Anemia and bone loss have been shown to be linked to each other. Both diseases have a global impact on public health burden, with a higher prevalence among females and an increase with age. We conducted a cohort study using the IQVIA Medical Research Database (IMRD, 2010–2023) in the UK with the aim to evaluate the risk of fracture among adult patients with anemia, compared to propensity score matched controls, stratifying by fracture site, age, and severity of anemia. Among 84,932 patients with anemia and 158,731 controls, we observed that anemia was associated with a significant 20% risk of any fracture (males with a 32% increased risk, females with increased risk only for hip and vertebral fractures). In patients ≥ 50 years, all fracture sites showed a significant association with anemia, while in patients < 50 years old, accounting for ~ 10% of fractures, only hip fracture was associated with anemia. Severe anemia defined as hemoglobin levels < 8 g/dl was associated with a higher risk of vertebral and humerus fractures. These data indicate that there is a close link between anemia and bone homeostasis with high clinical relevance and a significant public health burden that might enhance the efforts in the clinics to diagnose, prevent, and treat anemia and osteoporosis.

Our results are broadly in line with the results of previous studies. The 20% increased risk of fractures in patients presenting with anemia, compared to their respective matched controls has also been shown within this range by other colleagues [6, 7, 10, 11, 13, 16]. For example, Lee and colleagues found a 10–29% elevated risk for fractures in men and women aged ≥ 65 years in all typical fragility fracture sites [10]. In 2021, Kim et al. made similar findings, observing an 11–37% risk increase for anemic men and women aged ≥ 50 years, with men having a greater risk than women [12]. In the MrOS study in 2017, anemic men aged > 65 years presented with an increased fracture risk [17].

We further identified sex-specific differences that are in line with previous literature. In our analysis, the risk of fracture was higher among males than females at all fracture sites, with risks ranging between 27 and 89% for males, while among females the risk was largely non-significant, with the exception of hip and vertebral. Previous studies have also found higher associations between anemia and fractures in males [9, 10, 12]. Several factors can be viewed in connection to this finding. First, sex hormones, specifically low testosterone levels in males, have been shown to coincide with anemia and low BMD [24,25,26]. In addition, low testosterone levels in males have shown to lead to decreased muscle strength, less mobility, and increased risk for falling [27]. Unfortunately, we were unable to assess testosterone levels in our database to identify if this is a mediating factor for fracture risk among males with anemia. We also note the possibility of an inherent diagnosis bias should not be excluded. Anemia is commonly associated with, and therefore evaluated for more frequently, in women than men when presenting symptoms of anemia. This may lead to a longer time window between onset and diagnosis of the disease in men.

To the best of our knowledge, this is the first large population-based cohort to evaluate the association between anemia and fracture risk among younger patients (i.e., < 50 years). A small study including male and female military soldiers in Israel identified an association between the presence of anemia at recruitment (mean age 19.9 years) and subsequent development of stress fractures in the ankle among female soldiers in combat training [18]. Thus, while our results of an elevated fracture risk among patients aged ≥ 50 are consistent with prior literature [11], the inclusion of those < 50 is novel. However, there were fewer fractures among those aged < 50 (roughly 1/10th of the those observed in patients ≥ 50) limiting statistical power. Among patients < 50 years, anemia was only associated with a significantly elevated risk of hip fracture; however, these findings should be interpreted cautiously until future work can confirm these findings, either through larger cohort studies or a meta-analyses to increase statistical power and precision.

Additionally, we further stratified by anemia type, based on diagnosis codes. While we acknowledge the limitation of the diagnosis codes and the potential for misclassification, the results do provide interesting insights. First, we again found that the risk was highest among male patients and those ≥ 50 years in all strata. Additionally, we identified that patients with vitamin B12/folate anemia had the highest risk of experiencing any fracture compared to iron deficiency and general anemia, which was consistent among both sexes and age groups. Vitamin B12 and folate anemia are also considered as side effects of severe alcohol consumption, malnutrition, or malabsorption. While we include alcohol use in the propensity score, the impact of prolonged alcohol use or alcoholism is not well captured and we cannot exclude this as an important confounding factor.

The underlying mechanisms responsible for the association of anemia and fractures are yet to be fully understood. From the clinical point of view, trauma leads to fracture in many patients. However, those who do not experience a fracture to a—maybe similar—degree of trauma do not enter the hospitals. A study with certain strengths of stresses reflecting trauma on bones related to the BMD is not feasible to be performed in patients. Overall, we did find an elevated risk of hip fractures in younger patients with anemia. Given that we would expect the likelihood of having a major trauma leading to a hip fracture to be non-differential between our groups and we matched closely on age, the results suggest that anemia is an independent contributor to fracture risk. For example, if two individuals had similar traumatic accidents, the individual with anemia would have a higher likelihood of experiencing a hip fracture. Of course, this finding requires further confirmation in other studies. However, given the impact of hip fractures on a patient’s quality of life, this is an interesting finding and suggests we should further investigate how anemia impacts bone health in younger adults. In younger adults, there was no difference except at the hip. This could be a chance finding, but it could also be the first sign for a decrease in BMD and anemia. To our knowledge, there has been no study on fractures in younger adults relating BMD, fracture cause (such as falling and dizziness), and fracture to anemia. As stated above, there is also no study on certain stresses that lead to fracture in one bone but not in the other, as such a study is not feasible. We are aware of a study on fracture healing that recruited 94 patients with normal fracture healing and 88 patients with delayed fracture healing of the femoral neck. The authors found that serum miR-656-3p modulates osteoblast function by targeting BMP-2, offering novel therapeutic and diagnostic targets for the management of delayed fracture healing [19].

Regarding the mechanisms, the potential key mechanisms linking anemia and bone homeostasis are the bone morphogenetic signaling pathway (BMP) and Wnt signaling. Inhibition of bone morphogenetic protein signaling (BMP) [20] resulted in reduced osteoblast activation and increased hemoglobin levels. This was the first link we saw between iron homeostasis and bone generation and impairment of osteoblasts. In a murine model, inhibition of the BMP signaling pathway led to increased hemoglobin levels as well, and was also able to treat the anemia of inflammation [21, 22]. In 2019, we identified the transferrin receptor 2 (Tfr2), that is mainly expressed in the liver and controls iron homeostasis as a regulator of bone homeostasis that inhibits bone formation [23].

With respect to the cellular level, osteoclasts, responsible to remove damaged bone areas, and osteoblasts, responsible to build new bone, are the two key players in this turnover activity. Iron deficiency disturbs the process and equilibrium between osteoclast and osteoblast action, as the differentiation of osteoblasts is inhibited by impairment of collagen production and disturbance of vitamin D metabolism [24,25,26]. This requirement of iron for bone turnover and sufficient osteoblast differentiation via vitamin D metabolism and collagen production plays a role in retaining adequate BMD [24,25,26,27]. In addition, iron and bone share the bone morphogenetic pathway (BMP), which also regulates osteoblast differentiation [28]. The association of impaired bone formation via the BMP signalling pathway upon iron deficiency has been examined in animal models, such as zebra fish, and continues to be a subject of interest in the investigation of the association between iron and bone [29]. In addition, iron deficiency induces low bone turnover due to energy and cofactor deficiency, both of which require iron [30]. Furthermore, erythropoietic activity disturbed by low iron levels can also affect the microarchitecture of bone in its dynamic state [31, 32]. Thus, BMD can be negatively affected and potentially lead to an increased risk for fractures. Moreover, bone matrix properties not associated with BMD appear to be involved in the increased fracture risk for anemic men, aged ≥ 65 years, as suggested by the 2017 MrOS study [16]. Other BMD-independent mechanisms associated with anemia, such as decreased muscle strength, low muscle density and even falling due to dizziness could potentially increase the risk for fractures as well.

In addition to the direct association between iron and bone homeostasis, erythropoiesis also affects bone metabolism, mainly through erythropoietin (EPO). Specifically, the EPO-fibroblast growth factor 23 (FGF-23) axis has shown to play an important role in bone mineralization [31, 32]. The 2017 MrOs study showed a 1.57 to 1.72-fold increased hip fracture risk for multi-ethnic anemic men aged ≥ 65 years, independent of BMD, suggesting an association of other bone matrix properties with anemia [13]. This, in combination with the indirect association of iron and bone homeostasis via erythropoiesis, implies a greater and more complex interdependency of bone health and iron levels. In summary, disturbed iron homeostasis, as seen in anemic patients, not only shows an increased fracture risk in the context of observational studies but can also be connected on a biological basis.

Our study has several strengths and weaknesses. The large sample size and the long follow-up time are considered a major strength. Using the IQVIA Medical Research Database to gain access to patients’ records, allowed the inclusion of many highly relevant covariates and the subsequent statistically powerful analysis. The availability of laboratory values enabled us to identify patients based on both diagnosis codes and confirmed laboratory values, thereby increasing the validity of our identified cases [15]. Access to laboratory values for hemoglobin also provided the opportunity to stratify our cohort by anemia severity. However, the number of fractures was low when stratified by fracture site, particularly for the group of severe anemia, making interpretation challenging. Therefore, these findings need to be interpreted with caution. However, we have included the overall fracture numbers and risk estimates so they can be included in future meta-analyses. An additional strength is the inclusion of patients < 50-years and the stratification by the type of anemia, which was possible due to the large sample size. Nevertheless, the stratification by anemia type was limited to include only those patients with a diagnostic code. This restriction could potentially facilitate the inclusion of undiagnosed anemic patients into the control groups. However, in this scenario, we can assume that the observed elevated risk of 20% in anemic patients for fracture is a conservative estimate.

We further acknowledge the potential for residual and unmeasured confounding that is inherent to observational studies. While we did not include high-trauma fracture codes, both closed and open fracture codes were included in our outcome definition to avoid underestimating the burden of fracture, as falls from standing height or less (low-energy trauma) can lead to open compound fractures, particularly when a patient has osteoporosis [33]. Moreover, we acknowledge the work by Leslie and colleagues that found patients with low BMD had similar risks for both high- and low-trauma fractures, ultimately supporting the inclusion of all fractures when evaluating osteoporosis risks [34]. Thus, we are confident that if high-trauma fractures were misclassified and included in our analysis; we have not biased our estimates.

Nevertheless, we were unable to differentiate fractures occurring primarily due to (a) direct effects of anemia on bone density and quality, (b) indirect mechanisms such as anemia-related muscle weakness, sarcopenia, or dizziness leading to falls, or (c) a combination of both.

Similarly, we were unable to assess changes in BMD due to anemia or fall risk as we did not have information on BMD, bone marker, or fall-related information in our database. Instead, we used the rare primary outcome “fracture,” thereby limiting the power. Thus, it has to explicitly be acknowledged that the observed association may reflect a combination of skeletal fragility and increased fall susceptibility and appropriately powered longitudinal prospective studies are warranted to dissect the impact of bone fragility vs. the biomechanical impact of the fall.

Finally, it appears that in men, the “advantage” of being intrinsically protected against fragility fractures by their higher bone mass and strength, is counteracted by lifestyle (higher consumption of alcohol and smoking) and behavioral factors (risk-taking activities). However, we also believe that while men have the “advantage” of being intrinsically protected against fragility fractures by their higher bone mass and strength, they may be disadvantaged by anemia being perceived as a female condition. Thus, they may be less likely to be screened and detected earlier. As a result, it is possible that males had a longer disease duration and subsequently had a higher fracture risk. While we used a time-varying analysis, we did not account for disease duration as knowing the start of anemia is inherently challenging in real-world data. Additional work investigating, if disease duration could mediate the risk between men and women would be of interest to determine if there is a biological/mechanistic reason for the sex differences, or if earlier detection in men could minimize fracture risk. Again, longitudinal studies assessing lifestyle factors in more detail would be necessary to conclusively address these questions.