Vertebral fractures increase the risk of subsequent vertebral fractures: results from a large German health insurance dataset

Introduction In the DVO guidelines on osteoporosis, the 10-year hip or vertebral fracture (VF) risk represents the key outcome that guides decisions on diagnosis and therapy. Incident VFs induce a high risk for subsequent additional VFs with hazard ratios (HR) ranging from 2-10 depending on number, severity, and VF definition (clinical vs radiological, Genant vs other grading scores). These HR data were derived in prospective cohort studies but even the largest of those studies still has insufficient sample size to calculate HR as a function of sex and age including statistical interactions. Registry based analyses, on the other hand, have sufficient sample size and health insurance registries provide representative real-world data. On the other hand, the ambiguity of the ICD codes may lead to serious overestimation of HRs for VF, because often the same fracture is coded multiple times over time, i.e. repeat codes cannot be distinguished from real new fractures.

Methods We analyzed a representative German health insurance dataset large enough that restriction to vertebral level specific ICD-10 codes (S2202, S2203, S2204, S2205, S2206, S3201, S3202, S3203, S3204, S3205) still yields a substantial sample size. We identified the first incidence of a vertebral fracture during the follow-up time period with a specific VF ICD code that differed from any VF ICD code of that patient during the baseline period. This eliminates the problem of repeat fracture HR overestimation. Fracture risk was modelled with age-adjusted Cox proportional hazard models.

Results In a 7-year dataset of 658,175 patients (mean age=66.2, inter quartile range 19) we identified 4,828 patients (3,566 women and 1,262 men) with one incident vertebral fracture during a 4-year baseline period coded with specific ICD codes. During the subsequent 3-year follow-up period 5,816 patients (4,194 women and 1,622 men) suffered a VF at a vertebral level different from the baseline VF level. In women, 1 vs 0 baseline VF increased the risk for a subsequent VF with an average HR=4.0 (95%CI 3.5-4.5). In the age range of 70-90 years the average HR of men was higher by a factor of 1.5(1.0-2.1): HR=5.6 vs 3.7 in women (p<60;0.04). In women this HR decreased from HR=15.9 at age 50 to HR=2.6 at age 90 whereas for men the HR was largely independent of age.

Discussion Our data helps build more accurate VF risk models. Using representative German data, we quantified in detail to what extent VF increases the risk for subsequent VF. In women these HRs decrease with age, and they are higher in men than in women.

Keywords Frakturrisiko, Wirbelkörperfrakturen, Osteoporose

Korrespondenzadresse Claus-C. Glüer, Christian-Albrechts-Universität zu Kiel, Klinik für Radiologie und Neuroradiologie, Sektion Biomedizinische Bildgebung, Am Botanischen Garten 14, 24118 Kiel, Deutschland, E-Mail: [e-mail]

Publication History

Article published online:
08 September 2022