Differing determinants of disability trends among men and women aged 50 years and older

Abstract

Abstract Background Researchers have emphasized the importance of examining how different factors affect men’s and women’s functional status over time. To date, the literature is unclear about whether sex affects the rate of change in disability in middle to older age. Researchers have further emphasized the importance of examining how different factors affect men’s and women’s functional status over time. We examined (a) sex differences in disability trends and (b) the determinants of the rate of change in disability for men and women 50 years and older. Methods This study utilized the Taiwan Longitudinal Study on Aging Survey, a nationally representative database (four waves of survey data 1996–2007, N = 3429). We modeled and compared the differences in disability trends and the influences of determinants on trends among men and women using multiple-indicator and multiple-group latent growth curves modeling (LGCM). Equality constraints were imposed on 10 determinants across groups. Results Once disability began, women progressed toward greater disability 18% faster than men. Greater age added about 1.2 times the burden to the rate of change in disability for women than men (p

Full text

Chenetal. BMC Geriatrics (2022) 22:11
https://doi.org/10.1186/s12877-021-02574-3
RESEARCH ARTICLE
Diering determinants ofdisability trends
amongmen andwomen aged 50 years
andolder
Ya‑Mei Chen1* , Tung‑Liang Chiang1, Duan‑Rung Chen2, Yu‑Kang Tu3, Hsiao‑Wei Yu4 and Wan‑Yu Chiu1
Abstract
Background: Researchers have emphasized the importance of examining how different factors affect men’s and
women’s functional status over time. To date, the literature is unclear about whether sex affects the rate of change in
disability in middle to older age. Researchers have further emphasized the importance of examining how different
factors affect men’s and women’s functional status over time. We examined (a) sex differences in disability trends and
(b) the determinants of the rate of change in disability for men and women 50 years and older.
Methods: This study utilized the Taiwan Longitudinal Study on Aging Survey, a nationally representative database
(four waves of survey data 1996–2007, N = 3429). We modeled and compared the differences in disability trends and
the influences of determinants on trends among men and women using multiple‑indicator and multiple‑group latent
growth curves modeling (LGCM). Equality constraints were imposed on 10 determinants across groups.
Results: Once disability began, women progressed toward greater disability 18% faster than men. Greater age added
about 1.2 times the burden to the rate of change in disability for women than men (p < 0.001). More comorbidities
also added significantly more burden to baseline disability and rate of change in disability among women than men
(p < 0.001), but women benefited more from higher education levels in lower baseline disability and slower rate of
change. Having a better social network was associated with lower baseline disability among women only (p < 0.05).
For both men and women, physically active leisure‑time activities were beneficial in lower baseline disability (p men
and women < 0.001) and rate of change in disability (p men < 0.01; p women < 0.05), with no significant differences between
groups.
Conclusions: Age may widen the sex gap in the rate of change in disability. However, both sexes benefit from par‑
ticipating in leisure‑time activities. Promoting health literacy improves health outcomes and physical function among
women.
Keywords: Disability trajectory, Sex, Leisure‑time activities, Determinants
© The Author(s) 2021. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which
permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the
original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or
other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line
to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory
regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this
licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. The Creative Commons Public Domain Dedication waiver (http:// creat iveco
mmons. org/ publi cdoma in/ zero/1. 0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.
Background
Maintaining physical function has been a key public
health priority for many fast-aging societies for some
time. Over the past 10 years, researchers’ attention has
been drawn to identifying factors associated with changes
in physical function trends [10, 11, 16, 50]. Sex differ-
ences in the nature and range of health pathways over the
life course are among these factors, and there have been
Open Access
*Correspondence: chenyamei@ntu.edu.tw
1 Institute of Health Policy and Management, College of Public Health,
National Taiwan University, Room 633, No. 17, Xu‑Zhou Road, Taipei 100,
Taiwan
Full list of author information is available at the end of the article
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Page 2 of 10
Chenetal. BMC Geriatrics (2022) 22:11
calls to further delineate sex patterns and health-related
consequences [33].
Studies have now shown that sex differences in func-
tional status among older adults reflect not only biologi-
cal differences but also differences in privilege and power
based on sex identity and past decision making [13, 15,
27, 33, 39, 51]. Researchers have further emphasized the
importance of examining how different factors affect
men’s and women’s functional status over time [10, 27, 57,
58]. When Liang etal. [26] examined functional changes
over time among middle-aged and older men and women
from a life course perspective, they found that decreases
in functional status were more accelerated—in terms of
both baseline disability and rate of change in disabili-
ties—among women than men. Chen etal. [10] reported
that sex may not be a risk factor for developing initial
disability, yet women who do develop disability may be
at greater risk than men of faster increases in disability.
However, how much faster women’s rate of change in dis-
ability may be remains unclear.
Latent growth curves modeling (LGCM) has been e
recently advocated as a better method for addressing
questions related to individual change over time because
it provides estimates of an individual growth curve
for each subject, including estimated baseline values
and rates of change, while also taking individual varia-
tions into consideration [35, 36, 43]. In addition, LGCM
gives researchers more flexibility to estimate patterns of
change for its ability to establish nonlinear growth trajec-
tories [14, 36].
Only a few determinants have yet been examined for
their association with older adults’ disability trends.
ese determinants have included both mutable deter-
minants, such as health behaviors and social support,
and immutable determinants, such as age and number
of comorbidities [2, 9, 10, 27, 48, 54]. However, the cur-
rent literature remains unclear as to what extent these
determinants affect disability trends among men and
women, and especially how they affect rate of change
in disability [47, 56]. us, our study aimed to examine
both (1) sex differences in disability trends and (2) the
different determinants of the rate of change in disability
for men and women in middle age and older.
Methods
Data andsample
This study used the Taiwan Longitudinal Study on
Aging (TLSA), which was a national population-repre-
sentative survey launched in 1989, aged 50 and up, and
followed up in 1993, 1996, 1999, 2003, and 2007. It was
conducted by the Taiwan Provincial Institute of Fam-
ily Planning (which later became the Bureau of Health
Promotion of the Taiwan Department of Health) and
the University of Michigan, with support from Tai-
wan’s government and the U.S. National Institute on
Aging (Taiwan Provincial Institute of Family Planning
etal., 1989). A second cohort, aged 50–67 years, was
added in 1996 and followed in the subsequent waves.
Data quality and details of the survey have been pre-
sented previously [10, 27, 56]. We included four waves
of survey data—from the 1996 to 2007 surveys—in this
study’s analysis, due to certain key variables are availa-
ble only from the data collected in the 1996–2007 sur-
veys. This study included 3429 people who survived to
the 2007 survey and had completed at least one of the
four surveys for analysis (please see Fig.1 for details).
All subjects provided written informed consent, and
the ethical committee of the Bureau of Health Promo-
tion, Taiwan, approved the national survey study.
Fig. 1 Flow Diagram of the Taiwan Longitudinal Study on Aging Cohort Sample and Follow‑Up Surveys
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Page 3 of 10
Chenetal. BMC Geriatrics (2022) 22:11
Respondents were asked to choose between two
options for sex: Male or Female. Sample weights rep-
resenting Taiwan’s population aged 50 and older as of
1996 were included. Missing values were replaced using
the multiple imputation procedure in Mplus 7.3 [25].
Measures
Disability trends
In this study, we applied multiple-indicator latent growth
curve modeling (LGCM), and included a latent variable
for disability trends assessed by three indicators—activi-
ties of daily living (ADLs [22];), instrumental activities
of daily living (IADLs [21];), and Nagi’s functional limi-
tations [37]. ese three indicators were all measured at
1996, 1999, 2003, and 2007 four time points (Disability
1996 to Disability 2007; please see Fig.2 for illustration).
Using multiple functional outcome measures to
assess functional limitations in the older population
has been recommended in the literature [10, 23, 57].
e National Research Council has suggested includ-
ing functional limitations in addition to ADL and IADL
limitations to better enable researchers to understand
the disability process [38]. LGCM allows researchers to
include multiple indicators to estimate the growth curve
of the general process of functional disability and the
advantages of using multiple-indicator LCGM has been
addressed in previous studies [4, 10, 18, 44].
Details regarding the interview contents of TLSA data
have been presented previously [10, 11]. e three indi-
cators we included—Nagi’s functional limitations, ADL
disability, and IADL disability—assess physical function
from multiple perspectives [44]. e severity level for
each activity in these three indicators was assessed with
four grades, from 0 (no limitation) to 3 (unable to do).
e severity level for each category was then summed
(see Table1).
Factors thatinuence disability trends bysex
Our analysis examined 10 determinants—age, education
level, number of comorbidities, depression, alcohol con-
sumption (yes or no), recreational and physically active
leisure-time activities, social network, social relations,
and use of assistive devices—that have been reported in
earlier studies to influence older adults’ disability trends
[10]. Data on these factors were drawn from the baseline
TLSA survey (the 1996 survey).
Fig. 2 Multiple‑Group Latent Growth Curve Model for Disability and Disablement Factors Among Men and Women 50 Years and Older. Notes:
FLxxxx = Nagi’s functional limitation in xxxx (year); IADLxxxx = instrumental activities of daily living in xxxx (year); ADLxxxx = activitiesof daily living in
xxxx (year); GFDxxxx = general functional disability in xxxx (year). The indicators for each latent disability variable were illustrated for both men and
women
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Page 4 of 10
Chenetal. BMC Geriatrics (2022) 22:11
Age and education level were measured by the actual
year of age and education received. Comorbidities were
measured as number of reported chronic health condi-
tions (e.g., hypertension, diabetes mellitus, heart dis-
ease, stroke, cancer, pulmonary disease, arthritis, gastric
ulcer, liver disease, hip fracture, cataract, renal disease,
gout, and spinal spurs). Depression was assessed by the
10-item version of the Center for Epidemiologic Stud-
ies Depression Scale (CES-D), which represents levels
of depressive symptoms ranging from 0 to 30 [42]. Pres-
ence or absence of alcohol consumption was assessed by
a question about drinking habits.
Leisure-time activities include the following: (1) watch-
ing television, (2) listening to music or radio, (3) reading,
(4) playing mahjongg or chess, (5) gathering with friends
or family, (6) gardening, (7) taking a walk, (8) outdoor
activities such as tai chi, and (9) group activities. Factor
loadings ranged from .500 to .863 [8]. Based on infor-
mation from previous studies, we grouped the first five
activities into recreational leisure-time activities and the
latter four activities into physically active leisure-time
activities [1, 8, 19, 55]. Please see Supplementary Table1
for detailed information about our categorization of lei-
sure-time activities.
Social network was assessed by frequency of contact
with relatives and friends per week [59]. Social support
was assessed with four items measuring level of satisfac-
tion (1–5) with emotional support, resulting in a sum
score ranging from 4 to 20 (higher scores represent greater
satisfaction with support). e four items were “Someone
listens to me,” “Someone cares about me,” “My family cares
about me” (level of satisfaction), and “Someone will take
care of me if I become ill.” e internal consistency was
0.822. and the factor loading ranged from .733 to .800. Use
of assistive devices was assessed by individual’s use of four
types of devices (0–4): glasses, hearing aids, dentures, and
wheelchairs, resulting a sum score ranging from 0 to 4.
Latent growth curve modeling andanalysis
We used both multiple-indicator and multiple-group
LGCM to test the different influences of the determi-
nants for both groups (men and women) and applied
testing for partial invariance [6]. We applied the second-
order growth model [30] with the assumption that all
Table 1 Characteristics of the sample (N = 3249)
Note. *p < 0.05, **p < 0.01, ***p < 0.001. FLxxxx Nagi’s functional limitation in xxxx (year), IADLxxxx Instrumental activities of daily living in xxxx (year), ADLxxxx Activities
of daily living in xxxx (year)
Min Max Men (N = 1718) Women (N = 1711) P-value
Mean SD Mean SD
Determinants (1996)
Age 50 96 63.960 8.113 63.880 8.409 0.778
Education 0 17 6.950 4.609 3.200 3.897 < 0.001
Comorbidities 1 5 3.530 1.043 3.130 1.043 < 0.001
Depression 0 30 4.190 4.689 5.990 5.787 < 0.001
No alcohol consumption 0 1 .630 .484 .930 .252 < 0.001
Leisure‑time activities, recreational 0 5 2.900 1.0200 2.170 0.947 < 0.001
Leisure‑time activities, physically active 0 4 1.300 1.045 1.150 1.000 < 0.001
Social network 0 176 20.060 17.727 18.200 15.928 < 0.001
Social support 4 20 16.270 2.909 16.21 2.882 0.515
Use of assistive devices 0 3 1.280 0.728 1.250 0.707 0.153
FL1996 0 24 0.83 2.541 2.18 3.713 < 0.001
FL1999 0 24 1.42 3.115 3.52 4.712 < 0.001
FL2003 0 24 2.51 4.413 5.32 5.93 < 0.001
FL2007 0 24 4.01 6.191 6.99 7.024 < 0.001
ADL1996 0 18 0.08 0.941 0.12 0.997 0.200
ADL1999 0 18 0.13 1.051 0.23 1.333 0.010
ADL2003 0 18 0.34 1.873 0.72 2.719 < 0.001
ADL2007 0 18 1.19 3.795 1.9 4.589 < 0.001
IADL1996 0 18 0.42 1.657 1.22 2.671 < 0.001
IADL1999 0 18 0.55 1.826 1.67 3.252 < 0.001
IADL2003 0 18 1.24 3.102 2.87 4.567 < 0.001
IADL2007 0 18 2.62 4.944 4.43 5.766 < 0.001
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Page 5 of 10
Chenetal. BMC Geriatrics (2022) 22:11
indicators shared the same trait and the same trait growth
process, and all indicators shared the same state residual
component within scaling differences [4]. Figure2 shows
the setup of our multiple-group LGCM. Each latent vari-
able of disability was identified by three physical function
measures: ADLs, IADLs, and functional limitations. e
upper half and lower half of Fig.2 indicate the baseline
and rate of change in disability, which indicates speed of
progression toward disability in each group. is growth
process contains two latent factors of baseline disabil-
ity and two latent factors of disability slope (i.e., rate of
change in disability per year from 1996 to 2007) over the
11 years of the study period, for men and women respec-
tively. Baseline and rate of change in disability were thus
measured by four latent variables, Disability 1996 to Dis-
ability 2007. e determinants were included to assess
the impact on disability baselines and slopes across sex.
e measurement errors were set to be correlated.
Our analysis was based on comparisons of differ-
ent models in which parameters were constrained or
not constrained to be equal. e analysis procedure was
multi-stepped and included (1) testing unconditional
multiple-group nonlinear and linear growth models to dis-
ability trends among men and women and comparing the
model fit; (2) testing unconstrained models allowing all
parameters to be freely estimated across groups; (3) testing
constrained models assuming that parameters are equal
across groups, and comparing by using chi-square differ-
ence tests between fully constrained and unconstrained
models; and (4) comparing structural parameters by sys-
tematically constraining and unconstraining specific paths
to determine which paths contribute to significant differ-
ences between the two. Equality constraints were imposed
on the 10 determinants assessed across groups [40, 43].
In this study, the LGCM was fit to data using Mplus
(version 7.1) with a robust maximum likelihood estima-
tor. Four model fit indexes were applied to evaluate the
adequacy of model fit [24]: (a) chi-square statistics [20],
(b) the Bentler Comparative Fit Index (i.e., CFI ≥ 0.9 [3,
5];, and (c) root mean square error of approximation (i.e.,
RMSEA ≤0.05) with 90% confidence interval [45]. Signif-
icant chi-square difference (∆χ2) tests, which were used
to determine significant differences between constrained
and unconstrained models, indicated determinants that
showed significantly different influences on men’s and
women’s disability trends [36].
Results
Sample characteristics
e sample was about 50% women, with a mean age
in 1996 of 63.96 (SD = 8.113) years for men and 63.88
(SD = 8.409) years for women. Detailed information
regarding the sample included for analysis is presented
in Table1, which also shows that the level of disability
among men and women continually increased over time.
Both men and women started out with less severe dis-
abilities in 1996, with grades of 0.83 (SD = 2.54) and
2.18 (SD = 3.71) for Nagi’s functional limitations, 0.42
(SD = 1.66) and 1.22 (SD = 2.67) for IADLs, and 0.08
(SD = 0.94) and 0.12 (SD = 1.00) for ADLs. Baseline
grades for men and women were significantly different
for Nagi’s functional limitations (p < 0.001) and IADLs
(p < 0.001). Functional disabilities among these groups
increased over the years; in 2007, more severe disabil-
ity was measured in Nagi’s functional limitations (men:
1.04, SD = 6.19 vs. women: 6.99, SD = 7.04), IADLs (2.62,
SD = 4.94; 4.43, SD = 5.77), and ADLs (1.19, SD = 3.80;
1.9, SD = 4.59).
Latent growth curve model
e unconditional modeling results showed that the non-
linear models fit better to each group’s disability trends
(χ2 [66, N = 3429] = 768.275 [men 303.501 vs. women
464.774], p < 0.001; CFI = .941; RMSEA = .058). e
multiple-group model showed that disabilities increased
more slowly among women than men at Wave 3 of the
survey, but increased at a faster rate among women at
the Wave 4 survey. Baseline disability levels and rate of
change in disability were constrained in separate models
and compared to the unconditional and unconstrained
model. e baseline disability levels showed no signifi-
cant difference between the two groups, but once dis-
ability began, the progression toward greater disability
was almost 18% faster among women than men (B: 0.694
men vs. 0.817 women; p < 0.01). e detailed results of
the nonlinear LGCM for disability are presented in Sup-
plementary Table2 and Supplementary Fig.1.
Factors thatinuence men andWomen’s rate ofchange
indisability dierently
e conditional nonlinear LGCM also fit well to the dis-
ability trends (χ2 [303, N = 3429] = 1824.20 [745.343
men vs. 1078.857 women], p < 0.001; CFI = .928;
RMSEA = .054). Among women compared to men,
greater age added 1.23 times the burden to the rate of
change in disability (βAge slope: 0.380, p < 0.001 men vs.
0.467, p < 0.01 women, ∆χ2 = 11.997, p < 0.001). Higher
education level was associated with lower rate of change
in disability for women but not men, although the differ-
ential impact between the two groups was only margin-
ally significant (β Education slope: −0.061, p > 0.05 men vs .
-0.066, p < 0.01 women, ∆χ2 = 3.623, p = 0.057). Number
of comorbidities was found to add burden to the rate of
change in disability in both groups, but the impact of
not significantly different between groups (β Comorbidi-
ties slope: 0.108, p < 0.01 men vs. 0.146, p < 0.001 women,
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Page 6 of 10
Chenetal. BMC Geriatrics (2022) 22:11
∆χ = 2.236, p > 0.05). Finally, both men and women ben-
efited from the effect of physically active leisure-time
activity on slowing the rate of change toward greater
disability. Although the differential impact between the
two groups on the rate of change was again only margin-
ally significant, men tended to benefit more than women
from physically active leisure-time activities (β Physically
active LTA slope: −0.092, p < 0.01 men vs. -0.063, p < 0.05
women, ∆χ2 = 3.672, p = 0.055).
Other factors studied, such as depression, alcohol hab-
its, and having better social networks, showed differential
impacts between the two groups only on baseline dis-
abilities and not the rate of change in disability. Please see
Table2 for details.
Discussion
Past studies have returned inconsistent results on
whether sex is associated with different levels of burden
on disability trends among middle-aged and older adults
[10, 26, 27, 46, 56]. Our study findings advance this body
of knowledge by confirming that while middle-aged and
older men and women demonstrate no differences in
baseline disability, once disability has begun, the rate of
change in disability is faster among women than men—
18% faster in our study. However, it is necessary to be
cautious when interpreting difference in rate of change
between groups. In this case, since both men’s and wom-
en’s trends progressed in a curved manner, the differ-
ences in rate of change may also be different across time.
Another key contribution from this research lies in its
focus on how mutable and immutable determinants asso-
ciate with disability trends differently by sex. Age posed
greater risks to disability progression among women than
men, while women received marginally more benefit than
men from education. However, both women and men
benefited from engaging with physically active leisure-
time activities through a slower progression in disability.
Age adds more burden forwomen thanmen
Age and comorbidities are known to be significant fac-
tors for disability ([12, 13, 15, 27, 29, 46, 54]). Most deter-
minants identified in past studies [17, 32, 47, 48, 54, 56]
showed significantly different influences by sex only on
baseline disabilities in the current study. Age was the
only determinant that our study showed to have differ-
ent influences on change in disability among middle-aged
and older men and women. We found that age added
1.23 times the burden in rate of change in disability on
women. is indicates that age may also widen the exist-
ing gap between men and women in the rate of change in
disability.
Table 2 Differential impacts of determinants on men’s and women’s disability trends (N = 3429)
Note. *p < 0.05, **p < 0.01, ***p < 0.001
a Marginally signicant p-values of Χ2di test (p < 0.1) are presented when at least one of the estimates for men or women were signicant
Determinants Intercept (Baseline) Slope (Rate of Change)
Men Women Χ2di test Men Women Χ2di test
Estimate β (SE)
Standardize β Estimate β (SE)
Standardize β Estimateβ (SE)
Standardize β Estimateβ (SE)
Standardize β
Age 0.021 (0.008) 0.100* 0.083 (0.008)
0.265*** 22.227*** 0.017 (0.002)
0.380*** 0.026 (0.002)
0.467*** 11.997***
Education 0.003 (0.013) 0.009 −0.033 (0.012)
-0.055** 5.689* −0.004 (0.002) ‑0.061 −0.007 (0.003)
-0.066** 3.632 (0.057)a
Comorbidities 0.101 (0.036) 0.090** 0.242 (0.045)
0.157*** 6.686* 0.026 (0.009) 0.108** 0.041 (0.009)
0.146*** 2.263
Depression 0.073 (0.017)
0.220*** 0.101 (0.017)
0.240***
−0.990 −0.001 (0.002) ‑0.013 −0.001 (0.002)‑0.015 2.535
No alcohol consump‑
tion 0.271 (0.064)
0.086*** 0.09 (0.14) 0.024 4.785** −0.005 (0.019) ‑0.007 −0.034 (0.041) ‑0.020 3.444
Leisure‑time activities,
recreational
−0.163 (0.067)
-0.109**
−0.127 (0.054)
-0.069** 2.527 −0.004 (0.009) ‑0.012 −0.001 (0.012) ‑0.002 3.728
Leisure‑time activities,
physically active
−0.136 (0.04)
-0.094***
−0.291 (0.06)
-0.125*** 4.970* −0.029 (0.008)
-0.092***
−0.026 (0.011)
-0.063* 3.672 (0.055)a
Social network −0.001 (0.002) ‑0.013 −0.005 (0.003)
-0.037* 4.510* 0.00 (0.001) −0.004 0.000 (0.001) ‑0.007 3.178
Social support 0.023 (0.015) 0.043 0.036 (0.025) 0.043 2.490 ‑0.004 (0.003) ‑0.033 −0.003 (0.005) ‑0.021 3.021
Use of assistive devices 0.149 (0.064) 0.071** 0.108 (0.099) 0.032 2.123 −0.016 (0.015) ‑0.035 −0.02 (0.016) ‑0.033 2.979
Model fit χ2 [303, N = 3429] = 1824.20 [Men: 745.343 vs. Women: 1078.857], p < 0.001; CFI = .928; RMSEA = .054
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Page 7 of 10
Chenetal. BMC Geriatrics (2022) 22:11
Some may argue that women have longer life expec-
tancies, and therefore may experience faster increases
in disability simply due to their older age. However, in
our analysis, the mean age of the groups of men and
women in all four waves of data was not significantly
different (see Supplementary Table1). In addition, past
studies have suggested that if chronic illness is well
controlled, aging is not inevitably related to functional
decline [10, 46].
Our study further indicates that while age adds more
burden to women than to men in terms of rate of change
in disability, comorbidities add burden to both groups.
Our results showed that even with age controlled in the
model, number of chronic illnesses still added as much as
two times the burden to women’s baseline disability as to
men’s baseline disability. However, the number of chronic
illnesses added burden to rate of change in disability
equally for men and women. us, preventing the devel-
opment of chronic illness and decreasing the numbers of
chronic illnesses should be the first priority for maintain-
ing physical function for both men and women. Prevent-
ing disability during aging, especially for women, should
be a focus in future policy-making [12].
e influence of age on the overall disability trend may
also be different between adults in middle age and older.
An earlier study [56] that used the same dataset as our
study also examined trends among adults age 50 years
and older. at study found that those who were 50 to
59 years old at baseline showed similar patterns of dis-
ability trends as those in other age ranges, but had dif-
ferent probabilities of entering into different disability
trend patterns [56]. Yu etal.[54] also indicated that those
who are younger are more likely to enter a heathier
trend. Careful attention and explanation of participants’
age ranges is necessary, and further studies are recom-
mended to examine the influence of age on disability
trends between middle-aged and older adults.
Higher education may benet women butnotmen
Past studies have shown that a higher education can be
a protective factor against developing disability in later
life. More-educated older adults invest in late-life health
through healthier behaviors and are thus at less risk of
developing and increasing functional limitations or phys-
ical disabilities [11, 29]. However, the role of education on
disability for men versus women has been controversial
in the literature. In our study, women with higher educa-
tion levels not only had lower baseline disability but also
tended to show slower progression toward greater dis-
ability. No beneficial effects of education were observed
among men, on baseline or progression toward disability.
ese findings are not consistent with those of past
studies. Zimmer etal. [56] pointed out the possibility of
an intertwined influence between sex and education on
older adults’ disabilities, noting that education seems
to be less important to predicting disability trajectory
among women than it is among men, and that women
with less education than their husbands may benefit in
part from influences tied to the husbands’ characteristics.
In contrast, we found education had a beneficial effect
on disability only for women and not for men, though
the difference was only marginally significant. Other past
studies have emphasized that women’s health behaviors
are associated with their levels of education and health
literacy [28]. Women with higher education may particu-
larly benefit from such characteristics and therefore ben-
efit from lower baseline and slower progression toward
disabilities [11].
Based on our study findings, then, continuing to pro-
mote higher education levels for women in Taiwan
should be considered in future health policy-making.
Past studies have also suggested that promoting health
literacy among women promotes better health outcomes
and physical function, so this could also be considered a
policy priority [11, 31, 52].
As to why men did not benefit from higher education
in this study, a review study has pointed out that men
responded better toward male-specific health-related
information, rather than assuming all health education
efforts are equally effective with everyone [41]. Planning
different health education campaigns for women and
men is recommended.
Leisure-time activities benet bothwomen andmen
Many past studies have reported that being physically
active reduces disability in older adults and prevents
new-onset ADL disabilities [17, 47, 48, 54, 56]. Strobl
etal. [47] suggested that men benefit more than women
from physically active leisure-time activities in terms
of developing late-life disability, but that once disability
begins, there appears to be no further association with
the severity of disability. Our findings were thus partly in
line with the results of previous studies [11, 47].
Our study findings indicated that once disability began,
physically active leisure-time activities were strongly
associated with slower progress toward severe disability
among both men and women. Our study further showed
that men seemed to benefit more than women from
physically active leisure-time activities in terms of slower
progression toward disability. Past studies have suggested
that sex differences might contribute to different out-
comes from physical activities, such as non-fatal chronic
conditions, lower muscle strength, and lower bone den-
sity in women [39].. Past studies have also shown that
men and women age 50 and older prefer different physi-
cal activities [34, 49] and that the percentage of women
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Page 8 of 10
Chenetal. BMC Geriatrics (2022) 22:11
reporting high levels of physical activity was significantly
lower than the percentage of men reporting high activ-
ity levels [7, 47]. ese may also lead to fewer disadvan-
tages in making slower progression in disabilities among
women. Further research is needed to understand to
what extent the level of physical activity affects middle-
aged and older men’s and women’s rate of change in
disability.
Strobl etal. [47] have pointed out that their study sam-
ple cannot be representative of all older people, particu-
larly those who do not choose to participate in research
due to disabling conditions. Our study was based on a
representative survey of the population, which included
people who had and had not participated in leisure-time
activities. Different target samples might also contribute
different findings from the current study and past studies
[47]. However, promoting physically active leisure-time
activities for both sexes is a promising strategy.
Limitations
Several limitations need to be addressed. e first is that
for parsimony of the model, we investigated only 10 of
the commonly studied determinants of sex disparities. A
number of other variables known to influence the devel-
opment of disability (e.g., cognitive impairments and eco-
nomic status) were not included due to data availability.
e current study can still serve as a foundation for fur-
ther studies that examine a more comprehensive set of
determinants and their associations with different func-
tional outcomes in men and women. e second limita-
tion is that, as with many longitudinal studies, this study
had selective attrition. We included in the analysis only
those men and women who survived the 11-year period
from 1996 to 2007.
In addition, although differences in mean age of the
men and women included in our analysis remained non-
significant across all four waves of data, those who were
not included were more likely to be older and have more
severe disabilities. us, our results shall be interpreted
with caution.
e advantage of LGCM is in examining the distribu-
tion of trajectories that vary continuously across indi-
viduals [57]. e disadvantage is that including deceased
individuals may lead to sampling error and bias the esti-
mation of the disability trajectory, particularly for those
who experience early onset of disability [56]. As a result,
LGCM tends to favor separate estimates for surviving and
deceased respondents [53], and we decided not to include
the deceased in our analyses. However, this may limit our
ability to generalize our findings to those who died, and
our results should thus be interpreted with caution.
Conclusions
To date, very few population-based studies have aimed
to understand the issue of sex-specific differences in
the impact of determinants on the rate of change in
disability among men and women in middle aged and
older. We found that while women did not bear a larger
burden of baseline disability than men, once disability
began, women’s progression toward greater disabil-
ity occurred faster. Only age had a different impact by
sex on the rate of change in disability; while education
and physically active leisure-time activities marginally
benefited both women and men through slower pro-
gression toward disabilities. Physically active leisure-
time activities are mutable determinants that promise
to be beneficial for both sexes, though men seemed
to benefit more than women from participating in
these activities. Promoting physically active leisure-
time activities should be a priority for future policy
and interventions aimed at maintaining adults’ physi-
cal functioning over time—for both men and women.
Better control of chronic illness, preventing disability
at earlier ages, and promoting middle-aged and older
women’s education also remain important policy goals.
Abbreviations
FL : Functional limitation; IADL : Instrumental activities of daily living; ADL:
Activities of daily living.
Supplementary Information
The online version contains supplementary material available at https:// doi.
org/ 10. 1186/ s12877‑ 021‑ 02574‑3.
Additional le1: Supplementary Table1. Detailed Descriptions of the
Measures (N = 3249). Supplementary Table2. Descriptive Results and
Factor Loadings of Nonlinear Unconditional LGCM for Disability Trends in
Four Waves of Survey Data (N = 3429).
Additional le2: Supplementary Figure1. Disability Trends of Men and
Women Over 11 years of study period.
Acknowledgements
The authors wish to express their gratitude to the National Science Council for
its generous financial support and the Department of Statistics and the Social
and Family Affairs Administration at the Ministry of Health and Welfare for its
gracious help with data access.
Authors’ contributions
YC planned the study, performed all statistical analyses, interpreted the results,
and wrote the paper. YT supervised the data analysis, performed statistical
analyses, interpreted the results, helped write the paper, and contributed to
revising the paper. DC and TC helped plan the study and revise the manu‑
script. HY helped clear the data, performed part of the analysis, helped inter‑
pret the result, and contributing to revising the manuscript. WC performed
statistical analysis, interpreted the results. All authors have read and approved
the manuscript.
Funding
This work was supported by the Ministry of Science and Technology
(MOST105–2410‑H‑002‑214‑MY3, MOST108–2410‑H‑002‑123‑SS2, and MOST
109–2634‑F‑ 002‑044.). The funding body (MOST) has no influence on or
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Page 9 of 10
Chenetal. BMC Geriatrics (2022) 22:11
contribution to the study design, data collection, analysis and interpretation of
data, or writing the manuscript.
Availability of data and materials
The data that support the findings of this study are available from Taiwan’s
Health and Welfare Data Science Center but restrictions apply to the avail‑
ability of these data, which were used under license for the current study,
and so are not publicly available. Data are however available from the
authors upon reasonable request and with permission of Taiwan’s Health
and Welfare Data Science Center. Further contact information is available
through the following URL: https:// dep. mohw. gov. tw/ dos/ cp‑ 5119‑ 59201‑
113. html.
Declarations
Ethics approval and consent to participate
The Taiwan Longitudinal Study on Aging (TLSA) was a national population‑
representative survey launched in 1989, aged 50 and up, and followed
up in 1993, 1996, 1999, 2003, and 2007. The current study was approved
by the Research Ethics Committee of National Taiwan University Hospital
(2013HS064, 201503016 W); We had applied for accessing and using the TLSA
data to the Taiwan’s Health and Welfare Data Science Center, Ministry of Health
and Welfare (H102054).
Consent for publication
Not applicable.
Competing interests
The authors declare that they have no competing interests.
Author details
1 Institute of Health Policy and Management, College of Public Health, National
Taiwan University, Room 633, No. 17, Xu‑Zhou Road, Taipei 100, Taiwan.
2 Institute of Health Behaviors and Community Sciences, College of Public
Health, National Taiwan University, Room 636, No. 17, Xu‑Zhou Road, Tai‑
pei 100, Taiwan. 3 Institute of Epidemiology and Preventive Medicine, College
of Public Health, National Taiwan University, Room 539, No. 17, Xu‑Zhou Road,
Taipei 100, Taiwan. 4 Depar tment of Gerontology and Health Care Manage‑
ment, Chang Gung University of Science and Technology, Room 1406, No. 261,
Wenhua 1st Rd, Taoyuan 333, Taiwan.
Received: 19 January 2021 Accepted: 22 October 2021
References
1. Adams KB, Leibbrandt S, Moon H. A critical review of the literature
on social and leisure activity and wellbeing in later life. Ageing Soc.
2011;31(4):683–712. https:// doi. org/ 10. 1017/ S0144 686X1 00010 91.
2. Aday LA, Andersen R. A framework for the study of access to medical
care. Health Serv Res. 1974;9(3):208–20.
3. Bentler PM. Comparative fit indexes in structural models. Psychol Bull.
1990;107(2):238–46. https:// doi. org/ 10. 1037/ 0033‑ 2909. 107.2. 238.
4. Bishop J, Geiser C, Cole DA. Modeling latent growth with multiple indica‑
tors: a comparison of three approaches. Psychol Methods. 2014. https://
doi. org/ 10. 1037/ met00 00018.
5. Bollen KA. A new incremental fit index for general structural equations
models. Sociol Methods Res. 1989;17:303–16.
6. Byrne BM. Structural equation modeling with Mplus: Basic concepts,
applications, and programming. New York: Routledge; 2013.
7. Cawthon PM, Fink HA, Barrett‑Connor E, Cauley JA, Dam TT, Lewis CE,
et al. Alcohol use, physical performance, and functional limitations in
older men. J Am Geriatr Soc. 2007;55(2):212–20. https:// doi. org/ 10. 1111/j.
1532‑ 5415. 2007. 01062.x.
8. Chao S‑F. Changes in leisure activities and dimensions of depres‑
sive symptoms in later life: a 12‑year follow‑up. Gerontologist.
2014a;56(3):397–407. https:// doi. org/ 10. 1093/ geront/.
9. Chen CM, Mullan J, Su YY, Griffiths D, Kreis IA, Chiu HC. The longitudinal
relationship between depressive symptoms and disability for older
adults: a population‑based study. J Gerontol A Biol Sci Med Sci. 2012.
https:// doi. org/ 10. 1093/ gerona/ gls074.
10. Chen YM, Chen DR, Chiang TL, Tu YK, Yu HW. Determinants of rate of
change in functional disability: an application of latent growth curve
modeling. Arch Gerontol Geriatr. 2016;64:21–8. https:// doi. org/ 10. 1016/j.
archg er. 2015. 11. 012.
11. Chen YM, Tu YK, Yu HW, Chiu TY, Chiang TL, Chen DR, et al. Leisure time
activities as mediating variables in functional disability progression:
an application of parallel latent growth curve modeling. PLoS One.
2018;13(10):e0203757. https:// doi. org/ 10. 1371/ journ al. pone. 02037 57.
12. Chiu CJ, Wray LA, Ofstedal MB. Diabetes‑related change in physical dis‑
ability from midlife to older adulthood: evidence from 1996‑2003 survey
of health and living status of the elderly in Taiwan. Diabetes Res Clin
Pract. 2011;91(3):413–23. https:// doi. org/ 10. 1016/j. diabr es. 2010. 12. 003.
13. Chou KL, Leung JCB. Disability trends in Hong Kong community‑
dwelling Chinese older adults: 1996, 2000, and 2004. J Aging Health.
2008;20(4):385–404. https:// doi. org/ 10. 1177/ 08982 64308 315852.
14. Duncan TE, Duncan SC, Strycker LA. An introduction to latent variable
growth curve modeling: concepts, issues, and publication. 2nd ed.
Mahwah; 2006.
15. Freedman VA, Martin LG, Schoeni RF. Recent trends in disability and
functioning among older adults in the United States: a systematic review.
JAMA. 2002;288(24):3137–46 https:// doi. org/ jrv20 048 [pii].
16. Fried LP, Ferrucci L, Darer J, Williamson JD, Anderson G. Untangling the
concepts of disability, frailty, and comorbidity: implications for improved
targeting and care. J Gerontol A Biol Sci Med Sci. 2004;59(3):255–63
http:// www. ncbi. nlm. nih. gov/ pubmed/ 15031 310.
17. Hamer M, Lavoie KL, Bacon SL. Taking up physical activity in later life and
healthy ageing: the English longitudinal study of ageing. Br J Sports Med.
2014;48(3):239–43. https:// doi. org/ 10. 1136/ bjspo rts‑ 2013‑ 092993.
18. Hancock G, Kuo W‑L, Lawrence F. An illustration of second‑order latent
growth models. Struct Equ Model Multidiscip J. 2001;8(3):470–89. https://
doi. org/ 10. 1207/ s1532 8007s em0803_7.
19. Hu Y‑H, Chiu C‑J, Wong JD, Lin D‑C, Wray LA. The role of leisure activities
in the relationship between marital transition in later midlife and psycho‑
logical well‑being trajectories. Int J Aging Hum Dev. 2017;86(4):327–46.
https:// doi. org/ 10. 1177/ 00914 15017 729683.
20. Jöreskog KG, Sörbom D. Advanced in factor analysis and structural equa‑
tion models; 1979.
21. Kane RA, Kane RL. Assessing the elderly: a practical guide to measure‑
ment: Lexington Books; 1981.
22. Katz S, Akpom CA. Index of ADL. Med Care. 1976;14(5 Suppl):116–8.
23. Keysor J. Does late‑life physical activity or exercise prevent or minimize
disablement? A critical review of the scientific evidence. Am J Prev Med.
2003;25(3):129–36. https:// doi. org/ 10. 1016/ s0749‑ 3797(03) 00176‑4.
24. Kline RB. Principles and practice of structural equation modeling: Guil‑
ford; 2011.
25. Kmetic A, Joseph L, Berger C, Tenenhouse A. Multiple imputation to
account for missing data in a survey: estimating the prevalence of osteo‑
porosis. Epidemiology. 2002;13(4):437–44 http:// www. ncbi. nlm. nih. gov/
entrez/ query. fcgi? cmd= Retri eve& db= PubMe d& dopt= Citat ion& list_
uids= 12094 099.
26. Liang J, Bennett JM, Shaw BA, Quinones AR, Ye W, Xu X, et al. Gender
differences in functional status in middle and older age: are there any age
variations? J Gerontol B Psychol Sci Soc Sci. 2008;63(5):S282–92 http://
www. ncbi. nlm. nih. gov/ pubmed/ 18818 448.
27. Liang J, Wang CN, Xu X, Hsu HC, Lin HS, Lin YH. Trajectory of functional
status among older Taiwanese: gender and age variations. Soc Sci Med.
2010;71(6):1208–17. https:// doi. org/ 10. 1016/j. socsc imed. 2010. 05. 007.
28. Liu YB, Liu L, Li YF, Chen YL. Relationship between health literacy, health‑
related behaviors and health status: a survey of elderly Chinese. Int J
Environ Res Public Health. 2015;12(8):9714–25. https:// doi. org/ 10. 3390/
ijerp h1208 09714.
29. Martin LG, Zimmer Z, Hurng B‑S. Trends in late‑life disability in Taiwan,
1989–2007: the roles of education, environment, and technology. Popul
Stud. 2011;65(3):289–304. https:// doi. org/ 10. 1080/ 00324 728. 2011.
604730.
30. McArdle JJ. Causal modeling applied to psychonomic systems simulation.
Behav Res Methods Instrum. 1980;12:193–209.
31. McDougall GJ Jr, Mackert M, Becker H. Memory performance, health
literacy, and instrumental activities of daily living of community residing
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

Page 10 of 10
Chenetal. BMC Geriatrics (2022) 22:11
•
fast, convenient online submission
•
thorough peer review by experienced researchers in your field
•
rapid publication on acceptance
•
support for research data, including large and complex data types
•
gold Open Access which fosters wider collaboration and increased citations
maximum visibility for your research: over 100M website views per year
•
At BMC, research is always in progress.
Learn more biomedcentral.com/submissions
Ready to submit your research
Ready to submit your research
? Choose BMC and benefit from:
? Choose BMC and benefit from:
older adults. Nurs Res. 2012;61(1):70–5. https:// doi. org/ 10. 1097/ NNR.
0b013 e3182 3b18f4.
32. McLaughlin D, Leung J, Pachana N, Flicker L, Hankey G, Dobson A. Social
support and subsequent disability: it is not the size of your network that
counts. Age Ageing. 2012. https:// doi. org/ 10. 1093/ ageing/ afs036.
33. Moen P, Chermack K. Gender disparities in health: strategic selection,
careers, and cycles of control. J Gerontol B Psychol Sci Soc Sci. 2005;60
Spec No 2:99–108. https:// doi. org/ 10. 1093/ geronb/ 60. speci al_ issue_2.
s99.
34. Moschny A, Platen P, Klaassen‑Mielke R, Trampisch U, Hinrichs T. Physical
activity patterns in older men and women in Germany: a cross‑sec‑
tional study. BMC Public Health. 2011;11:559. https:// doi. org/ 10. 1186/
1471‑ 2458‑ 11‑ 559.
35. Muthen B, Asparouhov T. Growth mixture modeling: analysis with non‑
Gaussian random effects. In: Fitzmaurice MG, Davidian GV, Molenberghs
G, editors. Longitudinal data analysis: Chapman & Hall/CRC; 2009. p.
143–65.
36. Muthen L, Muthen B. Mplus User’s guide. 4th ed; 2012. Muthen & Muthen
37. Nagi SZ. An epidemiology of disability among adults in the United States.
Milbank Mem Fund Q Health Soc. 1976;54(4):439–67.
38. National Research Council. Improving the measurement of late‑life
disability in population surveys: beyond ADLs and IADLs, summary of a
workshop: The National Academies Press; 2009.
39. Oksuzyan A, Petersen I, Stovring H, Bingley P, Vaupel JW, Christensen K.
The male‑female health‑survival paradox: a survey and register study of
the impact of sex‑specific selection and information bias. Ann Epidemiol.
2009;19(7):504–11. https:// doi. org/ 10. 1016/j. annep idem. 2009. 03. 014.
40. Park I, Schutz RW. An introduction to latent growth models: analysis
of repeated measures physical performance data. Res Q Exerc Sport.
2005;76(2):176–92 http:// www. ncbi. nlm. nih. gov/ pubmed/ 16128 485.
41. Peak T, Gast JA. Aging Men’s health‑related behaviors. SAGE Open.
2014;4(4). https:// doi. org/ 10. 1177/ 21582 44014 558044.
42. Radloff LS. The use of the Center for Epidemiologic Studies Depression
Scale in adolescents and young adults. J Youth Adolesc. 1991;20(2):149–
66. https:// doi. org/ 10. 1007/ BF015 37606.
43. Rogosa D, Brandt D, SZimowski M. A growth curve approach to the meas‑
urement of change. Psychol Bull. 1982;92(3):726–48.
44. Spector WD, Fleishman JA. Combining activities of daily living with instru‑
mental activities of daily living to measure functional disability. J Gerontol
B Psychol Sci Soc Sci. 1998;53(1):S46–57. https:// doi. org/ 10. 1093/ geronb/
53b.1. s46.
45. Steiger JH. Structural model evaluation and modification. Multivar Behav
Res. 1990;25:173–80.
46. Stenholm S, Westerlund H, Head J, Hyde M, Kawachi I, Pentti J,
et al. Comorbidity and functional trajectories from midlife to old
age: the health and retirement study. J Gerontol A Biol Sci Med Sci.
2015;70(3):332–8. https:// doi. org/ 10. 1093/ gerona/ glu113.
47. Strobl R, Muller M, Thorand B, Linkohr B, Autenrieth CS, Peters A, et al.
Men benefit more from midlife leisure‑time physical activity than women
regarding the development of late‑life disability‑‑results of the KORA‑age
study. Prev Med. 2014;62:8–13. https:// doi. org/ 10. 1016/j. ypmed. 2014. 01.
017.
48. Tak E, Kuiper R, Chorus A, Hopman‑Rock M. Prevention of onset and
progression of basic ADL disability by physical activity in community
dwelling older adults: a meta‑analysis. Ageing Res Rev. 2013;12(1):329–38.
https:// doi. org/ 10. 1016/j. arr. 2012. 10. 001.
49. Thandi MKG, Phinney A, Oliffe JL, Wong S, McKay H, Sims‑Gould J, et al.
Engaging older men in physical activity: implications for health promo‑
tion practice. Am J Mens Health. 2018;12(6):2064–75. https:// doi. org/ 10.
1177/ 15579 88318 792158.
50. Tsay S. The long‑term care 2.0 version of the community‑based model in
Taiwan. Taipei: Ministry of Health and Welfare; 2016.
51. Johnsson‑Latham G. Power and Privileges: Gender Discrimination and
Poverty. Stockholm: The Ministry for Foreign Affairs, Sweden; 2004.
52. Wolf MS, Feinglass J, Thompson J, Baker DW. In search of ’low health
literacy’: threshold vs. gradient effect of literacy on health status and mor‑
tality. Soc Sci Med. 2010;70(9):1335–41. https:// doi. org/ 10. 1016/j. socsc
imed. 2009. 12. 013.
53. Yang Y, Lee LC. Dynamics and heterogeneity in the process of human
frailty and aging: evidence from the U.S. older adult population. J
Gerontol B Psychol Sci Soc Sci. 2010;65b(2):246–55. https:// doi. org/ 10.
1093/ geronb/ gbp102.
54. Yu HW, Chen DR, Chiang TL, Tu YK, Chen YM. Disability trajectories and
associated disablement process factors among older adults in Taiwan.
Arch Gerontol Geriatr. 2015;60(2):272–80. https:// doi. org/ 10. 1016/j. archg
er. 2014. 12. 005.
55. Yu HW, Chiang TL, Chen DR, Tu YK, Chen YM. Trajectories of leisure activity
and disability in older adults over 11 years in Taiwan. J Appl Gerontol.
2018;37(6):706–27. https:// doi. org/ 10. 1177/ 07334 64816 650800.
56. Zimmer Z, Martin LG, Jones BL, Nagin DS. Examining late‑life functional
limitation trajectories and their associations with underlying onset, recov‑
ery, and mortality. J Gerontol B Psychol Sci Soc Sci. 2014;69(2):275–86.
https:// doi. org/ 10. 1093/ geronb/ gbt099.
57. Zimmer Z, Martin LG, Nagin DS, Jones BL. Modeling disability trajectories
and mortality of the oldest‑old in China. Demography. 2012;49(1):291–
314. https:// doi. org/ 10. 1007/ s13524‑ 011‑ 0075‑7.
58. Zimmer Z, Martin LG, Nagin DS, Jones BL, Hrung BS. Disability trajectories
by age, sex, and education, among older adults in Taiwan. 2009. http://
www. ined. fr/ fichi er/t_ telec harge ment/ 37859/ telec harge ment_ fichi er_
fr_ zimmer. pdf
59. Zunzunegui MV, Alvarado BE, Del Ser T, Otero A. Social networks, social
integration, and social engagement determine cognitive decline in
community‑dwelling Spanish older adults. J Gerontol B Psychol Sci Soc
Sci. 2003;58(2):S93–S100 http:// www. ncbi. nlm. nih. gov/ pubmed/ 12646
598. http:// www. ncbi. nlm. nih. gov/ pmc/ artic les/ PMC38 33829/ pdf/ S93.
pdf.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in pub‑
lished maps and institutional affiliations.
Content courtesy of Springer Nature, terms of use apply. Rights reserved.

1.
2.
3.
4.
5.
6.
Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”), for small-
scale personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are maintained. By
accessing, sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use (“Terms”). For these
purposes, Springer Nature considers academic use (by researchers and students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or a personal
subscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription
(to the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the Creative Commons license used will
apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data internally within
ResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not
otherwise disclose your personal data outside the ResearchGate or the Springer Nature group of companies unless we have your permission as
detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that Users may
not:
use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to circumvent access
control;
use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil liability, or is
otherwise unlawful;
falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by Springer Nature in
writing;
use bots or other automated methods to access the content or redirect messages
override any security feature or exclusionary protocol; or
share the content in order to create substitute for Springer Nature products or services or a systematic database of Springer Nature journal
content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates revenue,
royalties, rent or income from our content or its inclusion as part of a paid for service or for other commercial gain. Springer Nature journal
content cannot be used for inter-library loans and librarians may not upload Springer Nature journal content on a large scale into their, or any
other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any information or
content on this website and may remove it or features or functionality at our sole discretion, at any time with or without notice. Springer Nature
may revoke this licence to you at any time and remove access to any copies of the Springer Nature journal content which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express or implied
with respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or warranties imposed by law,
including merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be licensed
from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other manner not
expressly permitted by these Terms, please contact Springer Nature at
onlineservice@springernature.com