This study showed that one out of three pathologic fractures in the Norwegian Hip Fracture Register had been incorrectly reported as pathologic. The underlying cancer diagnosis was incorrectly reported or missing in almost half of the cases.
After validation, 0.8% of fractures registered in the NHFR were pathologic fractures secondary to a malignant disease. Pathologic fractures secondary to a benign lesion were less common (0.1%). Prostate cancer (30%), breast cancer (20%), and lung cancer (17%) were the most common malignant diagnoses in our material. In a systematic literature review including 40 studies with a total of 3211 metastatic lesions in the complete femur, breast (35%), lung (15%), and prostate (10%) were the most common sites of primary tumour . An observational study from the Swedish Fracture Register reported metastasis from prostate cancer most common (23%), followed by unknown (23%) and breast cancer (18%) in both operated and unoperated fractures of the lower extremity . In an analysis of a large nationwide database from Japan, the most common primary sites of tumour were lung (19.2%), breast (16.6%), and prostate (10.3%) . A large number of small patient series have shown a great variation in these numbers, but most of them report breast, prostate, and lung to be the most common primary tumours in patients with pathologic fractures [12,13,14,15,16,17,18,19,20,21]. In our material, in contrast with previously published studies, prostate cancer was the most common primary tumour. This difference may partly be a result of our complete population material, in contrast with other studies reporting on smaller selections of patients. In addition, differences in cancer incidences between different countries and parts of the world probably play an important role. The Norwegian cancer registry reports higher incidence for prostate cancer (178.6/100,000) than breast cancer (138.3/100,000) and lung cancer (119.5/100000), and higher mortality rate for prostate than breast cancer . In contrast, the International Agency for Research on Cancer has reported a higher average rate for breast cancer (55.9/100,000) than for prostate cancer (37.5/100,000) and lung cancer (28.6/100,000) . Rates from Japan show higher incidence of lung cancer and lower for prostate and breast cancer .
During validation of data in the NHFR, we observed a large number of patients who had been falsely reported to have a pathologic hip fracture. Other variables as date, side, and implant type had been reported and transferred from the form correctly.
As up to now, the surgeons have reported hip fractures to the NHFR by filling in a paper form, with no mandatory information on diagnosis or primary site of tumour. There has not been a uniform or clear recommended way of how to report pathologic fractures leading to some register data of low value. It is clearly difficult for the orthopaedic surgeon to decide if a fracture is for sure pathologic or not in the acute setting at time of surgery. The surgeon can suspect that a fracture may be pathologic secondary to a malignant disease based on examination of preoperative X-rays or CT scans or based on findings during surgery. In some cases, the patient has a known cancer diagnosis. The exact diagnose of disseminated disease can sometimes, however, not be made until results from intra-operative biopsies are available, often several weeks after surgery. In other cases, a pathologic fracture can be the first sign of a malignant disease. These patients have often been investigated by CT scan prior to surgery in order to identify the primary malignant diagnose, but the exact diagnoses have not always been identified at time of surgery. As a consequence, there is normally some uncertainty as to whether the underlying diagnose of a pathologic fracture is correct and also difficult to decide the correct cancer diagnosis to report when the operation form to the NHFR is filled in. The high number of incorrectly diagnoses found in this study is, therefore, not surprising but shows the importance of validating data before further studies on pathologic fractures are performed using data from the NHFR. On the other hand, there are an uncertain number of pathologic fractures registered as normal, which we have not been able to detect/validate due to the large number of patients (n = 119,753).
The incidence of pathologic fractures secondary to cancer is uncertain, and large validated studies are missing. A retrospective cohort study from the Registry for Geriatric Trauma of the German Trauma Society (Deutsche Gesellschaft für Unfallchirurgie (DGU)) (ATR-DGU) 2016–2020 showed that 211 of 29,541 patients (> 70 years old) suffered from pathologic hip fractures corresponding to a rate of 0.7%. The register does not contain information regarding diagnosis, and the number can be to low due to age > 70 . A retrospective review of the American College of Surgeons—National Surgical Quality Improvement Program (ACS-NSQIP) database from 2011 to 2017 showed a total of 67,548 patients of which 378 (0.6%) patients had a pathologic fracture, but does not contain information regarding the histologic diagnosis . The results from both the German and the US studies correspond well to the portion of pathologic fractures secondary to a malignant disease found in our study.
Prevention of insufficient data quality through clear definitions, standard guidelines for collection, and adequate training and motivation of personnel is of great importance . Since 2021, the NHFR has started to implement electronically registration of hip fractures where the surgeon can tick off various benign and malign lesions. This will hopefully reduce the number of forms with missing information on any underlying malignant diagnosis. However, the uncertainty on the diagnose at time of surgery remains. In future, as a consequence of this study, the NHFR is planning for a follow-up form, which should be sent to the surgeon 4–6 weeks after surgery in cases where a pathologic fracture has been reported, to validate the diagnose after further investigation and pathology reports.
The strength of this register-based study is the large number of included patients. We included all pathologic hip fractures from all (43) hospitals operating these fractures in Norway. All diagnoses were thoroughly validated by one single orthopaedic surgeon with highly specialized competence in orthopaedic oncology. Both patient journals, pathology reports, advanced imaging as CT scans, scintigraphy, and PET CTs were reviewed making the diagnosis as accurate as possible.
A large weakness in our study is the possibility that pathologic fractures may have been wrongly reported as a non-pathologic fractures. These fractures could, unfortunately, not be detected and validated as it would have been far too time-consuming to validate all the 119,753 fractures reported as non-pathologic. Accordingly, some pathologic fractures are for sure missing, making the true incidence of pathologic fractures higher than 0.8%.