This study contributes to the awareness of musculoskeletal health and the impact on the aging person. In a community-dwelling, population-based cohort of older women, we show that probable osteosarcopenia clearly confers a higher overall risk of fracture and mortality more so even than low bone mass alone.

In this cohort, at age 75 the proportion of women with confirmed osteosarcopenia is three percent, which is comparable to a few other population-based studies of northern European women [10, 11]. Establishing an accurate estimate in community-dwelling individuals is difficult, since osteosarcopenia is related to age and sex and on the definitions used [8, 9, 15].

Two studies, using the same definition of sarcopenia as us, one Danish cross-sectional (n = 529, mean age 75; range 65–93) and one British longitudinal study (n = 405, median age 76) reported prevalence at 1.5% and 2%, respectively [10, 11]. This is most likely explained by the inclusion of both sexes, as the prevalence of osteosarcopenia has been reported higher among older women than men [2, 33].

Another longitudinal study (n = 1114 women, mean age 77.6, different ethnicities) reported prevalence of osteosarcopenia to be 12% [18]. The more than double prevalence compared to our study can, at least partly, be explained by the use of the original (EWGSOP1) criteria of sarcopenia, which have been shown to give a higher prevalence of osteosarcopenia compared to EWGSOP2 [34]. This emphasizes the effect of how osteosarcopenia is defined; in an Australian study (n = 1032), despite a lower mean age (62 years) the reported prevalence was 8.3–9.7% [13], when including low muscle mass or low muscle strength alone.

Prevalence of confirmed osteosarcopenia increases with age, in our cohort from 3% at 75 to ~ 9% at 85y (although these are the most conservative assessment – when we also consider those with probable osteosarcopenia, an estimated two to four times more women may be affected).

The higher prevalence from using the EWGSOP1 sarcopenia criteria was also obvious in a longitudinal study by Salech et al. with prevalence starting at 8.9% (age 60–69.9), rising to 18.3% (age 70–79.9), and 33.7% (> 80 years) [9]. However, the lower prevalence in our study may, at least partially, reflect the relatively good health of the cohort. While this may introduce a ‘healthy cohort bias,’ on the other hand, due to the unique study design whereby all participants are identically aged we are likely seeing a more “true” age-related prevalence of osteosarcopenia in older, community-dwelling women.

The relationship between bone and muscle is closely associated throughout life [35]. In the OPRA cohort all women with sarcopenia had low bone mass, while conversely only a small percentage (4%) of those with low bone mass had sarcopenia, as has also been noted in other studies [14]. Furthermore, osteoporosis is more prevalent among women with sarcopenia than among those in the low bone mass [10, 14].

Women in both the probable osteosarcopenia group and in the low bone mass group had a two-time greater risk of major osteoporotic fracture compared to women in the normal group. However, most importantly women with probable osteosarcopenia had a higher risk of hip fracture even compared to those with low bone mass alone. This is reflected in the clinical characteristics of women with probable osteosarcopenia, who were frailer and also had more falls than women with low bone mass [12, 14, 18, 33], indicating the need to consider sarcopenia (not just confirmed, but also probable) as part of fracture management.

Interestingly, the low bone mass group had more incident vertebral, proximal humerus, and distal radius fractures but fewer hip fractures than the probable osteosarcopenia group. A finding that might indicate a better physical capability to react to a fall. Hip fracture occurs earlier in those with probable osteosarcopenia and there appears to be a higher imminent risk of fracture based on the survival curves. Although the difference in median time to first hip and major osteoporotic fracture between probable osteosarcopenia and low bone mass group was less than half a year, this period of “gained” independence is of great importance in old age.

Our results on osteosarcopenia and fracture aligns with other longitudinal studies, although our risk estimates are higher [9, 13, 15, 18]. One with 5,640 person-years of follow-up reported an increased risk of fracture (HR 1.54 [1.13–2.08]) [9]. Studies of different designs, mixed sexes and lower ages, report similarly increased 10- or 5-year risk of fractures with osteosarcopenia but not compared to low bone mass or sarcopenia alone [13, 15], which draws attention to the difficulties in comparing studies.

Similar to our observations, other longitudinal studies report higher mortality risk with osteosarcopenia but not low bone mass or sarcopenia alone. But again, direct comparison is difficult because of younger and mixed populations and definitions [9, 13]. The higher mortality is not surprising, since these women have lower BMI, are weaker, more physically limited, have poorer kidney function, a less healthy lifestyle, and take more medications (an expression of co-morbidity). In short, these women are frailer, and this geriatric syndrome has undoubtedly proven to be related to increased mortality [36].

Taken together the data highlights that in older women, as important as screening for and treating low bone mass is the evaluation of muscle parameters and a focus on fall prevention in order to maintain musculoskeletal integrity. Illustrating this, we have previously shown that within a group categorized as ‘low risk’ based on FRAX score, those who are frail actually have a high risk of fracture [37]. While there is no current treatment for osteosarcopenia, interventions such as resistance training and anti-osteoporosis treatment [38] could directly or indirectly prevent further deterioration of muscle mass and strength that often follows fractures, particularly hip fracture.

Strengths of the study include the longitudinal, population-based design which allows for generalization to the target population. Uniquely, the women were all the same age allowing assessment of changes in prevalence and outcomes over time; giving a perspective on “chronological” and “biological” trajectory of change in musculoskeletal aging. Age and 10-year follow-up give the opportunity to investigate osteosarcopenia during a critical period in life where a decline in health with falls, fractures, and mortality are more likely and therefore provide important information on when and how to be more attentive to this disease. Second, all individuals were Caucasian, the same gender and age, therefore reducing confounding related to sex hormones, accumulation of co-morbidities, and ethnicity. Third, the participation rate was high (65%, 75%, and 76% at respective visits) and the cohort constituted as much as 33% of all 75-year-old women living in Malmö, Sweden at the time of inclusion. Fourth, this study includes the most used criteria of both osteopenia/osteoporosis (WHO) and sarcopenia (EWGSOP2) to define osteosarcopenia. Besides, EWGSOP2 is the preferred and most cited definition in Europe and Australia [10]. Fifth, in contrast to other studies, both overall fracture risk as well as risk of hip fracture were investigated; and characterization of all types of osteoporotic fractures in relation to the three comparative groups provided important practical information on possible mechanisms leading to fractures. In other studies, fractures were mostly self-reported or lacked detailed information on collection. Only one longitudinal, population-based study provided information on fracture type.

Study limitations are also acknowledged. First, we adopted the cut-offs from the EWGSOP2 definition and, while recognizing that locally derived reference data might have been useful, the advantage lies in facilitating comparison with the existing literature. Second, it may have been preferable to use the same osteosarcopenia classification for estimating both prevalence and its associated risks. However, we used the strictest meaning (‘confirmed osteosarcopenia’) to most accurately estimate prevalence of the condition among older community-dwelling women. For the risk analyses we included both ‘confirmed’ and ‘probable’ osteosarcopenia in the definition partly due to the low number of women with confirmed sarcopenia but also because low muscle strength (sufficient to assign probable osteosarcopenia) is considered the most important influence on clinical outcome [39, 40]. We assume that the observed elevated mortality risk would also be apparent in those with more severe muscle loss, given the overlap between sarcopenia, osteosarcopenia, and frailty [2, 41]. Third, when calculating the prevalence of osteosarcopenia, two or three variables were required (T-score and muscle strength, as well as muscle mass for ‘confirmed’ osteosarcopenia), which might lead to sampling bias. Fourth, the OPRA cohort did not have sufficiently detailed information about co-morbidities to create a co-morbidity index. Hence, as a proxy we adjusted for polypharmacy, which could be considered a limitation, although it is reported to be correlated with co-morbidity. In addition, while we know that some have used glucocorticoids ‘at some point,’ detailed information on date, duration, or dose was not available. Since glucocorticoid use was overall very low (n = 29 at age 75; n = 41 at age 80; n = 22 at age 85), we have not corrected for this. Fifth, those who chose not to participate might have worse health, but this is common in most studies of older populations [42] and might explain the low number with sarcopenia in the cohort. Sixth, knee strength was used since handgrip strength was not available at baseline. Although evidence for association between handgrip and knee strength in older individuals is conflicting [43], isometric torque method of the lower limb is a validated tool for determination of overall muscle strength [16]. And although handgrip strength is commonly used to predict physical function, knee strength might be a more appropriate proxy since there is a greater age-related loss of leg strength compared to arm strength [44, 45]. This could possibly relate to a higher degree of disuse of the lower extremities with age.