndrome (frailty phenotype approach) (Fried et al., 2001) or as a state (frailty index approach) (Mitnitski et al., 2002, Rockwood, 2005b). Although the frailty phenotype and frailty index have been compared (Kulminski et al., 2008, Malmstrom et al., 2014, Ravindrarajah et al., 2013, Rockwood et al., 2007, Theou et al., 2013, Woo et al., 2012), their similarities and differences are still not fully agreed upon. Many studies have demonstrated that frailty scores and characteristics from the two measures are comparable (Malmstrom et al., 2014, Woo et al., 2012, Mitnitski et al., 2011), while others have provided evidence that the frailty index can define the risk of adverse outcomes more precisely than does the phenotype (Hubbard et al., 2009, Malmstrom et al., 2014, Rockwood et al., 2007, Theou et al., 2013).

The majority of studies that compared the frailty phenotype and the frailty index have focused solely on people over the age of 65 (Theou et al., 2013, Kulminski et al., 2008, Rockwood et al., 2007, Woo et al., 2012, Mitnitski et al., 2011, Jung et al., 2014), despite emerging evidence suggesting that frailty begins much earlier than that (Rockwood, Song, & Mitnitski, 2011). Studies that have examined a younger age group have focused on a smaller sample that is not representative of the general population such as the European Male Aging Study (Ravindrarajah et al., 2013) or a the African American Health cohort (Malmstrom et al., 2014). Cesari, Gambassi, van Kan, and Vellas (2014) recently suggested that the two constructs should be considered complementary to one another rather than interchangeable and thus should not be compared. Even so, contrasting the two measures can help us understand the frailty construct, especially using data from a general population.

Fried et al. (2001) identify frailty as the presence of three or more of the five criteria: unintentional weight loss, low energy, slow gait, reduced grip strength and reduced physical activity. The frailty index (Rockwood et al., 1999) operationalizes frailty as the fraction of deficits present in an individual (Searle et al., 2008). Multiple studies have examined frailty in the National Health and Nutrition Examination Survey (NHANES), a large-scale series of cross-sectional surveys that have been used to extensively describe the health of the U.S population. The majority of these studies have followed an adapted version of the phenotype proposed by Wilhelm-Leen et al. (Bowling and Muntner, 2012, Eichholzer et al., 2013, Singh et al., 2012, Smit et al., 2012, Smit et al., 2013, Wilhelm-Leen et al., 2009, Wilhelm-Leen et al., 2010). This definition can no longer be used, as components of the phenotype were not measured in cohorts after 2002. Of note, no study has examined frailty using NHANES cohorts past 2002, suggesting that a validated adapted phenotype is necessary. Furthermore, although a frailty index is feasible, it has not yet been employed in the NHANES dataset.

This study compared two alternative measures of frailty using the NHANES data: a modified 4-item version of the frailty phenotype and a frailty index. The objectives were two-fold. First, to examine whether the frailty index and modified frailty phenotype demonstrate common characteristics in terms of distribution, mean score, sex differences in frailty scores, limit (99th percentile), relationship of frailty with age, prevalence of frailty, and second to examine the association between each definition of frailty and adverse health measures including disability, self-reported health, and healthcare utilization.

Section snippets

Sample and study design

We conducted secondary analysis of the cross-sectional data from the 2003–2004 and 2005–2006 cohorts of the United States National Health and Nutrition Examination Survey (NHANES) (CDC, 2014). This study was part of a larger study that examined the association between frailty and physical activity (Blodgett, Theou, Kirkland, Andreou, & Rockwood, 2014). Frailty was not directly measured, however frailty level could still be identified through the data available using the frailty index and

Results

Of 4874 participants over the age of 50, 4096 (mean age 63.4 ± 10.3; 53.5% women, 18.0% ADL disability) were included in this analysis. Characteristics were comparable for those who were excluded due to missing frailty data (mean age 64.6 ± 10.9, 52.3% women, 19.3% ADL disability). See Table 2 for the demographic characteristics of the sample. The distribution of scores on both frailty scales, measured as a proportion of deficits present out of total deficits possible, resembles a right skewed

Discussion

This secondary analysis of NHANES showed that the properties of both the frailty index and modified frailty phenotype were consistent with previously accepted characteristics of other frailty measures (Rockwood and Mitnitski, 2007, Theou et al., 2013) including the right skewed distribution of frailty, higher levels of frailty in women compared to men, an upper limit of deficit accumulation and an exponential increase with age. However the magnitude of these characteristics differed (e.g.

Funding

KR is supported by an operating grant from the Canadian Institutes of Health Research and receives funding from the Dalhousie Medical Research Foundation as Kathryn Allen Weldon Professor of Alzheimer Research. OT is supported by a Banting Postdoctoral Fellowship.

Conflict of interest

None.

References (41)

  • J. De Lepeleire et al.

    Frailty: An emerging concept for general practice

    British Journal of General Practice

    (2009)
  • M. Eichholzer et al.

    Urinary phytoestrogen levels and frailty in older American women of the National Health and Nutrition Examination Survey (NHANES) 1999–2002: A cross-sectional study

    Annals of Nutrition and Metabolism

    (2013)
  • L.P. Fried et al.

    Frailty in older adults: Evidence for a phenotype

    Journals of Gerontology Series A: Biological Sciences and Medical Sciences

    (2001)
  • L.P. Fried et al.

    Untangling the concepts of disability, frailty, and comorbidity: Implications for improved targeting and care

    Journals of Gerontology Series A: Biological Sciences and Medical Sciences

    (2004)
  • R.J. Gobbens et al.

    The predictive validity of the Tilburg Frailty Indicator: Disability, health care utilization, and quality of life in a population at risk

    Gerontologist

    (2012)
  • P. Green et al.

    Gait speed and dependence in activities of daily living in older adults with severe aortic stenosis

    Clinical Cardiology

    (2012)
  • J.M. Guralnik et al.

    Lower-extremity function in persons over the age of 70 years as a predictor of subsequent disability

    New England Journal of Medicine

    (1995)
  • M. Hoover et al.

    Validation of an index to estimate the prevalence of frailty among community-dwelling seniors

    Health Reports

    (2013)
  • R. Hubbard et al.

    Characterising frailty in the clinical setting—A comparison of different approaches

    Age and Aging

    (2009)
  • H.W. Jung et al.

    Prevalence and outcomes of frailty in Korean elderly population: Comparisons of a multidimensional frailty index with two phenotype models

    PLOS ONE

    (2014)
  • Cited by (249)

    View all citing articles on Scopus
    View full text