Abstract
Objectives:
New absolute and normalized handgrip strength (HGS) cut-points may not yield similar predictive value for cognitive performance. We sought to determine the associations of 1) each absolute and normalized weakness cut-point, and 2) compounding weakness on future cognitive impairment in older Americans.
Design:
Longitudinal-panel.
Setting and Participants:
The analytic sample included 11,116 participants aged ≥65-years from the 2006–2018 waves of the Health and Retirement Study. Participants from the Health and Retirement Study completed detailed interviews which included physical measures and core interviews.
Methods:
The modified Telephone Interview of Cognitive Status assessed cognitive function and persons scoring <11 were classified as having a cognitive impairment. A handgrip dynamometer measured HGS. Men were considered weak if their HGS was <35.5-kg (absolute), <0.45 kg/kg (body mass normalized), or <1.05 kg/kg/m2 (body mass index normalized); whereas, women were classified as weak if their HGS was <20.0-kg, <0.337 kg/kg, or <0.79 kg/kg/m2. Compounding weakness included those below 1, 2, or all 3 cut-points. Generalized estimating equations quantified the associations.
Results:
Persons considered weak under the absolute cut-point had 1.62 (95% confidence interval (CI): 1.34–1.96) greater odds for future cognitive impairment, but no significant associations were observed for those classified as weak under the body mass (odds ratio (OR): 1.12; CI: 0.91–1.36) body mass index normalized (OR: 1.17; CI: 0.95–1.43) cut-points. Older Americans below all 3-weakness cut-points had 1.47 (CI: 1.15–1.88) greater odds for future cognitive impairment, but no significant associations were found for persons classified as weak under 1 (OR: 1.08; CI: 0.83–1.42) or 2 (OR: 1.19; CI: 0.91–1.55) cut-points.
Conclusions and Implications:
Our findings suggest that each weakness cut-point has differential prognostic value for future cognitive impairment, and aggregating weakness cut-points may improve their predictive utility. Consideration should be given to how weakness categories are uniquely linked to cognitive function.
Keywords: Aging, Functional Status, Geriatrics, Muscle Strength, Muscle Strength Dynamometer
Brief Summary:
Odds for cognitive impairment may change when weakness definitions differ.
INTRODUCTION
Handgrip strength is a long-standing, viable assessment of muscle function.1 Weakness, which is operationalized as handgrip strength being below a pre-specified cut-point, is a risk factor for many age-related morbidities and disabilities.2 Accumulating evidence suggests that low handgrip strength is associated with cognitive performance.3–6 Measures of handgrip strength necessitate that persons engage in a voluntary isometric grip force task, and the ability of the finger flexors to synergistically contract is a marker of neuromuscular functioning.7 While there could be several factors that may help to explain why handgrip strength is associated with cognitive function,4 the neural systems that mediate the control of coordinated muscle contractions are also linked to the wide-ranging spectrum of indicators that are reflective of cognitive function.8 Accordingly, handgrip strength could be a useful screening tool for identifying cognitive impairment.9
Despite the prognostic value of handgrip strength for several health conditions including cognitive impairment, absolute handgrip strength may not fully capture how body size influences strength capacity. Therefore, normalizing handgrip strength to body mass,10 body mass index,11 or stature12 has been encouraged for elevating precision in measurement through partitioning the influence of body size, while minimally threatening feasibility. Although absolute and normalized handgrip strength measures have wide-spread usage, their inconsistent utilization may generate heterogeneity in handgrip strength protocols. For example, several absolute and normalized cut-points for determining weakness have been produced.13 Given the multitude of handgrip strength cut-points that exist, efforts have been made to create cut-point standardization.
The Sarcopenia Definitions and Outcomes Consortium generated separate weakness cutpoints for 1) absolute handgrip strength, 2) body mass normalized handgrip strength, and 3) body mass index normalized handgrip strength.14,15 While these normalized weakness cut-points help to account for how body size impacts strength, the prevalence of older adults considered weak under each cut-point is disproportionate.16 As such, the predictive utility of these weakness cutpoints is opaque, and investigations should examine the application of these cut-points for adverse health outcomes.16 Moreover, compounding weakness may exist, whereby persons considered weak under more than one cut-point could be truly classified as weak. Given that weakness is associated with cognitive dysfunction,17 evaluating how these absolute and normalized cut-points are associated with cognitive impairment may provide important insights into their prognostic value. Our study sought to examine the associations of 1) each absolute and normalized weakness cut-point, and 2) compounding weakness categories on future cognitive impairment in older Americans. We hypothesized that older Americans 1) below the absolute weakness cut-point, but not the body mass index or body mass normalized cut-points, would have greater odds for future cognitive impairment, and 2) under more weakness cut-points would have greater odds for future cognitive impairment.
METHODS
Participants
A secondary analysis of 11,242 participants aged at least 65-years from the 2006–2018 waves of the RAND Health and Retirement Study with data for at least one wave of handgrip strength, and one or more subsequent waves of cognitive function assessments were performed for this investigation. The Health and Retirement Study utilizes a longitudinal-panel design for monitoring health factors in Americans during aging.18 Participants in the Health and Retirement Study are followed biennially until death, and new participants are added for helping the Health and Retirement Study maintain a national sample. Interview response rates have regularly been >80%.19 Additional details about the Health and Retirement Study are available elsewhere.20
Starting in the 2006 wave, the Health and Retirement Study began conducting detailed face-to-face interviews with physical measures including handgrip strength.21 These face-to-face interviews occurred on a random half of the Health and Retirement Study sample and alternated completion at each wave, while the other half sample only engaged in the core interview so that participant burden could be reduced. Participants provided written informed consent before entering the Health and Retirement Study and the University’s Behavioral Sciences Committee Institutional Review Board approved study protocols.
Measures
Cognitive Function
Cognitive functioning was examined with a modified version of the Telephone Interview of Cognitive Status, which is a validated and well-utilized screening tool designed for epidemiological studies such as the Health and Retirement Study from the Mini-Mental State Examination.22 A 35-point composite scale was used. Assessments included immediate and delayed word recall from a list of 10 words, serial sevens subtraction test starting with the number 100, counting backward at maximal speed for 10 consecutive numbers beginning from the number 20, object naming, date naming, and stating the current president and vice president of the United States. Persons with scores <11 were classified as falling below the cognitive impairment threshold.23
Handgrip Strength
A Smedley spring-type handgrip dynamometer (Scandidact; Odder, Denmark) was used for handgrip strength measurements. Interviewers explained protocols, the dynamometer was fit to the hand size of each participant, and a practice trial was permitted as participants grasped the dynamometer with their arm at the side and elbow flexed at 90°. Beginning on the self-reported non-dominant hand, participants squeezed the dynamometer with maximal effort, twice on each hand, alternating between hands. Persons unable to stand or position their arm properly during handgrip strength testing could sit and support their arm on an object as appropriate. Participants did not engage in handgrip strength testing if they had a surgical procedure in the previous 6-months, swelling, inflammation, severe pain, or an injury to both hands in the month before the face-to-face interview. More details about the handgrip strength protocols from the Health and Retirement Study are available elsewhere.24
The maximum handgrip strength recorded regardless of hand was included in the analyses. Men were considered weak if their handgrip strength was <35.5-kg (absolute), <0.45 kg/kg (normalized to body mass), or <1.05 kg/kg/m2 (normalized to body mass index), while women were classified as weak if their handgrip strength was <20.0-kg, <0.337 kg/kg, or <0.79 kg/kg/m2.14–16 Further, compounding weakness was categorized as participants being below 1, 2, or all 3 cut-points. Although the Health and Retirement Study was not part of the eight cohort studies of community-dwelling older adults that were used to derive these cut-points,14,15 the Health and Retirement Study was included for assessing prevalence estimates of weakness from the cut-points developed by the Sarcopenia Definitions and Outcomes Consortium.16
Covariates
Age, sex, race, educational achievement, height, and body mass were self-reported. Persons with a body mass index ≥30 kg/m2 were considered obese. Participants also reported if they currently smoke cigarettes, or if they had ever smoked more than 100 cigarettes in their lifetime (previous smoker). Interviewers asked participants if a healthcare provider had diagnosed them with hypertension, diabetes, cancer (not including minor skin cancer), chronic lung disease, stroke, heart condition, emotional or psychiatric problems, and arthritis or rheumatism. Persons indicating an affirmative diagnosis for at least two conditions were considered as having multimorbidity. A single-item measure of perceived health was collected at each wave and participants self-rated their health as either “excellent”, “very good”, “good”, “fair”, or “poor”.
Depressive symptoms were assessed with the 8-item Center for the Epidemiologic Studies Depression scale. Participants reported if they experienced any negative or positive emotions (reverse scored) during the week before the interview date. Scores ranged from 0–8, with greater scores suggesting more depressive symptoms. Persons with scores ≥3 were classified as depressed.25 Participants engaging in moderate-to-vigorous physical activity at least “once a week” were considered as engaging in moderate-to-vigorous physical activity.26 Social engagement was examined with three items: 1) volunteer activities at religious, educational, health-related, or other organizations for at least an hour in the previous year, 2) at least weekly contact with parents or in-laws, and 3) current employment status. Scores ranged from 0–3 with greater scores suggesting more social engagement.27 Participants told interviewers about their ability to complete activities of daily living at each wave: dressing, eating, transferring in or out of bed, toileting, bathing, and walking across a small room. Persons suggesting difficulty or an inability in completing any basic self-care task were considered as having a limitation. Those with missing covariates were excluded (n=126).
Statistical Analysis
All analyses were conducted with SAS 9.4 software (SAS Institute; Cary, NC). The analytic sample included 11,116 participants. Each participant entered our investigation when handgrip strength was first measured (time t), and cognitive function status, along with other covariates, were examined at each wave in which handgrip strength was recorded. For a given wave, the outcome was cognitive impairment assessed at the next available wave (time t + 1). Length of follow-up between baseline handgrip strength, and the subsequent cognitive function assessment was accounted for in the analyses. Appendix 1 summarizes when participants first entered our investigation, and when cognitive function was subsequently evaluated for each wave. Of the 11,116 participants, 5,885 (52.9%) had their handgrip strength assessed during the 2006, 2010, and 2014 waves, while 5,231 (47.1%) had handgrip ascertained during the 2008, 2012, and 2016 waves. The outcome of cognitive function was determined at the next wave of the Health and Retirement Study for 20,137 of 20,336 (99.0%) participant-waves (i.e., 2 years). Additionally, 4,643 (41.8%) participants were in the analyses for one wave, 3,726 (33.5%) for two waves, and 2,747 (24.7%) for three waves. The full breakdown of the patterns in wave participation is available in Appendix 2.
The descriptive characteristics of the participants are presented as mean±standard deviation and frequency (percentage) for categorical variables. Means and 95% confidence intervals (CI) were likewise created for the baseline descriptive characteristics of the participants by weakness status from the individual cut-points. Misclassification characteristics (sensitivity, specificity, positive predictive value, negative predictive value) of each weakness cut-point (absolute, body mass normalized, body mass index normalized) and compounding weakness categories for future cognitive impairment were also generated.
Individual generalized estimating equations analyzed the associations of 1) absolute weakness (reference: no absolute weakness), 2) body mass index normalized weakness (reference: no body mass index normalized weakness), and 3) body mass normalized weakness (reference: no body mass normalized weakness) on future cognitive impairment. Moreover, a generalized estimating equation examined the association between persons considered weak under 1, 2, or 3 weakness cut-points (reference: below 0 weakness cut-points) and future cognitive impairment. Each model was first only adjusted for cognitive impairment at current wave and follow-up years (crude). The fully-adjusted models controlled for cognitive impairment at current wave, follow-up years, sex, race, multimorbidity, obesity, age, cigarette smoking status, social engagement, self-rated health, depression status, moderate-to-vigorous physical activity, activities of daily living limitations, and educational achievement. Covariates were assessed at the current wave. All generalized estimating equations accounted for repeated measures and the outcome for the next wave participated was utilized.
To evaluate the extent to which missing data (e.g., drop-out) may have influenced our findings, we conducted sensitivity analyses that limited the fully-adjusted generalized estimating equations to each participant’s first wave for the association of 1) individual weakness cutpoints, and 2) compounding weakness on future cognitive impairment. An alpha level of 0.05 was used for all analyses.
RESULTS
The descriptive characteristics of the participants are presented in Table 1. Participants were overall aged 72.3±6.5 years and were mostly female (58.1%). Of the participants, 4,921 (44.3%) were not weak, 2,089 (18.8%) were categorized as weak under 1 cut-point, 2,183 (19.6%) were weak under 2 cut-points, and 1,923 (17.3%) were weak under all 3 cut-points. Those below all 3-weakness cut-points were aged 75.5±7.7 years and 6.2% had a cognitive impairment at next wave; however, persons below none of the weakness cut-points were aged 70.9±5.5 years and 1.8% had a cognitive impairment at next wave. Appendix 3 presents the means and 95% confidence intervals for the baseline descriptive characteristics of the participants by weakness status from each cut-point.
Table 1.
Baseline Descriptive Characteristics of the Participants.
| Overall (n=11,116) | 0 Weakness Categories (n=4,921) | 1 Weakness Category (n=2,089) | 2 Weakness Categories (n=2,183) | 3 Weakness Categories (n=1,923) | |
|---|---|---|---|---|---|
| Age (years) | 72.3±6.5 | 70.9±5.5 | 72.1±6.2 | 72.9±6.7 | 75.5±7.7 |
| Body Mass Index (kg/m2) | 28.1±5.6 | 25.9±3.7 | 29.1±5.3 | 31.4±7.1 | 29.2±5.8 |
| Social Engagement | 1.2±0.8 | 1.3±0.8 | 1.3±0.8 | 1.1±0.8 | 0.9±0.8 |
| Female (n (%)) | 6,459 (58.1) | 2,810 (57.1) | 952 (45.5) | 1,389 (63.6) | 1,308 (68.0) |
| White (n (%)) | 9,053 (81.4) | 4,075 (82.8) | 1,703 (81.5) | 1,707 (78.2) | 1,568 (81.5) |
| Multimorbidity (n (%)) | 7,603 (68.4) | 2,858 (58.0) | 1,474 (70.6) | 1,721 (78.8) | 1,550 (80.6) |
| Cigarette Smoking Status (n (%)) | |||||
| Current Smoker | 1,128 (10.1) | 606 (12.3) | 213 (10.2) | 181 (8.3) | 128 (6.7) |
| Previous Smoker | 5,134 (46.2) | 2,197 (44.7) | 1,057 (50.6) | 1,006 (46.1) | 874 (45.4) |
| Never Smoked | 4,854 (43.7) | 2,118 (43.0) | 819 (39.2) | 996 (45.6) | 921 (47.9) |
| Self-Rated Health (n (%)) | |||||
| Excellent | 1,015 (9.1) | 638 (13.0) | 173 (8.2) | 121 (5.5) | 83 (4.3) |
| Very Good | 3,488 (31.4) | 1,851 (37.6) | 630 (30.2) | 571 (26.2) | 436 (22.7) |
| Good | 3,734 (33.6) | 1,594 (32.4) | 743 (35.6) | 783 (35.8) | 614 (32.0) |
| Fair | 2,248 (20.2) | 702 (14.2) | 428 (20.5) | 546 (25.1) | 572 (29.7) |
| Poor | 631 (5.7) | 136 (2.8) | 115 (5.5) | 162 (7.4) | 218 (11.3) |
| Depressed (n (%)) | 2,024 (18.2) | 689 (14.0) | 339 (16.2) | 490 (22.4) | 506 (26.3) |
| MVPA Participation (n (%)) | 6,396 (57.5) | 3,326 (67.6) | 1,168 (55.9) | 1,037 (47.5) | 865 (45.0) |
| ADL Limitation (n (%)) | 1,734 (15.6) | 384 (7.8) | 292 (13.9) | 472 (21.6) | 586 (30.5) |
| Obesity (n (%)) | 3,584 (32.2) | 710 (14.4) | 950 (45.5) | 1,152 (52.7) | 772 (40.1) |
| High School Graduate or Above (n (%)) | 8,815 (79.3) | 4,033 (82.0) | 1,694 (81.1) | 1,170 (76.5) | 1,418 (73.7) |
| Follow-Up Years | 2.1±0.4 | 2.1±0.5 | 2.1±0.4 | 2.1±0.5 | 2.1±0.5 |
| Cognitive Impairment (n (%)) | 165 (1.5) | 38 (0.8) | 30 (1.4) | 41 (1.8) | 56 (2.9) |
| Cognitive Impairment at Next Wave (n (%)) | 352 (3.2) | 89 (1.8) | 62 (2.9) | 82 (3.7) | 119 (6.2) |
Note: ADL=activities of daily living, MVPA=moderate-to-vigorous physical activity.
Table 2 shows the proportions of participants considered weak for each cut-point individually and collectively. Table 3 outlines the misclassification characteristics of the specific weakness categories on future cognitive impairment. The sensitivity was 67.1% (95% confidence interval (CI): 66.4–67.7) for persons below the absolute weakness cut-point, 66.7% (CI: 66.0–67.3) for those classified as weak under the BMI normalized cut-point, and 43.6% (CI: 42.9–44.3) for older adults below the body mass normalized weakness cut-point. Alternatively, the specificity was 60.5% (CI): 57.0–64.0) for persons considered weak under the absolute cut-point, 48.6% (CI: 45.1–52.2) for those below the BMI normalized weakness cut-point, and 69.3% (CI: 65.9–72.5) for older adults classified as weak under the body mass normalized cut-point.
Table 2.
Proportions of Participants Considered Weak for Each Cut-Point.
| n (%) | |
|---|---|
| Overall | |
| Not-Weak | 4,921 (44.3) |
| Absolute Weakness | 346 (3.1) |
| Body Mass Normalized Weakness | 1,719 (15.5) |
| Body Mass Index Normalized Weakness | 24 (0.2) |
| Absolute Weakness + Body Mass Normalized Weakness | 823 (7.4) |
| Absolute Weakness + Body Mass Index Normalized Weakness | 12 (0.1) |
| Body Mass Normalized Weakness + Body Mass Index Normalized Weakness | 1,348 (12.1) |
| Absolute Weakness + Body Mass Normalized Weakness + Body Mass Index Normalized Weakness | 1,923 (17.3) |
| Future Cognitive Impairment | |
| Not-Weak | 89 (25.3) |
| Absolute Weakness | 23 (6.5) |
| Body Mass Normalized Weakness | 39 (11.1) |
| Body Mass Index Normalized Weakness | 0 (0.0) |
| Absolute Weakness + Body Mass Normalized Weakness | 49 (13.9) |
| Absolute Weakness + Body Mass Index Normalized Weakness | 3 (0.9) |
| Body Mass Normalized Weakness + Body Mass Index Normalized Weakness | 30 (8.5) |
| Absolute Weakness + Body Mass Normalized Weakness + Body Mass Index Normalized Weakness | 119 (33.8) |
Table 3.
Misclassification Characteristics of the Individual Weakness Cut-Points on Future Cognitive Impairment.
| Estimate (%) | 95% Confidence Interval | |
|---|---|---|
| Individual Cut-Points | ||
| Absolute Weakness | ||
| Sensitivity | 67.1 | 66.4, 67.7 |
| Specificity | 60.5 | 57.0, 64.0 |
| Positive Predictive Value | 97.7 | 97.5, 98.0 |
| Negative Predictive Value | 6.6 | 6.0, 7.1 |
| Body Mass Index Normalized Weakness | ||
| Sensitivity | 66.7 | 66.0, 67.3 |
| Specificity | 48.6 | 45.1, 52.2 |
| Positive Predictive Value | 97.1 | 96.8, 97.4 |
| Negative Predictive Value | 5.3 | 4.7, 5.8 |
| Body Mass Normalized Weakness | ||
| Sensitivity | 43.6 | 42.9, 44.3 |
| Specificity | 69.3 | 65.9, 72.5 |
| Positive Predictive Value | 97.3 | 97.0, 97.7 |
| Negative Predictive Value | 4.5 | 4.1, 4.8 |
| Compounding Categories | ||
| ≥1 Cut-Point | ||
| Sensitivity | 39.8 | 39.1, 40.5 |
| Specificity | 76.6 | 73.5, 79.6 |
| Positive Predictive Value | 97.8 | 97.4, 98.1 |
| Negative Predictive Value | 4.6 | 4.2, 5.0 |
| ≥2 Cut-Points | ||
| Sensitivity | 58.4 | 57.8, 59.1 |
| Specificity | 61.7 | 58.2, 65.1 |
| Positive Predictive Value | 97.5 | 97.2, 97.8 |
| Negative Predictive Value | 5.4 | 4.9, 5.8 |
| ≥3 Cut-Points | ||
| Sensitivity | 79.1 | 78.5, 79.7 |
| Specificity | 40.1 | 36.6, 43.6 |
| Positive Predictive Value | 97.1 | 96.9, 97.4 |
| Negative Predictive Value | 6.8 | 6.1, 7.6 |
Table 4 shows the results for the associations of the specific weakness cut-points on future cognitive impairment. Persons considered below the absolute weakness cut-points had 1.62 (CI: 1.34–1.96) greater odds for future cognitive impairment. However, no significant associations were observed for persons under the BMI normalized (odds ratio (OR): 1.17; CI: 0.95–1.43) or body mass normalized weakness cut-points (OR: 1.12; CI: 0.91–1.36). Appendix 4 presents the results of the sensitivity analyses for the associations of the individual weakness cut-points on future cognitive impairment limited to each participant’s first wave. The results for the associations of compounding weakness on future cognitive impairment are presented in Table 5. Older Americans below all 3-weakness cut-points had 1.47 (CI: 1.15–1.88) greater odds for future cognitive impairment, but no significant associations were observed for those considered weak under 1 (OR: 1.08; CI: 0.83–1.42) or 2 (OR: 1.19; CI: 0.91–1.55) cut-points. Appendix 5 shows the results for the associations of the individual weakness cut-points on future cognitive impairment limited to each person’s first wave.
Table 4.
Results for the Associations of the Individual Weakness Cut-Points on Future Cognitive Impairment.
| Crude | Fully-Adjusted | |||
|---|---|---|---|---|
| Odds Ratio | 95% Confidence Interval | Odds Ratio | 95% Confidence Interval | |
| Absolute Weakness† | 2.88 | 2.44, 3.39 | 1.62 | 1.34, 1.96 |
| Body Mass Index Normalized Weakness‡ | 1.55 | 1.31, 1.83 | 1.17 | 0.95, 1.43 |
| Body Mass Normalized Weakness¥ | 1.53 | 1.28, 1.81 | 1.12 | 0.91, 1.36 |
Reference: no absolute weakness;
Reference: no body mass index normalized weakness;
Reference: no body mass normalized weakness.
Note: Crude models controlled for cognitive impairment at current wave and follow-up years. Fully-adjusted models controlled for cognitive impairment at current wave, follow-up years, sex, race, multimorbidity, obesity, age, smoking status, social engagement, self-rated health, depressive status, moderate-to-vigorous physical activity participation, basic self-care limitation, and educational achievement.
Table 5.
Results for the Associations of the Compounding Weakness Cut-Points on Future Cognitive Impairment.
| Crude | Fully-Adjusted | |||
|---|---|---|---|---|
| Odds Ratio | 95% Confidence Interval | Odds Ratio | 95% Confidence Interval | |
| 1 Weakness Category | 1.28 | 0.99, 1.64 | 1.08 | 0.83, 1.42 |
| 2 Weakness Categories | 1.54 | 1.21, 1.97 | 1.19 | 0.91, 1.55 |
| 3 Weakness Categories | 2.68 | 2.18, 3.29 | 1.47 | 1.15, 1.88 |
Reference: 0 weakness categories.
Note: Crude models controlled for cognitive impairment at current wave and follow-up years. Fully-adjusted models controlled for cognitive impairment at current wave, follow-up years, sex, race, multimorbidity, obesity, age, smoking status, social engagement, self-rated health, depressive status, moderate-to-vigorous physical activity participation, basic self-care limitation, and educational achievement.
DISCUSSION
The principal results of this investigation revealed that older Americans considered weak with only the absolute cut-point had 62% greater odds for future cognitive impairment, whereas no significant associations were observed for the body mass index or body mass normalized cut-points. Further, those below all 3 absolute and normalized weakness cut-points had 47% greater odds for future cognitive impairment; however, no significant associations were observed for persons below 1 or 2 cut-points. Our findings suggest that while each of the absolute and normalized cut-points are meant to be equivalent for how weakness is categorized, their associations with future cognitive impairment differ.
Body size influences strength capacity;28 however, absolute handgrip strength does not directly account for body dimension characteristics such as body mass index and body mass. The creation of body mass index and body mass normalized weakness cut-points presents healthcare providers with a potentially more precise measure of strength capacity that remains feasible. Normalizing handgrip strength to body size could be especially useful for certain health outcomes that are linked to contributing characteristics of chronic cardiometabolic morbidities.10,29 While directly adjusting for body dimension in strength capacity measurements may help to improve precision, normalizing strength when, for example, body mass is high should be met with caution. The relationship between body mass and strength is not linear, and high body mass may skew how handgrip strength is normalized because the proportion of strength to body mass will drive the denominator of the equation, and thereby artificially deflate normalized handgrip strength measures.30
Although persons below the individual absolute and normalized weakness cut-points had differential associations with future cognitive impairment in our study, these cut-points could be used aggregately for providing a more comprehensive weakness assessment. Single cut-points are often utilized for determining weakness status,31 but the potential of misclassification for weakness and other age-related health conditions may exist. Moreover, consideration should be given to how different cut-points influence weakness cases,13 such that higher cut-points may elevate weakness cases, thereby inflating the risk for false positives. Examining weakness in this regard alongside other physical measures may also have utility in cognitive prognosis.32
Indeed, our findings showed that persons only below the absolute weakness cut-points had greater odds for future cognitive impairment, and that absolute weakness could be key in driving the association between those below all 3 cut-points and future cognitive impairment. Accordingly, absolute weakness could be emerging as a simple assessment method of strength capacity to assess cognitive impairment risk, and we overall recommend the absolute weakness cut-points for helping to determine future cognitive impairment. However, weakness is associated with several adverse health outcomes aside from cognitive dysfunction.33 Future research should seek to examine methods for separating weakness cases that are specific to cognitive impairment from other age-related health conditions. Weakness cut-points that are unique to cognitive dysfunction in this regard may help to elevate handgrip strength as a single assessment method for cognitive impairment. Evaluating other muscle function attributes with electronic handgrip dynamometry and accelerometry may also provide novel insights.34
Some limitations should be noted. Handgrip strength measurements alternated completion at each wave because they were part of the enhanced face-to-face interviews. Proxy respondents may have had low cognitive functioning, but their physical measurements such as handgrip strength may not have been collected. Participants were included in our analyses if they had at least two observations, and persons that died shortly after being interviewed may have had sudden declines in handgrip strength and cognitive function. Genetic-related data such as Apolipoprotein E4 were not publicly available. While the Telephone Interview of Cognitive Status is a well-utilized assessment tool of cognitive screening for population-based studies such as Health and Retirement Study, more precise assessments may provide additional insights into cognitive domains most sensitive to handgrip strength.
Conclusions and Implications
Our findings revealed that older Americans considered weak under the absolute weakness cut-point had greater odds for future cognitive impairment. Persons considered weak under all 3 cut-points also had greater odds for future cognitive impairment. These results indicate that the absolute and normalized weakness cut-points have different predictive utility for cognitive impairment, and the use of the 3 cut-points collectively may help to better identify weakness and future cognitive impairment, which will possibly inform clinical conversations with older adults in post-acute and long-term care. Acknowledgement for how each weakness cut-point is used should be considered, such that low burden interventions (e.g., physical activity) may allow for greater weakness false positives, as there are still benefits for having older adults participate in such interventions when they may not actually have weakness. However, high burden interventions may benefit from cut-points with little false positive because impact could be poor. Advancing our methods for determining weakness may help our rapidly growing older adult population retain independence, preserve health, and extend longevity.
Funding:
Research reported in this publication was supported by the National Institute on Aging of the National Institutes of Health under Award Numbers R15AG072348 (to RM) and R01AG075117 (to JMH). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. AES is supported by a Henry Brodaty Dementia Australia Mid-Career Fellowship.
Appendix 1. Participant Breakdown for Wave in Which Handgrip Strength and Future Cognitive Impairment Status Were Measured.
| A) | Outcome Assessed | ||||||
|---|---|---|---|---|---|---|---|
| 2008 wave | 2010 wave | 2012 wave | 2014 wave | 2016 wave | 2018 wave | ||
| First HGS Measured | 2006 wave | 3,788 | 30 | 8 | 3 | 0 | 0 |
| 2008 wave | - | 3,602 | 51 | 11 | 3 | 0 | |
| 2010 wave | - | - | 1,122 | 11 | 1 | 0 | |
| 2012 wave | - | - | - | 952 | 15 | 3 | |
| 2014 wave | - | - | - | - | 906 | 16 | |
| 2016 wave | - | - | - | - | - | 594 | |
| B) | Outcome Assessed | ||||||
| 2008 wave | 2010 wave | 2012 wave | 2014 wave | 2016 wave | 2018 wave | ||
| HGS Measured | 2006 wave | 3,788 | 30 | 8 | 3 | 0 | 0 |
| 2008 wave | - | 3,602 | 51 | 11 | 3 | 0 | |
| 2010 wave | - | - | 3,607 | 25 | 1 | 0 | |
| 2012 wave | - | - | - | 3,423 | 35 | 6 | |
| 2014 wave | - | - | - | - | 3,417 | 26 | |
| 2016 wave | - | - | - | - | - | 2,300 | |
Note: A) first HGS measured and subsequent cognitive function outcomes assessed; B) all waves where HGS was measured and subsequent cognitive function outcomes were assessed. HGS=handgrip strength.
Appendix 2. Patterns of Random Half Sample Alternating Wave Participation for Handgrip Strength and Cognitive Impairment in the Generalized Estimating Equations.
| B) | |||
|---|---|---|---|
| 2006 Wave | 2010 Wave | 2014 Wave | Frequency (%) |
| X | X | X | 1,574 (14.2) |
| X | X | O | 925 (8.3) |
| X | O | X | 175 (1.6) |
| X | O | O | 1,155 (10.4) |
| O | X | X | 772 (6.9) |
| O | X | O | 362 (3.3) |
| O | O | X | 922 (8.3) |
| B) | |||
| 2008 Wave | 2012 Wave | 2016 Wave | Frequency (%) |
| X | X | X | 1,173 (10.6) |
| X | X | O | 1,321 (11.9) |
| X | O | X | 95 (0.9) |
| X | O | O | 1,078 (9.7) |
| O | X | X | 438 (3.9) |
| O | X | O | 532 (4.8) |
| O | O | X | 594 (5.3) |
Note: A) first random half sample; B) second random half sample. X=participated; O=did not participate.
Appendix 3. Means and 95% Confidence Intervals for the Baseline Descriptive Characteristics of the Participants by Weakness Status from the Individual Cut-Points.
| Absolute Weakness (n=3,104) | No Absolute Weakness (n=8,012) | BMI Normalized Weakness (n=3,307) | No BMI Normalized Weakness (n=7,809) | Body Mass Normalized Weakness (n=5,813) | No Body Mass Normalized Weakness (n=5,303) | |
|---|---|---|---|---|---|---|
| Age (years) | 75.7 (75.5, 76.0) | 71.0 (70.9, 71.1) | 73.7 (73.5, 74.0) | 71.7 (71.6, 71.9) | 73.3 (73.1, 73.5) | 71.3 (71.1, 71.4) |
| Body Mass Index (kg/m2) | 27.3 (27.1, 27.5) | 28.5 (28.4, 28.6) | 31.7 (31.4, 31.9) | 26.7 (26.6, 26.8) | 30.5 (30.3, 30.6) | 25.6 (25.5, 25.7) |
| Social Engagement | 1.0 (1.0, 1.1) | 1.3 (1.3, 1.4) | 1.0 (1.0, 1.1) | 1.3 (1.3, 1.4) | 1.1 (1.1, 1.2) | 1.3 (1.3, 1.4) |
| Female | 50.4 (48.7, 52.2) | 61.0 (59.9, 62.1) | 78.4 (77.0, 79.8) | 49.4 (48.3, 50.6) | 60.0 (58.8, 61.3) | 55.9 (54.5, 57.2) |
| White | 82.0 (80.6, 83.3) | 81.2 (80.3, 82.0) | 79.1 (77.7, 80.5) | 82.4 (81.5, 83.2) | 80.1 (79.1, 81.1) | 82.8 (81.8, 83.8) |
| Multimorbidity | 76.1 (74.6, 77.6) | 65.3 (64.3, 66.4) | 81.1 (79.8, 82.5) | 62.9 (61.9, 64.0) | 77.7 (76.6, 78.7) | 58.1 (56.8, 59.5) |
| Cigarette Smoking Status | ||||||
| Current Smoker | 8.6 (7.6, 9.6) | 10.7 (10.6, 11.4) | 6.9 (6.0, 7.7) | 11.5 (10.8, 12.2) | 8.0 (7.2, 8.6) | 12.5 (11.6, 13.4) |
| Previous Smoker | 48.5 (46.7, 50.2) | 45.3 (44.1, 46.3) | 42.9 (41.2, 44.6) | 47.6 (46.4, 48.6) | 47.6 (46.2, 48.8) | 44.7 (43.3, 46.0) |
| Never Smoked | 42.9 (41.1, 44.5) | 44.0 (42.9, 45.0) | 50.2 (48.4, 51.9) | 40.9 (39.8, 41.9) | 44.4 (43.1, 45.7) | 42.8 (41.4, 44.1) |
| Self-Rated Health | ||||||
| Excellent | 5.9 (5.1, 6.7) | 10.4 (9.7, 11.0) | 4.2 (3.4, 4.8) | 11.2 (10.5, 11.9) | 5.9 (5.2, 6.4) | 12.7 (11.8, 13.5) |
| Very Good | 24.8 (23.2, 26.2) | 34.0 (32.9, 34.9) | 23.4 (21.9, 24.8) | 34.7 (33.6, 35.8) | 26.4 (25.3, 27.5) | 36.8 (35.4, 38.0) |
| Good | 32.7 (31.0, 34.3) | 33.9 (32.9, 34.9) | 33.8 (32.1, 35.4) | 33.5 (32.4, 34.5) | 34.7 (33.4, 35.9) | 32.4 (31.1, 33.6) |
| Fair | 27.0 (25.4, 28.5) | 17.6 (16.7, 18.4) | 28.6 (27.0, 30.1) | 16.7 (15.8, 17.5) | 25.0 (23.8, 26.0) | 15.0 (14.0, 15.9) |
| Poor | 9.6 (8.5, 10.6) | 4.1 (3.7, 4.5) | 10.0 (8.9, 11.0) | 3.9 (3.4, 4.2) | 8.0 (7.3, 8.7) | 3.1 (2.6, 3.6) |
| Depressed | 23.1 (21.6, 24.6) | 16.3 (15.4, 17.1) | 26.1 (24.6, 27.6) | 14.8 (14.0, 15.6) | 21.5 (20.5, 22.6) | 14.5 (13.5, 15.4) |
| MVPA Participation | 49.6 (47.8, 51.3) | 60.6 (59.5, 61.6) | 43.7 (42.0, 45.4) | 63.3 (62.2, 64.4) | 49.0 (47.7, 50.3) | 66.8 (65.6, 68.1) |
| ADL Limitation | 25.0 (23.5, 26.5) | 11.9 (11.2, 12.6) | 28.1 (26.5, 29.6) | 10.3 (9.6, 10.9) | 22.1 (21.0, 23.2) | 8.4 (7.6, 9.1) |
| Obesity | 27.3 (25.7, 28.8) | 34.1 (33.1, 35.1) | 56.3 (54.6, 58.0) | 22.0 (21.1, 22.9) | 49.1 (47.8, 50.4) | 13.6 (12.7, 14.6) |
| High School Graduate or Above | 74.3 (72.8, 75.8) | 81.2 (80.3, 82.0) | 74.9 (73.4, 76.3) | 81.1 (80.3, 82.0) | 77.4 (76.3, 78.5) | 81.3 (80.2, 82.3) |
| Follow-Up Years | 2.1 (2.1, 2.2) | 2.1 (2.1, 2.2) | 2.1 (2.1, 2.2) | 2.1 (2.1, 2.2) | 2.1 (2.1, 2.2) | 2.1 (2.1, 2.2) |
| Cognitive Impairment | 2.5 (1.9, 3.1) | 1.0 (0.8, 1.3) | 2.5 (2.0, 3.0) | 1.0 (0.8, 1.2) | 2.0 (1.6, 2.3) | 0.9 (0.6, 1.1) |
| Cognitive Impairment at Next Wave | 6.2 (5.4, 7.1) | 1.9 (1.6, 2.2) | 4.6 (3.8, 5.3) | 2.5 (2.2, 2.9) | 4.0 (3.5, 4.5) | 2.1 (1.7, 2.5) |
Note: ADL=activities of daily living, MVPA=moderate-to-vigorous physical activity.
Appendix 4. Results of the Sensitivity Analyses for the Associations of the Individual Weakness Cut-Points on Future Cognitive Impairment Limited to Each Participant’s First Wave.
| Odds Ratio | 95% Confidence Interval | |
|---|---|---|
| Absolute Weakness† | 1.66 | 1.27, 2.18 |
| Body Mass Index Normalized Weakness‡ | 1.19 | 0.89, 1.60 |
| Body Mass Normalized Weakness¥ | 1.12 | 0.91, 1.36 |
Reference: no absolute weakness;
Reference: no body mass index normalized weakness;
Reference: no body mass normalized weakness.
Note: Models controlled for cognitive impairment at current wave, follow-up years, sex, race, multimorbidity, obesity, age, smoking status, social engagement, self-rated health, depressive status, moderate-to-vigorous physical activity participation, basic self-care limitation, and educational achievement.
Appendix 5. Results of the Sensitivity Analyses for the Associations of the Individual Weakness Cut-points on Future Cognitive Impairment Limited to Each Participant’s First Wave.
| Odds Ratio | 95% Confidence Interval | |
|---|---|---|
| 1 Weakness Category | 1.31 | 0.90, 1.90 |
| 2 Weakness Categories | 1.32 | 0.90, 1.92 |
| 3 Weakness Categories | 1.66 | 1.17, 2.35 |
Reference: 0 weakness categories.
Note: Model controlled for cognitive impairment at current wave, follow-up years, sex, race, multimorbidity, obesity, age, smoking status, social engagement, self-rated health, depressive status, moderate-to-vigorous physical activity participation, basic self-care limitation, and educational achievement.
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
Conflicts of Interest: None.
REFERENCES
- 1.Beaudart C, Rolland Y, Cruz-Jentoft AJ, et al. Assessment of Muscle Function and Physical Performance in Daily Clinical Practice : A position paper endorsed by the European Society for Clinical and Economic Aspects of Osteoporosis, Osteoarthritis and Musculoskeletal Diseases (ESCEO). Calcif Tissue Int. 2019;105(1):1–14. doi: 10.1007/s00223-019-00545-w. [DOI] [PubMed] [Google Scholar]
- 2.McGrath RP, Kraemer WJ, Snih SA, Peterson MD. Handgrip Strength and Health in Aging Adults. Sports Med. 2018;48(9):1993–2000. doi: 10.1007/s40279-018-0952-y. [DOI] [PubMed] [Google Scholar]
- 3.Bohannon RW. Grip Strength: An Indispensable Biomarker For Older Adults. Clin Interv Aging. 2019;14:1681–1691. doi: 10.2147/CIA.S194543. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Zammit AR, Robitaille A, Piccinin AM, Muniz-Terrera G, Hofer SM. Associations Between Aging-Related Changes in Grip Strength and Cognitive Function in Older Adults: A Systematic Review. J Gerontol A Biol Sci Med Sci. 2019;74(4):519–527. doi: 10.1093/gerona/gly046. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Shaughnessy KA, Hackney KJ, Clark BC, et al. A Narrative Review of Handgrip Strength and Cognitive Functioning: Bringing a New Characteristic to Muscle Memory. J Alzheimers Dis. 2020;73(4):1265–1278. doi: 10.3233/JAD-190856. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Duchowny KA, Ackley SF, Brenowitz WD, et al. Associations Between Handgrip Strength and Dementia Risk, Cognition, and Neuroimaging Outcomes in the UK Biobank Cohort Study. JAMA Netw Open. 2022;5(6):e2218314. doi: 10.1001/jamanetworkopen.2022.18314. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Clark BC, Carson RG. Sarcopenia and Neuroscience: Learning to Communicate. J Gerontol A Biol Sci Med Sci. 2021;76(10):1882–1890. doi: 10.1093/gerona/glab098. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Carson RG. Get a grip: individual variations in grip strength are a marker of brain health. Neurobiol Aging. 2018;71:189–222. doi: 10.1016/j.neurobiolaging.2018.07.023. [DOI] [PubMed] [Google Scholar]
- 9.Garcia-Cifuentes E, Botero-Rodríguez F, Ramirez Velandia F, et al. Muscular Function as an Alternative to Identify Cognitive Impairment: A Secondary Analysis From SABE Colombia. Front Neurol. 2022;13:695253. doi: 10.3389/fneur.2022.695253. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Whitney DG, Peterson MD. The Association Between Differing Grip Strength Measures and Mortality and Cerebrovascular Event in Older Adults: National Health and Aging Trends Study. Front Physiol. 2019;9:1871. doi: 10.3389/fphys.2018.01871. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Cawthon PM, Travison TG, Manini TM, et al. Establishing the Link Between Lean Mass and Grip Strength Cut Points With Mobility Disability and Other Health Outcomes: Proceedings of the Sarcopenia Definition and Outcomes Consortium Conference. J Gerontol A Biol Sci Med Sci. 2020;75(7):1317–1323. doi: 10.1093/gerona/glz081. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Nevill AM, Tomkinson GR, Lang JJ, Wutz W, Myers TD. How Should Adult Handgrip Strength Be Normalized? Allometry Reveals New Insights and Associated Reference Curves. Med Sci Sports Exerc. 2022;54(1):162–168. doi: 10.1249/MSS.0000000000002771. [DOI] [PubMed] [Google Scholar]
- 13.McGrath R, Cawthon PM, Clark BC, Fielding RA, Lang JJ, Tomkinson GR. Recommendations for Reducing Heterogeneity in Handgrip Strength Protocols. J Frailty Aging. 2022;11(2):143–150. doi: 10.14283/jfa.2022.21. [DOI] [PubMed] [Google Scholar]
- 14.Manini TM, Patel SM, Newman AB, et al. Identification of Sarcopenia Components That Discriminate Slow Walking Speed: A Pooled Data Analysis. J Am Geriatr Soc. 2020;68(7):1419–1428. doi: 10.1111/jgs.16524. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Cawthon PM, Manini T, Patel SM, et al. Putative Cut-Points in Sarcopenia Components and Incident Adverse Health Outcomes: An SDOC Analysis. J Am Geriatr Soc. 2020;68(7):1429–1437. doi: 10.1111/jgs.16517. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Patel SM, Duchowny KA, Kiel DP, et al. Sarcopenia Definition & Outcomes Consortium Defined Low Grip Strength in Two Cross-Sectional, Population-Based Cohorts. J Am Geriatr Soc. 2020;68(7):1438–1444. doi: 10.1111/jgs.16419. [DOI] [PubMed] [Google Scholar]
- 17.Cui M, Zhang S, Liu Y, Gang X, Wang G. Grip Strength and the Risk of Cognitive Decline and Dementia: A Systematic Review and Meta-Analysis of Longitudinal Cohort Studies. Front Aging Neurosci. 2021;13:625551. doi: 10.3389/fnagi.2021.625551. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Sonnega A, Faul JD, Ofstedal MB, Langa KM, Phillips JW, Weir DR. Cohort Profile: the Health and Retirement Study (HRS). Int J Epidemiol. 2014;43(2):576–585. doi: 10.1093/ije/dyu067. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Sample Sizes and Response Rates. Health and Retirement Study. https://hrs.isr.umich.edu/sites/default/files/biblio/ResponseRates_2017.pdf. Accessed May 5, 2023.
- 20.Health and Retirement Study. HRS Data Book. https://hrs.isr.umich.edu/about/data-book. Accessed May 5, 2023.
- 21.Fisher GG, Ryan LH. Overview of the Health and Retirement Study and Introduction to the Special Issue. Work Aging Retire. 2018;4(1):1–9. doi: 10.1093/workar/wax032. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Plassman BL, Newman TT, Welsh KA, et al. Application in epidemiological and longitudinal studies. Cogn Behav Neurol. 1994;7:235–41. [Google Scholar]
- 23.Langa KM, Larson EB, Karlawish JH, et al. Trends in the prevalence and mortality of cognitive impairment in the United States: is there evidence of a compression of cognitive morbidity?. Alzheimers Dement. 2008;4(2):134–144. doi: 10.1016/j.jalz.2008.01.001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Crimmins E, Guyer H, Langa K, Ofstedal MB, Wallace R, Weir D. Documentation of Physical Measures, Anthropometrics and Blood Pressure in the Health and Retirement Study. https://hrs.isr.umich.edu/sites/default/files/biblio/dr-011.pdf. Accessed May 5, 2023.
- 25.Turvey CL, Wallace RB, Herzog R. A revised CES-D measure of depressive symptoms and a DSM-based measure of major depressive episodes in the elderly. Int Psychogeriatr. 1999;11(2):139–148. doi: 10.1017/s1041610299005694. [DOI] [PubMed] [Google Scholar]
- 26.Feng X, Croteau K, Kolt GS, Astell-Burt T. Does retirement mean more physical activity? A longitudinal study. BMC Public Health. 2016;16:605. Published 2016 Jul 20. doi: 10.1186/s12889-016-3253-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Howrey B, Avila JC, Downer B, Wong R. Social Engagement and Cognitive Function of Older Adults in Mexico and the United States: How Universal Is the Interdependence in Couples?. J Gerontol B Psychol Sci Soc Sci. 2021;76(Suppl 1):S41–S50. doi: 10.1093/geronb/gbaa025. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Jaric S. Muscle strength testing: use of normalisation for body size. Sports Med. 2002;32(10):615–631. doi: 10.2165/00007256-200232100-00002. [DOI] [PubMed] [Google Scholar]
- 29.Brown EC, Buchan DS, Madi SA, Gordon BN, Drignei D. Grip Strength Cut Points for Diabetes Risk Among Apparently Healthy U.S. Adults. Am J Prev Med. 2020;58(6):757–765. doi: 10.1016/j.amepre.2020.01.016. [DOI] [PubMed] [Google Scholar]
- 30.McGrath R. Comparing absolute handgrip strength and handgrip strength normalized to body weight in aging adults. Aging Clin Exp Res. 2019;31(12):1851–1853. doi: 10.1007/s40520-019-01126-5. [DOI] [PubMed] [Google Scholar]
- 31.Alley DE, Shardell MD, Peters KW, et al. Grip strength cutpoints for the identification of clinically relevant weakness. J Gerontol A Biol Sci Med Sci. 2014;69(5):559–566. doi: 10.1093/gerona/glu011. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Orchard SG, Polekhina G, Ryan J, et al. Combination of gait speed and grip strength to predict cognitive decline and dementia. Alzheimers Dement (Amst). 2022;14(1):e12353. doi: 10.1002/dad2.12353. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.McGrath R, Johnson N, Klawitter L, et al. What are the association patterns between handgrip strength and adverse health conditions? A topical review. SAGE Open Med. 2020;8:2050312120910358. doi: 10.1177/2050312120910358. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.McGrath R, Tomkinson GR, Clark BC, et al. Assessing Additional Characteristics of Muscle Function With Digital Handgrip Dynamometry and Accelerometry: Framework for a Novel Handgrip Strength Protocol. J Am Med Dir Assoc. 2021;22(11):2313–2318. doi: 10.1016/j.jamda.2021.05.033. [DOI] [PMC free article] [PubMed] [Google Scholar]