Osteoporosis and fragility fractures constitute a major public health burden, and its prevalence continues to rise in our ageing population, with postmenopausal women being at the highest risk [1, 2]. Over the past three decades, several pharmacotherapeutic approaches (antiresorptive agents e.g. bisphosphonates, denosumab and osteoanabolic agents e.g. romosozumab, teriparatide) have been developed and proven to reduce fracture risk in postmenopausal osteoporosis [3]. Despite this, the incidence of fragility fractures is still rising, and lifetime prevalence remains very high [1, 2]. In developed countries, one-third of women over the age of 50 years will experience an osteoporotic fracture during their remaining lifetime, commonly associated with devastating consequences such as pain and worse quality of life, mobility, independence and survival [1, 4,5,6,7].

Exercise, has the potential to target fracture risk reduction through numerous avenues, including i) improvements in balance and falls risk reduction, ii) maintenance or gains in bone mineral density (BMD) and iii) beneficial changes in bone structure. There is emerging clinical trial evidence that high-intensity resistance and impact training (HiRIT) is more beneficial to lumbar spine BMD than traditionally prescribed low to moderate-intensity regimens [8]. Further, HiRIT can be performed safely by postmenopausal women with osteoporosis or osteopenia at elevated risk of fracture, when ensuring certain safety principles are in place, including supervision, individualisation and progressive intensity increments [4]. Overall, these skeletal benefits of exercise may translate to fracture-risk reduction given i) majority of fragility fractures, particularly hip fractures, occur secondary to a fall [1, 9] and ii) given BMD improvement is a surrogate marker for fracture risk reduction [10].

In this review, we summarise the latest evidence on skeletal benefits of exercise in postmenopausal women (falls prevention, BMD, fracture-risk reduction), discuss potential mechanisms driving the relationship between skeletal loading and increased BMD and critically discuss future directions in the field, including whether multi-modal combinations between exercise and osteoporosis medication should be further explored. Comprehensive reviews on the effects of exercise on bone health throughout the lifespan [11], in older men [12] and in individuals with sarcopenia [13] and frailty [14] can be found elsewhere.

Evidence for Exercise as a Falls Prevention Strategy in Postmenopausal Women

The majority (> 90%) of fragility hip fractures occur secondarily to falls [1, 9]. Hence, exercise interventions which can reduce the risk of falls in postmenopausal women may impart meaningful benefit on falls-related fractures. A recent Cochrane review of 116 randomised controlled trials (RCTs) in healthy adults ≥ 60 years (n = 25,160, 74% female) found an overall 23% reduction in rate of falls with exercise interventions compared to usual care or control exercise (rate ratio 0.77, 95% CI 0.71–0.83, high-certainty evidence) [15]. A ~ 20–30% reduction in fall rates and risk of falling have consistently been reported in other meta-analyses of exercise RCTs in community-dwelling older adults [16, 17]. Programs that involve balance and functional exercises, or are multi-component in nature (balance, functional and resistance exercises) are more beneficial for falls prevention than any individual exercise component [15,16,17]. A total weekly dose of at least 1.5–3.0 h of multimodal exercise for a minimum of three to six months is required to observe an effect however the most effective exercise prescription for falls prevention is not established due to lack of head-to-head trials [15, 17, 18]. Crucial elements of falls prevention programs are incorporation of exercises that challenge balance [17] and simulate real-life scenarios which carry particular risk for falling. Limitations of these meta-analyses include substantial heterogeneity, unclear risk of bias, poor adherence to exercise and ascertainment bias (with retrospective recall of falls history) [16, 17].

Finnegan et al. assessed long-term persistence of falls reduction by examining trials in community-dwelling older adults (> 65 years) who underwent predominantly gait, balance and functional training with greater than 12-months follow-up [19]. The few trials with over 24-months follow-up achieved a 17% lower risk of falling but no sustained effect on falls reduction. This apparent diminishing effect on falls reduction with longer follow-up may relate to attrition affecting statistical power, limited adherence/persistence to exercise, or a waning of effect over time [19]. It is unclear whether supervision mediates efficacy and safety during falls prevention exercise programs, as this detail is often not reported in trials [17]. However, Sherrington et al. suggested no additional benefit with group exercise and that delivery by a healthcare professional was not required to achieve a reduction in falls risk [15]. In a meta-analysis, telehealth-based exercise interventions in older adults (combination of strength and balance, low-to-moderate intensity, median three sessions/week for 4-months) to improve physical function appeared to be well-tolerated with reasonable adherence but improvements in mobility and balance were modest with insufficient data on falls prevention [20]. Telehealth-based exercise interventions require further investigation and may have a crucial role in optimising engagement with falls prevention strategies in postmenopausal women in the workforce or frail individuals.

Effects of Exercise on Bone Mineral Density in Postmenopausal WomenHigh-intensity Resistance and Impact Training (HiRIT) is the Most Efficacious Exercise for Improving Bone Mass

According to the principles of osteogenic loading derived from animal studies, applying higher strains at a rapid frequency in a weightbearing position provides the most robust stimulus for load-induced adaptations in bone strength [21]. Recently, there has been emerging interest in harnessing these concepts in clinical care by testing the efficacy and safety of higher-intensity exercise protocols in postmenopausal women with osteopenia/osteoporosis [22,23,24,25]. Such programs were traditionally under-studied due to concerns of tolerability and precipitating fractures in this high-risk population. In a meta-analysis, Kistler-Fischbacher et al. identified only a small proportion of exercise trials (4/63; < 10%) that incorporated high intensity protocols in studies in postmenopausal women [8]. High intensity was defined as loads greater than 80% 1RM (one repetition maximum) at frequency of < 8 repetitions for resistance training and ground reaction forces > 4 × bodyweight for impact training. High-intensity resistance interventions, particularly when combined with impact training, were shown to be most effective for increasing lumbar spine (LS) BMD, although only one-quarter of studies were in women with osteopenia/osteoporosis. Other meta-analyses in this field have consistently shown that LS BMD gains (~ 2–3%) can be achieved with high-intensity exercise protocols, however evidence is often of low-certainty due to considerable heterogeneity [26, 27]. This limitation in collating existing data may be partly explained by heterogeneity amongst moderate-intensity exercise protocols and variation between analyses in how ‘high-intensity’ is defined [8]. Certainly, LS BMD gains were not seen in a meta-analysis of moderate-to-low intensity progressive resistance exercise trials in mostly postmenopausal women [28]. Unsurprisingly, low-impact exercises (walking), and non-weightbearing exercises (cycling, swimming) do not exert a positive effect on BMD [29, 30].

Emerging Clinical Trial Evidence Supports Greater Osteogenic Effect at the Lumbar Spine with High-intensity Training

Exercise frequency of at least two sessions per week have been shown to positively influence LS BMD response, particularly if the duration of the program exceeds 12-months [31]. However, hip BMD has consistently been shown to be stable, at best, in response to even high-intensity resistance training. In an RCT in predominantly older women (n = 162, mean age 67-years) who were osteopenic or at high falls risk (Osteocise), a 12-month multimodal exercise program incorporating high-velocity progressive resistance training, moderate-impact weightbearing exercise and highly challenging balance exercises resulted in modest net BMD gains of 1.0–1.1% at the LS and femoral neck (FN), with no change at the total hip (TH) [22]. A greater net bone density response at the LS (+ 2–4%) and maintenance effect compared to low-intensity exercise at FN BMD (+ 0.3% vs −2.0%, net difference 2.3%, p = 0.025) were demonstrated with an 8-month high intensity resistance and impact training (HiRIT) program studied in postmenopausal women with mild-moderately low bone mass in two separate RCTs (LIFTMOR, MEDEX-OP, n = 106 total randomised to exercise intervention) [23, 24]. This HiRIT program incorporated resistance exercises conducted in weightbearing positions and targeting muscles in the mid-lower back including deadlifts, back squats and overhead presses, and high-impact jump exercises (landing after dropping from a chin-up bar) [32]. HiRIT was well-tolerated with average adherence > 80% [16, 17]. Maximal weight lifted during deadlift was significantly positively associated with LS BMD response (MEDEX-OP), supporting a dose–response relationship between skeletal load magnitude and BMD gains [17]. A 13-month trial in early postmenopausal women with osteoporosis/osteopenia similarly found a significant BMD effect at the LS but not the TH with a high-impact, high-intensity resistance program [25].

In LIFTMOR and MEDEX-OP, HiRIT was associated with structural changes in the proximal femur using 3D-modelling software to conduct post-hoc analyses of hip DXA scans [33], including improvements in FN cortical thickness, and TH trabecular volumetric BMD [23, 34]. Hence, the proximal femur may adapt to loading by altering its morphology (rather than bone density) and so the complete picture of benefits of skeletal loading at the hip (structure, geometry) may not be fully captured by DXA (areal BMD) [23, 34, 35]. This may be of substantial clinical relevance as cortical bone parameters such as thickness are major contributors to femoral neck strength and fracture-resistance [36, 37]. However values representing least significant change are not reported for these 3D hip DXA parameters so conclusions from this data should be extrapolated with caution. In Osteocise, there was no benefit of multimodal exercise on distal femoral or proximal tibial trabecular bone microarchitecture (TbM) [38], whilst a meta-analysis indicated limited evidence to assess effects of exercise on TbM at the distal tibia and radius in postmenopausal women potentially due to insufficient power [39].

Uncertainties Around Maintenance of Bone Density Gains with Sustained Exercise and Loss of Bone Density Gains During De-training

Data is scarce investigating persistence of osteogenic effects over sustained long-term participation in an exercise intervention. In a 6-month real-world extension of a 12-month RCT (Osteocise), there was persistence of net femoral neck BMD gains (~ 2%) despite a slight decline in mean exercise adherence to < 50%, but lumbar spine BMD effect was lost [32]. It is uncertain whether progressive BMD gains can be achieved with continuous incremental skeletal loading, or at what timepoint a plateau in osteogenic response may occur. The previous study suggests a lower maintenance frequency of exercise may be able to maintain the initial benefits achieved with skeletal mechano-adaptation [38]. However, a minimum threshold of exercise frequency and/or intensity to maintain BMD gains is not established and would be of great relevance particularly in older populations who may find it challenging and/or cost-prohibitive to maintain an intensive supervised training program. Prior studies exploring this in postmenopausal women have been conflicting, limited by imbalance in long-term attrition rate, lack of randomisation to exercise groups and poor exercise adherence [40,41,42]. An abstract explored whether BMD gains after 8-months of HiRIT were maintained during a 3-year extension of the LIFTMOR trial which included ~ 50% of initial participants [43]. Those who continued HiRIT after the trial exhibited significant ongoing improvement in BMD at the LS (8.63 ± 5.29% vs 2.18 ± 5.65%, p = 0.042) and FN (3.67 ± 4.45% vs 2.85 ± 5.79%, p = 0.014) compared to those who did not. Another unanswered question is whether, and how rapidly, interruption or cessation of training results in loss of BMD gains. Few small studies in postmenopausal women showed BMD can return to baseline levels or approach that of controls after 12-months of detraining, although earlier follow-up has not been investigated [44,45,46].

Effects of Exercise on Risk of Fragility Fractures in Postmenopausal WomenLimitations in Existing Data Supporting Fracture Risk Reduction with Exercise

It is plausible that exercise may also reduce fragility fracture risk, particularly when combining i) balance/functional exercises to address falls prevention with ii) resistance/impact training to promote bone strength. However, there is no definitive evidence that any specific exercise program can reduce fracture risk, as no RCT has been sufficiently powered to demonstrate this outcome. This is likely due to difficulty testing an exercise program consistently across a sufficiently large population over adequate duration to show a difference in fracture rates. Meta-analyses of trials in community-dwelling older adults (predominantly women) have indicated exercise interventions can reduce falls-related fractures and major osteoporotic fractures by ~ 20–30% compared with control groups [47,48,49,50,51]. These trials mostly focussed on balance and functional exercises alone or in combination with resistance exercise. However, evidence-based recommendations for any specific type or characteristics of an exercise program for optimal fracture risk reduction are limited as subgroup analyses have not demonstrated significant fracture prevention with any single mode of training [47, 48]. Supervised exercise may exert a greater effect on fracture risk reduction, however longer duration > 12-months and progressive intensity have not been shown to influence fracture outcomes [48,