Monitoring 6-Month Trajectory of Grip Strength Improves the Prediction of Long-Term Change in Grip Strength in Disabled Older Women

Abstract

Background.

This study aims to assess the degree and heterogeneity of decline in grip strength in the short term (6 months) and the clinical relevance of short-term decline to long-term decline (2.5 years) in grip strength in disabled older women.

Methods.

Eighty-four community-dwelling disabled women aged 65 years or older were evaluated on a weekly basis over 6 months, followed by an additional five semi-annual visits for a total follow-up of 3 years. The study outcome is maximum isometric handgrip strength in the nondominant hand. Linear random effects models were used to characterize population- as well as individual-level change in grip strength over time.

Results.

In the weekly assessments over the initial 6 months, individual-level short-term change in grip strength was detectable (mean = −0.12kg/month; p = .06), heterogeneous (range: −2.03±0.95kg/month), and independent of absolute grip strength at baseline (mean = 18.4kg). Additionally, among women with grip strength greater than 16.7kg at baseline, the long-term rate of decline in grip strength was accelerated by 0.15kg/year (p = .014) for every 0.5 standard deviation (0.25kg/month) increase in the short-term rate of decline. The baseline absolute grip strength, however, was not significantly associated with the long-term rate of decline (correlation = −0.36, p = .105).

Conclusions.

Our findings suggest that short-term change in grip strength is detectable and meaningful for disabled older women and it is the individual’s past trajectory of grip strength relative to her current strength level, not her current absolute strength per se, that is important for predicting future strength decline.

Aging is associated with the progressive loss of skeletal muscle mass and strength referred to as sarcopenia. Muscle weakness as determined by a single measurement of grip strength has proven repeatedly to correlate with subsequent adverse health outcomes (1), even when measured in mid-life to predict physical disability decades later (2). Using longitudinal data from a sample of initially high functioning women aged 70–79 at baseline, we recently reported that a decline in grip strength over time is a stronger predictor of a greater variety of subsequent adverse outcomes than is a single observation of grip strength, suggesting that “becoming weaker” is important in addition to “being weak” (3,4). In addition, we found low grip strength to be the most common first manifestation of physical frailty (5,6), suggesting that muscle weakness may serve as a warning sign of increasing vulnerability for frailty. Altogether, these findings have led to the recent efforts to define clinically meaningful cut points for grip strength in the diagnosis of sarcopenia and growing interest in using grip strength as a key endpoint in clinical trials of new interventions for sarcopenia and frailty (7–9). However, the success of such efforts requires improved knowledge in two areas. First, whether a single measurement of strength can sufficiently capture the underlying risk? Second, given that most of the pilot trials have a short follow-up, typically 6 months, it is important to know the degree and heterogeneity of detectable change in the short term and the clinical relevance of such change to long-term change in grip strength and health outcomes. So far, these topics have not been adequately studied.

Building on our earlier longitudinal work on muscle strength and health in older adults, we hypothesize that short-term trajectory of grip strength, characterized by rate of change in grip strength, predicts long-term trajectory of grip strength in older adults. To address this hypothesis, we used data from the Weekly Disability Substudy of the Women’s Health and Aging Study I (WHAS I). The study sample is comprised of 113 women aged 65 years and older who were evaluated on a weekly basis over 6 months, followed by additional five visits 6 months apart for a total follow-up of 3 years. These frequently collected and high quality data on grip strength provide a unique opportunity to (i): characterize the between-person heterogeneity of short-term (6 months) trajectory of grip strength; and (ii) examine the relative impact of baseline grip strength versus the short-term trajectory of grip strength on long-term (2.5 years) trajectory of grip strength in disabled older women.

Methods

Study Population

WHAS I is a prospective, observational study of the causes and course of disability in the one-third most disabled older women living in the community (10). Eligibility criteria included self-reported difficulty in at least 2 of 4 domains of physical function (mobility, upper extremity, high functioning, and self-care tasks) and a Mini-Mental Status Exam score of 18 or higher. Of the 1,409 eligible screener respondents, 1,002 (71.1%) participated in the full study. A comprehensive questionnaire-based interview, physical examination, and blood testing were conducted in participants’ homes at baseline (November 1992–February 1995) and six follow-up exams 6 months apart over 3 years.

The Weekly Disability Substudy sample consisted of 113 women randomly selected with approximately equal numbers of subjects in nine groups, defined by age (65–74, 75–84, 85+ years) and three levels of severity of disability (disability in two, three, and four domains) (11). Participants were interviewed and examined weekly in their homes, on the same day of the week and at the same time of day, for 24 weeks. Weekly substudy visits began 1–2 weeks after the baseline assessments were completed for the main study. The studies were approved by the Johns Hopkins University Institutional Review Board. The analytic sample consists of 84 women who were alive and had grip strength measurement at the 1-year visit.

Measure of Grip Strength

Isometric grip strength was measured in kilograms using a JAMAR hand dynamometer (Model #PC5030J1, J.A. Preston Co., Jackson, MI). The test was performed three times on each hand with the participant in a sitting position with the arm to be tested pressing against her side at a right angle. Study participants were instructed to grab the metal handles of the dynamometer and squeeze as hard as they could. The maximum measurement in the nondominant hand was used in the analyses (4).

Covariates

Baseline covariates include age, race (white vs non-white), years of education, and body mass index (BMI < 18.5, 18.5–24.9, 25–29.9, or ≥30kg/m²), smoking status (current, former, or never smoker), and presence or absence of 14 major chronic diseases determined by predefined algorithms that utilized self-report, physical examination, laboratory tests, and medical records (see Table 1) (10). The number of “definite” diseases, out of 14, was used as a measure of disease burden. Depressive symptoms were assessed using the 30-item Geriatric Depression Scale (GDS) (12) and depressive mood is deemed present if the number of depressive symptoms is greater than 9.

Table 1.

Univariate Effects of Baseline Demographic and Health Characteristics on Baseline and Rate of Change in Grip Strength in the Initial 6 Months of the Study: Weekly Substudy of the Women’s Health and Aging Study I (n = 84)

	N (%)	Baseline Grip Strength (kg)	p Value*	Rate of Change in Grip Strength (kg/month)	p Value†
Age (years)
65–74	28 (33.3)	20.6 (19.0, 22.3)	REF	−0.087 (−0.285, 0.112)	REF
75–84	30 (35.7)	17.8 (16.2, 19.4)	.016	−0.114 (−0.315, 0.087)	.850
≥85	26 (31.0)	16.7 (15.0, 18.4)	.001	−0.151 (−0.360, 0.059)	.665
Education (years)
0–8th grade	35 (41.7)	18.8 (17.2, 20.3)	REF	−0.041 (−0.218, 0.136)	REF
9–11th grade	21 (25.0)	18.4 (16.4, 20.4)	.780	−0.165 (−0.399, 0.069)	.407
12th grade	11 (13.1)	18.2 (15.4, 21.0)	.732	−0.233 (−0.545, 0.079)	.294
>12th grade	17 (20.2)	17.8 (15.5, 20.0)	.467	−0.137 (−0.407, 0.133)	.560
Race
White	54 (64.3)	17.3 (16.1, 18.5)	REF	−0.203 (−0.345, −0.061)	REF
Black	30 (35.7)	20.4 (18.8, 22.0)	.003	0.049 (−0.144, 0.243)	.040
Body mass index
Underweight: <18.5	4 (5.1)	14.0 (9.6, 18.4)	.064	−0.131 (−0.683, 0.421)	.524
Normal: ≥18.5, <25	16 (20.3)	18.7 (16.5, 20.8)	REF	−0.329 (−0.587, −0.072)	REF
Overweight: ≥25, <30	27 (34.2)	16.5 (14.9, 18.2)	.135	0.034 (−0.162, 0.231)	.028
Obese: ≥30	32 (40.5)	20.3 (18.8, 21.8)	.229	−0.125 (−0.311, 0.062)	.207
Smoking
Never	32 (38.1)	16.1 (14.5, 17.6)	REF	−0.106 (−0.294, 0.082)	REF
Former	34 (40.5)	20.0 (18.5, 21.5)	<.001	−0.070 (−0.257, 0.116)	.791
Current	18 (21.4)	19.5 (17.5, 21.6)	.007	−0.217 (−0.465, 0.031)	.485
Number of diseases‡
0–1	8 (9.5)	17.0 (13.7, 20.2)	.349	−0.023 (−0.410, 0.363)	.620
2	34 (40.5)	18.4 (16.8, 20.0)	.788	−0.121 (−0.306, 0.063)	.947
≥3	42 (50.0)	18.7 (17.2, 20.1)	REF	−0.130 (−0.294, 0.035)	REF
Depressive symptoms
<10	51 (60.7)	18.0 (16.7, 19.3)	REF	0.027 (−0.115, 0.168)	REF
≥10	33 (39.3)	19.0 (17.4, 20.6)	.94	−0.343 (−0.521, −0.165)	.001

Notes: REF = reference category.

*Test of difference in baseline grip strength between reference and each of the other categories of the predictor.

^†Test of difference in rate of change in grip strength between reference and each of the other categories of the predictor.

^‡Of 14 “definite” diseases, adjudicated by physicians based on predefined criteria, including: coronary artery disease (angina pectoris and/or myocardial infarction), congestive heart failure, degenerative disc disease, spinal stenosis, hip fracture, osteoporosis, osteoarthritis (of knee, hip, or hand), rheumatoid arthritis, stroke, Parkinson’s disease, pulmonary disease, diabetes mellitus, peripheral arterial disease, and cancer.

Statistical Analyses

Linear random effects models (REM) were first used to estimate separately the average rate of decline in the initial 6 months (termed “short-term decline” henceforth) and in the last 2.5 years of the study (termed “long-term decline” henceforth). We excluded grip strength measurements between baseline and 6 months in the REM fitting of the long-term trajectory to avoid data overlap with the short-term trajectory characterization. The REM allows us to evaluate the between-person heterogeneity in baseline grip strength (ie, random intercept) and rate of change in grip strength over time (ie, random slope) while accounting for within-person correlation among the repeated strength measurements. Test–retest reliability of grip strength measurement was calculated using the variance of the random intercept and the residual variance estimated from the REM as measures of interperson and intraperson variability, respectively. Next, we evaluated the univariate and multivariate-adjusted effects of baseline demographic and health characteristics on the baseline and the short-term rate of decline in grip strength. To assess the relative impact of baseline grip strength versus the short-term grip strength decline on long-term trajectories of grip strength, we jointly modeled the short-term and the long-term trajectory using a multivariate REM in which the rate of long-term decline is regressed on the rate of short-term decline and baseline grip strength. To address the concern of floor effect in strength decline among women with low grip strength at baseline, the fitting of the joint model was stratified by baseline strength using the lowest tertile (≤16.7kg) versus top two tertiles (>16.7kg), termed “weak strength” and “normal strength” respectively henceforth, while constraining the confounding effects to be the same for both strata. Because the associational trends were similar for the top two tertiles, they were combined to increase study power due to small sample size. The model was fit using the full information maximum likelihood estimator that is unbiased under the assumption of data missing at random (13). The Bayesian Information Criterion and regression residuals were used to assess model fit. MPLUS, version 7.11 (Muthén & Muthén, Los Angeles, California) were used for model fitting.

Results

Of the 84 women in this study, the mean age was 79 years; 36% were African American; 32%, 35%, and 33% reported having difficulty in two, three, and four domains of physical function; and 75% were either overweight or obese. On average, they had 10 years of education and 3 diseases. Compared to the 1,002 women from the main study, those in the weekly substudy had higher average disease count (mean [SD]: 2.8 [1.2] vs 2.4 [1.3]; of 14), higher prevalence of former or current smokers (61.9% vs 45.4%), but otherwise comparable with respect to other demographics and health characteristics in Table 1.

As shown in Figure 1, the short-term trajectory of grip strength appears to be quite heterogeneous among a random sample of eight women from the weekly substudy. The average baseline grip strength was 18.4kg (range: 8.0–29.9, SD = 4.7), and the average rate of decline during the initial 6 months (ie, “short-term decline”) was -0.12kg/month (range: −2.03 to +0.95, SD = 0.49kg/month; p = .06). A womon’s short-term rate of change was not affected by her baseline strength (correlation (ρ) = 0.034, p = .785). The test–retest reliability of the grip strength test was 0.90.

Figure 1.

A random sample of eight subject-specific trajectories of grip strength (represented by segmented line) from the weekly substudy, with a separate least-squares regression fitted to the data of each person (dashed line) to revel overall time trend.

Older age, being White, and never-smoker were associated with weaker grip strength at baseline (p < .05); and being white and having depressive mood were associated with faster short-term decline in grip strength after adjusting for baseline strength (p < .05). In addition, overweight women experienced no decline in grip strength compared to normal weight women (p = .028; Table 1). The effects of age, race, and smoking on baseline grip strength remained significant after multivariate adjustment that included these three and education, BMI and number of diseases. The effects of BMI and depressive symptoms on short-term strength decline remained significant after adjusting for age, race, education, smoking, and number of diseases.

Grip strength declined at an average rate of 0.80kg/year (range: −1.92 – +0.08, SD = 0.67kg/year; p < .001) in the last 2.5 years of the study (ie, “long-term decline”). The absolute grip strength at the 6-month visit (ie, the baseline of the long-term trajectory) was not significantly associated with the long-term decline (ρ = −0.36, p = .105). The correlation between short-term and long-term decline was 0.23 for the entire sample (p = .577); the association however varied by baseline grip strength (Figure 2). For women with normal grip strength at baseline, faster short-term decline was associated with faster long-term decline; and the association remained significant after adjusting for the absolute grip strength at the 6-month visit. Specially, the long-term decline was accelerated by 0.15kg/year (95% confidence interval (CI) = 0.03~0.27; p = .014) for every 0.5 standard deviation (0.25kg/month) increase in the short-term decline, after adjustment for age, race, education, depressive symptoms, number of diseases, smoking status, and strength at baseline. In contrast, the relationship became negative for women with weak grip strength at baseline such that the long-term decline was decreased by 0.90kg/year (95% confidence interval (CI) = 0.07~1.74; p = .034) for every 0.5 standard deviation increase in the short-term decline. While the effect remained significant for those with weak strength after further adjustment for baseline BMI (1.02kg/year; p = .049), the effect was attenuated for those with normal strength (0.10kg/year; p = .08). Among women with normal strength, obese and overweight women were more likely to maintain their grip strength in the long term compared to underweight or normal weight women (p = .019, .385).

Figure 2.

Differential association between initial 6-month (ie, short-term) rate of decline (kg/month) in grip strength and subsequent 30-month (i.e., long-term) rate of decline (kg/year) in grip strength by baseline grip strength (kg) (top two tertiles on the left versus bottom tertile on the right); Smooth splines with three degrees of freedom are added to reveal systematic trends.

Discussion

This is the first longitudinal study to characterize short-term dynamics of change in grip strength in older women. Despite the fact that the women in the study represented the one-third most physically disabled community-dwelling older women, the individual-level change in grip strength over 6 months was detectable and heterogeneous. More importantly, absolute baseline grip strength was not informative for predicting short-term change in grip strength. Additionally, we found that the short-term and the long-term decline in grip strength was positively correlated among women with grip strength above 16.7kg at baseline; and it is the individual’s past trajectory of grip strength relative to her current strength level, not her current absolute strength per se, that is important for predicting future change. Taken together, these findings provide further evidence that “becoming weaker” is as important as “being weak.”

Although there has been a growing epidemiological literature linking muscle weakness to adverse outcomes in older adults, delineating the benefits likely to result from the clinical translation of such findings requires a modern approach that can successfully tackle at least three issues. First, we need to have better knowledge on the reliability of the strength measurements, the magnitude of change in grip strength detectable in the short-term, as well as the impact of such change on long-term change in grip strength. This information is particularly important for older adults with comorbidities and disability because they are arguably the people who have the most to gain from interventions that delay or reverse muscle strength decline. While data from this study provided evidence of good reliability in grip strength measurements among older disabled women, it is generally inadequate to rely on a single measurement of grip strength when selecting intervention targets if the goal is to impact the long-term trajectory of grip strength and its health consequences. In this study, we found that, for two older women with the same absolute grip strength at the present time, the risk of strength decline in the future is greater for a woman whose strength has been on a rapid decline, such that her current value reflects an already compromised level relative to her past peak strength, than it is for a woman whose strength has been maintained at her usual “normal” level, albeit at a “norm” that is lower than that for other age-matched individuals.

Second, the answer to the question of whether to pay attention to absolute strength, amount of change, or dropping below a threshold may be different in people in different parts of the strength spectrum. It could be argued that absolute strength becomes less informative for frail and/or disabled women because the variability of their grip strength at baseline is truncated compared to that of a normal population. In this study, however, the age-adjusted mean baseline grip strength was only slightly lower than that previously seen among the least disabled women in the Women’s Health and Aging Study II (WHAS II) (21.3 vs 23.5kg at age 70) (4), and the variability was comparable at baseline (SD = 4.95, 4.85 kg, respectively). Interestingly, we found instead that the association between short-term and long-term change in grip strength varied by baseline strength; namely, there was a positive correlation between faster short-term decline and faster long-term decline among women with normal grip strength, but a negative correlation among women with weak grip strength. While the negative correlation may have resulted from a combination of floor effect and selective loss of follow-up of women with weak and declining strength, the long-term health consequences of women with weak and declining strength remain to be studied.

Third, in order to justify the resources needed for clinical monitoring of change in grip strength, the data collected must benefit older patients and caregivers in terms of predicting outcomes that are important to them such as functional independence. Using data from WHAS II with up to 12 years of follow-up, we previously reported that faster rates of decline in grip strength predicted elevated relative risks of falling, mobility and IADL disability, frailty, and mortality after adjustment for baseline grip strength; and the magnitude of the trajectory association was comparable to that of the baseline strength association (3). It remains unclear, however, whether the magnitude and clinical consequences of grip strength decline observed in the relatively healthy older women of WHAS II can be generalized to moderately to severely disabled women. Such knowledge will be critical for developing projections of the impact of interventions in treating strength decline and adverse outcomes in clinical populations, thereby helping inform the choice of intervention. We are pursuing this aim separately from this study.

Consistent with prior studies, older age and being white were associated with weaker strength (14,15). The greater handgrip strength found in black women is believed to be related to their greater muscle mass and known racial differences in body dimensions (15). In our study, the rate of grip strength decline also differed by race with white women experiencing greater rate of decline compared to black women. The difference, however, became nonsignificant after adjusting for age, smoking, BMI, and number of diseases. In addition, we found that depressive mood was associated with greater rate of decline in grip strength even after multivariate adjustment. A similar relationship was previously reported in older men (16). Potential mechanisms underlying the relationship between depression and muscle strength decline may include physical inactivity, malnutrition, weight loss, impaired voluntary activation of muscles, and low testosterone (in men) (17–20). Interestingly, overweight women also had greater maintenance of grip strength compared to normal weight or obese women, which is consistent with the reported protective effects of being overweight in older persons (21–25).

Major strengths of this study include its prospective design with both high-quality, weekly collected data over a 6-month period and subsequent semiannual follow-ups up to 3 years, and representative sample approximating clinical populations of older women with disability and comorbidity. However, there are several limitations. First, our analyses were restricted to a cohort of disabled women living in the community. Therefore, the results cannot be generalized to men or less disabled women. Second, because the analysis excluded those who dropped out between the end of the weekly substudy and the first annual follow-up visit, there might be a selection bias in the association between short-term and long-term changes in grip strength, and the selection bias could be more prominent among women with weak strength at baseline such that the pattern seen in the weak strength group might be the result of survival bias. In fact, of the 29 substudy women excluded from this analysis, 15 died during the study; of the 15 who died, 10 died within the first year of follow-up and eight of the ten had weak strength at baseline. Third, there is a concern that it may be difficult in the clinical setting to acquire, standardize, locate, and store repeated measurement of strength over long periods of time when people change physicians, health care systems, and living locations frequently, especially while health status is declining. However, we believe, as new generations of electronic medical record systems increasingly allow integrated record keeping and individual-level tracking, monitoring of strength change over time will soon be quite feasible. Further, such monitoring will help answer the call for personalized medicine if strength decline indeed can explain meaningful between-person heterogeneity in outcomes differently than can absolute strength alone.

In summary, we were able to obtain for the first time a precise estimate of the magnitude of short-term change in grip strength and its impact relative to absolute grip strength on long-term grip strength decline in disabled older women. Demonstrating so supports a new paradigm of examining relationships between muscle strength and health, and suggests grip strength trajectory as a previously undiscovered window into late-life vulnerability. The ultimate test of our findings will require randomized controlled trials to determine whether clinical care and outcomes of geriatric patients can be improved by monitoring grip strength and instituting interventions in clinical practice.

Funding

For data acquisition, management, and analysis, this research was supported, in part, by N01AG012112 and R03AG041992 from the National Institute on Aging (NIA), National Institutes of Health (NIH), and the Johns Hopkins Older Americans Independence Center under NIA/NIH contract P30-AG02133.