The Effectiveness of Combined Exercise Interventions For Preventing Postmenopausal

The Effectiveness of Combined Exercise Interventions for Preventing Postmenopausal
Bone Loss: A Systematic Review and Meta-Analysis
Renqing Zhao1*2, PhD; Mengyi Zhang2, MS; and Qi Zhang2, MS
1 College of Physical Education, Yangzhou University, Yangzhou, Jiangsu, China

Downloaded from www.jospt.org at Univ of New England on March 5, 2017. For personal use only. No other uses without permission.
2 College of Physical Education and Health Sciences, Zhejiang Normal University, Jinhua,
Zhejiang, China
*Address all correspondence and requests for reprints to: Renqing Zhao, Yangzhou
Copyright © ${year} Journal of Orthopaedic & Sports Physical Therapy®. All rights reserved.
University, College of Physical Education, 88 South University Avenue, Yangzhou, Jiangsu
225009, China. E-mail: zhaorenqing@hotmail.com
Funding: This work was supported in part by Zhejiang Provincial Natural Science
Journal of Orthopaedic & Sports Physical Therapy®
Foundation of China under Grant (No. LY14H070001).
Conflict of interest: None
Words counts: 3516
Running title: Exercise for preserving BMD in Older Women
1
1 ABSTRACT
2 STUDY DESIGN: Systematic review and meta-analysis.
3 BACKGROUND: It remains unclear whether exercise combining different types of physical
4 activities (combined exercise interventions) would effectively preserve postmenopausal
5 women’s bone mineral density (BMD) at different sites.
6 OBJECTIVES: To examine the impact of combined exercise interventions on lumbar spine,

7 femoral neck, total hip and total body BMD in postmenopausal women.
8 METHODS: An electronic database search was conducted in PubMed, EMBASE,
9 SPORTDiscus, and Web of Science up to Jan1, 2016. Randomized controlled trials (RCTs)
10 that conducted combined exercise interventions and reported BMD values in postmenopausal
11 women were included. Two authors independently extracted the data from individual studies.
12 The primary endpoint was the change in BMD values from baseline to follow-up. The effect
13 sizes were estimated by the standardized mean difference (SMD) methods using fixed-effects
14 models.
15 RESULTS: Eleven RCTs including 1061 postmenopausal women met the inclusion criteria.
16 The levels of between-study heterogeneity were relatively low (I2<50%). Exercise integrating
17 different physical activities significantly increased lumbar spine (SMD=0.170,
18 95%=0.027-0.313, p=0.019), femoral neck (SMD=0.177, 95%CI=0.030-0.324, p=0.018),
19 total hip (SMD=0.198, 95%CI=0.037-0.3590, p=0.016) and total body (SMD=0.257,
20 95%CI=0.053-0.461, p=0.014) BMD. Combined exercise interventions generated a beneficial
21 effect on femoral neck BMD (SMD=0.219, 95%CI=0.034-0.404, p=0.020) in groups with
22 participants’ age <60 years, and significantly improved lumbar spine BMD (SMD=0.349,
2
23 95%CI=0.064-0.634, p=0.016) in groups with women’s age≥60 years.
24 CONCLUSION: Our findings suggest that combined exercise interventions appear to be
25 effective in preserving postmenopausal women’s BMD at lumbar spine, femoral neck, total
26 hip and total body.
27 Level of Evidence: Therapy, level 1a.
28 Key words: Exercise; Women; Menopause; Bone loss; Meta-analysis

29
30
3
31 INTRODUCTION
32 Fracture is a major source of public health problems1, and the incidence of fractures appears
33 to be increasing worldwide2-5. The hips and spine are the most common sites for fracture. The
34 estimates of worldwide osteoporotic fractures for all populations were approximately 1.6
35 million at the hip and 1.4 million at the spine in the year 20005. Lifetime risk of osteoporotic
36 fractures is very high for women ranging from 40% to 50%4. Currently, exercise has been
recognized as an effective strategy for preventing postmenopausal bone loss6 and reducing
37
38 the risk of fracture7,8. However, single exercise training mode frequently generates
39 site-specific effects on bone mineral density (BMD). Recently, a study9 reported that impact
40 exercise often affected the hip BMD, whereas another review10 demonstrated that resistance
41 training was usually effective in preventing spine bone loss; aerobic activities seemed to have
42 inconsistent results on BMD11,12. Combined exercise interventions, which mix at least two
43 different types of exercise, e.g. resistance training combined with impact exercise, are
44 assumed to generate various mechanical strains and affect different skeletal loading sites.
45 Several clinical trials have been conducted to investigate the efficiency of combined exercise
46 protocols in preventing postmenopausal bone loss13-23. However, the outcomes were
47 inconsistent. Some studies observed increment of BMD at lumbar spine, femoral neck, total
48 hip and total body13,15,16 or reduction of bone loss rate at lumbar spine and femoral neck14,16
49 after combined exercise interventions, whereas others either did not find a positive effect17,22
50 or revealed a negative outcome20. With regard to the inconsistent results, wide variations
51 existed in the sample size, population age and exercise interventions14,16,18,19,21. Therefore, a
52 general consensus remains unreached. Using a systematic review and meta-analysis approach,
4
53 we primarily aimed to examine the influence of combined exercise protocols on BMD at
54 lumbar spine, femoral neck, total hip and total body in postmenopausal women compared
55 with controls maintaining their daily routine activities.
56 METHODS
57 Search Strategy and Inclusion Criteria
58 We conducted a systematic review and meta-analysis in accordance with PRISMA
recommendations and the criteria of the reporting of meta-analysis guidelines24, and used a
59
60 predetermined protocol. An electronic database search was conducted in PubMed, EMBASE,
61 SPORTDiscus, and Web of Science up to Jan 1, 2016 to identify all of the relevant studies.
62 Terms used for database searching were “mixed exercise/loading”, “combined exercise”,
63 “resistance training/exercise”, “strength training/exercise”, “aerobic exercise”,
64 “weight-bearing exercise”, “impact exercise”, “jumping”, “skipping”, and “bone density”,
65 while the search was restricted to the female. No language restriction was used.
66 The included studies were RCTs, which compared the change in BMD between the exercise
67 intervention group and control group (non-exercise group) in postmenopausal women. The
68 included participants were healthy or osteoporotic postmenopausal women who had not
69 participated in regular exercise (i.e. they had been exercising for <2 h per week) prior to
70 enrollment, and had no history of medicine administrations (e.g. glucocorticoid use and
71 hormone replacement therapy) or disease experience (e.g. heart transplantation and cancer)
72 known to affect bone metabolism. But participants consuming calcium and vitamin D were
73 eligible for inclusion. The duration of exercise interventions lasted for at least six months, in
74 view of the fact that bone remodeling requires approximately that length of time.
5
75 The exercise interventions included in our meta-analysis were combined exercise
76 interventions. Frequently, single exercise mode only includes a single type of exercise, e.g.
77 resistance training, impact exercise, or dynamic aerobic activities. When several single
78 exercise modes are combined to augment beneficial effects on bone, e.g. resistance training
79 combined with impact exercise, they are frequently referred to as combined exercise
80 protocols/interventions. Participants in control group were frequently required to maintain

81 their usual care and levels of physical activity.
82 Data Extraction and Quality Assessment
83 Two authors (MZ and ZQ) independently extracted the data from individual studies according
84 to the methods provided by the Cochrane Reviewers’ Handbook25.The details that were
85 extracted included the following: participant characteristics, sample size, exercise intensity,
86 frequency and duration, supplements, attrition, compliance, regions of interest (ROIs), and
87 BMD values with standard deviations (SDs). Usually, we extracted data at the final follow-up
88 time point from longitudinal studies, but if data at 12 month were available, we preferred to
89 extract data at 12 month follow-up time point to reduce variation of exercise interventions
90 between studies.
91 Risk of bias was assessed on the basis of individual study characteristics using the Cochrane
92 Collaboration risk of bias assessment tool25. Generally, we concentrated on the following
93 issues: sequence generation, allocation concealment, blinding, incomplete outcome data,
94 selective outcome reporting, and other potential sources of bias. We were especially
95 interested in whether randomization was accomplished according to the guideline. We also
96 evaluated whether optimal blinding was used to avoid control participants’ enthusiasm in
6
97 increasing their exercise level. Moreover, we evaluated whether the intention-to-treat (ITT)
98 rather than per-protocol approach was used in analysis of the data in original studies, as the
99 issues related to participant withdrawals. Likewise, we checked the selective outcome
100 reporting of specific results. Finally, we evaluated other source of risk of bias, including
101 imbalance in bone baseline variables, small sample size (< 100 participants), and short
102 follow-up time (< 6 months). Each item was classified as low risk if the definition of high
103 quality was fulfilled, high risk if the definition was not fulfilled, and unclear if the
104 information was not available in the original article. The final assessment for all of the
105 included studies was presented in a “risk of bias” table.
106 Data Synthesis, Analysis, and Assessment of Heterogeneity

107 The primary endpoint was the change in BMD from baseline to follow-up at lumbar spine,
108 femoral neck, total hip and total body, respectively; the change in BMD measurements
109 between the exercise intervention and control groups were pooled and estimated by the
110 standardized mean difference (SMD) methods. SMD is the ratio of the mean difference to the
111 pooled standard deviation.
112 Heterogeneity of results between studies was determined using the Cochran’s Q-test (p<0.10
113 for statistical significance) and I2 (I2>50% used as a threshold for significant heterogeneity).
114 I2 is the percentage of variation attributable to heterogeneity, which is frequently calculated
115 by the following formula25:
116 I2=(Q - df)/Q × 100%,
117 where Q is the chi-squared statistic and df is its degrees of freedom.
118 Funnel plots were used to examine publication bias or other potential sources of bias. They
7
119 are simple scatter plots of the treatment effects estimated from individual studies against a
120 measure of study size. In the absence of bias, results from small studies will widely scatter at
121 the bottom of the graph, and publication bias may lead to asymmetrical funnel plots. Funnel
122 plots were produced for the treatment effects on lumbar spine, femoral neck, total hip, and
123 total body BMD, respectively. The test for the overall effects (Z score) was regarded as
124 significant at p<0.05. STATA version 12 (Stata Corp, TX, USA) was used to perform the
125 meta-analysis and produce the graph.
126 Subgroup Analyses
127 A subgroup analysis was conducted by the age of participants (<60 years vs. ≥60 years) to
128 determine whether skeletal response to combined exercise training differed in age. The
129 subgroup analysis were performed only in the lumbar spine and femoral neck outcomes due
130 to the limited number of trials that measured BMD at total hip and total body sites.
131 RESULTS
132 Study Characteristics

133 The systematic search identified and screened 1253 potential abstracts, of which 1202 were
134 excluded because they were unrelated to the specific topic and duplicate studies (FIGURE 1).
135 Finally, forty-one of 52studies were excluded, with reasons: (1) not RCTs; (2) not a BMD
136 study; (3) not postmenopausal women; (4) inappropriate interventions. Eleven trials met the
137 inclusion criteria. TABLE 1 shows the descriptive data for the 11 qualifying trials. Overall, a
138 total of 1061 postmenopausal women (between 55.3±6.3 and 73.2±4.9 years) were included
139 in our meta-analysis. The sample size varied from 20 to 174 participants. Eight trials
13,14,16,17,19-21,23
140 reported findings on the basis of the ITT approach; 3 studies15,18,22 only
8
141 provided the per-protocol data. Three trials16,18,22 conducted mixed interventions of exercise
142 and estrogen or isoflavone treatment, but the data of participants conducting exercise only
143 and taking placebo alone were included in the meta-analysis (Table 1).Nine studies (n=759)
144 measured the lumbar spine BMD; 8 trials (n=723) reported the femoral neck outcomes; 6
145 studies (n=601) provided the total hip findings; 4 trials (n=375) performed the total body
146 BMD measurements (TABLE 1). The studies were conducted in 7 countries; they were
Australia, Canada, Germany, Finland, Sweden, UK, and USA. There were two studies16,20
147
148 having at least two follow-up measures and data at 12 month follow-up time point were
149 available; therefore, we extracted the data of 12 month follow-up time point for analysis to
150 reduce the variation between exercise interventions. Finally, 813-18,20,22 of 11 trials reported
151 data on intervention period of 12 months.
152 Exercise Interventions
153 Five hundred and forty-four participants completed the exercise interventions; 517 as the
154 control maintained their routine physical activities. The training regimens included in our
155 meta-analysis were combined exercise protocols. The exercise interventions were delivered at
156 center or both center and home, and were supervised or partially supervised. The duration of
157 the exercise interventions ranged from 8 to 30 months. Compliance with the exercise program
158 was assessed by attendance at the center and logs for the exercises done at home, and ranged
159 from 67 to 95% (TABLE 1). No exercise-related injuries were reported in the intervention
160 groups.
161 Risk of Bias Assessment
162 TABLE 2 shows the results of the analysis of risk of bias in individual studies. Three studies
9
163 were judged at low risk in any domain. Five studies were judged high risk in one domain, and
164 2 trials were judged high risk in 2 domains. However, there were 4 trials with unclear in some
165 domains that could not be judged with certainty. The domains that frequently were judged as
166 high risk were incomplete outcome data, selective reporting, and small sample size.
167 Effects of Exercise Interventions on BMD
168 The primary analysis aimed to compare the change in BMD between the exercise and the
169 control groups. For effects of exercise interventions on lumbar spine and total body BMD,
170 combined resistance training protocols were the major training mode practiced, and combined
171 impact exercise protocols and combined resistance training interventions were the common
172 exercise types conducted for preservation of femoral neck and total hip BMD. The BMD
173 measurements of the selected trials were consistent at lumbar spine (I2=0.0%), femoral neck
174 (I2=15.1%), total hip (I2=43.7%) and total body (I2=0.0%) (TABLE 3).Therefore, only
175 fixed-effects models were used for combining the data. The pooled effect sizes associated
176 with exercise interventions were significant for lumbar spine BMD (SMD=0.170,
177 95%=0.027-0.313, p=0.019), femoral neck BMD (SMD=0.177, 95%CI=0.030-0.324,
178 p=0.018), total hip BMD (SMD=0.198, 95%CI=0.037-0.3590, p=0.016) and total body BMD
179 (SMD=0.257, 95%CI=0.053-0.461, p=0.014), respectively(FIGURE 2).
180 We conducted subgroup analysis by age (<60 years vs. ≥60 years) to determine whether
181 skeletal response to exercise training differed in age. In groups with participant’s age<60
182 years, 6 (I2=0.0%) and 4(I2=0.0%) consistent studies conducted combined exercise
183 interventions, and determined the lumbar spine and femoral neck BMD, respectively.
184 Combined exercise protocols generated positive effects on the femoral neck BMD
10
185 (SMD=0.219, 95%CI=0.034-0.404, p=0.020).In groups with women’s age ≥60 years, 4 trials
186 (I2=0.0%) with a population of 97 participants and 4 trials (I2=44.7%) including 137
187 individuals conducted combined exercise interventions, and measured lumbar spine and
188 femoral neck BMD, respectively. The pooled effect sizes associated with combined exercise
189 interventions were significant for the lumbar spine BMD (SMD=0.349, 95%CI=0.064-0.634,
190 p=0.016) in relative older participants (≥60 years) (TABLE 3).

191 Funnel Plots
192 Funnel plots were produced for the influence of combined exercise protocols on lumbar spine,
193 femoral neck, total hip, and total body BMD from all of the included trials (FIGURE3).Visual
194 inspection of the plots implicated that there existed some degree of asymmetry. It seemed that
195 some small studies were missing on the left hand area (outcomes with low statistical
196 significance) of the plots for lumbar spine and total hip outcomes, which indicated
197 smaller-study effects might exist and asymmetry of plots might be caused by risk of
198 publication bias. However, there were several small studies was lacking on the right hand
199 area (outcomes with high statistical significance) of plot for femoral neck, which indicated
200 there was weak evidence for presence of small-study effects and publication bias appeared
201 not to be the source of asymmetry of plots. Most of studies lie within the 95% confidence
202 limits, indicating that the asymmetry is less due to high heterogeneity between studies.
203 DISCUSSION
204 We pooled the data of 1061 postmenopausal women from 11 RCTs to examine the effects of
205 combined exercise protocols on BMD in postmenopausal women. Combined exercise
206 interventions positively affected the lumbar spine, femoral neck, total hip and total body
11
207 BMD compared with the control group maintaining usual levels of physical activity.
208 Postmenopausal women younger than 60 years seemed to be more sensitive to combined
209 exercise interventions at femoral neck, while individuals older than 60 years were responsive
210 to combined exercise interventions at lumbar spine.
211 Our study provided pooled evidence for the concept that combined exercise interventions
212 were effective for preventing postmenopausal bone loss at different skeletal loading sites.
According to the Frost’s mechanostat theory26 and animal studies27,28, exercise protocols that
213
214 produce high strain can increase strength of bone. Therefore, high intensity resistance training
215 has been recognized as an effective strategy for stimulating osteogenic response. However,
216 current evidence suggests that high intensity resistance training frequently generates benefit
217 effects only on the spine10. For impact exercise, such as jumping, high mechanical strain
218 generated against the ground during exercise is frequently attenuated before transmission to
219 the spine, and may only generate an osteogenic stimulus for the hips9. Therefore, single
220 exercise training mode usually produces a site-specific effect on bone. This raises an
221 important question what types of mechanical loading are optimum for promoting BMD in
222 postmenopausal women. In a review of factors that interact with physical activity to affect
223 bone in postmenopausal women, Borer et al29 indicate that the increment of BMD in elderly
224 women following exercise is usually modest and question whether principles shown to
225 increase BMD using animal models have been appropriately applied to human studies. It has
226 been reported that adaptive skeletal response requires dynamic rather than static mechanical
227 stimulation, and exercise intensity, strain frequency and the pattern of mechanical loading are
228 critical factors for increasing skeletal response29. Therefore, exercise protocols such as
12
229 combined exercise programs that integrate those importance factors together to exert a
230 mixing stimulation on bone may be the optimum type of exercise for improving
231 postmenopausal bone health. Our findings demonstrated that combined exercise protocols
232 integrating different exercise modes were able to preserve BMD on the lumbar spine, femoral
233 neck, total hips and total body in postmenopausal women.
234 There were two meta-analyses by Martyn-St James et al30 and Zhao et al31 reporting the
235 benefits of mixed loading exercise programs in preventing postmenopausal bone loss.
236 However, the two reviews were different from ours in many aspects. Firstly, reviews of
237 Martyn-St James and Zhao only conducted a subgroup analysis on this specific topic, while
238 our study primarily focused on the topic of interest and made a systematic search of studies
239 relevant. Findings only from a subgroup analysis did not draw a definitive conclusion on the
240 topic of interest. Secondly, their meta-analyses also incorporated non-RCTs, while our study
241 included RCTs only. It is reported that inadequate randomization in non-RCTs tends to
242 increase risk of bias32,33 and subsequently weakens the robustness of evidence. In addition,
243 Martyn-St James’s and Zhao’s meta-analyses only examined the hip and spine BMD, whereas
244 our review determined the lumbar spine, femoral neck, total hip and total body BMD.
245 Cummings et al34 suggested that adults who preserved each 1 SD in hip BMD would decrease
246 hip fracture risk by approximately 2.6 times. Therefore, our findings that combined exercise
247 protocols preserved femoral neck BMD by 0.177 SD relative to controls implied that
248 combine exercise training modes were effective in reducing hip fracture risk. Currently, a
249 number of studies reported the beneficial effects of exercise interventions in preventing falls
250 and fractures in elderly adults35-37. The overall effects could be greater, considering the
13
251 benefits of increased muscle mass, strength improvement and good balance - all of which are
252 recognized as independent factors that reduce the risk of fracture. Most included trials had
253 examined adverse events, and no training-related injuries were reported in those studies,
254 suggesting that combined exercise protocols were relatively safe for practice.
255 Subgroup analysis suggested that postmenopausal women aged < 60 years seemed to be more
256 sensitive to combined exercise interventions at femoral neck than women aged ≥ 60 years.
257 The possible explanation is that estrogen level is the critical modulator for skeletal response
258 to mechanical strain38,39. Femoral neck frequently sustains more mechanical stimulus during
259 mixed exercise and, therefore, femoral neck may more sensitive to change of estrogen level.
260 Postmenopausal women aged < 60 years is regarded to have higher levels of estrogen and
261 therefore may be more sensitive to mechanical strain than those aged ≥ 60 years. Therefore,
262 combined exercise was more likely to increase femoral neck BMD in postmenopausal women
263 aged < 60 years than those aged ≥ 60 years. Lumbar spine BMD of postmenopausal women
264 aged ≥60 years was still sensitive to exercise, which indicated that other factors other than
265 mechanical stimulus might contribute the beneficial effects, such as exercise-related
266 increment of calcium absorption40. However, it should be careful to elucidate the findings
267 because subgroup analysis only included a small number of studies.
268 Funnel plots for lumbar spine, femoral neck, and total hip BMD appeared to be somewhat
269 asymmetry (FIGURE 3). The evidence indicated that small-study effects were a likely cause
270 of asymmetry for lumbar spine and total hip outcomes, which indicated that some studies
271 with low statistical significance might not be published and had somewhat publication bias.
272 But the estimate of femoral neck BMD was less affected by small-study effects, which
14
273 indicated publication bias might not pose much a threat on the outcomes. We did not check
274 funnel plots of total body BMD due to only 4 trials included. However, small number of
275 study group comparisons available for funnel plot interpretation probably was not sufficient
276 to distinguish real asymmetry.
277 The aspects of methodological quality, including six domains, were assessed following the
278 recommendations of the Cochrane Collaboration25. Some trials did not address the methods
279 used in the sequence generation or allocation concealment, though they declared that the
280 participants of the studies were randomly allocated, which probably prevented the authors
281 from making a clear judgment of sequence generation and allocation concealment. Several
282 studies also failed to provide the information of blinding. The domains that frequently were
283 judged as high risk were incomplete outcome data, selective reporting, and small sample size.
284 Those questions indicated that some trials had limitations in study design or data reporting.
285 Eight studies reported ITT data; 3 studies failed to provide a valid ITT strategy when attrition
286 occurred and, therefore, were analyzed by a per-protocol approach. The ITT analysis is
287 regarded unbiased in addressing clinically relevant research questions; some trials were not
288 qualified for an ITT analysis, which probably increased the risk of bias as a result of attrition
289 failing to be accounted for.
290 Study Limitations
291 Our findings provide pooled evidence to support the idea that combined excise protocols
292 produce beneficial effects on postmenopausal women bone health. However, there exist
293 limitations in our study. The primary outcome of our meta-analysis is the change in BMD.
294 However, BMD is probably not the optimal means to determine bone strength, because it
15
295 only represents approximately 60-70 % of variation in bone strength41; It does not incorporate
296 other aspects of bone quality, for example microarchitecture. Additionally, some trials14,17,19,21
297 only included a smaller study population, which tended to weaken the individual study
298 quality and then posed a threat to risk of bias of our meta-analysis.
299 CONCLUSIONS
300 The general conclusions of the present meta-analysis were that combined exercise protocols
301 that integrated different exercise training modes tended to be effective in improving or
302 preserving the lumbar spine, femoral neck, total hip and total body BMD in postmenopausal
303 women, which supported the concept that combined exercise protocols frequently generate
304 non-site-specific effects on mechanical loading sites. The present findings were clinically
305 significant because the increment of BMD effectively prevented bone loss and may be helpful
306 for reducing risk of fracture. Methodological quality assessment suggested that some trials
307 had limitations in study design or data reporting, which indicated that future clinical
308 investigations should include a large number of populations and meet the strict standards of
309 conducting clinical trials required by current days. Summarily, combined exercise protocols
310 provide an alternative strategy for preventing postmenopausal bone loss.

311
16
312 KEY POINTS:
313 FINDINGS: Single exercise training modes, e.g. resistance training or impact exercise, were
314 known to offer a limited beneficial effect on postmenopausal bone loss. Our findings suggest
315 that combined exercise interventions(e.g. resistance training, impact exercise and dynamic
316 aerobic activities) are also effective in preserving postmenopausal women’s bone mineral
317 density (BMD) at the lumbar spine, femoral neck, total hip and total body.
318 IMPLICATIONS: These findings are clinically significant because the BMD increment
319 effectively prevents postmenopausal bone loss and may reduce the risk of fracture. Combined
320 exercise protocols can be recommended to improve postmenopausal women’s BMD,.
321 CAUTION: Some included trials only incorporated a smaller study population, which tended
322 to weaken the individual study quality and then posed a threat to risk of bias.
323
324 References
325 1. Lorrain J, Paiement G, Chevrier N, et al. Population demographics and socioeconomic impact of
326 osteoporotic fractures in Canada. Menopause. May-Jun 2003;10(3):228-234.
327 2. Kanis JA, Oden A, McCloskey EV, et al. A systematic review of hip fracture incidence and probability of
328 fracture worldwide. Osteoporosis international : a journal established as result of cooperation
329 between the European Foundation for Osteoporosis and the National Osteoporosis Foundation of the
330 USA. Sep 2012;23(9):2239-2256.
331 3. MacIntyre NJ, Dewan N. Epidemiology of distal radius fractures and factors predicting risk and
332 prognosis. Journal of hand therapy : official journal of the American Society of Hand Therapists.
333 Apr-Jun 2016;29(2):136-145.
334 4. Johnell O, Kanis J. Epidemiology of osteoporotic fractures. Osteoporosis international : a journal
335 established as result of cooperation between the European Foundation for Osteoporosis and the
336 National Osteoporosis Foundation of the USA. Mar 2005;16 Suppl 2:S3-7.
337 5. Johnell O, Kanis JA. An estimate of the worldwide prevalence and disability associated with
338 osteoporotic fractures. Osteoporosis international : a journal established as result of cooperation
340 USA. Dec 2006;17(12):1726-1733.
341 6. Howe TE, Shea B, Dawson LJ, et al. Exercise for preventing and treating osteoporosis in
342 postmenopausal women. The Cochrane database of systematic reviews. 2011(7):CD000333.
17
343 7. de Kam D, Smulders E, Weerdesteyn V, Smits-Engelsman BC. Exercise interventions to reduce
344 fall-related fractures and their risk factors in individuals with low bone density: a systematic review of
345 randomized controlled trials. Osteoporosis international : a journal established as result of cooperation
347 USA. Dec 2009;20(12):2111-2125.
348 8. Kemmler W, Haberle L, von Stengel S. Effects of exercise on fracture reduction in older adults: a
349 systematic review and meta-analysis. Osteoporosis international : a journal established as result of
350 cooperation between the European Foundation for Osteoporosis and the National Osteoporosis
351 Foundation of the USA. Jul 2013;24(7):1937-1950.
352 9. Zhao R, Zhao M, Zhang L. Efficiency of jumping exercise in improving bone mineral density among
353 premenopausal women: a meta-analysis. Sports medicine. Oct 2014;44(10):1393-1402.
354 10. Martyn-St James M, Carroll S. High-intensity resistance training and postmenopausal bone loss: a
355 meta-analysis. Osteoporosis international : a journal established as result of cooperation between the
356 European Foundation for Osteoporosis and the National Osteoporosis Foundation of the USA.
357 2006;17(8):1225-1240.
358 11. Kelley G. Aerobic exercise and lumbar spine bone mineral density in postmenopausal women: a
359 meta-analysis. Journal of the American Geriatrics Society. Feb 1998;46(2):143-152.
360 12. Kelley GA. Aerobic exercise and bone density at the hip in postmenopausal women: a meta-analysis.
361 Preventive medicine. Nov-Dec 1998;27(6):798-807.
362 13. Bergstrom I, Landgren B, Brinck J, Freyschuss B. Physical training preserves bone mineral density in
363 postmenopausal women with forearm fractures and low bone mineral density. Osteoporosis
364 international : a journal established as result of cooperation between the European Foundation for
365 Osteoporosis and the National Osteoporosis Foundation of the USA. Feb 2008;19(2):177-183.
366 14. Bolton KL, Egerton T, Wark J, et al. Effects of exercise on bone density and falls risk factors in
367 post-menopausal women with osteopenia: A randomised controlled trial. J Sci Med Sport. Mar
368 2012;15(2):102-109.
369 15. Bravo G, Gauthier P, Roy PM, et al. Impact of a 12-month exercise program on the physical and
370 psychological health of osteopenic women. Journal of the American Geriatrics Society. Jul
371 1996;44(7):756-762.
372 16. Chilibeck PD, Vatanparast H, Pierson R, et al. Effect of exercise training combined with isoflavone
373 supplementation on bone and lipids in postmenopausal women: a randomized clinical trial. Journal of
374 bone and mineral research : the official journal of the American Society for Bone and Mineral Research.
375 Apr 2013;28(4):780-793.
376 17. Englund U, Littbrand H, Sondell A, Pettersson U, Bucht G. A 1-year combined weight-bearing training
377 program is beneficial for bone mineral density and neuromuscular function in older women.
378 Osteoporosis international : a journal established as result of cooperation between the European
379 Foundation for Osteoporosis and the National Osteoporosis Foundation of the USA. Sep
380 2005;16(9):1117-1123.
381 18. Going S, Lohman T, Houtkooper L, et al. Effects of exercise on bone mineral density in calcium-replete
382 postmenopausal women with and without hormone replacement therapy. Osteoporosis international :
383 a journal established as result of cooperation between the European Foundation for Osteoporosis and
384 the National Osteoporosis Foundation of the USA. Aug 2003;14(8):637-643.
385 19. Jessup JV, Horne C, Vishen RK, Wheeler D. Effects of exercise on bone density, balance, and
386 self-efficacy in older women. Biological research for nursing. Jan 2003;4(3):171-180.
18
387 20. Korpelainen R, Keinanen-Kiukaanniemi S, Heikkinen J, Vaananen K, Korpelainen J. Effect of impact
388 exercise on bone mineral density in elderly women with low BMD: a population-based randomized
389 controlled 30-month intervention. Osteoporosis international : a journal established as result of
390 cooperation between the European Foundation for Osteoporosis and the National Osteoporosis
391 Foundation of the USA. Jan 2006;17(1):109-118.
392 21. Marques EA, Wanderley F, Machado L, et al. Effects of resistance and aerobic exercise on physical
393 function, bone mineral density, OPG and RANKL in older women. Experimental gerontology. Jul
394 2011;46(7):524-532.
395 22. Milliken LA, Going SB, Houtkooper LB, et al. Effects of exercise training on bone remodeling,
396 insulin-like growth factors, and bone mineral density in postmenopausal women with and without
397 hormone replacement therapy. Calcified tissue international. Apr 2003;72(4):478-484.
398 23. von Stengel S, Kemmler W, Engelke K, Kalender WA. Effects of whole body vibration on bone mineral
399 density and falls: results of the randomized controlled ELVIS study with postmenopausal women.
400 Osteoporosis international : a journal established as result of cooperation between the European
401 Foundation for Osteoporosis and the National Osteoporosis Foundation of the USA. Jan
402 2011;22(1):317-325.
403 24. Moher D, Liberati A, Tetzlaff J, Altman DG, Group P. Preferred reporting items for systematic reviews
404 and meta-analyses: the PRISMA statement. Annals of internal medicine. Aug 18 2009;151(4):264-269,
405 W264.
406 25. Higgins J, Green S. Cochrane Reviewers' Handbook 5.0.1 (updated Septemper 2008). Chichester, UK:
407 Wiley: The Cochrane library; 2008.

408 26. Frost HM. Bone "mass" and the "mechanostat": a proposal. The Anatomical record. Sep
409 1987;219(1):1-9.
410 27. Notomi T, Okazaki Y, Okimoto N, Saitoh S, Nakamura T, Suzuki M. A comparison of resistance and
411 aerobic training for mass, strength and turnover of bone in growing rats. European journal of applied
412 physiology. Dec 2000;83(6):469-474.
413 28. Schulte FA, Ruffoni D, Lambers FM, et al. Local mechanical stimuli regulate bone formation and
414 resorption in mice at the tissue level. PloS one. 2013;8(4):e62172.
415 29. Borer KT. Physical activity in the prevention and amelioration of osteoporosis in women : interaction
416 of mechanical, hormonal and dietary factors. Sports medicine. 2005;35(9):779-830.
417 30. Martyn-St James M, Carroll S. A meta-analysis of impact exercise on postmenopausal bone loss: the
418 case for mixed loading exercise programmes. British journal of sports medicine. Dec
419 2009;43(12):898-908.
420 31. Zhao R, Zhao M, Xu Z. The effects of differing resistance training modes on the preservation of bone
421 mineral density in postmenopausal women: a meta-analysis. Osteoporosis international : a journal
423 National Osteoporosis Foundation of the USA. May 2015;26(5):1605-1618.
424 32. Schulz KF, Chalmers I, Hayes RJ, Altman DG. Empirical evidence of bias. Dimensions of methodological
425 quality associated with estimates of treatment effects in controlled trials. Jama. Feb 1
426 1995;273(5):408-412.
427 33. Pildal J, Hrobjartsson A, Jorgensen KJ, Hilden J, Altman DG, Gotzsche PC. Impact of allocation
428 concealment on conclusions drawn from meta-analyses of randomized trials. International journal of
429 epidemiology. Aug 2007;36(4):847-857.
430 34. Cummings SR, Black DM, Nevitt MC, et al. Bone density at various sites for prediction of hip fractures.
19
431 The Study of Osteoporotic Fractures Research Group. Lancet. Jan 9 1993;341(8837):72-75.
432 35. El-Khoury F, Cassou B, Charles MA, Dargent-Molina P. The effect of fall prevention exercise
433 programmes on fall induced injuries in community dwelling older adults: systematic review and
434 meta-analysis of randomised controlled trials. Bmj. 2013;347:f6234.
435 36. Gillespie LD, Robertson MC, Gillespie WJ, et al. Interventions for preventing falls in older people living
436 in the community. The Cochrane database of systematic reviews. 2012;9:CD007146.
437 37. Zhao R, Feng F, Wang X. Exercise interventions and prevention of fall-related fractures in older people:
438 a meta-analysis of randomized controlled trials. International journal of epidemiology. Jul 31 2016.
439 38. Tobias JH. At the crossroads of skeletal responses to estrogen and exercise. Trends in endocrinology
440 and metabolism: TEM. Dec 2003;14(10):441-443.
441 39. Zhao R, Xu Z, Zhao M. Effects of Oestrogen Treatment on Skeletal Response to Exercise in the Hips and
442 Spine in Postmenopausal Women: A Meta-Analysis. Sports medicine. Aug 2015;45(8):1163-1173.
443 40. Nelson ME, Fisher EC, Dilmanian FA, Dallal GE, Evans WJ. A 1-y walking program and increased dietary
444 calcium in postmenopausal women: effects on bone. The American journal of clinical nutrition. May
445 1991;53(5):1304-1311.
446 41. Ammann P, Rizzoli R. Bone strength and its determinants. Osteoporosis international : a journal
448 National Osteoporosis Foundation of the USA. 2003;14(supplement3):S13-18.
449
450
451
20
457
456
455
454
453
452
Figure legends
denotes bone mineral density.
21
FIGURE 1Flow chart for selection of studies. RCTs represent randomized controlled trials. BMD
458
459 FIGURE 2Forest plots for assessing the change in bone mineral density (BMD) associated with
460 combined exercise interventions. The dotted line represents the mean treatment effect. The diamond
461 denotes overall treatment effect with 95% confidence interval (CI). SMD indicates standardized mean
462 difference.
22
467
466
465
464
463
23
with 95% confidence interval (CI) lines. SMD denotes standardized mean difference.
FIGURE 3 Funnel plots for bone mineral density (BMD) outcomes from all of the included studies
468 TABLE 1Characteristics of included trials
Study Devs
Age (yrs) Sample Setting of
author and Exercise interventions Controls Supplements and
[mean ± SD] size (n) interventions
country ROIs
Bergstrom E:58.9±4.3 E:60 Three times of walking, 1 to 2 Continue Center-based, Vitamin D DXA,
et al13 2008 C:59.6±3.6 C:52 sessions of upper and lower body their usual supervised and calcium Ls,
care
Sweden strengthening exercises, and aerobic Ht
exercise, 4-5 days per week.
Duration: 12 mo
Compliance: 95%
Bolton et E:60.3±5.6 E:19 Two sets of 8 reps and 1 set of 12 Continue Center/home- No DXA,
al142012 C:56.3±4.7 C:20 reps of resistance training, plus their usual based, supplements Ls,
care and partially
Australia daily jumping and balance exercise, Ht
levels of supervised
6 days per week.
physical
Duration: 52 weeks
activity
Compliance: 88 %
Bravo et E:59.6±5.8 E:70 Three times of upper body Continue Center-based, Not DXA,
al151996 C:59.9±6.4 C:72 strengthening exercises, plus their daily supervised statement Fn,
routine
Canada aerobic dancing and stepping, 3 Ls
activities
days per week.
Duration: 12 mo
Compliance: Not statement
Chilibeck et E:55.3±6.3 E:86 Two sets of 8 reps of upper and Take Center/home- No DXA,
al16 2013 C:56.4±7.1 C:88 lower body resistance exercises at placebo based, supplements Fn,
only partially
Canada 80% 1 RM, plus 2 times of walking, Ls,
supervised
4 days per week. Bt, Ht
Duration : 24 mo (data on 12 mo)
Compliance: 77%
Englund et E:72.8±3.6 E:24 Two sets of 8-12 reps of lower body Maintain Center-based, No DXA,
al172005 C:73.2±4.9 C:24 strengthening exercises, plus normal supervised supplements Fn,
physical
Sweden walking, jogging or stepping, 2 days Ls, Bt
activity
per week.
Duration: 12 mo
Compliance: 67%
Going et E:55.8±4.7 E:91 Two sets of 6-8 reps of upper and Take Center-based, 800 mg DXA,
al18 2003 C:57.1±5.0 C:70 lower body resistance training at placebo supervised calcium Fn,
only
USA 70% or 80% 1 RM, plus citrate daily Ls, Bt
weight-bearing circuit-jogging,
24
skipping, hopping and weight vest
stepping, 3 days per week.
Duration: 12 mo
Compliance: 79.9%
Jessup et E:69.1±2.8 E:10 Three times of 8-10 reps of Sedentary Center-based, 1000 mg DXA,
19
al 2003 C:69.4±4.2 C:10 resistance training at 50% and 75% participant supervised calcium and Fn,
s
USA 1 RM, plus weighted vest walking, 400 IU Ls
stair-climbing, and balance-training vitamin D
exercises, 3 days per week. daily
Duration: 8 mo
Korpelainen T:72.9±1.1 T:84 A combined programme of Continue Center/home- No DXA,

20
et al 2006 C:72.8±1.2 C:76 jumping, walking, lower body their daily based, supplements Fn,
Finland exercises, dropping, dancing, routine partially Ht
stamping, and stair climbing, etc, 6 activities supervised
days per week.
Duration: 30 mo (data on 12 mo)
Compliance: 75%
Marques et E:67.3±5.2 E:24 Dynamic aerobic activities and Continue Center-based, Not DXA,
al212011 C:67.9±5.9 C:24 lower body strength exercises, their daily supervised statement Fn,
routines
UK involving stepping, skipping, graded Ht
and usual
walking, jogging, dancing, aerobics
physical
and step choreographies, etc, 3 days
activity
per week. levels
Duration : 8 mo
Compliance: 78.4%
Milliken et E/C:56.9 ± E:26 Two sets of 6-8 reps of upper and Take Center-based, No DXA,
al22 2003 4.6 C:30 lower body resistance training at placebo supervised supplements Fn,
only
USA 70–80% 1 RM, and aerobic Ls, Bt
weight-bearing activity including
walking, jumping, skipping, and
stepping while wearing weight
vests, 3 days per week.
Duration: 12 mo
von Stengel E:68.6±3.0 E:50 Two times of dancing aerobic Perform Center/home- 1500 mg DXA,
et a23 2011 C:68.1±2.7 C:51 exercise, balance training, 1-3 sets light based, calcium and Ls,
physical partially
Germany of lower body strength training, and 400 IU Ht
exercises supervised
15 reps of upper body stretching vitamin D
and a
25
exercises, plus 2 times of home relaxation daily
exercise involving strength and program
weekly
stretching exercises, 4 days per
week.
Duration: 18 mo
Compliance: 75%
469 Note: The sample size listed is the number of participants originally allocated in the individual studies.
470 E exercise training group; C control group; yrs years; mo months; reps repetitions; RM repetition maximal; Devs
471 devices; ROIs regions of interest; DXA dual energy X-ray absorptiometry; excl excluded; Ls lumbar spine; Fn
472 femoral neck; Ht total hip; Bt total body.

26
473 TABLE 2Risk of bias assessment for individual studies
Sequence Allocation Blinding Incomplete Selective Other potential

generation concealment outcome data reporting sources of bias
Bergstrom et al13 Low Unclear Unclear Low Low Low
Bolton et al14 Low Low Low Low Low High 2
Bravo et al15 Low Low Low High Low Low
Chilibeck et al16 Low Low Low Low Low Low
Englund et al17 Unclear Unclear Unclear Low Low High 2
Going et al18 Unclear Unclear Unclear High Low Low

Jessup et al19 Low Low Unclear Low Low High2
Korpelainen et al20 Low Low Low Low Low Low
Marques et al21 Low Low Unclear Low High High2
Milliken et al22 Unclear Unclear Unclear Low High High2
von Stengel et al23 Low Low Low Low Low Low

474 Notes: Other potential sources of bias include the following items: 1. imbalance in bone baseline variables; 2.
475 small sample size (<100 participants) in relation to the measurement precision; and 3. short follow-up time (< 6
476 months).
27
477 TABLE 3 Primary and subgroup analyses.
Test for
Studies Subjects Heterogeneity Inconsistency Statistical SMD
Analysis overall
(n) (n) (p value) (I2) methods 95%CI
effect
All Studies
Lumbar 9 E:389 0.854 0.0% Fixed-effects 0.170 z=2.34

spine C:370 methods (0.027 (p=0.019)
0.313)
Femoral 8 E:371 0.312 15.1% Fixed-effects 0.177 z=2.37
neck C:352 methods (0.030 (p=0.018)
0.324)
Total hip 6 E:310 0.114 43.7% Fixed-effects 0.198 z=2.41
C:291 methods (0.037 (p=0.016)
0.359)
Total body 4 E:195 0.828 0.0% Fixed-effects 0.257 z=2.47
C:180 methods (0.053 (p=0.014)
0.461)
Women
younger than
60 yr
spine C:274 methods (-0.055 (p=0.191)
0.275)

neck C:222 methods (0.034 (p=0.020)
0.404)
Women older
than 60 yr
spine C:96 methods (0.064 (p=0.016)
0.634)
neck C:130 methods (-0.204 (p=0.373)
0.544)
478 E exercise training group; C control group; yr years; SMD standardized mean difference
479
480
28

The Effectiveness of Combined Exercise Interventions For Preventing Postmenopausal

Caricato da

Informazioni sul documento

Titolo originale

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

The Effectiveness of Combined Exercise Interventions For Preventing Postmenopausal

Caricato da

Copyright:

Formati disponibili

The Effectiveness of Combined Exercise Interventions for Preventing Postmenopausal

Bone Loss: A Systematic Review and Meta-Analysis

Renqing Zhao1*2, PhD; Mengyi Zhang2, MS; and Qi Zhang2, MS

1 College of Physical Education, Yangzhou University, Yangzhou, Jiangsu, China

University, College of Physical Education, 88 South University Avenue, Yangzhou, Jiangsu

225009, China. E-mail: zhaorenqing@hotmail.com

Foundation of China under Grant (No. LY14H070001).

Conflict of interest: None

Words counts: 3516

Running title: Exercise for preserving BMD in Older Women

2 STUDY DESIGN: Systematic review and meta-analysis.

3 BACKGROUND: It remains unclear whether exercise combining different types of physical

4 activities (combined exercise interventions) would effectively preserve postmenopausal

5 women’s bone mineral density (BMD) at different sites.

6 OBJECTIVES: To examine the impact of combined exercise interventions on lumbar spine,

8 METHODS: An electronic database search was conducted in PubMed, EMBASE,

17 different physical activities significantly increased lumbar spine (SMD=0.170,

18 95%=0.027-0.313, p=0.019), femoral neck (SMD=0.177, 95%CI=0.030-0.324, p=0.018),

19 total hip (SMD=0.198, 95%CI=0.037-0.3590, p=0.016) and total body (SMD=0.257,

20 95%CI=0.053-0.461, p=0.014) BMD. Combined exercise interventions generated a beneficial

21 effect on femoral neck BMD (SMD=0.219, 95%CI=0.034-0.404, p=0.020) in groups with

24 CONCLUSION: Our findings suggest that combined exercise interventions appear to be

26 hip and total body.

27 Level of Evidence: Therapy, level 1a.

28 Key words: Exercise; Women; Menopause; Bone loss; Meta-analysis

46 protocols in preventing postmenopausal bone loss13-23. However, the outcomes were

55 with controls maintaining their daily routine activities.

57 Search Strategy and Inclusion Criteria

58 We conducted a systematic review and meta-analysis in accordance with PRISMA

60 predetermined protocol. An electronic database search was conducted in PubMed, EMBASE,

63 “resistance training/exercise”, “strength training/exercise”, “aerobic exercise”,

64 “weight-bearing exercise”, “impact exercise”, “jumping”, “skipping”, and “bone density”,

80 protocols/interventions. Participants in control group were frequently required to maintain

81 their usual care and levels of physical activity.

82 Data Extraction and Quality Assessment

92 Collaboration risk of bias assessment tool25. Generally, we concentrated on the following

93 issues: sequence generation, allocation concealment, blinding, incomplete outcome data,

95 interested in whether randomization was accomplished according to the guideline. We also

99 issues related to participant withdrawals. Likewise, we checked the selective outcome

105 included studies was presented in a “risk of bias” table.

106 Data Synthesis, Analysis, and Assessment of Heterogeneity

111 pooled standard deviation.

114 I2 is the percentage of variation attributable to heterogeneity, which is frequently calculated

115 by the following formula25:

116 I2=(Q - df)/Q × 100%,

117 where Q is the chi-squared statistic and df is its degrees of freedom.

125 meta-analysis and produce the graph.

126 Subgroup Analyses

132 Study Characteristics

151 data on intervention period of 12 months.

152 Exercise Interventions

161 Risk of Bias Assessment

167 Effects of Exercise Interventions on BMD

177 95%=0.027-0.313, p=0.019), femoral neck BMD (SMD=0.177, 95%CI=0.030-0.324,

179 (SMD=0.257, 95%CI=0.053-0.461, p=0.014), respectively(FIGURE 2).

190 p=0.016) in relative older participants (≥60 years) (TABLE 3).

191 Funnel Plots

205 combined exercise protocols on BMD in postmenopausal women. Combined exercise

210 to combined exercise interventions at lumbar spine.

233 neck, total hips and total body in postmenopausal women.

267 because subgroup analysis only included a small number of studies.

276 to distinguish real asymmetry.

289 failing to be accounted for.