Effects of Pilates Exercise on Bone Mineral Density in Postmenopausal Women: A Systematic Review and Meta-analysis

Abstract

Background and Purpose: 

Despite the popularity of Pilates exercises among postmenopausal women, few studies have devoted attention to verifying the effects of the technique on bone mineral density (BMD), and, to date, no systematic review and meta-analysis have been conducted on this topic. Our objective was to conduct a systematic review and meta-analysis of randomized controlled trials examining the effect of Pilates on BMD.

Methods: 

Randomized controlled trials were considered eligible, with follow-up of 6 months and more, which verified the effects of Pilates exercise on the BMD of postmenopausal women. The calculations of the meta-analysis were performed through the weighted mean difference between the Pilates exercise and control groups, through the absolute change between pre- and postintervention in the areal bone mineral density.

Results: 

Three randomized controlled trials met the inclusion criteria and were included in the meta-analysis. Only 1 study presented satisfactory methodological quality. Pilates exercises did not offer significant effects to improve areal bone mineral density of the lumbar spine (0.019 g/cm² [95% confidence interval (CI), −0.018 to 0.057], P = .32), total hip (0.012 g/cm² [95% CI, −0.002 to 0.027], P = .10), or femoral neck (0.000 g/cm² [95% CI, −0.021 to 0.022], P = .97).

Conclusions: 

Pilates exercises had no significant effects on BMD in postmenopausal women. However, the few studies included in the meta-analysis and the low methodological quality of the majority of the studies do not allow safe extrapolation of the results at this time. More robust randomized controlled trials with high methodological quality are needed so that the results of this meta-analysis can be confirmed.

INTRODUCTION

Physical exercise is considered an important adjunct in the prevention and treatment of osteoporosis.^1,2 The exact mechanism by which mechanical stimulation enabled by exercise is converted into anabolic or catabolic signaling (mechanotransduction) in bone tissue is not yet fully understood. Bone mass is regulated, among other factors, by mechanical forces derived from skeletal muscle; thus, changes in strength and muscle mass provided by regular and systematic exercise practice, especially with greater overload, seem to contribute to the mechanical load on bone.¹ Another aspect concerns the endocrine and metabolic changes that occur as a result of exercise, through biochemical responses that provide anabolic effects directly on bone tissue.²

A systematic review and meta-analysis of randomized controlled trials (RCTs) showed that progressive resistance exercises performed with low repetition and increased overload were effective for increasing bone mineral density (BMD) of the lumbar spine and femoral neck in postmenopausal women.³ On the other hand, the same study showed that no significant results were observed for muscle resistance exercises that used low overload and a higher number of repetitions. Other forms of intervention, such as supporting static weight, had an effect only on BMD when considering women older than 70 years, with analysis involving only 1 study. On the other hand, the combination of different forms of exercise allowed an increase in BMD of the lumbar spine, femoral neck, and trochanter.³

One possibility for progressive resistance exercises that use few repetitions and greater overload is Pilates.^4,5 This technique uses specific equipment, often consisting of springs of different intensities that provide resistance. Pilates exercises can also be performed with accessories such as elastic bands and balls to provide resistance. It is also common to perform the technique using only 1 mat, in which movements are performed with body weight resistance.⁶ Pilates aims to work on global body strengthening, focusing primarily on the abdominal and paravertebral muscles, postural alignment, and development of flexibility, stimulating the performance of movements through 6 principles (concentration, centering, control, breath, fluidity, and precision).⁵

Over the past 2 decades, Pilates exercises have become increasingly popular among postmenopausal women, which has aroused the interest of the scientific community to further investigate the effects of the technique on different outcomes that impact the health of this population.⁷ Regarding bone health, as Pilates is a technique that works on muscle strengthening, Pilates was quoted alongside weight training and yoga in the “Clinician's Guide to Prevention and Treatment of Osteoporosis,” published in 2014, as an adjunct in the prevention/treatment of osteoporosis.⁸

The first RCT that investigated the effects of Pilates exercise on BMD in postmenopausal women was published in 2015.⁹ Thus, to date, there are few published studies on this subject^9–11 and no systematic reviews and meta-analysis, despite the relevance of this outcome, especially at this stage of life. In typical Pilates exercises, there is a great demand from the muscles that perform flexion, extension, and stabilization of the trunk and the hip.⁵ In addition, the contraction of deep muscles is stimulated, such as obliques, transverse abdomen, multifidus, and the entire pelvic floor region,^5,6 which supposedly has the potential to enable effects on BMD of the spine and the hip. These factors, in a way, differentiate Pilates exercises from traditional dynamic resistance exercises. However, it is necessary to investigate whether this factor can offer additional effects on BMD.

With this in mind, the present systematic review and meta-analysis study of RCTs aimed to identify the effects of Pilates exercise on BMD in postmenopausal women compared with no treatment or other forms of intervention.

METHODS

This study followed the recommendations of the PRISMA protocol¹² and is registered in PROSPERO (https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=156609, CRD42020156609). Inclusion criteria were (a) randomized controlled clinical trials, which investigated the effects of Pilates exercises on BMD in postmenopausal women (there were no restrictions on ethnicity or level of physical activity); (b) studies with an intervention period of 6 months or more; and (c) studies in which the Pilates training took place using equipment, accessories, or mat.

The exclusion criteria were (a) studies with duplicate information in another RCT; (b) not using Pilates exercises; (c) Pilates exercises not associated with the observation of BMD; (d) women who were not in the postmenopausal period; (e) women who had pathological conditions or secondary causes of osteoporosis; (f) studies with male participants; (g) studies in which participants who used medication to improve BMD were not divided equally between the Pilates and control groups; and (h) studies that associated Pilates exercises with another form of intervention.

Databases and Search Strategies

The search was carried out in the following databases: PubMed, Web of Science, LILACS, SciELO, the Cochrane Library, and PEDro, without the use of a filter that limited the publication date or language. Searches of clinical trial registration databases (clinicaltrials.gov and apps.who.int/trialsearch/) were also carried out to find unpublished studies. The final search took place on May 2, 2020.

As a search strategy, the following key words were selected: (“postmenopausal” OR “menopause” OR “perimenopause” OR “climacteric” OR “postmenopausal women” OR “aged” OR “aging” OR “ageing” OR “elderly” OR “older people” OR “older adults” OR “older adult” OR “older women” OR “geriatric” OR “geriatrics”) AND (“Pilates” OR “Pilates method” OR “Pilates-based exercises” OR “Pilates exercise” OR “clinical Pilates” OR “clinic Pilates” OR “Pilates training” OR “mat Pilates” OR “mat-based Pilates” OR “equipment-based Pilates”) AND (“bone mineral density” OR “bone density” OR “bone” OR “bones” OR “bone mass” OR “bone mineral content” OR “BMC” OR “BMD” OR “bone strength” OR “osteoporosis” OR “osteopenia”). The search strategy was adapted for each database used when necessary.

Study Selection

One reviewer (R.G.O.) carried out the initial search strategy in the databases, extracting the titles and abstracts. Subsequently, the selection of studies, evaluation, and data extraction was conducted blindly by 2 authors (G.E.U.A. and E.A.C.), based on the reading of titles and abstracts. Potentially eligible articles were read in full. A manual search was performed in the reference lists of all eligible articles in an attempt to find new references. Disagreements, when not resolved between the 2 reviewers, were passed on to a third reviewer (L.C.O.) who decided on the issue. The same form for data extraction was used by all reviewers.

The PICO method was used to structure the bibliographic search and data extraction: P (population) = postmenopausal women; I (intervention) = Pilates exercises; C (comparison) = no intervention, or other forms of intervention; and O (outcome) = BMD.

Data Extraction

The following information was extracted from each study: (a) name of the first author and year of publication; (b) number of volunteers allocated to each group; (c) country where the study was carried out; (d) mean and standard deviation of age in each group; (e) duration of the study, weekly frequency, and time of each session; (f) exercises used (protocols, control of exercise intensity, and number of sets and repetitions); (g) sequence of exercises; (h) equipment and accessories used; (i) activities carried out by other intervention groups; (j) activities of the control group; (k) BMD assessment instrument and assessed region; (l) condition of the volunteers regarding changes in BMD (without changes, osteopenic or osteoporotic); (m) use of calcium, vitamin D, or medications; (n) results reported for the BMD in the comparison between groups; (o) percentage of volunteers who completed the Pilates program; and (p) adverse events.

Evaluation of the Methodological Quality of the Studies

Methodological quality was assessed using the PEDro scale (Physiotherapy Evidence Database)^13,14 by 2 blind reviewers (G.E.U.A. and E.A.C.). A third reviewer (L.C.O.) was requested in case of disagreement. The PEDro scale takes into account the internal validity and the sufficiency of statistical information from the studies and presents 11 questions, with 3 items on the Jadad scale¹⁵ and 9 items on the Delphi list.¹⁶ The first question is not scored (related to the external validity of the study), and the other 10 questions are scored. Each item that meets the required criteria receives 1 point, making it possible to classify the quality of each study as excellent (9-10), good (6-8), fair (4-5), or poor (<4). Studies with a score of less than 6 were considered as low quality. Maher et al¹⁴ demonstrated good reliability between raters, with an intraclass correlation coefficient of 0.68 when using consensus ratings, generated by 2 or 3 independent raters on the PEDro scale.

Definition of Pilates

Pilates was defined by muscle-strengthening exercises, which respect 6 traditional basic principles (concentration, centering, control, breath, fluidity, and precision) during execution,^5,6 performed on specific equipment (eg, Cadillac, Reformer, Chair, and Ladder Barrel), accessories (eg, thera-band and ball), or a gym mat, in which the weight of the body itself is used as resistance.

Synthesis of Results

For the meta-analysis, the measurement of effect was the weighted mean difference between the Pilates groups versus control, in absolute alteration of the bone mineral density area (aBMD) between pre- and postintervention, for each segment (lumbar spine, total hip, and femoral neck) evaluated using dual-energy x-ray absorptiometry expressed in grams per square centimeter (g/cm²).

The Cochran Q test for heterogeneity was performed and considered statistically significant if P ≤ .10. Heterogeneity was also quantified with the statistical I², whereby 0% to 40% may not be important, 30% to 60% may represent moderate heterogeneity, 50% to 90% may represent high heterogeneity, and 75% to 100% is defined as considerable heterogeneity.¹⁷ Fixed-effects models were used when there was no statistically significant heterogeneity; otherwise, random-effects models were used. A meta-analysis was carried out only if 2 or more studies evaluated the same outcome. The values for the effect of treatment were considered statistically significant when P < .05. Because of the number of studies included in the meta-analysis, it was not possible to assess the risk of publication bias by the funnel plot. In addition, also due to the low number of studies, it was not possible to perform sensitivity and subgroup analyses. All analyses were performed using the program Review Manager (RevMan) 5.3, Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration.

Statistical Manipulation of BMD Data

For all procedures, for data manipulation and combination of results, the recommendations described in the Cochrane Handbook for Systematic Reviews of Interventions were followed.¹⁷ If a study did not report the mean and standard deviation of the difference between pre- and postintervention, the authors were contacted.¹⁰ If there was no answer, we performed statistical inference to obtain the BMD results, in the form of mean and standard deviation of the difference between pre- and postintervention, using the raw data provided in the original publication. When the absolute values of mean and standard deviation for the moments before and after intervention¹⁰ were available, the mean of the difference was calculated by subtracting the average of the postintervention by the average of the preintervention. In this case, the standard deviation corresponding to the absolute change in BMD was estimated using Follmann's method.¹⁷ A correlation coefficient (r) value of 0.95 was used in the calculation, since this correlation coefficient proved to be conservative in simulations that we performed with data from one of the studies included in this review.¹¹

RESULTS

Qualitative Synthesis of the Studies

Initially, 63 potentially relevant titles and abstracts were identified. After removing duplicates, the titles and abstracts of 47 studies were read, of which 37 did not meet the inclusion criteria. Subsequently, 12 studies were read in full, of which 9 did not meet the eligibility criteria (a complete list of studies excluded at this stage is available in Supplemental Digital Content Table 1, available at: https://links.lww.com/JGPT/A68). The main reason for exclusion was the failure to assess BMD (5 studies), followed by the fact that the study was not an RCT (4 studies). Thus, 3 studies met the inclusion criteria and could be included in the systematic review and meta-analysis (Figure 1).

Figure 1.: PRISMA flowchart presenting the summary of searches carried out in the literature. BMD indicates bone mineral density; RCT, randomized controlled trial.

The RCTs included in this systematic review (Table 1)^9–11 were published between the years 2015 and 2019, and the total number of participants was 116 (ranging from 24¹⁰ to 51 women¹¹). The groups in each study ranged from 2^9,10 to 3,¹¹ with an average age of the volunteers ranging from 54.1¹¹ to 65.2¹⁰ years and interventions lasting 6 months.^9–11 The weekly frequency of Pilates ranged from 2¹⁰ to 3 times,^9,11 with session durations between 30¹⁰ and 60 minutes.^10,11 Over the 6 months of intervention, studies used 2¹¹ or 3^9,10 different exercise protocols. The number of sets and repetitions for each exercise was reported in only 1 study,¹¹ while the exercise sequence ranged from just strengthening the lower limbs and core¹⁰ to performing a complete sequence, involving initial warm-up/stretching, strengthening whole body, and final cooling/stretching.^9,11 Two studies used traditional equipment to practice Pilates, consisting mainly of springs as a form of resistance,^10,11 while another study used accessories, such as a ball and thera-band.⁹

Table 1. - Summary of the Articles Included in the Systematic Review

Authors	Number of Volunteers and Location	Mean and SD of the Age of the Groups	Pilates Exercise Protocol	Activities of the Second Intervention Group	Activities of the Control Group (CON)	BMD Evaluation Tool and Evaluated Region	Condition of the Volunteers Regarding Alterations in BMD	Administration of Calcium, Vitamin D, or Medication	BMD Results Reported in the Original Publication (P < .05)
Angin et al (2015)9	41: Pilates = 22 CON = 19 Cyprus	Pilates: 58.2 ± 5.4 CON: 55.9 ± 9.2	6 mo; 3× wk.; 60': Protocol 1 (6 wk): 22-23 exercises (clinical Pilates exercises) Protocol 2 (12 wk): 22-27 exercises (resistive exercises with thera-band) Protocol 3 (6 wk): 24-25 exercises (resistive exercise with Pilates ball) Series × repetitions: not reported Sequence: warm-up, main exercise program, and cooling phases Equipment or accessories: mat, thera-band, and ball	...	Maintain their normal routine	DXA: lumbar spine	Osteoporotic	Fosamax 70 mg, once a week	Pilates: improved intergroup (vs CON) for lumbar spine
Liposcki (2016)10	24: Pilates = 9 CON = 11 Brazil	Pilates: 63.7 ± 3.3 CON: 65.2 ± 3.0	6 mo; 2× week.; 30': Protocol 1 (8 wk): 12 exercises (adaptation) Protocol 2 (9 wk): 12 exercises (increase in intensity) Protocol 3 (8 wk): 12 exercises (maintenance) Series × repetitions: not reported Sequence: strengthening of lower limbs and core Equipment or accessories: Cadillac, Reformer, Chair, and Mat	...	Maintain their normal routine	DXA: lumbar spine, femoral neck, total hip	No alterations, osteopenic or osteoporotic	Not used	No results found
Oliveira et al (2019)11	51: Pilates = 17 WBV: 17 CON = 17 Brazil	Pilates: 55.6 ± 6.8 WBV: 56.4 ± 6.5 CON: 54.1 ± 5.3	6 mo; 3× week; 60': Protocol 1 (3 mo): 21 basic exercises (Borg CR10: 5-6) Protocol 2 (3 mo): 21 intermediate exercises (Borg CR10: 5 to 6) Series × repetitions: 1 × 10 Sequence: initial stretches; strengthening of lower limbs, core, and upper limbs; final stretching Equipment or accessories: Chair, Wall Unit, Cadillac, Ladder Barrel, Spine Corrector, Reformer, and Ball	WBV: 6 mo, 3× week; 5': Frequency: 20 Hz Magnitude: 3.2 g Equipment: side-alternating vibration	Maintain their normal routine	DXA: lumbar spine, femoral neck, total hip, trochanter, intertrochanter, and Ward's area	No alterations, osteopenic or osteoporotic	Not used	Pilates and WBV: improved intergroup (vs CON) for lumbar spine and trochanter

Abbreviations: BMD, bone mineral density; CON, control group; DXA, dual-energy x-ray absorptiometry; WBV, whole body vibration group.

Only 1 study used a second intervention group, involving whole body vibration training¹¹ and all included a control group that maintained the usual routine.^9–11 Bone mineral density assessment was performed using dual-energy x-ray absorptiometry in all cases, involving the regions of the lumbar spine,^9–11 femoral neck,^10,11 total hip,^10,11 trochanter, intertrochanter, and Ward's area.¹¹ Regarding changes in BMD, 2 studies included volunteers regardless of classification (without changes, osteopenic, or osteoporotic)^10,11 and 1 included only women with osteoporosis.⁹ Only 1 study administered medication, which was offered to volunteers of both the Pilates and the control groups.⁹

Of the 3 RCTs included in this systematic review, 1 found no differences between the Pilates and the control groups.¹⁰ The other 2 studies identified significant effects on increased lumbar spine aBMD in favor of the Pilates group,^9,11 in addition to the trochanter region for 1 study,¹¹ compared with the control groups. Compliance with Pilates programs was reported by only 1 study,¹¹ which reported a frequency of 92.6%.

Methodological Quality of the Studies

Table 2 shows the methodological quality of the studies according to the PEDro scale. One study¹¹ scored 8 points, which indicates high methodological quality; however, the other 2 studies scored 5 points and were considered of low methodological quality.^9,10

Table 2. - Methodological Quality of the Studies Included in the Systematic Review, as Evaluated by the PEDro Scale

Authors	Eligibility Criteria	Random Allocation	Concealed Allocation	Baseline Comparability	Blind Subjects	Blind Therapists	Blind Assessor	Adequate Follow-up Dropout: <15%	Intention-to-Treat Analysis	Between-Group Comparisons	Point Estimates and Variability	Score
Angin et al (2015)9	Yes	Yes	No	Yes	No	No	No	Yes	No	Yes	Yes	5
Liposcki (2016)10	Yes	Yes	No	Yes	No	No	Yes	No	No	Yes	Yes	5
Oliveira et al (2019)11	Yes	Yes	Yes	Yes	No	No	Yes	Yes	Yes	Yes	Yes	8

Quantitative Synthesis of the Studies (Meta-analysis)

There was no significant effect for any segment, comparing Pilates and control groups, for the aBMD of the lumbar spine (0.019 g/cm² [95% confidence interval (CI), −0.018 to 0.057], P = .32, n = 95, studies = 3, I² = 82%; Figure 2a), total hip (0.012 g/cm² [95% CI, −0.002 to 0.027], P = .10, n = 54, studies = 2, I² = 0%; Figure 2b), and femoral neck (0.000 g/cm² [95% CI, −0.021 to 0.022], P = .97, n = 54, studies = 2, I² = 0%; Figure 2c).

Figure 2.: Analyses of Pilates effect on areal bone mineral density (g/cm²) in postmenopausal women: (a) Lumbar Spine; (b) Total hip; (c) Femoral neck. Forest plots show the weighted mean difference between the Pilates and control groups in absolute pre-post change. Note: The RevMan 5.3 software reported all bone mineral densities in terms of two decimal places in the forest plots. In the text, areal bone mineral densities are reported in terms of three decimal places.

Adverse Events

Only 1 study reported adverse events.¹¹ In this case, 2 falls were reported in the Pilates group, 2 falls in the group that performed whole body vibration, and 1 fall in the control group (which led to the fracture of the wrist of the control group participant). All falls occurred outside the exercise sessions. The main complaint reported by the participants in the Pilates group was delayed-onset muscle soreness, especially in the first weeks of intervention. Other less serious adverse events, such as pain in specific regions of the body, muscle spasms, and cramps, occurred less frequently.

DISCUSSION

Summary of the Main Results

When analyzing the qualitative synthesis of the studies available in Table 1, it can be observed that the Pilates exercises demonstrate potential for increasing the lumbar spine aBMD, since 2^9,11 of the 3 studies included in this systematic review identified significant results in the comparisons with the control groups, although it is not possible to establish this relationship for the other bone regions. However, our meta-analysis found no significant effect for Pilates exercises, in comparison with the control groups, for aBMD of the lumbar spine, total hip, and femoral neck. Sensitivity analysis that could be performed to exclude low-quality studies was not feasible, considering that only 1 study¹¹ presented satisfactory methodological quality. Likewise, different subgroup analyses, which could explain factors capable of impacting significant effects to improve aBMD, such as weekly frequency, session duration, and use of equipment or accessories, among others, were not feasible due to the low number of studies included in the meta-analysis.

Agreements and Disagreements With Other Studies

This is the first systematic review and meta-analysis that aimed to identify the effects of Pilates exercises on BMD in postmenopausal women. Other systematic review studies that sought to investigate the effects of different types of physical exercise on BMD in postmenopausal women found conflicting results. Howe et al³ included different exercise modalities broken down by categories in the meta-analysis. The authors identified that muscle resistance exercises with greater overload and few repetitions provided significant effects on aBMD of the lumbar spine and femoral neck, while exercises with a high number of repetitions and low overload did not provide any effect. On the other hand, in the meta-analysis carried out by Zhao et al,¹⁸ it was identified that resistance exercises alone did not promote significant effects on the lumbar spine and femoral neck aBMD in postmenopausal women. Significant effects occurred only when resistance exercises were combined with high-impact or weight-bearing exercises.

In the present study, Pilates exercises were considered exclusively. This technique uses more overload and fewer repetitions (typically between 5 and 10 repetitions); however, no significant results were observed. One factor that must be taken into account is that only 1 study¹¹ reported the number of repetitions and claimed to have controlled the intensity of effort through the Borg scale (CR10),¹⁹ in which the participants maintained a perception of high effort (Borg scale score between 5 and 6). This factor is relevant, since the control of physical effort in Pilates sessions is not very objective, with the overload provided by body weight, thera-bands, or springs attached to the equipment.

Contrary to what happens with Pilates exercises, in conventional resistance training, free weights and specific machines are used, in which it is possible to objectively establish the amount of effort of the performer, typically in pounds or kilograms. In this case, the control of the intensity of the physical effort and consequent progression of the load can occur through the percentage of the maximum effort in each exercise (eg, 80% of 1 maximum repetition), as presented in the meta-analysis by Zhao et al.¹⁸ Another factor to be considered is that the studies included in this review^9–11 used only Pilates exercises, that is, not combined with high-impact or weight-bearing exercise, which may justify the results found.

It should also be considered that the 3 studies included in this systematic review and meta-analysis had a total intervention time of 6 months, which has been considered in the literature as the minimum time necessary for changes in bone tissue to be observed.^3,18 Regarding the weekly frequency and duration of each session, the 2 studies^9,11 that reported significant effects on the lumbar spine aBMD performed 3 weekly sessions, lasting 60 minutes each, while the study that found no significant results¹⁰ included 2 weekly sessions, lasting only 30 minutes each. Thus, for the first 2 studies,^9,11 the total cumulative dose of exercise (total number of sessions over the study × duration in minutes of each session) was 4680 minutes, while in the study by Liposcki,¹⁰ the total was only 1560 minutes, a factor that should be considered in future studies. Other meta-analyzes^3,18 did not perform subgroup analyzes for these variables, which makes it impossible to determine whether in conventional resistance training, the weekly frequency, and duration of each session are capable of impacting the measures of effect for BMD in postmenopausal women. However, the literature has highlighted that the cumulative dose of exercise is a determining factor in the definition of protocols that aim to prevent osteoporosis in postmenopausal women.²⁰

Another important point is the failure to establish relationships between Pilates exercises and other forms of exercises, such as those used to strengthen the core, or dynamic resistance exercises used in conventional physiotherapy, in relation to the effects on BMD. The only comparison made possible at this time was between Pilates and whole body vibration exercises, through one of the studies included in this review.¹¹ In this case, both interventions offered an equal effect on BMD.

Quality of Evidence

The 3 studies included in our meta-analysis were from Cyprus⁹ and Brazil,^10,11 which limits generalizations regarding the applicability of evidence in other locations. The methodological quality of 2 studies^9,10 was low and should be taken into account when interpreting the results of our review. No study blinded the participants and therapists and 2 studies blinded the evaluators.^10,11 In therapies in which the stimuli are noticeable, it is very difficult, if not impossible, to blind participants and therapists¹³ as is the case with interventions involving Pilates exercises. However, it is unlikely that this bias may have influenced BMD measures. Two RCTs^9,11 had follow-up from at least 85% of the participants; however, all studies used small samples.

Potential Biases in the Review Process

Our review included only RCTs, which reduces the risk of bias. However, most studies^9,10 did not hide the distribution of volunteers in each group. Only 1 study adequately dealt with incomplete data in the follow-up, performing analysis of intention to treat.¹¹ For 2 studies,^9,11 we had the mean and standard deviation of the absolute difference in BMD, between pre- and postintervention. However, in 1 study,¹⁰ it was necessary to perform statistical manipulations to determine the standard deviation of the difference, which could decrease the accuracy of the data inserted in the meta-analysis. Finally, the search did not extend to all existing databases. However, we performed searches in 4 primary databases (PubMed, Web of Science, LILACS, and SciELO), 2 secondary databases (the Cochrane Library and the PEDro), and 2 clinical trial registration databases (clinicaltrials.gov and apps.who.int/trialsearch/), aiming to find unpublished studies. In addition, we conducted a thorough search of all bibliographic references of the studies included in the review in an attempt to find other RCTs.

CONCLUSIONS

Implications for Practice

Our results suggest that Pilates exercises need to be further investigated in relation to their potential effects on BMD in postmenopausal women, so that they can eventually be recommended for this purpose. Physiotherapists, when opting for Pilates exercises for postmenopausal women aiming at increasing BMD, should do so as an adjunctive treatment but not exclusive. Although our study contributed to preliminary discussions, it was not possible to determine which factors may represent an intervention most likely to have effects on BMD in postmenopausal women. This is mainly due to the low number of studies available at this time and the low methodological quality of 2 of the 3 studies included in the meta-analysis.

Implications for Research

Our results collaborate with the definition of future research protocols involving interventions with Pilates exercises on BMD in postmenopausal women. We suggest future studies with greater methodological care, especially regarding the confidential allocation and adequate treatment of postintervention data, and including analysis of intention to treat. In addition, studies should prioritize a greater cumulative dose of exercise, with adequate control of effort intensity. Finally, studies with a larger number of participants and longer follow-up are necessary, which may help clarify the long-term effects of Pilates exercises on BMD in postmenopausal women.