Osteoporotic fractures and subsequent fractures: imminent fracture risk from an analysis of German real-world claims data

Osteoporosis is a progressive, systemic, skeletal disorder characterized by low bone mass, an increase in bone fragility, and susceptibility to fracture. In 2010, 22 million women and 5.5 million men were affected by osteoporosis in the European Union (EU) [1]. In Germany specifically, the estimated prevalence of osteoporosis (based on ICD-10-GM [International Classification of Diseases, 10th revision, German Modification] code M80.*/M81.*) was 4.4% in 2016, affecting approximately 3.63 million patients, 83% of whom were female [2, 3].

Osteoporosis patients are vulnerable to fragility fractures, defined as fractures that result from mechanical forces that would not normally cause a fracture, such as a fall from a standing height [4]. Fragility fractures most often affect the vertebrae of the spine, the hip/femur, and the distal radius near the wrist [4]. It is estimated that ~ 334,000 osteoporosis patients in Germany suffered new vertebral or hip/femur fractures in 2016 [3]. Fragility fractures are associated with chronic pain, disability, reduced quality of life and increased mortality [1, 5‐9]. In Germany, they accounted for approximately €11 billion in healthcare-related costs and a loss of over 300,000 quality-adjusted life years in 2017 [10]. As the average age of Germany’s population increases, the incidence of fragility fractures is predicted to rise 18.5% between 2017 and 2030, while the associated annual costs are forecasted to increase by 23.2% [10]. Over the same period, the related quality-adjusted life year losses are expected to rise by 22.4% [10].

Patients who sustain an initial osteoporotic fracture are at increased risk of subsequent fractures [10‐13]. In women aged 50–90 years, the risk of sustaining a fracture within a year of an initial fracture is roughly five times greater than in individuals with no prior fracture [14]. After the first year, fracture risk declines, but does not return to pre-fracture levels, remaining higher than in the general population for at least a decade [14‐16]. Overall, 10–18% of individuals suffer a subsequent fracture within the 1–2 years following a prior fracture, which is known as the period of imminent risk [11, 14, 17‐19]. Imminent fracture risk is particularly high in older patients and those with index vertebral fractures [20]. Despite this, many patients in Germany fail to receive osteoporosis treatment in the immediate aftermath of a fracture [10]. By minimizing this treatment gap, particularly in the subpopulations at greatest risk, the social and economic burden of osteoporosis could be reduced [20].

This study aimed to: (a) assess the imminent risk of subsequent fracture in osteoporosis patients in Germany that experience an initial (“index”) fracture; (b) characterize the incidence of different fracture types; and (c) identify clinical and demographic variables that are independently associated with subsequent fracture risk. We hypothesized that risk of fracture would be associated with the incidence of a recent fracture, amongst other factors (e.g. age, type of index fracture) that have been identified as increasing the risk of subsequent fracture in other populations [20].

This is one of the first studies to characterize subsequent fracture risk following an index fracture in individuals with osteoporosis in Germany. Of the 18,354 osteoporotic patients in this study that had a coded index fracture between 2010 and 2014, around one in six went on to sustain a subsequent fracture during the 1-year follow-up period.

The concept of imminent risk following an initial osteoporotic fracture is well established, based on analyses of patient data from North America, Europe, East Asia and Australasia [13, 17, 20, 26, 27]. Our results indicate that imminent risk is similarly present in the German population. Here, 15.9% of osteoporosis patients with a history of at least one fracture suffered a subsequent fracture during the 1-year follow-up period, which is substantially higher than fracture rates reported in samples of osteoporotic patients not limited to individuals with a prior fracture. For instance, an earlier German study found that 10.7% of male osteoporosis patients aged ≥ 60 years and 9.5% of female osteoporosis patients aged ≥ 55 years sustained new vertebral or hip/femur fractures in 2016 [2]. Our results also suggest that the imminent risk of subsequent fracture is somewhat higher in Germany than in other countries, although this may reflect differences in the patient demographics and/or fracture coding practices used in these studies [11, 14, 19, 20, 28].

The annual incidence of osteoporosis-associated femoral fractures in females aged 75–79 years in Germany is approximately 0.6% [29]. In this study, the 1-year cumulative incidence of subsequent hip/femur fracture was between 2.0 and 2.8% for female osteoporosis patients with an average age of 77. Hence, index fracture is associated with a considerable increase in hip/femur fracture risk.

Subsequent fractures result in poor health, social, and economic outcomes; a subsequent hip fracture, for example, is associated with decreased mobility and social independence, as well as increased mortality [28, 30‐32]. The heightened fracture risk within the 1–2 years following an index fracture highlights the need for immediate treatment after an index fracture. In Germany, however, around 60% of women aged ≥ 50 years remain untreated during the first year after an osteoporotic fracture [10], highlighting a missed opportunity to treat patients at imminent risk of fracture.

The incidence of subsequent fractures was greatest in older patients, echoing earlier findings that each year of life increases subsequent MOF risk [17]. Patients sustaining an index vertebral fracture were at elevated risk of subsequent fractures compared to patients with other index fracture types. This finding is consistent with evidence suggesting that vertebral fractures often precede additional fractures as part of a “fracture cascade” [16].

The number of men and women were imbalanced as typically found in osteoporosis populations. As the multivariate analysis did not comprise sex as an independent risk factor, we believe that excluding men would not have a significant impact on the overall results.

Index fracture type, age and sex were all identified as risk factors for subsequent fracture in our univariate analyses. However, the female patient group was on average older than the male group (77 vs. 73 years, respectively), and sex was not significantly associated with subsequent fracture risk when considered as part of our final multivariate model. In contrast, index fracture type, age, epilepsy/use of antiepileptics and heart failure were all found to be independently associated with subsequent fracture risk. These risk factors are already featured in the DVO guidelines as variables that are predictive for osteoporotic fractures [25], and our latest results confirm their impact on subsequent fracture risk [33]. Therefore, it is likely that epilepsy/use of antiepileptics contribute to an imminent risk of fracture by an increased risk of falls. In contrast, fall-inducing medications were not identified as an independent risk factor for imminent risk of fracture in our multivariate analysis (Table 3).

Treatment of such as use of Bisphosphonates were not associated with lower risk for a subsequent fracture. In our database, we did not have access to the exact date of treatment initiation for all of the patients. Based on previous reports in Germany, we assume that, only a minority of patients receive a treatment directly after an osteoporosis-related fracture, [34] which would explain the small impact of treatment on subsequent fracture risk. In addition, the small impact of osteoporosis treatments on subsequent fracture risk could also be explained by the documented low persistence to treatment. As previously shown, in several German-based studies, persistency with oral bisphosphonates, the first-line treatment in Germany, were reported to be as low as 20% after 12 months of follow-up [35‐37].

Previous evidence suggests that numerous other variables are associated with osteoporotic fractures, including RA, type I diabetes, chronic obstructive pulmonary disease, Cushing’s syndrome, androgen therapy, glucocorticoids, antidepressants and aromatase inhibitors [25], yet none were identified as independent risk factors for subsequent fracture in the present study. Glucocorticoid-induced osteoporosis is known to increase patients’ fracture risk [25, 28, 38‐40], but the association between glucocorticoid use and subsequent fracture risk did not approach significance (p = 0.728) in our univariate analysis. In addition, RA confers heightened fracture risk independently of glucocorticoid use [25], yet RA was not significantly associated with subsequent fracture risk in our multivariate model. It is conceivable that the key risk factors for subsequent fractures differ from the established list of osteoporotic fracture risk factors, and/or that the importance of specific risk factors is dependent on timing within the baseline period. However, glucocorticoid use, fall-inducing medications, rheumatic disease and CCI were all identified as independent risk factors in a recent Swedish study of subsequent fracture risk [20]. Again, different fracture coding practices could account for these findings.

This study had several limitations. First, data for several established risk factors were not available for inclusion in our regression analyses, including bone mineral density, body mass index, smoking, fall history, total number of previous fractures and parental history of hip fracture. Second, while imminent risk was assessed in a large cohort of osteoporotic patients sampled from the German general population, the increase in fracture risk associated with a prior fracture could only be estimated through comparison with incidence rates from the published literature. Third, previous research has demonstrated that subsequent fracture risk is highest within the 1–2 years following an initial osteoporotic fracture [14, 17, 18], but in the present study, subsequent fracture risk was only assessed over a 1-year follow-up period. Consequently, subsequent fractures that occurred after this point were not captured in our dataset, limiting the power and scope of the study and its conclusions. Furthermore, the average time interval between index and subsequent fractures was limited by the 1-year duration of the follow-up period; as such, it is not anticipated to reflect the average fracture interval in the wider osteoporotic population. In addition, while distal forearm/wrist fracture as well as fractures of the humerus are typical osteoporotic fracture and also contribute to the count as major osteoporotic fractures, all patients also had an OP diagnosis or a prescription for OP medication. These, for OP patients, rare fractures of shoulder and were included for reasons of completeness. However, since the absolute and relative number of hand and shoulder is very low, we are confident that the results are not affected by this broader definition.

In our analyses, patients were categorized according to three- rather than four-character ICD-10-GM codes. While three-character codes have been used in other real-world analyses of osteoporotic fracture frequency in Germany [34, 41], four-character ICD-10-GM codes offer greater specificity; in particular, they allow the differentiation of clinical and non-clinical fractures. It is, therefore, possible that some non-clinical vertebral fractures were captured here. Consistent with this suggestion, the ratio of vertebral-to-hip/femur index fractures in our dataset was considerably higher than documented elsewhere [42], although other studies have reported a substantially greater number of vertebral than hip/femur index fractures [43‐45]. The use of three-character ICD-10-GM codes may also have resulted in the inclusion of some non-osteoporotic fractures, although all patients were diagnosed with osteoporosis and/or prescribed osteoporosis medication prior to index fracture, so it is likely that most were osteoporosis-related. Lastly, the algorithms we used to identify incident fractures are not yet validated; some subsequent fractures may actually have been existing fractures recorded at follow-up visits, resulting in an overestimation of subsequent fracture incidence.

Osteoporosis patients in Germany sustaining an initial fracture are at imminent risk of subsequent fracture, with approximately one in six patients with a coded subsequent fracture during the 1-year follow-up period. Immediate treatment following an index fracture could prevent subsequent fractures, reducing the associated negative health, social, and economic consequences. Clinicians should consider the recency of patients’ previous fracture(s) when making treatment decisions, paying particular attention to older adults and patients with comorbidities that may put them at increased risk of subsequent fracture.