Predictors of functional recovery for hip fractured elders during 12 months following hospital discharge: a prospective study on a Taiwanese sample.

Abstract

To examine the longitudinal changes in overall and individual physical activities of daily living (PADLs) and instrumental activities of daily living (IADLs), at 1, 3, 6, and 12 months after hospital discharge in elderly subjects, 110 hip fractured elders (mean+/-SD age, 79.4+/-7.5; 60.9% females) were enrolled in a prospective study. At 12 months following hospital discharge, 56.1% subjects had recovered their overall PADLs, 37.9% had recovered their overall IADLs, and 74.2% could walk independently or with the aide of a cane. When analyzed by generalized estimating equations (GEE), for individual PADL, bathing and climbing stairs had consistently improved at every time point throughout the 1-year follow-up period, while transferring, toileting, and walking ability only improved significantly in the first 3 months after discharge. For individual IADL, the proportion of recovery for mobility appeared to increase significantly during the first 6 months post-discharge, and the remaining IADLs appeared to be stable. These results indicated that the recovery rate is varied for performance of different activities according to the complexity and the involvement of the lower extremities. We also found that less concomitant diseases, and a shorter hospital stay could predict a better recovery trend of overall and of most individual PADLs. These findings may be applicable to other countries with Chinese populations, and could provide a reference for health care providers to develop specific interventions for Chinese hip fractured elders.

Full text

Page 1

ORIGINAL ARTICLE
Predictors of functional recovery for hip fractured elders during
12 months following hospital discharge: a prospective study
on a Taiwanese sample
Yea-Ing Lotus Shyu Æ Min-Chi Chen
Jersey Liang Æ Chi-Chuan Wu Æ Juin-Yih Su
Received: 28 May 2003/ Accepted: 17 November 2003/Published online: 28 February 2004
? International Osteoporosis Foundation and National Osteoporosis Foundation 2004
Abstract To examine the longitudinal changes in overall
and individual physical activities of daily living (PADLs)
and instrumental activities of daily living (IADLs), at 1,
3, 6, and 12 months after hospital discharge in elderly
subjects, 110 hip fractured elders (mean±SD age,
79.4±7.5; 60.9% females) were enrolled in a prospective
study. At 12 months following hospital discharge,
56.1% subjects had recovered their overall PADLs,
37.9% had recovered their overall IADLs, and 74.2%
could walk independently or with the aide of a cane.
When analyzed by generalized estimating equations
(GEE), for individual PADL, bathing and climbing
stairs had consistently improved at every time point
throughout the 1-year follow-up period, while transfer-
ring, toileting, and walking ability only improved sig-
nificantly in the first 3 months after discharge. For
individual IADL, the proportion of recovery for
mobility appeared to increase significantly during the
first 6 months post-discharge, and the remaining IADLs
appeared to be stable. These results indicated that the
recovery rate is varied for performance of different
activities according to the complexity and the involve-
ment of the lower extremities. We also found that less
concomitant diseases, and a shorter hospital stay could
predict a better recovery trend of overall and of most
individual PADLs. These findings may be applicable to
other countries with Chinese populations, and could
provide a reference for health care providers to develop
specific interventions for Chinese hip fractured elders.
Keywords Elderly persons Æ Health related quality of
life Æ Hip fracture Æ Longitudinal change
Introduction
Because of its association with excess mortality of 5–
20%, and morbidity that results in huge economic costs,
hip fracture is identified as the most dramatic conse-
quence of osteoporosis in the elderly who are 65 years
old or older [1]. Previous studies found not only a sus-
tained decline in physical function early after a hip
fracture, but also a steady ‘‘unrecovery’’ of a large per-
centage (26–76%) of elderly persons who survive the
first or even the second year following a hip fracture [2,
3, 4, 5, 6, 7, 8, 9, 10, 11, 12]. It is estimated that the
incidence rate (per 100,000) of hip fractures for the
elderly population in Taiwan in 1993 was 211, and that
this number is 10 times the incidence rate for the general
population [13]. Despite the use of advanced treatment,
Osteoporos Int (2004) 15: 475–482
DOI 10.1007/s00198-003-1557-2
Y.-I.L. Shyu &
Center for Gerontological Research,
Chang Gung University,
259 Wen-Hua 1st Road, Kwei-Shan,
Tao-Yuan 333, Taiwan, ROC
E-mail: yeaing@mail.cgu.edu.tw
Tel.: +886-3-2118800 ext. 5275
Fax: +886-3-2118400
Y.-I.L. Shyu
School of Nursing, Chang Gung University,
259 Wen-Hua 1st Road, Kwei-Shan,
Tao-Yuan 333, Taiwan, ROC
M.-C. Chen
Department of Public Health,
Chang Gung University,
259 Wen-Hua 1st Road, Kwei-Shan,
Tao-Yuan 333, Taiwan, ROC
J. Liang
Health Management and Policy,
School of Public Health,
University of Michigan,
M3234 School of Public Health II,
Ann Arbor, MI 48109-2029, USA
C.-C. Wu
Department of Orthopedics,
Chang Gung Memorial Hospital,
5 Fu-Hsin Street, Kwei-Shan,
Tao-Yuan 333, Taiwan, ROC
J.-Y. Su
Department of Trauma & Emergency Surgery,
Chang Gung Memorial Hospital, 5 Fu-Hsin Street,
Kwei-Shan, Tao-Yuan 333,
Taiwan, ROC

Page 2

the mortality rate remains significant, and a majority of
the patients will never recover completely in terms of
physical activities of daily living (PADLs). In addition,
the economic cost of hip fracture in the elderly in Tai-
wan is huge [13]. As in many other countries in the
world, the aging population in Taiwan has increased
rapidly, and will continue to grow. The population of
Taiwan aged 65 and over is projected to increase from
1.56 million (7.37%) in 1994 to 3.4 million (13.83%) in
2020, and 5.6 million (21.65%) in the year 2036 [14]. As
the aging population increases, the problems caused by
hip fracture are aggravated, and it has become a major
concern for health care providers in Taiwan.
There is extensive literature addressing the conse-
quences of hip fracture in terms of the impact and
recovery in physical functioning [2, 4, 6, 8, 10, 11, 15,
16]. However, there are several gaps in the literature.
First of all, few previous studies on functional recovery
after hip fracture [11, 15, 17, 18] used longitudinal data
with more than two time points, and few used methods
of longitudinal data analysis to explore the trends and
identify risk factors associated with change in functional
recovery over time. The trajectory of functional recovery
after hip fracture can not only provide information on
the consequences of hip fracture at certain points in
time, but also on the changes in outcomes from an
earlier to a more long-term time period. This informa-
tion may help health care providers predict the prog-
nosis of the patients more precisely, provide a reference
for timing of outcome assessment, and assist them in
developing a timely management approach to shape,
control or change the course of the functional recovery
after hip fracture.
Secondly, prior observations are almost exclusively
based on data from Western developed nations, but little
is known about the functional recovery of hip fractured
elders in Asian countries. It would be critical to evaluate
the external validity of these observations with data
derived from other societies and cultures. Taiwan differs
substantially from the Western countries in health care
system, clinical practice, case-mix, culture, and social
organization. For example, anthropometrical measure-
ments including body mass index, height, weight, hip
axis and femoral neck length are different among dif-
ferent ethnic groups [19, 20, 21, 22]. The housing envi-
ronment in Taiwan is different from that in Western
countries. Moreover, the existence of a national health
insurance and the choice for people to visit clinics in a
medical center without referrals in Taiwan is different
from health care systems in Western countries. These
differences might make the functional recovery of elders
with hip fracture different from their counter parts in
Western countries. Thus, the findings from Western
countries need to be validated in Taiwan. The similari-
ties and differences between findings from Taiwan and
those from Western countries are useful for further sci-
entific understanding and clinical implementation.
This study has two major objectives: a) to describe
the return to previous levels of PADLs and IADLs for
hip fractured elders during 1 year following their hos-
pital discharge; and b) to explore the effects of hospital
stay, concomitant illness and pre-fracture dependency in
PADLs on patterns of recovery for overall and indi-
vidual PADL/IADL. These variables were selected
because they were identified as important clinical pre-
dictors for prognosis after hip fracture [11, 15].
Materials and methods
A prospective, correlational study design was used to explore the
functional recovery following the discharge of hip fractured elders.
Data were collected from a sample of elderly persons with hip
fracture from a medical center before hospital discharge, and then
further at 1, 3, 6 and 12 months after hospital discharge via face-to-
face interviews at clinics, or in a home setting or via telephone
interviews. For those elderly persons who did not return to the
clinics and refused to receive home visits, telephone interviews were
conducted. For elderly persons who were unable to come to the
phone or were cognitively impaired, information from proxy
respondents was obtained. Proxies are considered reliable sources
of information pertaining to facts and observable behaviors, as
reported by Maganizer and colleagues [23].
Study setting and sample
This study was conducted in the trauma wards of a 3800-bed
medical center in northern Taiwan and in home settings. The cri-
teria includes was 1) aged 65 years or older; 2) hospitalized due to
hip fracture and having received surgery for internal fixation or
arthroplasty; and 3) living geographically in northern Taiwan. A
total of 110 subjects were enrolled in the study and interviewed
before hospital discharge. Eight (7.3%) subjects died within 1
month, 5 (4.5%) between the 1st and 3rd month, 2 (1.8%) between
the 3rd and 6th month, and 3 (2.7%) between the 6th and 12th
month following discharge. One (0.9%) subject dropped out before
the 1st month, 4 (3.6%) between the 1st and 3rd month, 6 (5.5%)
between the 3rd and 6th month, and 14 (12.7%) between the 6th
and 12th month following discharge due to losing contact caused
by changing of address or refusal to participate further.
Procedures
Human subject approval was obtained from the institution before
collecting the data. Research assistants identified potential subjects
after they received surgery and before they were discharged from
the hospital, and invited them to participate in this study. Elderly
patients who agreed to participate received face-to-face interviews.
Training programs for interviewers and regular research team
meetings were used to monitor the quality of the data obtained.
Two registered nurses were hired as interviewers for the study.
They attended two training sessions. The first session focused on
introduction of the instrument and an interview demonstration.
The second training session included an actual interview practice.
Problems raised in the data collection were discussed and resolved
on a regular basis.
Instruments
Physical functioning
We used the Chinese Barthel Index (CBI) to assess the physical
functioning of hip fracture elders, including dependencies in eating,
transferring, grooming, toileting, bathing, walking, climbing stairs,
dressing, bowel and bladder control. The CBI was translated by
476

Page 3

Chen et al. [24] and is reported as having established reliability and
validity, and as being appropriate for assessing frail elders in Tai-
wan. The score of the scale ranged from 0 to 100, with 0 repre-
senting total dependence and 100 representing total independence.
In this study, Cronbach’s alpha of the CBI was 0.87. We then
recoded the score of each item and the total score of the CBI as 1, if
the score recovered to the individual’s pre-fracture level, and coded
it as 0 if the score did not recover to the pre-fracture level.
Instrumental function
We used the Chinese version of Lawton and Brody’s [25] measure
for instrumental activities of daily living (IADLs) to assess the
instrumental functioning for hip fractures elders, including ability
to use the telephone, do shopping, cooking, housekeeping, laundry,
transportation, taking medicine and handling money. Lawton and
Brody’s IADL dependency scale was also translated by Chen
and colleagues [24], and is reported as having established reliability
and validity, and as being appropriate for assessing frail elders in
Taiwan. In this study, a Cronbach’s alpha of 0.88 was found. We
also examined whether the instrumental function recovered to the
individual’s pre-fracture level (1) or not (0) in this report.
Predictors of physical function recovery
Predictors examined in this study included concomitant illness, pre-
fracture walking ability, education (being illiterate or not), whether
living with spouse, age, type of surgery, and length of hospital stay.
For concomitant illness, the elderly person was considered to have
a serious concomitant illness (coded as 1) if any one of the fol-
lowing conditions were noted on their chart: cancer, stroke, heart,
renal or liver disease, diabetes mellitus, osteoporosis, or dementia.
For pre-fracture walking ability, we used the item for assessing
independence in the CBI to assess the pre-fracture walking ability.
These variables were included because they were significant in the
univariate analysis, and were consistent predictors for functional
recovery in many previous studies [11, 15].
Data analysis
The logistic regression was first used to explore the predictors of
recovery at each time point (3rd, 6th and 12th month after hospital
discharge). The possibility of recovery was modeled as a function of
illiterate, age, hospital stay, concomitant disease, pre-fracture
walking ability, living with spouse and type of surgery. The sig-
nificant predictors were further included in the longitudinal anal-
ysis.
The generalized estimating equations (GEE) [26, 27] approach
was used to determine the risk factors related to the probabilities of
hip fractured elders recovering their living ability baseline (1st
month) at different time points (3rd, 6th and 12th month after
hospital discharge). The GEE approach is able to account for
possible correlations in repeated measures over time and to explore
the differences among different time points. In contrast with re-
peated measures with the ANOVA approach, one advantage of
using the GEE model is that partial information can also be uti-
lized. That is to say, for those who drop out during the study
period, their information before dropping out has made a contri-
bution to the estimating of the parameters relative to the available
data. Therefore, it is especially useful for longitudinal studies where
sample attrition is an inevitable issue. Its application in our study
was to model dependent dichotomous variables (recovered or not
in overall PADLs and individual PADL) as a function of time,
hospital stay, whether being illiterate or not, type of surgery and
concomitant diseases. These predictors were selected because they
were significant for overall PADL in the cross-sectional logistic
analysis. Because none of the predictors for the overall IADL was
significant in the cross-sectional analysis, we modeled recovery of
overall IADL and IADL simply as a function of time in the GEE
approach. In our study, the physical functioning at 1 month after
hospital discharge was used for comparison instead of the pre-
fracture physical functioning as a baseline, since the pattern of
recovery after hip fracture was specifically of interest and data after
hip fracture were collected prospectively, whereas pre-fracture
functioning was collected retrospectively after hospital admission.
For the elders who died (n=18, 16.4%) or dropped out (n=25,
22.7%) within a year following discharge, the data before death or
dropping out was used for the analysis. The largest number of
subjects that dropped out occurred during the 6th month to the
12th month after discharge. We found that there is little or no
difference between the group that pass away or dropped out during
the 6th month to the 12th month and the group that stayed on
throughout the full 12 months after discharge. In other words, we
modeled the dependent binary variables (recovered or not) as
function of ‘‘time’’ (baseline, 3rd month, 6th month), ‘‘group’’
(pass away or dropped out between 6 and 12 months after follow-
up versus completed in the 12 months of follow-up) and interaction
between time and ‘‘group.’’ Note that time is treated as a cate-
gorical variable—’’baseline,’’ ‘‘3rd month’’ and ‘‘6th month.’’
Hence, the interaction between ‘‘time’’ and ‘‘group’’ includes two
terms, that is 3rd month·group and 6th month·group. It was
found that (i) the non-significant interactions indicating the trend
of recovery over time were similar in two groups (P=0.70, 0.30 for
eating; P=0.52, 0.71 for transferring; P=0.67, 0.18 for grooming;
P=0.65, 0.58 for toileting; P=0.26, 0.96 for bathing; P=0.55, 0.14
for walking; P=0.91, 0.59 for climbing stairs; P=0.18, 0.45 for
dressing; P=0.76, 0.74 for bowel control; P=0.23, 0.18 for bladder
control; P=0.73, 0.02 for overall IADL); (ii) after removing
interaction, the ‘‘group’’ effect was also ignorable except for
bladder control (P=0.16 for eating; P=0.76 for transferring;
P=0.68 for grooming; P=0.63 for toileting; P=0.45 for bathing;
P=0.18 for walking; P=0.94 for climbing stairs; P=0.47 for
dressing; P=0.07 for bowel control; P=0.05 for bladder control;
P=0.26 for overall IADL). We also carried out a parallel analysis
using the GEE approach to explore the trends of recovery in
PADLs and IADLs only for subjects who remained in the study
throughout the whole 12 months. The trends were very similar to
the trends for recovery in PADLs and IADLs for the whole sample.
As a result, individuals with incomplete information were included
for our analysis.
Results
Proportion of functional independence
In the original sample (n=110), there were 67 (60.9%)
females and 43(39.1%) males. The age of the subjects
ranged from 65 to 96 with a mean age of 79.4 (SD=7.5)
years, among them, 54 (49.1%) aged 80 or over. Thirty-
seven (33.6%) of the subjects were married, and 73
(66.4%) were widowed. Seventy-eight (70.9%) had no
formal education, 18 (16.4%) completed primary school
(1–6 years), 12 (10.9%) completed high school (7–13
years), and 2 (1.8%) had a completed college (>13
years) education. Before the admission, 102 (92.8%)
elderly persons lived with their family members, one
(0.9%) lived with friends, five (4.5%) lived in an insti-
tution and two (1.8%) lived alone. Seventy (63.6%)
subjects had received internal fixation and 40 (36.4%)
received arthroplasty. Average length of hospital stay
was 12.40 (SD=5.43) days; among them, 21 (19.1%)
stayed in the hospital for over 2 weeks.
Before their fracture (n=110), 90 (81.8%) of these
elderly patients were fully independent in PADLs, 34
477

Page 4

(30.9%) in IADLs, and 106 (96.4%) could walk inde-
pendently or with the aid of a cane. In contrast, at 1
month following hospital discharge, for the remaining
subjects (n=100), only four (4%) were fully independent
in PADLs, one (1%) in IADLs and 45 (45%) could walk
independently or with the aid of a cane. At the 3rd
month following hospital discharge, for the remaining
subjects (n=91), 12 (13.2 %) were fully functionally
independent in PADLs, three (3.3%) in IADLs and 60
(65.9%) could walk independently with the aid of a
cane. At 6 months following hospital discharge, for the
remaining subjects (n=83), 21 (25.3%) were fully func-
tionally independent in PADLs, eight (7.3%) in IADLs
and 55 (66.3%) could walk independently with the aid of
a cane. At 12 months following hospital discharge, for
the remaining subjects (n=66), 34 (51.5%) were fully
functionally independent in PADLs, two (3.0%) in
IADLs and 48 (72.7%) could walk independently or
with the aid of a cane.
Recovery proportion of individual PADLs and IADLs
The longitudinal changes in recovery proportion of
individual PADL and IADL are presented in Fig. 1. It
appeared that hip fracture had more impact on activities
involving the lower extremities, such as climbing stairs,
transferring, bathing, walking and toileting. The longi-
tudinal changes in individual PADL were evaluated by a
GEEmodel (Table 1).Some
including grooming and bowel and bladder control,
were quite stable over the 1-year follow-up period.
Bathing and climbing stairs consistently improved at
every time point throughout the 1-year follow-up period.
Several PADLs, including eating, transferring, toileting,
and walking ability, only improved significantly in the
first 3 months after discharge. Dressing improved
significantly from 1 to 6 months, and remained stable
from the 6th until the 12th month. The proportion of
functional recovery at each time point for each indi-
vidual PADL is listed in Table 1. By the end of 12
months, over 75%, in most instances over 80%, of the
subjects fully recovered to their pre-fracture functional
individual PADLs,
level in most of the individual PADLs, except for
activities involving lower extremities such as toileting,
walking and climbing stairs. A little more than 70% of
the subjects were able to recover to their pre-fracture
ability in performing toileting, and walking. The poorest
recovery proportion was found regarding the climbing
of stairs. Less than 60% of the subjects fully recovered
their ability in climbing stairs.
The longitudinal changes of individual IADL are
presented in Fig. 2. It appeared that hip fracture had
more impact on activities involving the lower extremi-
ties, such as shopping and mobility. The recovery pro-
portion of mobility appeared to increase significantly
during the first 6 months post-discharge. The ability to
use a telephone, take medication, or manage money was
less impacted by hip fracture. However, the ability to
manage money appeared to decline at the 12th month
post-discharge, and the ability to take medication
seemed to decline at the 6th month post-discharge. The
proportion of functional recovery at each time point for
each individual IADL is listed in Table 2. By the end of
12 months, over 70% of the subjects were able to fully
regain their pre-fracture functional level in activities
that did not involve the lower extremities. For activities
involving the lower extremities, including shopping
and mobility, around 50–60% of the subjects fully
recovered.
Recovery proportion of overall PADLs and IADLs
Longitudinal changes of proportion of elders returning
to their pre-fracture levels of overall PADLs, and
IADLs are shown in Fig. 3. Only four (4.0 %) patients
had regained their pre-fracture overall PADLs, and 31
(39%) had recovered their overall IADL performance
level, at one month after discharge. At 3 months fol-
lowing hospital discharge, 14 (13.2%) patients had
recovered their overall PADLs, and 36 (39.6%) had
recovered their overall IADLs performance level, among
the remaining subjects (n=91). At 6 months following
hospital discharge, 24 (28.9%) patients had recovered
their PADLs and 35 (42.2%) had recovered their overall
Fig. 1 Longitudinal changes in
individual PADL
478

Page 5

IADLs performance level among the remaining subjects
(n=83). At twelve months following hospital discharge,
37 (56.1%) patients had recovered their overall PADLs
and 25 (37.9%) had recovered their overall IADLs
performancelevel,among
(n=66).
theremainingsubjects
The significance of the differences between the trends
of recovery proportion of overall IADLs and PADLs
was examined using the GEE separately. The recovery
proportion of overall PADL among these elders con-
sistently improved throughout the follow-up period
from the 1st month to the 1st year following hospital
discharge. Among 56% of the elders who recovered to
their previous level of PADL performance, almost half
of them recovered between the 6th and 12th month
following hospital discharge. The recovery proportion of
overall IADLs did not change significantly throughout
the follow-up period.
Predictors for functional recovery
The cross-sectional logistic regressions modeled the
PADL at each time point. For the overall PADL, the
predictors for recovery at the 1st month after discharge
could not be calculated because too few subjects (n=4)
had recovered to their pre-fracture level of PADL.
Whether living with a spouse, with at least a concomi-
tant disease, with independent pre-fracture walking
ability, being illiterate, with hospital stay longer than 14
Table 1 The number of functional recovery and the total number
of observed subjects at each time point for each PADL subscale.
The significance in ‘‘Time’’ denotes the difference between the 1st
month (baseline) and the 3rd and 6th months and 1 year, respec-
tively. Co-variates entered included hospital stay, whether being
illiterate, type of surgery and concomitant disease
Fig. 2 Longitudinal changes in
individual IADL
Fig. 3 Longitudinal changes in patient’s functional recovery by
overall PADLs and IADLs
Variables Recovered subjects (total observed subjects; percentage)
1st month (baseline)3rd month6th month 1 year
PADL subscale (n; %)
Eating
Transferring
Grooming
Toileting
Bathing
Walking
Climbing stairs
Dressing
Bowel control
Bladder control
83 (100; 83%)
38 (100; 38%)
77 (100; 77%)
46 (100; 46%)
30 (100; 30%)
45 (100; 45%)
14 (100; 14%)
61 (100; 61%)
86 (100; 86%)
84 (100; 84%)
83 (91; 91.2%)*
71 (91; 78%)b
78 (91; 85.7%)
66 (91; 72.5%)b
49 (91; 53.8%)b
61 (91; 67%)b
24 (91; 26.4%)a
63 (91; 69.2%)
80 (91; 87.9%)
75 (91; 82.4%)
74 (83; 89.2%)
68 (83; 81.9%)b
71 (83; 85.5%)
60 (83; 72.3%)b
56 (83; 67.5%)b
56 (83; 67.5%)b
28 (83; 33.7%)b
65 (83; 78.3%)a
72 (83; 86.7%)
71 (83; 85.5%)
56 (66; 84.8%)
56 (66; 84.8%)b
52 (66; 78.8%)
48 (66; 72.7%)b
51 (66; 77.3%)b
49 (66; 74.2%)b
39 (66; 59.1%)b
54 (66; 81.8%)a
54 (66; 81.8%)
54 (66; 81.8%)
*P<0.05;aP<0.01;bP<0.001
479