Cautionary tails of grip strength in health inequality studies: An analysis from the Canadian longitudinal study on aging
Introduction
Health inequalities have been extensively documented for a wide variety of measures of health within community-based populations. With a few notable exceptions (e.g., motor vehicle accidents among men and breast cancer among women) (James et al., 2007; Tjepkema et al., 2013) inequalities in measures of health exhibit a socioeconomic gradient whereby the socioeconomically advantaged are in better health (Marmot et al., 2013). One of the most widely used measures of health in such studies is self-rated health, assessed by the question, “How would you rate your health in general – excellent, very good, good, fair, or poor?” (Hu et al., 2016; Moor et al., 2017; Jusot and Tubeuf, 2019) Health assessed by this simple question has been a staple in numerous population health surveys (e.g., Gateway to Global Aging Data (https://g2aging.org/?section=page&pageid=182)), and despite its simplicity and subjectivity, longitudinal studies show that it correlates well with future all-cause (Idler and Benyamini, 1997; Benyamini and Idler, 1999; DeSalvo et al., 2006) and cause-specific (Bamia et al., 2017; Mavaddat et al., 2014) mortality, and, to a lesser extent, with future functional limitation (Idler and Benyamini, 1997; Lee, 2000), hospitalization (Kennedy et al., 2001; Tamayo-Fonseca et al., 2015), and the incidence of chronic conditions (Latham and Peek, 2013). Its subjectivity is viewed both positively and negatively. On one hand, the subjective assessment can reveal aspects of health that are not accessible by external and/or objective means. On the other hand, the subjective assessment might bring biases in health inequality analyses if it varies systematically with characteristics of interest for inequality. Correlations between self-rated health and mortality vary across socioeconomic status, but with inconsistent directions (Huisman et al., 2007; Regidor et al., 2010). Studies also find both pessimistic (Adams and White, 2006; Dowd and Zajacova, 2010; Layes et al., 2012) or optimistic (Humphries and Van Doorslaer, 2000) assessments by the socioeconomically advantaged compared to the socioeconomically disadvantaged. These findings have raised a concern about the possibility of under- or over-estimation of a true underlying socioeconomic gradient using self-rated health (Quesnel-Vallée, 2007; Subramanian and Ertel, 2008). This concern, in turn, has increased interest in alternative measures of health that preserve attractive features of self-rated health yet are objective.
Hand grip strength—an estimate of isometric strength in the upper extremity—has been suggested as an attractive alternative to self-rated health for community-based older populations because it appears to share key strengths of, and overcome major limitations of, self-rated health (Frick and Ziebarth, 2013; Leopold and Engelhartdt, 2012; Leopold, 2019; Ziebarth, 2010). Grip strength is easy to measure by a portable, inexpensive, non-invasive, and simple device (Roberts et al., 2011). It is one among a set of objective physical capability measures (i.e., standing balance, chair rises, walking speed, etc.) that are increasingly studied and included in numerous population-based health and aging studies globally due to their simplicity (Cooper et al., 2010, 2014). As an objective rather than subjective measure, grip strength avoids the need to correct for optimism and pessimism as is the case for self-rated health. In addition, robust evidence from studies with long follow-up shows that grip strength, after adjustment for age, sex, and body size, correlates strongly with future mortality among community-dwelling mid-life and older adults from various populations (Cooper et al., 2010; Rijk et al., 2015). It also correlates with other future health events, such as cardiovascular incidence (Celis-Morales et al., 2018; Leong et al., 2015; Beyer et al., 2018; Silventoinen et al., 2009) and mortality (Celis-Morales et al., 2018; Leong et al., 2015; Sasaki et al., 2007; Kishimoto et al., 2014; Strand et al., 2016), cognitive decline (Cooper et al., 2014; Rijk et al., 2015; Alfaro-Acha et al., 2007), and functional limitation (Rijk et al., 2015). Grip strength has traditionally been understood as a function of leverage, muscle mass, and motor unit recruitment, a weak grip indicating recent health problems such as atrophy and neurological impairment. More recently, it is thought to indicate chronic lack of muscle mass related to malnutrition or the absence of stimulus (i.e., a sedentary lifestyle), therefore, capturing frailty or even overall physiological health (Bohannon, 2008; Sayer and Kirkwood, 2015; Carson, 2018). For example, based on the observation that grip strength is correlated with a greater number of markers of frailty (e.g., cognitive function, lens opacity, and number of teeth) than is chronological age, Syddall et al. propose grip strength as a single marker of frailty (Syddall et al., 2003). Since frailty is known to be graded socially (Stolz et al., 2017; Hajizadeh et al., 2016), it can be expected that grip strength could also be graded socially. A further analytical advantage is that grip strength is a continuous variable with a normal distribution, while self-rated health is an ordinal, categorical variable with five categories, thus grip strength avoids the challenges of analyzing variation in a categorical, ordinal variable within a regression framework (Long, 1997).
As a potential measure of health in community-based health inequality studies, a growing number of studies examined the social patterning of grip strength. These studies suggest the socioeconomic gradient of grip strength is not as clear as that of self-rated health with mixed results likely influenced by the study design (e.g., cross-sectional (Syddall et al., 2003; Carney and Benzeval, 2018; Hairi et al., 2010; Welmer et al., 2013; Arokiasamy and Selvamani, 2018), birth cohort (Strand et al., 2016; Starr and Deary, 2011; Hurst et al., 2013; Kuh et al., 2005), panel (Kröger et al., 2016), or systematic review and meta analysis (Birnie et al., 2011)), the choice of socioeconomic measures (Syddall et al., 2003; Hairi et al., 2010; Arokiasamy and Selvamani, 2018), sex (Carney and Benzeval, 2018; Kröger et al., 2016), and population (Hairi et al., 2010). Furthermore, contrasting self-rated health as a subjective measure of health and grip strength as an objective measure of health, a handful of studies examined similarities and differences in the social patterning of these measures. Using a two-year follow-up of participants in the Survey of Health, Ageing and Retirement in Europe (SHARE), Leopold and Engelhartdt (2012) found a stable education-related gap across age in self-rated health but an increasing education-related gap by age in grip strength (Leopold and Engelhartdt, 2012). Adding the physical component of the 12-item Short Form Survey (SF12) as a quasi/semi-objective measure of physical health to the comparison, and using five waves of the German Socio-Economic Panel Study (SOEP), Leopold (2019) found the following: among men, an increasing education-related gap by age in self-rated health and SF12 but a stable education-related gap across age in grip strength; among women, the gap was stable across age in self-rated health and SF12 but increasing by age in grip strength (Leopold, 2019). Using a single wave of the SOEP and standardizing for age and sex, Ziebarth (2010) and Frick and Ziebarth (2013) found a greater degree of income- and wealth-related inequality in self-rated health than in either grip strength or SF12 (Frick and Ziebarth, 2013; Ziebarth, 2010). These four studies—Leopold and Engelhartdt (2012), Leopold (2019), Ziebarth (2010), and Frick and Ziebarth (2013)24—compared self-rated health and grip strength in the same study contexts. Thus, unlike the aforementioned many other studies examining each measure separately across diverse study settings and designs (Strand et al., 2016; Carney and Benzeval, 2018; Hairi et al., 2010; Arokiasamy and Selvamani, 2018; Starr and Deary, 2011; Hurst et al., 2013; Kuh et al., 2005; Kröger et al., 2016; Birnie et al., 2011; Syddall et al., 2009), these studies allow for direct comparison. These four studies show that the social patterning of self-rated health and grip strength most likely differs for reasons beyond influences of study design issues. In addition, by introducing the quasi/semi-objective measure of SF12, these studies imply differences in the social patterning of these measures may derive from both subjectivity of the assessment and the health measurement construct (overall health vs. physical health).
Health is a multidimensional concept, and a growing number of studies examining determinants of grip strength reveal a potentially unique health measurement construct for grip strength (Isen et al., 2014; Frederiksen et al., 2002; Dodds et al., 2012). For example, while associations between health damaging behaviours and lower self-rated health are consistently shown (Shields and Shooshtari, 2001; Abu-Omar et al., 2004; Okosun et al., 2005; Galán et al., 2010; Shankar et al., 2011; Kim et al., 2013; Abuladze et al., 2017), such associations are inconsistently reported for grip strength (Cooper et al., 2010; Strand et al., 2011a, 2011b). In addition, Cesari et al. (2008) show potentially different pathways between mortality and grip strength and between mortality and self-rated health (Cesari et al., 2008). In a two-year follow-up study of over 300 persons aged 80 or older in Italy, both of these measures independently predicted mortality after adjustment for each other as well as age, sex, objectively assessed physical performance measures, and self-reported basic and instrumental activities of daily living (ADL and IADL), with no statistically significant interaction effect between them (Cesari et al., 2008).
To interpret why the social patterning of self-rated health and grip strength might diverge and to gain insight as to whether grip strength can be used as alternative to self-rated health in health inequality studies require an understanding of what grip strength is measuring vis-à-vis self-rated health. In an effort to provide such understanding, this study first analyzes how much self-rated health and grip strength are in accord—whether they co-vary and how much variation in one explains variation in the other—and then compares their associations with socioeconomic status and health behaviours for the same set of respondents. Unlike the aforementioned studies, which used these measures of health in the assessment of health inequalities and compared results of health inequalities, this study directly assesses the relationship between self-rated health and grip strength and compares their associations with socioeconomic status and health behaviours.
This study brings the following three strengths. First, this study focuses on grip strength adjusted for age, sex, and body size. It is age-sex-body-size-adjusted-grip strength, rather than unadjusted-grip strength, that is associated with various future health events such as mortality, cardiovascular disease, functional limitation, cognitive decline (Cooper et al., 2010, 2014; Rijk et al., 2015; Celis-Morales et al., 2018; Leong et al., 2015; Beyer et al., 2018; Silventoinen et al., 2009; Sasaki et al., 2007; Kishimoto et al., 2014; Strand et al., 2016; Alfaro-Acha et al., 2007). For this reason, adjusted rather than unadjusted-grip strength is most comparable to self-rated health. Furthermore, from a perspective of the measurement construct, unadjusted-grip strength is, foremost, a measure of physical health, whereas age-sex-body-size-adjusted-grip strength, in its association with various future health events, is multidimensional, corresponding to an understanding of grip-strength as an indicator of overall, multidimensional health (e.g., a “biomarker of ageing across the life course” (Sayer and Kirkwood, 2015) or a “vital sign for middle-aged and older adults” (Bohannon, 2008)). Second, to assess the relationship between adjusted-grip-strength and self-rated health, this study goes beyond the comparison of means. Specifically, rather than examining the mean adjusted-grip-strength of each of the five categories of self-rated health, this study investigates the distribution of adjusted-grip-strength within each of these five categories. Often referred to as discriminatory accuracy (Merlo et al., 2017), the degree of overlap between the distributions across these five categories adds further information regarding the association between self-rated health and grip strength not evident in the conventional mean comparison. Finally, this study uses a rich data source, the Canadian Longitudinal Study on Aging (CLSA) (Raina et al., 2009, 2019), which consists of a large sample (n = 26,754) of older adults with a wide age range (45–86 years) and provides extensive information on demographic, social, economic, and health behavioural factors known to be associated with health measured by self-report and clinical and physical assessments. Thus, the CLSA allows a broad assessment of factors associated with self-rated health and adjusted-grip strength in line with existing, comprehensive frameworks of multiple determinants of health (Solar and Irwin, 2010; Evans and Stoddart, 1990).
Section snippets
Study population
We use baseline, cross-sectional data from the CLSA, a population-based, longitudinal study following approximately 50,000 non-institutionalized Canadians aged 45–85 at the time of recruitment in 2012–2015 until 2033 or death (Raina et al., 2009, 2019). A subset of 30,097 participants, referred to as the CLSA Comprehensive, contains information collected through in-home face-to-face computer-assisted interviews and respondent visits to data collection sites for additional computer-assisted
Discussion
Against the backdrop of the desire of some analysts for an objective alternative to self-rated health in health inequality studies and the increasing attention in the aging literature to grip strength as one such measure, this study compared self-rated health and grip strength and their associations with socioeconomic status and health behaviours. Our direct comparison of self-rated health and adjusted-grip strength showed extremely poor discriminatory accuracy between these measures,
Funding
This study was supported by the Canadian Institutes of Health Research (MOP-142242).
Declaration of competing interest
Authors declare no conflict of interest.
Acknowledgments
We wish to thank Kathleen MacNabb and Stefan Phipps-Burton for providing research assistance.
This research was made possible using the data/biospecimens collected by the Canadian Longitudinal Study on Aging (CLSA). Funding for the Canadian Longitudinal Study on Aging (CLSA) is provided by the Government of Canada through the Canadian Institutes of Health Research (CIHR) under grant reference: LSA 9447 and the Canada Foundation for Innovation. This research has been conducted using the CLSA
References (91)
- et al.
Is the disease risk associated with good self-reported health constant across the socio-economic spectrum?
Publ. Health
(2006) - et al.
Age, socioeconomic patterns and regional variations in grip strength among older adults (50+) in India: evidence from WHO's Study on Global Ageing and Adult Health (SAGE)
Arch. Gerontol. Geriatr.
(2018) - et al.
Self-rated health and all-cause and cause-specific mortality of older adults: individual data meta-analysis of prospective cohort studies in the CHANCES Consortium
Maturitas
(2017) Get a grip: individual variations in grip strength are a marker of brain health
Neurobiol. Aging
(2018)- et al.
Physical activity across adulthood and physical performance in midlife
Am. J. Prev. Med.
(2011) - et al.
Birth weight and muscle strength: a systematic review and meta-analysis
J. Nutr. Health Aging
(2012) - et al.
Does self-rated health mean the same thing across socioeconomic groups? Evidence from biomarker data
Ann. Epidemiol.
(2010) - et al.
Producing health, consuming health care
Soc. Sci. Med.
(1990) - et al.
Unfair inequalities in health and health care
J. Health Econ.
(2009) - et al.
Self-rated health according to amount, intensity and duration of leisure time physical activity
Prev. Med.
(2010)