- Email: [email protected]
Motor functioning differentially predicts mortality in men and women
Archives of Gerontology and Geriatrics 72 (2017) 6–11
Contents lists available at ScienceDirect
Archives of Gerontology and Geriatrics journal homepage: www.elsevier.com/locate/archger
Full Length Article
Motor functioning differentially predicts mortality in men and women a,⁎
b
a,c
d
Marie Ernsth Bravell , Deborah Finkel , Anna Dahl Aslan , Chandra A. Reynolds , Jenny Hallgrena, Nancy L. Pedersenc,e
MARK
a
Institute of Gerontology, School of Health and Welfare, Jönköping University, Sweden Department of Psychology, Indiana University Southeast, USA Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm, Sweden d Department of Psychology, University of California, Riverside, USA e Department of Psychology, University of Southern California, Los Angeles, USA b c
A R T I C L E I N F O
A B S T R A C T
Keywords: Motor function Gender differences Survival
Introduction: Research indicates gender differences in functional performance at advanced ages, but little is known about their impact on longevity for men and women. Objective: To derive a set of motor function factors from a battery of functional performance measures and examine their associations with mortality, incorporating possible gender interactions. Method: Analyses were performed on the longitudinal Swedish Adoption/Twin Study of Aging (SATSA) including twenty-four assessments of motor function up to six times over a 19-year period. Three motor factors were derived from several factor analyses; fine motor, balance/upper strength, and flexibility. A latent growth curve model was used to capture longitudinal age changes in the motor factors and generated estimates of intercept at age 70 (I), rates of change before (S1) and after age 70 (S2) for each factor. Cox regression models were used to determine how gender in interaction with the motor factors was related to mortality. Results: Females demonstrated lower functional performance in all motor functions relative to men. Cox regression survival analyses demonstrated that both balance/upper strength, and fine motor function were significantly related to mortality. Gender specific analyses revealed that this was true for women only. For men, none of the motor factors were related to mortality. Conclusion: Women demonstrated more difficulties in all functioning facets, and only among women were motor functioning (balance/upper strength and fine motor function) associated with mortality. These results provide evidence for the importance of considering motor functioning, and foremost observed gender differences when planning for individualized treatment and rehabilitation.
1. Introduction Studies of disability and function among older individuals are common, due to the changes that usually appear with advanced age and the effect these physical changes have on daily living. Disability is typically measured by self-reports of function in Personal Activities in Daily Life (PADL) and/or function in Instrumental Activities in Daily Life (IADL) (Fauth, Zarit, & Malmberg, 2008). Self-reports of disability are partly confounded by social and psychological health factors such as gender roles, labor, and interests (Larsson & Thorslund, 2002). Assessments of functional impairments via observed performance of motor function are considered to more accurately capture true physiological impairments (Avlund, 1999; Guralnik et al., 1994). The two most commonly examined measures are grip strength and walking speed (Cooper et al., 2011). Studies comparing self-reported disability and ⁎
measured functional impairments show modest correlations, ranging from 0.17 to 0.54 (Ernsth Bravell, Zarit, & Johansson, 2011; Farag et al., 2012). Therefore, studies using self-reported versus assessed physical function will not necessarily yield the same results. Nevertheless, both self-reported ADL and observed functional ability decline with age, and they also demonstrate associations with longevity (Cooper, Kuh, & Hardy, 2010; Gallucci, Ongaro, Amici, & Regini, 2009; Hirsch, Bůžková, Robbins, Patel, & Newman, 2012; Stineman et al., 2012; Taekema, Gussekloo, Westendorp, de Craen, & Maier, 2012; Tiainen, Luukkaala, Hervonen, & Jylhä, 2013; White et al., 2013). In addition, there are documented gender differences in disability in late life, where women tend to have more problems at the same time as they live longer (e.g. Avlund, Vass, & Hendriksen, 2003; Crimmins, Kim, & Sole-Auro, 2010) and have poorer physical function (Daly et al., 2013; Orfila et al., 2006). Most current studies focus on single measurements such as grip
Corresponding author. E-mail address: [email protected] (M.E. Bravell).
http://dx.doi.org/10.1016/j.archger.2017.05.001 Received 16 September 2016; Received in revised form 24 February 2017; Accepted 1 May 2017 Available online 02 May 2017 0167-4943/ © 2017 Elsevier B.V. All rights reserved.
Archives of Gerontology and Geriatrics 72 (2017) 6–11
M.E. Bravell et al.
2.2. Measures
strength (Oksuzyan et al., 2010; Wu et al., 2012) or gait speed (Studenski et al., 2011). Few studies have considered gender differences in the relationship between mortality and multiple measures of physical motor functioning tapping different functional modalities (Cooper et al., 2011). The aim of the current study is therefore to evaluate a set of performance-based motor function factors and to explore their relationship to mortality, incorporating possible gender interactions.
Twenty-four measures of functional ability were collected at each IPT. They are listed in Table 2. The functional ability tasks were timed but motor functioning measured in seconds was skewed (most individuals perform the tasks successfully into their mid-60s). Previous studies have found problems using time to complete a task due to lack of variation (e.g. Ernsth Bravell et al., 2011). Performance on the 24 motor functioning tasks was also categorized by the nurses administrating the testing as without difficulty (1), with some difficulty (2), or impossible (3). Thus, even when a participant could, for example, balance for 10 s with eyes closed (achieving the maximum score), the nurses were able to report whether the participant demonstrated any difficulty with the balance task, regardless of time. The analyses described below made use of the qualitative nurse ratings for each of the 24 tasks.
2. Method and material 2.1. Sample As a part of the longitudinal Swedish Adoption/Twin Study of Aging (SATSA), twenty-four different assessments of motor functioning were collected six times over a 19-year period. Accrual procedures for SATSA have been described previously (Finkel & Pederson, 2004). In-person testing (IPT) took place in a location convenient to the participants and was completed during a single 4-h visit. The second (IPT2) and third (IPT3) waves of in-person testing occurred at three-year intervals. Inperson testing did not occur during wave 4; therefore, the next wave of in-person testing is labeled IPT5 and occurred after a 7-year interval (see Finkel & Pedersen, 2004). Regular 3-year testing continued after IPT5; therefore, the total time span from IPT1 to IPT7 was 19 years. For IPT1, motor function data was available for 645 persons. In addition, respondents were added during the study as they turned 50 years; 97 entered the study at IPT2, another 21 at IPT3, and finally 101 at IPT5. Not surprisingly, there was very little variance in motor function at IPT1; therefore, IPT2 was used as the baseline assessment for this study. Since almost all of the respondents performed fairly well with little variance before the age of 60, which agrees e.g. with Wu et al. (2012), only respondents aged 60 and above at IPT2 were included in the survival analyses. Baseline data (IPT2) on motor function were available from 436 twins ranging in age from 60 to 91 years. From these, 76% participated in three or more IPTs. Table 1 provides a description of the total sample, from which descriptive and factor analyses were performed (total sample). It also describes the sample from which the survival analyses were performed (age 60+).
2.3. Analyses To assess the effects of relatedness (twinship), the sample was divided into two groups: twin A from each pair in one group and twin B from each pair in another group. All analyses were conducted separately in each group to ensure that the results were the same. Only minor differences were found; thus, the descriptive statistics reported (Table 1) are from the full sample to maximize power. Cox regression survival models were applied using STATA/IC 12.1 and the robust sandwich estimation option to control for twinship and thus provide appropriate standard errors. 2.3.1. Factor structure To begin with, several factor analyses were performed in order to create motor factors. Due to skew evident in motor functioning in the first IPTs, where most individuals performed well, the first factor analyses (Principal Component Analysis with Variamax rotation) were performed on the motor function measures from IPT7. The factor analysis converged in three iterations that could be interpreted as; 1. Fine motor ability (explained 33% of variance); 2. Balance and strength motor ability (explained 27% of variance); 3. Flexibility (explained 14% of the variance). The extractions in the factor analysis on IPT7 are based on eigenvalues, and explain a total of 71% of the variance in motor function, see Table 2.
Table 1 Sample characteristics and descriptive Mean (SD). Fine motor N
Men
Women
IPT 2
585
242
343
IPT 2a
436
174
262
IPT3
566
233
333
IPT3a
378
149
229
IPT5
541
213
328
IPT5a
252
90
162
IPT6
447
183
264
IPT6a
185
67
118
IPT7
379
155
224
138
47
91
IPT7
a
Age 66.0 (9.0) 70.2 (6.1) 68.8 (9.2) 73.1 (5.8) 70.6 (10.0) 78.5 (5.1) 72.2 (9.3) 80.5 (4.6) 74.3 (9.0) 83.23 (4.1)
Total
Balance and strength Men
Women
Total
Men
Flexibility Women
**
Men **
Women
2.1
11.1** (2.5) 11.2* (2.5)
2.2 (0.6)
2.2* (0.4)
2.3* (0.6)
11.0 (2.1)
10.5** (1.4) 10.7* (1.5)
2.2* (0.5)
2.3* (0.6)
11.1 (2.1)
10.9* (1.5)
11.2* (2.4)
2.3 (0.6) 2.3 (0.6)
2.2* (0.5)
11.7 (3.7)
11.3* (3.1)
12.0* (4.1)
2.3 (0.8)
2.2** (0.6)
2.3* (0.7) 2.4** (0.9)
12.6 (4.5)
12.3 (4.3)
12.7 (4.6)
2.4 (0.9)
2.4 (0.8)
11.8 (3.3)
11.5 (3.1)
11.9 (3.4)
2.3 (0.7)
2.2*** (0.5)
2.5 (1.0) 2.4*** (0.8)
10.2** (3.3)
12.8 (4.1)
12.7 (4.1)
12.9 (4.1)
2.5 (0.9)
2.3** (0.6)
2.6** (1.0)
9.8 (3.4) 10.5 (3.7)
12.3 (4.2)
12.11 (4.2)
12.5 (4.2)
2.5 (0.9)
2.4 (0.8)
14.2 (5.2)
14.5 (6.2)
14.0 (4.6)
2.7 (1.1)
2.5 (1.0)
2.5 (0.9) 2.9 (1.2)
8.7 (1.4)
8.6 (1.5)
8.7 (1.3)
8.6 (1.4)
8.8 (1.5)
8.8 (1.6)
9.0 (2.5)
8.8 (1.9)
9.4 (2.8)
9.2 (2.6)
9.3 (2.2)
9.1* (1.5)
8.6 (1.2) 8.7 (1.7) 8.7 (1.3) 8.8 (1.3) 9.2 (2.8) 9.5 (2.9) 9.5* (2.6)
9.9 (2.9)
9.5** (1.8)
9.8 (3.4)
9.8 (3.3)
10.7 (3.9)
11.1 (4.3)
7
10.7 (2.0)
10.4
10.9 (2.2)
(0.4)
2.2** (0.6)
2.2 (0.5)
8.6 (1.3)
10.9
Total
(2.3)
8.6 (1.2)
(1.3)
**
Archives of Gerontology and Geriatrics 72 (2017) 6–11
M.E. Bravell et al.
Table 2 Results from factor analysis at the IPT7-sample including all 20 motor function variables (difficulties). Description of the motor function variables included in the factor analysis
Component
Difficulties to: Pour water from one glass to another, dominant hand (Pour water glass, dominant hand) Insert a key into a lock and turn (Insert and turn key) Screw in a light bulb (Screw in light bulb) Pour water from one glass to another, nondominant hand (Pour water glass, nondominant hand Insert a plug into an electrical socket (Put plug in socket) Put coins of various sizes into appropriate slots (Put coins in slot) Pour water from a jug into a glass (Pour water from jug into glass) Dial numbers 1–9 on a rotary phone (Dial a number) Stand up from a chair, arms crossed in front (Chairstand) Lift a drinking glass (Lifting glass) Touch right toes with left hand (Left fingers to right toe) Balance, feet side by side, eyes open (max 10 s) (Stand side by side looking) Lift a kilogram packet (Lift one kg) Time to pick pen off floor from standing position, max 30 s (Pick up pen) Touch left toes with right hand (Right fingers to left toe) Balance time, feet together, arms extended front, eyes closed, max 10 s (Rombergs test looking) Walk 3 m and return, at participant’s own pace (Gait and turn 3m) Touch left earlobe with right hand (Right hand to left earlobe) Touch right earlobe with left hand (Left hand to right earlobe) Stand up from a chair 5 times, arms crossed (Repeat chair stand)
1
2
3
,887 ,887 ,875 ,871 ,870 ,866 ,837 ,805 ,054 ,256 ,162 ,117 ,246 ,175 ,085 ,147 ,257 ,133 ,051 ,072
,063 ,181 ,164 ,233 ,174 ,252 ,113 ,153 ,788 ,764 ,752 ,735 ,732 ,712 ,706 ,641 ,571 ,093 ,156 ,558
-,189 ,158 ,234 -,046 ,292 ,109 -,020 ,340 ,362 -,196 ,133 ,121 -,028 ,350 ,144 ,430 ,540 ,777 ,742 ,574
Measures not included in the final motor factor solution Difficulties to: Feet semi-tandem, eyes open (max 10 s) Walk 10 steps, heel to toe Feet heel to toe, eyes open (max 10 s) Feet heel to toe, eyes closed (max 10 s) Extraction Method: Principal Component Analysis. Rotation Method: Varimax with Kaiser Normalization.
alpha did not increase by removing them from factor 2. There were significant correlations between the motor factors at all IPTs, with a range from 0.39 to 0.58 between fine motor and flexibility; from 0.31 to 0.49 between fine motor and balance/upper strength; and from 0.26 to 0.35 between flexibility and balance/upper strength. The strength of the correlations increased with increasing age.
Factor analyses were then performed for each IPT where the solutions explained between 61% to 69% of variance (IPT6 = 65%, IPT5 = 65%, IPT3 = 65%, IPT2 = 69%, IPT1 = 61%). Factor structures at each IPT were quite similar and a three-factor solution could be imposed at each wave. Additionally, possible age differences in factor structures were examined: at each wave, factor structures were compared for older adults (age ≥60) and younger adults (age < 60). Again, results were similar, but not identical, across age groups and interpretable 3-factor solutions emerged, but consistently with less variance explained in the younger age groups. Measures with sufficient variance (factor loadings > 0.4) were included in the motor factor and measures without sufficient variance (factor loading < 0.4) were excluded. Thus, based on the extracted solutions that were the most repeatable across IPTs, three motor scores were computed by summing the respective items:
2.3.2. Latent growth curve model A latent growth curve model (LGCM) was used to capture longitudinal age changes in the motor factors (McArdle, Prescott, Hamagami, & Horn, 1998; Reynolds, Finkel, McArdle, Gatz, & Pedersen, 2003). The structural model can be considered as a random coefficients model (Bryk & Raudenbush, 1992; Laird & Ware, 1982) where individual regression models are fitted to each subject’s longitudinal profile of data as well as an average model of growth estimated over the entire sample. In other words, the model provides estimation of fixed effects, i.e. fixed population parameters as estimated by the average growth model of the entire sample, and random effects, i.e., individual variation in growth model parameters. The variation in individual regression coefficients from the group model is then available for further analysis. Latent growth curve models take into account missing data by giving more weight to individuals with the most time points. Based on previous analyses (Finkel, Ernsth Bravell, & Pedersen, 2015; Finkel, Ernsth Bravell, & Pedersen, 2016), a growth curve model with two linear slopes was employed: change with age up to age 70 and change with age after age 70. The intercept thus reflects an estimate of motor performance at age 70. The random and fixed effects estimates were obtained using PROC Mixed in SAS 8.0 (SAS Institute, 1999). Empirical Bayes (EB) estimates of intercept and slope values were produced by the PROC Mixed procedure for each participant. EB estimates are weighted estimates based on both the fixed and random
• Fine motor ability, Cronbach’s alpha at IPT7 = 0.94. Best performance = 8; worst performance = 24. • Balance motor ability, Cronbach’s alpha at IPT7 = 0.91. Best performance = 10; worst performance = 30. • Flexibility, Cronbach’s alpha at IPT7 = 0.73. Best performance = 2; worst performance = 6.
Four motor function measures that did not load on any of the factors were excluded from factor construction. Because factor loadings varied somewhat across waves and ages, but were all of significant magnitude (> 0.4), unit weighting was used to calculate composite scale scores. Higher values indicate greater disability. It should also be noted that repeated chair stand, Romberg’s test and walk 3 m loaded highly in both factor 2 (balance and upper strength) and factor 3 (flexibility) in some of the factor analyses, but they were placed in factor 2 to maximize interpretability. Cronbach’s 8
Archives of Gerontology and Geriatrics 72 (2017) 6–11
M.E. Bravell et al. Table 3a Cox Regression Models with age as the time variable and motor factors intercept and slopes as covariates. Robust
95% CI for Exp(B)
Hazard ratio
Std. Error
Lower
Upper
Log pseudolikelihood = −2384.0701 Fine motor (I) Balance/upper strength (I) Flexibility intercept (I)
1.232*** 1.128** 0.991ns
0.057 0.050 0.428
1.125 1.034 0.425
1.348 1.1230 2.313
Log pseudolikelihood = −729.81129 Fine motor (S1) Balance/upper strength (S1) Flexibility (S1)
1.222* 1.872** 0.256ns
0.114 0.342 0.348
1.018 1.308 0.018
1.467 2.680 3.685
Log pseudolikelihood = −1239.4172 Fine motor (S2) Balance/upper strength (S2) Flexibility (S2)
985ns 1.045ns 0.921ns
0.040 0.032 0.150
0.910 0.984 0.668
1.068 1.110 1.268
Note. Motor factors intercepts and slopes index difficulty levels and change, respectively. I = Intercept, S1 = change up to age 70, S2 = change after age 70. * p < 0.05. ** p < 0.01. *** p < 0.001.
Because of the wide range in ages (up to 35 years) at participation in each IPT, an age-based survival analysis was used instead of a time- or wave-based model. That is, time-to-death is measured by age through subtracting the date of death from date of birth. Analyses were performed with SPSS Statistics 21 except the Cox Regression Survival Models that were analyzed using STATA/IC 12.1.
effects from the latent growth model (Bryk & Raudenbush, 1992). EB intercept estimates (I) were generated for all participants. For participants with 2 IPTs at age 70 or less, an EB estimate of slope 1 (S1) was generated. For participants with 2 IPTs at age 70 or more, an EB estimate of slope 2 (S2) was generated. EB estimates were used in the Cox regression survival model as independent variables. 2.3.3. Gender differences Independent sample t-tests were performed to explore cross-sectional gender differences in motor function. Cox regression survival analysis was performed with mortality (event = 1) as the outcome variable, age as the time variable, and gender and the EB estimates of the three motor factors (I, S1 and S2) as independent variables in a first step, and with gender in interaction with the EB estimates of three motor factors (I, S1 and S2) in a second step in order to determine the role of gender and function on mortality. Mortality data were collected from the Swedish Population Registry in February 2016, nine years after IPT7 testing was completed in 2007. By then 337 out of the 436 respondents, relevant for the Cox regression model, had passed away.
Table 3c Separate Cox Regression Models for men and women with age as the time variable and motor factors intercept and slopes as covariates. Robust Men
Log pseudolikelihood = −901.6837 Fine motor function (I) 1.084 Balance/upper strength (I) 1.142 Flexibility (I) 0.772 Log pseudolikelihood = −365.1121 Fine motor function (S1) 1.152 Balance/upper strength (S1) 1.483 Flexibility (S1) 0.103
Table 3b Cox Regression Models with age as the time variable and motor factors intercept and slopes in interaction with gender as covariates. Robust Hazard ratio Log pseudolikelihood = −2375.147 Gender x Fine motor (I) 1.087** Gender x Balance/upper 1.107** strength (I) Gender x Flexibility (I) 0.318** Log pseudolikelihood = −729.7645 Gender x Fine motor (S1) 1.162** Gender x Balance/upper 1.519** strength (S1) Gender x Flexibility (S1) 0.149* Log pseudolikelihood = −1239.2946 Gender x Fine motor (S2) 0.980ns Gender x Balance/upper 1.018ns strength (S2) Gender x Flexibility (S2) 0.904ns
Log pseudolikelihood = −394.3006 Fine motor function (S2) 1.045 Balance/upper strength (S2) 0.994 Flexibility (S2) 1.195
95% CI for Exp(B) Std. Error
Lower
Upper
0.034 0.033
1.022 1.044
1.156 1.175
0.054
0.228
0.446
0.066 0.159
1.039 1.237
1.230 1.866
0.130
0.027
0.822
0.022 0.017
0.937 0.984
1.024 1.053
0.085
0.751
1.087
Hazard ratio
ns ns ns
ns ns ns
ns ns ns
95% CI for Exp(B) Std. Error
Lower
Upper
0.085 0.088 0.508
0.930 0.982 0.212
1.264 1.329 2.804
0.143 0.412 0.246
0.903 0.860 < 0.001
1.471 2.558 10.973
0.068 0.043 0.268
0.919 0.912 0.770
1.188 1.083 1.854
Robust Women
Note. Motor factors intercepts and slopes index difficulty levels and change, respectively. I = Intercept, S1 = change up to age 70, S2 = change after age 70. ***p < 0.001. * p < 0.05. ** p < 0.01.
Hazard ratio
95% CI for Exp(B) Std. Error
Lower
Upper
Log pseudolikelihood = −1190.3985 Fine motor function (I) 1.040*** Balance/upper strength (I) 1.161** Flexibility (I) 1.040 ns
0.098 0.072 0.660
1.130 1.029 0.300
1.517 1.310 3.610
Log pseudolikelihood = −257.6896 Fine motor function (S1) 1.300 ns Balance/upper strength (S1) 2.412*** Flexibility (S1) .160 ns
0.201 0.581 0.355
0.958 1.504 0.002
1.757 3.868 12.408
Log pseudolikelihood = −680.8204 Fine motor function (S2) 0.959ns Balance/upper strength (S2) 1.078ns Flexibility (S2) 0.901ns
0.049 0.042 0.194
0.868 0.997 0.590
1.060 1.163 1.377
Note. Motor factors intercepts and slopes index difficulty levels and change, respectively. I = Intercept, S1 = change up to age 70, S2 = change after age 70. *p < 0.05.
9
Archives of Gerontology and Geriatrics 72 (2017) 6–11
M.E. Bravell et al.
had worse motor function, but motor function was also related to mortality, which may be somewhat difficult to interpret. Diseases such as arthritis and osteoporosis may most certainly contribute to poorer motor function in women, but it is not considered as a life threating disease. They may however cause a cascade of outcomes that lead to mortality, such falls, fractures, and hospitalizations. Results that confirm that men are physically stronger and have fewer disabilities, but have substantially higher mortality and less healthy life style at all ages compared with women, are common (i.e. Cooper et al., 2011; Deeg & Kriegsman, 2003). A number of explanations have been proposed for this male-female health paradox, rooted in biological, social, and psychological interpretations. It may be due to multiple causes that include fundamental biological differences between the sexes such as genetic factors, immune system responses, hormones, and disease patterns (Cooper et al., 2011). In a twin analysis of the motor factors created with these data, evidence for significant genetic and environmental influences on trajectories was found. Moreover, genetic influences on longitudinal changes in the balance/upper strength factor after age 70 were significantly greater for women than for men (Finkel et al., 2016). Other proposed explanations are behavioral gender differences such as risk-taking and reluctance to seek and comply with medical treatment. However, Crimmins et al. (2010) found that gender differences were less pronounced in abilities to provide self-care, even if females demonstrated more problems in IADL. The construction of the motor factors deserves mention. Even though the three factors were somewhat surprising in their construction, similar factors emerged from each IPT, even if less variance could be explained at IPT2 and IPT3. Also, these factors are similar to previous studies (Ernsth Bravell et al., 2011; Leitsch, 2000) but not completely identical as the measures vary across different studies. The surprising aspect from this study was that only two motor items loaded highly on the flexibility factor. One would assume that e.g. “pick up pen” or “reach toes with fingers” should be in that factor. Instead, it was the repeated chair-stand, walk 3 m and Romberg’s test that cross-loaded in some of the factor analyses. We chose to add those items into balance/upper strength, both due to better interpretation but also due to the fact that they regularly loaded the highest on these factors. Furthermore, Cronbach’s alpha was not improved by removing them. Some of the expected balance motor items (e.g. feet heel to toe; feet semi-tandem; walk 10 steps) did not load consistently over the IPTs and they were therefore excluded when constructing the motor factors.
2.4. Ethical considerations The SATSA-study was approved by Ethical committee at Karolinska Institutet and the Regional Ethics Review Board in Stockholm. 3. Results 3.1. Description of motor function Cross-sectional gender differences were found at early IPTs in balance/upper strength and flexibility, and in IPT6 in fine motor function (Table 1). Females had more problems at each of these IPTs. 3.2. Survival analysis Cox regressions with the empirical Bayes (EB) estimates were used to determine whether motor functions, and motor function in interaction with gender, were related to survival (Tables 3a and 3b). The patterns differed across the estimates where intercept (I) and slope 1 (S1) for fine motor function and balance/upper strength was related to mortality in the total sample (Table 3a). The same patterns were found when motor factors were evaluated in an interaction with gender (Table 3b), with the addition of significant associations in flexibility intercept (I) and slope 1 (S1). With the aim to understand the gender interactions, Cox regression models were performed separately for men and women (Table 3c). No significant associations were found for men. In women on the other hand, fine motor function (I), balance/upper strength (I) and balance/upper strength (S1) were related to higher risk for mortality. 4. Discussion This study confirms findings that different types of functional ability play different roles in age and ageing, but foremost that there are important gender differences. From the twenty-four measures of functional ability that were collected at each wave of assessment, three motor factors could be defined using principal component factor analysis. Cross-sectionally, there were gender differences across motor functions over time, where women had more difficulties. These gender differences are found in both the younger and older sample (60+ years). However, in IPT7, in advanced ages, men have as many motor function problems. The results are similar to Oksuzyan et al. (2010), and Finkel et al. (2003), where cross-sectional trajectories suggested that the decline in grip strength was steeper among men. It can therefore be concluded that those men who survive to live as long as women will suffer from the same disabilities as women, and experience the disabilities as chronic conditions in the same way that most older women do. The survival analyses with the total sample demonstrated that motor function difficulties were related to mortality, which agrees with other research (e.g. Gallucci et al., 2009; Studenski et al., 2011). The major finding from the separate Cox regression models, however, is that motor function (fine motor function and balance/upper strength difficulties) is predictive of mortality only for women. The different patterns of aging and mortality for men and women, in general, is consistent with the hypothesis that women have more chronic conditions that will impact motor functioning and as it turned out, mortality as well. Men, on the other hand, are less likely to have chronic conditions that impact motor functioning, but suffer more from acute, fatal, diseases. This reasoning is supported by other studies. Rozzini, Sleiman, Maggi, Noale and Trabucchi (2009) found that among reasons for admission to acute care, respiratory problems (pneumonia, acute dyspnea, and other infections) were more common in males than in females. Diseases that are not acute but still have a negative impact on physical functions, such as arthritis, are significantly more prevalent among females (Crimmins et al., 2010). In our study, females not only
5. Conclusion Different types of motor functions play different roles in age and ageing, but foremost there are important gender differences. Women have more difficulties in all three motor factors already at early ages and they are also important factors related to mortality for women. For men, who have better motor function at early ages, none of the motor factors were related to mortality. These results provide further evidence for the importance of the health care system to encourage improvement of both fine motor and the balance and upper strength function among older people, but foremost there is a need for individualized programs due to the obvious gender differences. Acknowledgements This work was supported by National Institute on Aging (AG04563, AG10175, AG08724); The MacArthur Foundation Research Network on Successful Aging; the Swedish Council for Working Life and Social Research (97:0147:1B, 2009-0795); Forte (2013-2292); and Swedish Research Council (825-2007-7460, 825-2009-6141, 521-2013-8689). Financial sponsors played no role in the design, execution, analysis and interpretation of data, or writing of the study. 10
Archives of Gerontology and Geriatrics 72 (2017) 6–11
M.E. Bravell et al.
Study. Archives of Gerontology and Geriatrics, 48, 281–283. Guralnik, J. M., Simonsick, E. M., Ferrucci, L., Glynn, R. J., Berkman, L. F., Blazer, D. G., et al. (1994). A short physical performance battery assessing lower extremity function: Association with self-reported disability and prediction of mortality and nursing home admission. Journal of Gerontology, 49, M85–94. Hirsch, C. H., Bůžková, P., Robbins, J. A., Patel, K. V., & Newman, A. B. (2012). Predicting late-life disability and death by the rate of decline in physical performance measures. Age and Ageing, 41, 155–161. Laird, N. M., & Ware, J. H. (1982). Random-effects models for longitudinal data. Biometrics, 38, 963–974. Larsson, K., & Thorslund, M. (2002). Does gender matter? Differences in patterns of informal support and formal services in a Swedish urban population. Research on Aging, 24, 308–336. Leitsch, S. A. (2000). Perceived control and functional health in the lives of the oldest-old. Doctoral thesis. USA: Human Development and Family Studies, Pennsylvania State University, State College. McArdle, J. J., Prescott, C. A., Hamagami, F., & Horn, J. L. (1998). A contemporary method for developmental-genetic analyses of age changes in intellectual abilities. Developmental Neuropsychology, 14, 69–114. Oksuzyan, A., Crimmins, E., Saito, S., O’Rand, A., Vaupel, J. W., & Christensen, K. (2010). Cross-national comparison of sex differences in health and mortality in Denmark, Japan and the US. European Journal of Epidemiology, 25, 471–480. Orfila, F., Ferrer, M., Lamarca, R., Tebe, C., Domingo-Salvany, A., & Alonso, J. (2006). Gender differences in health-related quality of life among the elderly: The role of objective functional capacity and chronic conditions. Social Science & Medicine, 63, 2367–2380. Reynolds, C. A., Finkel, D., McArdle, J. J., Gatz, M., & Pedersen, N. L. (2003). Quantitative genetic analysis of latent growth curve models of cognitive abilities in adulthood. Developmental Psychology, 41, 3–16. Rozzini, R., Sleiman, I., Maggi, S., Noale, M., & Trabucchi, M. (2009). Gender differences and health status in old and very old patients. JAMDA, 10, 554–558. Stineman, M. G., Xie, D., Pan, Q., Kurichi, J. E., Zhang, Z., Saliba, D., et al. (2012). Allcause 1-, 5-, and 10-year mortality in elderly people according to Activities of Daily Living stage. Journal of the American Geriatrics Society, 60, 485–492. Studenski, S., Perera, S., Patel, K., Rosano, C., Faulkner, K., Inzitari, M., et al. (2011). Gait speed and survival in older adults. JAMA, 305, 50–58. Taekema, D. G., Gussekloo, J., Westendorp, R. G. J., de Craen, A. J. M., & Maier, A. B. (2012). Predicting survival in oldest old people. American Journal of Medicine, 125, 1188–1194. Tiainen, K., Luukkaala, T., Hervonen, A., & Jylhä, M. (2013). Predictors of mortality in men and women aged 90 and older: A nine-year follow-up study in the Vitality 90+ study. Age and Ageing, 42, 468–475. White, D. K., Neogi, T., Nevitt, M. C., Peloquin, C. E., Zhu, Y., Boudreau, R. M., et al. (2013). Trajectories of gait speed predict mortality in well-functioning older adults: The Health, Aging and Body Composition Study. The Journals of Gerontology. Biological Sciences and Medical Sciences, 68, 456–464. Wu, Y., Zhang, D., Pang, Z., Oksuzyan, A., Jiang, W., Wang, S., et al. (2012). Genderspecific patterns in age-related decline in general health among Danish and Chinese: A cross-national comparative study. Geriatrics & Gerontology International, 12, 431–439.
References Avlund, K., Vass, M., & Hendriksen, C. (2003). Onset of mobility disability among community- dwelling old men and women. The role of tiredness in daily activities. Age and Ageing, 32, 579–584. Avlund, K. (1999). Methodological challenges in measurements of functional ability in gerontological research. A review. Aging Clinical and Experimental Research, 9, 164–174. Bryk, S., & Raudenbush, S. W. (1992). Hierarchical linear models: Applications and data analysis methods. Advanced qualitative techniques in the social sciences, 1. Sage Publications, Inc [Hierarchical linear models: Applications and data analysis methods Thousand Oaks]. Cooper, R., Kuh, D., & Hardy, R. (2010). Objectively measured physical capability levels and mortality: Systematic review and meta-analysis. BMJ, 341, c4467. http://dx.doi. org/10.1136/bmj.c4467. Cooper, R., Kuh, D., Cooper, C., Gale, C. R., Lawlor, D. A., Matthews, F., et al. (2011). 2011. The Falcon and Halcyon study teams. Objective measures of physical capability and subsequent health: A systematic review. Age and Ageing, 40, 14–23. Crimmins, E. M., Kim, J. K., & Sole-Auro, A. (2010). Gender differences in health: Results from SHARE, ELSA and HRS. European Journal of Public Health, 21(1), 81–91. Daly, R. M., Rosengren, B. E., Alwis, G., Ahlborg, H. G., Sernbo, I., & Karlsson, M. K. (2013). Gender specific age-related changes in bone density, muscle strength and functional performance in the elderly: A-10 year prospective population-based study. BMC Geriatrics, 13, 71. Deeg, D. H. J., & Kriegsman, D. M. W. (2003). Concepts of self-Rated health: Specifying the gender difference in mortality risk. Gerontologist, 43, 376–386. Ernsth Bravell, M., Zarit, S., & Johansson, B. (2011). Self reported activities of daily living and performance based functional ability: A study of congruence among the oldest old. Eur J Ageing, 8, 199–209. Farag, I., Sherrington, C., Kamper, S. J., Ferreira, M., Moseley, A. M., Lord, S. R., et al. (2012). Measures of physical functioning after hip fracture: Construct validity and responsiveness of performance-based and self-reported measures. Age and Ageing, 41, 659–664. Fauth, E. R., Zarit, S. H., & Malmberg, B. (2008). Mediating relationships within the disablement process model: A cross-sectional study of the oldest-old. Eur J of Ageing, 5, 161–179. Finkel, D., & Pedersen, N. L. (2004). Processing speed and longitudinal trajectories of change for cognitive abilities: The Swedish Adoption/Twin Study of Aging. Aging, Neuropsychology, and Cognition, 11, 325–345. Finkel, D., Pedersen, N. L., Reynolds, C. A., Berg, S., de Faire, U., & Svartengren, M. (2003). Genetic and environmental influences on decline in biobehavioral markers of aging. Behavior Genetics, 33, 107–123. Finkel, D., Ernsth Bravell, M., & Pedersen, N. L. (2015). Sex differences in genetic and environmental influences on longitudinal change in functional ability in late adulthood. The Journals of Gerontology. Series B, Psychological Sciences and Social Sciences, 70, 709–717. Finkel, D., Ernsth Bravell, M., & Pedersen, N. L. (2016). Temporal dynamics of motor functioning and cognitive aging. Journal of Gerontology Medical Science, 71, 109–116. Gallucci, M., Ongaro, F., Amici, G. P., & Regini, C. (2009). Frailty, disability and survival in the elderly over the age of seventy: Evidence from the Treviso Longeva (TRELONG)
11
Contents lists available at ScienceDirect
Archives of Gerontology and Geriatrics journal homepage: www.elsevier.com/locate/archger
Full Length Article
Motor functioning differentially predicts mortality in men and women a,⁎
b
a,c
d
Marie Ernsth Bravell , Deborah Finkel , Anna Dahl Aslan , Chandra A. Reynolds , Jenny Hallgrena, Nancy L. Pedersenc,e
MARK
a
Institute of Gerontology, School of Health and Welfare, Jönköping University, Sweden Department of Psychology, Indiana University Southeast, USA Department of Medical Epidemiology and Biostatistics, Karolinska Institutet, Stockholm, Sweden d Department of Psychology, University of California, Riverside, USA e Department of Psychology, University of Southern California, Los Angeles, USA b c
A R T I C L E I N F O
A B S T R A C T
Keywords: Motor function Gender differences Survival
Introduction: Research indicates gender differences in functional performance at advanced ages, but little is known about their impact on longevity for men and women. Objective: To derive a set of motor function factors from a battery of functional performance measures and examine their associations with mortality, incorporating possible gender interactions. Method: Analyses were performed on the longitudinal Swedish Adoption/Twin Study of Aging (SATSA) including twenty-four assessments of motor function up to six times over a 19-year period. Three motor factors were derived from several factor analyses; fine motor, balance/upper strength, and flexibility. A latent growth curve model was used to capture longitudinal age changes in the motor factors and generated estimates of intercept at age 70 (I), rates of change before (S1) and after age 70 (S2) for each factor. Cox regression models were used to determine how gender in interaction with the motor factors was related to mortality. Results: Females demonstrated lower functional performance in all motor functions relative to men. Cox regression survival analyses demonstrated that both balance/upper strength, and fine motor function were significantly related to mortality. Gender specific analyses revealed that this was true for women only. For men, none of the motor factors were related to mortality. Conclusion: Women demonstrated more difficulties in all functioning facets, and only among women were motor functioning (balance/upper strength and fine motor function) associated with mortality. These results provide evidence for the importance of considering motor functioning, and foremost observed gender differences when planning for individualized treatment and rehabilitation.
1. Introduction Studies of disability and function among older individuals are common, due to the changes that usually appear with advanced age and the effect these physical changes have on daily living. Disability is typically measured by self-reports of function in Personal Activities in Daily Life (PADL) and/or function in Instrumental Activities in Daily Life (IADL) (Fauth, Zarit, & Malmberg, 2008). Self-reports of disability are partly confounded by social and psychological health factors such as gender roles, labor, and interests (Larsson & Thorslund, 2002). Assessments of functional impairments via observed performance of motor function are considered to more accurately capture true physiological impairments (Avlund, 1999; Guralnik et al., 1994). The two most commonly examined measures are grip strength and walking speed (Cooper et al., 2011). Studies comparing self-reported disability and ⁎
measured functional impairments show modest correlations, ranging from 0.17 to 0.54 (Ernsth Bravell, Zarit, & Johansson, 2011; Farag et al., 2012). Therefore, studies using self-reported versus assessed physical function will not necessarily yield the same results. Nevertheless, both self-reported ADL and observed functional ability decline with age, and they also demonstrate associations with longevity (Cooper, Kuh, & Hardy, 2010; Gallucci, Ongaro, Amici, & Regini, 2009; Hirsch, Bůžková, Robbins, Patel, & Newman, 2012; Stineman et al., 2012; Taekema, Gussekloo, Westendorp, de Craen, & Maier, 2012; Tiainen, Luukkaala, Hervonen, & Jylhä, 2013; White et al., 2013). In addition, there are documented gender differences in disability in late life, where women tend to have more problems at the same time as they live longer (e.g. Avlund, Vass, & Hendriksen, 2003; Crimmins, Kim, & Sole-Auro, 2010) and have poorer physical function (Daly et al., 2013; Orfila et al., 2006). Most current studies focus on single measurements such as grip
Corresponding author. E-mail address: [email protected] (M.E. Bravell).
http://dx.doi.org/10.1016/j.archger.2017.05.001 Received 16 September 2016; Received in revised form 24 February 2017; Accepted 1 May 2017 Available online 02 May 2017 0167-4943/ © 2017 Elsevier B.V. All rights reserved.
Archives of Gerontology and Geriatrics 72 (2017) 6–11
M.E. Bravell et al.
2.2. Measures
strength (Oksuzyan et al., 2010; Wu et al., 2012) or gait speed (Studenski et al., 2011). Few studies have considered gender differences in the relationship between mortality and multiple measures of physical motor functioning tapping different functional modalities (Cooper et al., 2011). The aim of the current study is therefore to evaluate a set of performance-based motor function factors and to explore their relationship to mortality, incorporating possible gender interactions.
Twenty-four measures of functional ability were collected at each IPT. They are listed in Table 2. The functional ability tasks were timed but motor functioning measured in seconds was skewed (most individuals perform the tasks successfully into their mid-60s). Previous studies have found problems using time to complete a task due to lack of variation (e.g. Ernsth Bravell et al., 2011). Performance on the 24 motor functioning tasks was also categorized by the nurses administrating the testing as without difficulty (1), with some difficulty (2), or impossible (3). Thus, even when a participant could, for example, balance for 10 s with eyes closed (achieving the maximum score), the nurses were able to report whether the participant demonstrated any difficulty with the balance task, regardless of time. The analyses described below made use of the qualitative nurse ratings for each of the 24 tasks.
2. Method and material 2.1. Sample As a part of the longitudinal Swedish Adoption/Twin Study of Aging (SATSA), twenty-four different assessments of motor functioning were collected six times over a 19-year period. Accrual procedures for SATSA have been described previously (Finkel & Pederson, 2004). In-person testing (IPT) took place in a location convenient to the participants and was completed during a single 4-h visit. The second (IPT2) and third (IPT3) waves of in-person testing occurred at three-year intervals. Inperson testing did not occur during wave 4; therefore, the next wave of in-person testing is labeled IPT5 and occurred after a 7-year interval (see Finkel & Pedersen, 2004). Regular 3-year testing continued after IPT5; therefore, the total time span from IPT1 to IPT7 was 19 years. For IPT1, motor function data was available for 645 persons. In addition, respondents were added during the study as they turned 50 years; 97 entered the study at IPT2, another 21 at IPT3, and finally 101 at IPT5. Not surprisingly, there was very little variance in motor function at IPT1; therefore, IPT2 was used as the baseline assessment for this study. Since almost all of the respondents performed fairly well with little variance before the age of 60, which agrees e.g. with Wu et al. (2012), only respondents aged 60 and above at IPT2 were included in the survival analyses. Baseline data (IPT2) on motor function were available from 436 twins ranging in age from 60 to 91 years. From these, 76% participated in three or more IPTs. Table 1 provides a description of the total sample, from which descriptive and factor analyses were performed (total sample). It also describes the sample from which the survival analyses were performed (age 60+).
2.3. Analyses To assess the effects of relatedness (twinship), the sample was divided into two groups: twin A from each pair in one group and twin B from each pair in another group. All analyses were conducted separately in each group to ensure that the results were the same. Only minor differences were found; thus, the descriptive statistics reported (Table 1) are from the full sample to maximize power. Cox regression survival models were applied using STATA/IC 12.1 and the robust sandwich estimation option to control for twinship and thus provide appropriate standard errors. 2.3.1. Factor structure To begin with, several factor analyses were performed in order to create motor factors. Due to skew evident in motor functioning in the first IPTs, where most individuals performed well, the first factor analyses (Principal Component Analysis with Variamax rotation) were performed on the motor function measures from IPT7. The factor analysis converged in three iterations that could be interpreted as; 1. Fine motor ability (explained 33% of variance); 2. Balance and strength motor ability (explained 27% of variance); 3. Flexibility (explained 14% of the variance). The extractions in the factor analysis on IPT7 are based on eigenvalues, and explain a total of 71% of the variance in motor function, see Table 2.
Table 1 Sample characteristics and descriptive Mean (SD). Fine motor N
Men
Women
IPT 2
585
242
343
IPT 2a
436
174
262
IPT3
566
233
333
IPT3a
378
149
229
IPT5
541
213
328
IPT5a
252
90
162
IPT6
447
183
264
IPT6a
185
67
118
IPT7
379
155
224
138
47
91
IPT7
a
Age 66.0 (9.0) 70.2 (6.1) 68.8 (9.2) 73.1 (5.8) 70.6 (10.0) 78.5 (5.1) 72.2 (9.3) 80.5 (4.6) 74.3 (9.0) 83.23 (4.1)
Total
Balance and strength Men
Women
Total
Men
Flexibility Women
**
Men **
Women
2.1
11.1** (2.5) 11.2* (2.5)
2.2 (0.6)
2.2* (0.4)
2.3* (0.6)
11.0 (2.1)
10.5** (1.4) 10.7* (1.5)
2.2* (0.5)
2.3* (0.6)
11.1 (2.1)
10.9* (1.5)
11.2* (2.4)
2.3 (0.6) 2.3 (0.6)
2.2* (0.5)
11.7 (3.7)
11.3* (3.1)
12.0* (4.1)
2.3 (0.8)
2.2** (0.6)
2.3* (0.7) 2.4** (0.9)
12.6 (4.5)
12.3 (4.3)
12.7 (4.6)
2.4 (0.9)
2.4 (0.8)
11.8 (3.3)
11.5 (3.1)
11.9 (3.4)
2.3 (0.7)
2.2*** (0.5)
2.5 (1.0) 2.4*** (0.8)
10.2** (3.3)
12.8 (4.1)
12.7 (4.1)
12.9 (4.1)
2.5 (0.9)
2.3** (0.6)
2.6** (1.0)
9.8 (3.4) 10.5 (3.7)
12.3 (4.2)
12.11 (4.2)
12.5 (4.2)
2.5 (0.9)
2.4 (0.8)
14.2 (5.2)
14.5 (6.2)
14.0 (4.6)
2.7 (1.1)
2.5 (1.0)
2.5 (0.9) 2.9 (1.2)
8.7 (1.4)
8.6 (1.5)
8.7 (1.3)
8.6 (1.4)
8.8 (1.5)
8.8 (1.6)
9.0 (2.5)
8.8 (1.9)
9.4 (2.8)
9.2 (2.6)
9.3 (2.2)
9.1* (1.5)
8.6 (1.2) 8.7 (1.7) 8.7 (1.3) 8.8 (1.3) 9.2 (2.8) 9.5 (2.9) 9.5* (2.6)
9.9 (2.9)
9.5** (1.8)
9.8 (3.4)
9.8 (3.3)
10.7 (3.9)
11.1 (4.3)
7
10.7 (2.0)
10.4
10.9 (2.2)
(0.4)
2.2** (0.6)
2.2 (0.5)
8.6 (1.3)
10.9
Total
(2.3)
8.6 (1.2)
(1.3)
**
Archives of Gerontology and Geriatrics 72 (2017) 6–11
M.E. Bravell et al.
Table 2 Results from factor analysis at the IPT7-sample including all 20 motor function variables (difficulties). Description of the motor function variables included in the factor analysis
Component
Difficulties to: Pour water from one glass to another, dominant hand (Pour water glass, dominant hand) Insert a key into a lock and turn (Insert and turn key) Screw in a light bulb (Screw in light bulb) Pour water from one glass to another, nondominant hand (Pour water glass, nondominant hand Insert a plug into an electrical socket (Put plug in socket) Put coins of various sizes into appropriate slots (Put coins in slot) Pour water from a jug into a glass (Pour water from jug into glass) Dial numbers 1–9 on a rotary phone (Dial a number) Stand up from a chair, arms crossed in front (Chairstand) Lift a drinking glass (Lifting glass) Touch right toes with left hand (Left fingers to right toe) Balance, feet side by side, eyes open (max 10 s) (Stand side by side looking) Lift a kilogram packet (Lift one kg) Time to pick pen off floor from standing position, max 30 s (Pick up pen) Touch left toes with right hand (Right fingers to left toe) Balance time, feet together, arms extended front, eyes closed, max 10 s (Rombergs test looking) Walk 3 m and return, at participant’s own pace (Gait and turn 3m) Touch left earlobe with right hand (Right hand to left earlobe) Touch right earlobe with left hand (Left hand to right earlobe) Stand up from a chair 5 times, arms crossed (Repeat chair stand)
1
2
3
,887 ,887 ,875 ,871 ,870 ,866 ,837 ,805 ,054 ,256 ,162 ,117 ,246 ,175 ,085 ,147 ,257 ,133 ,051 ,072
,063 ,181 ,164 ,233 ,174 ,252 ,113 ,153 ,788 ,764 ,752 ,735 ,732 ,712 ,706 ,641 ,571 ,093 ,156 ,558
-,189 ,158 ,234 -,046 ,292 ,109 -,020 ,340 ,362 -,196 ,133 ,121 -,028 ,350 ,144 ,430 ,540 ,777 ,742 ,574
Measures not included in the final motor factor solution Difficulties to: Feet semi-tandem, eyes open (max 10 s) Walk 10 steps, heel to toe Feet heel to toe, eyes open (max 10 s) Feet heel to toe, eyes closed (max 10 s) Extraction Method: Principal Component Analysis. Rotation Method: Varimax with Kaiser Normalization.
alpha did not increase by removing them from factor 2. There were significant correlations between the motor factors at all IPTs, with a range from 0.39 to 0.58 between fine motor and flexibility; from 0.31 to 0.49 between fine motor and balance/upper strength; and from 0.26 to 0.35 between flexibility and balance/upper strength. The strength of the correlations increased with increasing age.
Factor analyses were then performed for each IPT where the solutions explained between 61% to 69% of variance (IPT6 = 65%, IPT5 = 65%, IPT3 = 65%, IPT2 = 69%, IPT1 = 61%). Factor structures at each IPT were quite similar and a three-factor solution could be imposed at each wave. Additionally, possible age differences in factor structures were examined: at each wave, factor structures were compared for older adults (age ≥60) and younger adults (age < 60). Again, results were similar, but not identical, across age groups and interpretable 3-factor solutions emerged, but consistently with less variance explained in the younger age groups. Measures with sufficient variance (factor loadings > 0.4) were included in the motor factor and measures without sufficient variance (factor loading < 0.4) were excluded. Thus, based on the extracted solutions that were the most repeatable across IPTs, three motor scores were computed by summing the respective items:
2.3.2. Latent growth curve model A latent growth curve model (LGCM) was used to capture longitudinal age changes in the motor factors (McArdle, Prescott, Hamagami, & Horn, 1998; Reynolds, Finkel, McArdle, Gatz, & Pedersen, 2003). The structural model can be considered as a random coefficients model (Bryk & Raudenbush, 1992; Laird & Ware, 1982) where individual regression models are fitted to each subject’s longitudinal profile of data as well as an average model of growth estimated over the entire sample. In other words, the model provides estimation of fixed effects, i.e. fixed population parameters as estimated by the average growth model of the entire sample, and random effects, i.e., individual variation in growth model parameters. The variation in individual regression coefficients from the group model is then available for further analysis. Latent growth curve models take into account missing data by giving more weight to individuals with the most time points. Based on previous analyses (Finkel, Ernsth Bravell, & Pedersen, 2015; Finkel, Ernsth Bravell, & Pedersen, 2016), a growth curve model with two linear slopes was employed: change with age up to age 70 and change with age after age 70. The intercept thus reflects an estimate of motor performance at age 70. The random and fixed effects estimates were obtained using PROC Mixed in SAS 8.0 (SAS Institute, 1999). Empirical Bayes (EB) estimates of intercept and slope values were produced by the PROC Mixed procedure for each participant. EB estimates are weighted estimates based on both the fixed and random
• Fine motor ability, Cronbach’s alpha at IPT7 = 0.94. Best performance = 8; worst performance = 24. • Balance motor ability, Cronbach’s alpha at IPT7 = 0.91. Best performance = 10; worst performance = 30. • Flexibility, Cronbach’s alpha at IPT7 = 0.73. Best performance = 2; worst performance = 6.
Four motor function measures that did not load on any of the factors were excluded from factor construction. Because factor loadings varied somewhat across waves and ages, but were all of significant magnitude (> 0.4), unit weighting was used to calculate composite scale scores. Higher values indicate greater disability. It should also be noted that repeated chair stand, Romberg’s test and walk 3 m loaded highly in both factor 2 (balance and upper strength) and factor 3 (flexibility) in some of the factor analyses, but they were placed in factor 2 to maximize interpretability. Cronbach’s 8
Archives of Gerontology and Geriatrics 72 (2017) 6–11
M.E. Bravell et al. Table 3a Cox Regression Models with age as the time variable and motor factors intercept and slopes as covariates. Robust
95% CI for Exp(B)
Hazard ratio
Std. Error
Lower
Upper
Log pseudolikelihood = −2384.0701 Fine motor (I) Balance/upper strength (I) Flexibility intercept (I)
1.232*** 1.128** 0.991ns
0.057 0.050 0.428
1.125 1.034 0.425
1.348 1.1230 2.313
Log pseudolikelihood = −729.81129 Fine motor (S1) Balance/upper strength (S1) Flexibility (S1)
1.222* 1.872** 0.256ns
0.114 0.342 0.348
1.018 1.308 0.018
1.467 2.680 3.685
Log pseudolikelihood = −1239.4172 Fine motor (S2) Balance/upper strength (S2) Flexibility (S2)
985ns 1.045ns 0.921ns
0.040 0.032 0.150
0.910 0.984 0.668
1.068 1.110 1.268
Note. Motor factors intercepts and slopes index difficulty levels and change, respectively. I = Intercept, S1 = change up to age 70, S2 = change after age 70. * p < 0.05. ** p < 0.01. *** p < 0.001.
Because of the wide range in ages (up to 35 years) at participation in each IPT, an age-based survival analysis was used instead of a time- or wave-based model. That is, time-to-death is measured by age through subtracting the date of death from date of birth. Analyses were performed with SPSS Statistics 21 except the Cox Regression Survival Models that were analyzed using STATA/IC 12.1.
effects from the latent growth model (Bryk & Raudenbush, 1992). EB intercept estimates (I) were generated for all participants. For participants with 2 IPTs at age 70 or less, an EB estimate of slope 1 (S1) was generated. For participants with 2 IPTs at age 70 or more, an EB estimate of slope 2 (S2) was generated. EB estimates were used in the Cox regression survival model as independent variables. 2.3.3. Gender differences Independent sample t-tests were performed to explore cross-sectional gender differences in motor function. Cox regression survival analysis was performed with mortality (event = 1) as the outcome variable, age as the time variable, and gender and the EB estimates of the three motor factors (I, S1 and S2) as independent variables in a first step, and with gender in interaction with the EB estimates of three motor factors (I, S1 and S2) in a second step in order to determine the role of gender and function on mortality. Mortality data were collected from the Swedish Population Registry in February 2016, nine years after IPT7 testing was completed in 2007. By then 337 out of the 436 respondents, relevant for the Cox regression model, had passed away.
Table 3c Separate Cox Regression Models for men and women with age as the time variable and motor factors intercept and slopes as covariates. Robust Men
Log pseudolikelihood = −901.6837 Fine motor function (I) 1.084 Balance/upper strength (I) 1.142 Flexibility (I) 0.772 Log pseudolikelihood = −365.1121 Fine motor function (S1) 1.152 Balance/upper strength (S1) 1.483 Flexibility (S1) 0.103
Table 3b Cox Regression Models with age as the time variable and motor factors intercept and slopes in interaction with gender as covariates. Robust Hazard ratio Log pseudolikelihood = −2375.147 Gender x Fine motor (I) 1.087** Gender x Balance/upper 1.107** strength (I) Gender x Flexibility (I) 0.318** Log pseudolikelihood = −729.7645 Gender x Fine motor (S1) 1.162** Gender x Balance/upper 1.519** strength (S1) Gender x Flexibility (S1) 0.149* Log pseudolikelihood = −1239.2946 Gender x Fine motor (S2) 0.980ns Gender x Balance/upper 1.018ns strength (S2) Gender x Flexibility (S2) 0.904ns
Log pseudolikelihood = −394.3006 Fine motor function (S2) 1.045 Balance/upper strength (S2) 0.994 Flexibility (S2) 1.195
95% CI for Exp(B) Std. Error
Lower
Upper
0.034 0.033
1.022 1.044
1.156 1.175
0.054
0.228
0.446
0.066 0.159
1.039 1.237
1.230 1.866
0.130
0.027
0.822
0.022 0.017
0.937 0.984
1.024 1.053
0.085
0.751
1.087
Hazard ratio
ns ns ns
ns ns ns
ns ns ns
95% CI for Exp(B) Std. Error
Lower
Upper
0.085 0.088 0.508
0.930 0.982 0.212
1.264 1.329 2.804
0.143 0.412 0.246
0.903 0.860 < 0.001
1.471 2.558 10.973
0.068 0.043 0.268
0.919 0.912 0.770
1.188 1.083 1.854
Robust Women
Note. Motor factors intercepts and slopes index difficulty levels and change, respectively. I = Intercept, S1 = change up to age 70, S2 = change after age 70. ***p < 0.001. * p < 0.05. ** p < 0.01.
Hazard ratio
95% CI for Exp(B) Std. Error
Lower
Upper
Log pseudolikelihood = −1190.3985 Fine motor function (I) 1.040*** Balance/upper strength (I) 1.161** Flexibility (I) 1.040 ns
0.098 0.072 0.660
1.130 1.029 0.300
1.517 1.310 3.610
Log pseudolikelihood = −257.6896 Fine motor function (S1) 1.300 ns Balance/upper strength (S1) 2.412*** Flexibility (S1) .160 ns
0.201 0.581 0.355
0.958 1.504 0.002
1.757 3.868 12.408
Log pseudolikelihood = −680.8204 Fine motor function (S2) 0.959ns Balance/upper strength (S2) 1.078ns Flexibility (S2) 0.901ns
0.049 0.042 0.194
0.868 0.997 0.590
1.060 1.163 1.377
Note. Motor factors intercepts and slopes index difficulty levels and change, respectively. I = Intercept, S1 = change up to age 70, S2 = change after age 70. *p < 0.05.
9
Archives of Gerontology and Geriatrics 72 (2017) 6–11
M.E. Bravell et al.
had worse motor function, but motor function was also related to mortality, which may be somewhat difficult to interpret. Diseases such as arthritis and osteoporosis may most certainly contribute to poorer motor function in women, but it is not considered as a life threating disease. They may however cause a cascade of outcomes that lead to mortality, such falls, fractures, and hospitalizations. Results that confirm that men are physically stronger and have fewer disabilities, but have substantially higher mortality and less healthy life style at all ages compared with women, are common (i.e. Cooper et al., 2011; Deeg & Kriegsman, 2003). A number of explanations have been proposed for this male-female health paradox, rooted in biological, social, and psychological interpretations. It may be due to multiple causes that include fundamental biological differences between the sexes such as genetic factors, immune system responses, hormones, and disease patterns (Cooper et al., 2011). In a twin analysis of the motor factors created with these data, evidence for significant genetic and environmental influences on trajectories was found. Moreover, genetic influences on longitudinal changes in the balance/upper strength factor after age 70 were significantly greater for women than for men (Finkel et al., 2016). Other proposed explanations are behavioral gender differences such as risk-taking and reluctance to seek and comply with medical treatment. However, Crimmins et al. (2010) found that gender differences were less pronounced in abilities to provide self-care, even if females demonstrated more problems in IADL. The construction of the motor factors deserves mention. Even though the three factors were somewhat surprising in their construction, similar factors emerged from each IPT, even if less variance could be explained at IPT2 and IPT3. Also, these factors are similar to previous studies (Ernsth Bravell et al., 2011; Leitsch, 2000) but not completely identical as the measures vary across different studies. The surprising aspect from this study was that only two motor items loaded highly on the flexibility factor. One would assume that e.g. “pick up pen” or “reach toes with fingers” should be in that factor. Instead, it was the repeated chair-stand, walk 3 m and Romberg’s test that cross-loaded in some of the factor analyses. We chose to add those items into balance/upper strength, both due to better interpretation but also due to the fact that they regularly loaded the highest on these factors. Furthermore, Cronbach’s alpha was not improved by removing them. Some of the expected balance motor items (e.g. feet heel to toe; feet semi-tandem; walk 10 steps) did not load consistently over the IPTs and they were therefore excluded when constructing the motor factors.
2.4. Ethical considerations The SATSA-study was approved by Ethical committee at Karolinska Institutet and the Regional Ethics Review Board in Stockholm. 3. Results 3.1. Description of motor function Cross-sectional gender differences were found at early IPTs in balance/upper strength and flexibility, and in IPT6 in fine motor function (Table 1). Females had more problems at each of these IPTs. 3.2. Survival analysis Cox regressions with the empirical Bayes (EB) estimates were used to determine whether motor functions, and motor function in interaction with gender, were related to survival (Tables 3a and 3b). The patterns differed across the estimates where intercept (I) and slope 1 (S1) for fine motor function and balance/upper strength was related to mortality in the total sample (Table 3a). The same patterns were found when motor factors were evaluated in an interaction with gender (Table 3b), with the addition of significant associations in flexibility intercept (I) and slope 1 (S1). With the aim to understand the gender interactions, Cox regression models were performed separately for men and women (Table 3c). No significant associations were found for men. In women on the other hand, fine motor function (I), balance/upper strength (I) and balance/upper strength (S1) were related to higher risk for mortality. 4. Discussion This study confirms findings that different types of functional ability play different roles in age and ageing, but foremost that there are important gender differences. From the twenty-four measures of functional ability that were collected at each wave of assessment, three motor factors could be defined using principal component factor analysis. Cross-sectionally, there were gender differences across motor functions over time, where women had more difficulties. These gender differences are found in both the younger and older sample (60+ years). However, in IPT7, in advanced ages, men have as many motor function problems. The results are similar to Oksuzyan et al. (2010), and Finkel et al. (2003), where cross-sectional trajectories suggested that the decline in grip strength was steeper among men. It can therefore be concluded that those men who survive to live as long as women will suffer from the same disabilities as women, and experience the disabilities as chronic conditions in the same way that most older women do. The survival analyses with the total sample demonstrated that motor function difficulties were related to mortality, which agrees with other research (e.g. Gallucci et al., 2009; Studenski et al., 2011). The major finding from the separate Cox regression models, however, is that motor function (fine motor function and balance/upper strength difficulties) is predictive of mortality only for women. The different patterns of aging and mortality for men and women, in general, is consistent with the hypothesis that women have more chronic conditions that will impact motor functioning and as it turned out, mortality as well. Men, on the other hand, are less likely to have chronic conditions that impact motor functioning, but suffer more from acute, fatal, diseases. This reasoning is supported by other studies. Rozzini, Sleiman, Maggi, Noale and Trabucchi (2009) found that among reasons for admission to acute care, respiratory problems (pneumonia, acute dyspnea, and other infections) were more common in males than in females. Diseases that are not acute but still have a negative impact on physical functions, such as arthritis, are significantly more prevalent among females (Crimmins et al., 2010). In our study, females not only
5. Conclusion Different types of motor functions play different roles in age and ageing, but foremost there are important gender differences. Women have more difficulties in all three motor factors already at early ages and they are also important factors related to mortality for women. For men, who have better motor function at early ages, none of the motor factors were related to mortality. These results provide further evidence for the importance of the health care system to encourage improvement of both fine motor and the balance and upper strength function among older people, but foremost there is a need for individualized programs due to the obvious gender differences. Acknowledgements This work was supported by National Institute on Aging (AG04563, AG10175, AG08724); The MacArthur Foundation Research Network on Successful Aging; the Swedish Council for Working Life and Social Research (97:0147:1B, 2009-0795); Forte (2013-2292); and Swedish Research Council (825-2007-7460, 825-2009-6141, 521-2013-8689). Financial sponsors played no role in the design, execution, analysis and interpretation of data, or writing of the study. 10
Archives of Gerontology and Geriatrics 72 (2017) 6–11
M.E. Bravell et al.
Study. Archives of Gerontology and Geriatrics, 48, 281–283. Guralnik, J. M., Simonsick, E. M., Ferrucci, L., Glynn, R. J., Berkman, L. F., Blazer, D. G., et al. (1994). A short physical performance battery assessing lower extremity function: Association with self-reported disability and prediction of mortality and nursing home admission. Journal of Gerontology, 49, M85–94. Hirsch, C. H., Bůžková, P., Robbins, J. A., Patel, K. V., & Newman, A. B. (2012). Predicting late-life disability and death by the rate of decline in physical performance measures. Age and Ageing, 41, 155–161. Laird, N. M., & Ware, J. H. (1982). Random-effects models for longitudinal data. Biometrics, 38, 963–974. Larsson, K., & Thorslund, M. (2002). Does gender matter? Differences in patterns of informal support and formal services in a Swedish urban population. Research on Aging, 24, 308–336. Leitsch, S. A. (2000). Perceived control and functional health in the lives of the oldest-old. Doctoral thesis. USA: Human Development and Family Studies, Pennsylvania State University, State College. McArdle, J. J., Prescott, C. A., Hamagami, F., & Horn, J. L. (1998). A contemporary method for developmental-genetic analyses of age changes in intellectual abilities. Developmental Neuropsychology, 14, 69–114. Oksuzyan, A., Crimmins, E., Saito, S., O’Rand, A., Vaupel, J. W., & Christensen, K. (2010). Cross-national comparison of sex differences in health and mortality in Denmark, Japan and the US. European Journal of Epidemiology, 25, 471–480. Orfila, F., Ferrer, M., Lamarca, R., Tebe, C., Domingo-Salvany, A., & Alonso, J. (2006). Gender differences in health-related quality of life among the elderly: The role of objective functional capacity and chronic conditions. Social Science & Medicine, 63, 2367–2380. Reynolds, C. A., Finkel, D., McArdle, J. J., Gatz, M., & Pedersen, N. L. (2003). Quantitative genetic analysis of latent growth curve models of cognitive abilities in adulthood. Developmental Psychology, 41, 3–16. Rozzini, R., Sleiman, I., Maggi, S., Noale, M., & Trabucchi, M. (2009). Gender differences and health status in old and very old patients. JAMDA, 10, 554–558. Stineman, M. G., Xie, D., Pan, Q., Kurichi, J. E., Zhang, Z., Saliba, D., et al. (2012). Allcause 1-, 5-, and 10-year mortality in elderly people according to Activities of Daily Living stage. Journal of the American Geriatrics Society, 60, 485–492. Studenski, S., Perera, S., Patel, K., Rosano, C., Faulkner, K., Inzitari, M., et al. (2011). Gait speed and survival in older adults. JAMA, 305, 50–58. Taekema, D. G., Gussekloo, J., Westendorp, R. G. J., de Craen, A. J. M., & Maier, A. B. (2012). Predicting survival in oldest old people. American Journal of Medicine, 125, 1188–1194. Tiainen, K., Luukkaala, T., Hervonen, A., & Jylhä, M. (2013). Predictors of mortality in men and women aged 90 and older: A nine-year follow-up study in the Vitality 90+ study. Age and Ageing, 42, 468–475. White, D. K., Neogi, T., Nevitt, M. C., Peloquin, C. E., Zhu, Y., Boudreau, R. M., et al. (2013). Trajectories of gait speed predict mortality in well-functioning older adults: The Health, Aging and Body Composition Study. The Journals of Gerontology. Biological Sciences and Medical Sciences, 68, 456–464. Wu, Y., Zhang, D., Pang, Z., Oksuzyan, A., Jiang, W., Wang, S., et al. (2012). Genderspecific patterns in age-related decline in general health among Danish and Chinese: A cross-national comparative study. Geriatrics & Gerontology International, 12, 431–439.
References Avlund, K., Vass, M., & Hendriksen, C. (2003). Onset of mobility disability among community- dwelling old men and women. The role of tiredness in daily activities. Age and Ageing, 32, 579–584. Avlund, K. (1999). Methodological challenges in measurements of functional ability in gerontological research. A review. Aging Clinical and Experimental Research, 9, 164–174. Bryk, S., & Raudenbush, S. W. (1992). Hierarchical linear models: Applications and data analysis methods. Advanced qualitative techniques in the social sciences, 1. Sage Publications, Inc [Hierarchical linear models: Applications and data analysis methods Thousand Oaks]. Cooper, R., Kuh, D., & Hardy, R. (2010). Objectively measured physical capability levels and mortality: Systematic review and meta-analysis. BMJ, 341, c4467. http://dx.doi. org/10.1136/bmj.c4467. Cooper, R., Kuh, D., Cooper, C., Gale, C. R., Lawlor, D. A., Matthews, F., et al. (2011). 2011. The Falcon and Halcyon study teams. Objective measures of physical capability and subsequent health: A systematic review. Age and Ageing, 40, 14–23. Crimmins, E. M., Kim, J. K., & Sole-Auro, A. (2010). Gender differences in health: Results from SHARE, ELSA and HRS. European Journal of Public Health, 21(1), 81–91. Daly, R. M., Rosengren, B. E., Alwis, G., Ahlborg, H. G., Sernbo, I., & Karlsson, M. K. (2013). Gender specific age-related changes in bone density, muscle strength and functional performance in the elderly: A-10 year prospective population-based study. BMC Geriatrics, 13, 71. Deeg, D. H. J., & Kriegsman, D. M. W. (2003). Concepts of self-Rated health: Specifying the gender difference in mortality risk. Gerontologist, 43, 376–386. Ernsth Bravell, M., Zarit, S., & Johansson, B. (2011). Self reported activities of daily living and performance based functional ability: A study of congruence among the oldest old. Eur J Ageing, 8, 199–209. Farag, I., Sherrington, C., Kamper, S. J., Ferreira, M., Moseley, A. M., Lord, S. R., et al. (2012). Measures of physical functioning after hip fracture: Construct validity and responsiveness of performance-based and self-reported measures. Age and Ageing, 41, 659–664. Fauth, E. R., Zarit, S. H., & Malmberg, B. (2008). Mediating relationships within the disablement process model: A cross-sectional study of the oldest-old. Eur J of Ageing, 5, 161–179. Finkel, D., & Pedersen, N. L. (2004). Processing speed and longitudinal trajectories of change for cognitive abilities: The Swedish Adoption/Twin Study of Aging. Aging, Neuropsychology, and Cognition, 11, 325–345. Finkel, D., Pedersen, N. L., Reynolds, C. A., Berg, S., de Faire, U., & Svartengren, M. (2003). Genetic and environmental influences on decline in biobehavioral markers of aging. Behavior Genetics, 33, 107–123. Finkel, D., Ernsth Bravell, M., & Pedersen, N. L. (2015). Sex differences in genetic and environmental influences on longitudinal change in functional ability in late adulthood. The Journals of Gerontology. Series B, Psychological Sciences and Social Sciences, 70, 709–717. Finkel, D., Ernsth Bravell, M., & Pedersen, N. L. (2016). Temporal dynamics of motor functioning and cognitive aging. Journal of Gerontology Medical Science, 71, 109–116. Gallucci, M., Ongaro, F., Amici, G. P., & Regini, C. (2009). Frailty, disability and survival in the elderly over the age of seventy: Evidence from the Treviso Longeva (TRELONG)
11