Epidemiology and reporting of osteoporotic vertebral fractures in patients with long-term hospital records based on routine clinical CT imaging

The present study was approved by the local institutional review board (ethics committee’s reference number 5022/11-A2) and was conducted in accordance with the Declaration of Helsinki. The requirement for informed consent was waived by the institutional review board due to the retrospective character of imaging data collection and post hoc analysis.

Study patients were identified from the local database of the Klinikum rechts der Isar of the Technical University Munich, a 1.161 bed university hospital located in southern Germany. The hospital is annually treating more than 60,000 hospitalized patients and additionally around 250,000 outpatients. All patients aged 45 years and older were eligible that had an index CT scan of the chest, abdomen, or pelvis performed in the outpatient clinic or during admission between September 1, 2008, and May 31, 2017, and that had a prior medical record of at least 5 years in the hospital database. In case of recurrent patients or patients with multiple CT scans in the study period, the first CT scan satisfying the inclusion criteria (see below) was assessed. In very few exceptions (n = 4 of final study population), patients had CT scans performed in external institutions prior to clinic visit or hospital admission, and imaging data was imported from CDROM into the institutional database. The study population was then selected based on imaging criteria and exclusion criteria. Imaging criteria for inclusion were a CT scan with sagittal reformations showing at least vertebrae T6 to L4, because sagittal reformations are required for reliable detection of VFs [18] and fractures commonly affect these levels [19]. Patients with high-energy fractures due to trauma were excluded. Moreover, patients with multiple myeloma were excluded, because this disease shares a defining characteristic with osteoporosis and fractures due to deteriorated bone strength cannot be clearly attributed to either one of these diseases [20].

All CT scans of included patients were acquired on multi-detector CT scanners, partly after administration of either both oral (Barilux Scan; Sanochemia Diagnostics, Neuss, Germany) and intravenous (Iomeron 400; Bracco, Konstanz, Germany) contrast medium or only intravenous contrast material. Scans were performed in the hospital on a Philips Brilliance 64, Philips iCT 256, Philips Ingenuity Core 128, Philips IQON (Philips, Best, The Netherlands), Siemens Biograph 64, Siemens Biograph 128, Siemens Definition AS, Siemens Definition AS + , or Siemens Sensation Cardiac 64 scanner (Siemens Healthineers, Erlangen, Germany)—with the exception of four scans that were performed on a Siemens Emotion 16 scanner in another institution. Scans were acquired in helical mode with a slice thickness of 0.9–1 mm using adaptive tube load at a peak tube voltage of 120 kVp—with few exceptions using 100 kVp, 130 kVp, or 140 kVp.

CT scans were evaluated for prevalent fractures at each vertebral level. Only thoracolumbar vertebrae were evaluated, as fractures are rare and usually of non-osteoporotic origin at the cervical spine [19]. Fracture reading was performed in consensus by one trained medical student (MH) and one radiologist (MTL) with 5 years of experience. Prevalent VFs were classified using the semi-quantitative method by Genant [21]. VFs were graded as mild for a height loss ≥ 20% and < 25% (grade 1), as moderate for a height loss of ≥ 25% and < 40% (grade 2), and as severe for a height loss ≥ 40% (grade 3). With reference to the recent discussion on grading osteoporotic fractures [22], it shall explicitly be mentioned that non-osteoporotic deformities were not graded as osteoporotic fractures, in line with the original description by Genant [21]. Moreover, in patients with neoplastic disease and suspicion of spinal lesions, all available imaging (CT and MRI) was additionally reviewed by an attending radiologist (JSK) in order to distinguish between osteoporotic or pathologic fractures [23]. Any patient with at least one VF that could be attributed to osteoporosis was counted as a patient with ≥ 1 osteoporotic VF irrespective of the presence of additional pathologic VFs.

For all included patients, any discharge or consultation notes in the hospital database closest to the time of the CT scan were reviewed for primary diagnosis (free text and ICD-10 codes), current medications, chemotherapy, and level of immobility. In knowledge of primary diagnoses, we summarized indications for CT imaging into 5 categories: staging of solid cancer, non-solid cancer, CT angiography, assessment of spinal stenosis, and other indications. Based on prescriptions noted in medical records, we assessed intake of medications for at least 3 months during a period of 5 years prior to the index CT scan. Assessed medications included nonsteroidal anti-inflammatory drugs, metamizole, opioids, immunosuppressive drugs, glucocorticoids, vitamin D, and bisphosphonates. Due to the retrospective study and sparse data, we summarized chemotherapeutic medications with adverse effects on bone metabolism in order to get sufficient statistical power. This mainly included chemotherapeutics and other medications for breast cancer treatment (e.g. tamoxifen, aromatase inhibitors, and GnRH antagonist drugs). Moreover, bisphosphonates were the most consistently prescribed anti-osteoporotic drugs we could find in medical notes and, for instance, Denosumab was only prescribed in 3 patients. Therefore, we only calculated statistics for bisphosphonates. The level of immobility was stratified into an ordinal scale of short (postoperative immobility, fatigue), medium (recurrent postoperative immobility, intensive care stay for longer periods of time, dementia, reduced general condition, help needed with walking [Barthel-Index], Karnovsky-Index < 70%, ECOG-Index 2, NYHA III), and prolonged immobility (cachexia, paraplegia, decubitus, Parkinson’s disease, muscular dystrophy, ankylosing spondylitis, help needed with walking, grooming, and toilet use [Barthel-Index], Karnovsky-Index < 50%, COPD GOLD IV, peripheral artery disease stadium III/IV, NYHA IV, ECOG-Index 3–4). We also reviewed the available records of all patients for established diagnoses of osteoporosis or low bone mass. Furthermore, imaging reports of CT scans of fractured patients were reviewed for any mention of deteriorated bone quality or non-traumatic spinal fractures. Valid terminology for deteriorated bone quality was “osteopenia”, “osteopenic bone structure”, or “decrease bone mineral content” (German: “Osteopenie”, “osteopene Knochenstruktur”, or “Knochen(mineral)salzminderung”). The terminology used to describe the fracture related abnormalities was analyzed in more detail. We distinguished between the primary term “fracture” (German: “Fraktur”, “Deckplatteneinbruch”, or “Grundplatteneinbruch”) and secondary terms such as “deformity”, “sintering”, “impression”, “height loss”, “height reduction”, or “Schmorl’s node” (German: “Deformität”, “Sinterung”, “Impression”, “Höhenminderung”, “Keilwirbelbildung”, or “Schmorlscher Knoten”).

Patients were stratified into age groups with < 50, 50–59, 60–69, 70–79, and ≥ 80 years of age, respectively. Association of selection status (included or excluded from all eligible patients) and age groups was tested using chi-squared statistics. Descriptive statistics such as mean, standard deviation, and frequency were used to summarize prevalence of VFs, clinical characteristics, and information from CT reports. The proportions of CT indications were compared between men and women using chi-squared test. Any osteoporotic fracture of grade 1 or higher was counted as an osteoporotic VF. Absolute numbers, frequencies, and 95% confidence intervals (CI) of prevalent VFs were given. The frequency of VFs was compared between men and women using independent samples t-test. Clinical characteristics associated with prevalent VFs were identified in univariate logistic regression models. Multivariate logistic regression models were then used to identify predictors of prevalent VF, adjusting for clinical variables that were significant in univariate analysis with a two-sided p-value < 0.05. Statistical analyses were performed using SPSS (version 26; IBM) and RStudio (version 1.4.1106; RStudio). Level of statistical significance was set at p < 0.05.

There were 2057 patients aged 45 years or older seen in the outpatient clinic or admitted to the hospital during the study period, which had a CT scan of the chest, abdomen, or pelvis and a well-documented medical record in the hospital over the last 5 years. Of these, 718 patients were included that met the imaging criteria and had no diagnosis of multiple myeloma (Fig. 1). The study population had a mean age of 69.3 ± 10.1 years ranging from 45.8 to 93.4 years. The mean age did not significantly differ between 228 women (69 ± 9.4 years) and 490 men (69.4 ± 10.5 years; p = 0.58). The age distribution of 718 included patients did not significantly differ from 1339 patients that were excluded due to imaging criteria and multiple myeloma (p = 0.168; Table S1). Indications for CT were staging of solid cancer (n = 502), CT angiography (n = 147), staging of non-solid cancer (n = 53), spinal stenosis (n = 11), or other reasons (n = 5; Table 1). The proportion of women who had CT for staging of solid cancer was significantly higher and for CT angiography was significantly lower compared to men, respectively (both p < 0.001; Table S2).

A total of 12.167 thoracolumbar vertebrae were assessed for fractures with most scans covering the entire thoracolumbar spine (Fig. 2). Of the 718 patients, 219 patients showed ≥ 1 osteoporotic VFs (30.5%) and 25 patients showed only pathologic VFs (3.5%). The overall prevalence of osteoporotic VFs was 26.3% in women and 32.5% in men, without significant difference between sexes (p = 0.097). Most fractures occurred at the thoracolumbar junction with the peak at level L1 and at the mid-thoracic spine with a local maximum at level T8 (Fig. 2). An example case of a patient with osteoporotic VFs is shown in Fig. 3.

Stratified into age groups, fracture prevalence steadily increased from 19.8% (CI 13.2–28.0%) at age 50–59 years up to 38.9% (CI 29.3–49.2%) at age 80 years and older (Table 2, Fig. 4). In women, fracture prevalence was 15.6% (CI 6.2–30.9%) at age 50–59 years, increased to 28.2% (CI 20.2–37.4%) at age 70–79 years, and continued to rise to 50% (CI 28.4–71.6%) at age 80 years and older. In men, fracture prevalence was 21.5% (CI 13.6–31.5%) at aged 50–59 years and seemingly levelled at age 60–69 years with 35.8% (CI 28.1–44.2%) with minimal increase until age 80 years and older with 36.1% (CI 25.7–47.6%).

Age was significantly associated with the prevalence of osteoporotic VFs (p = 0.001) and remained to be in multivariate analysis (p = 0.002; Table 1). There was no significant association between a certain CT indication and the prevalence of an osteoporotic VF (all p > 0.05; Table 1). However, patients with CT angiography showed a tendency for a higher prevalence rate of VFs (36.7%) compared to other CT imaging indications (p = 0.066). Moreover, the prevalence of osteoporotic VFs was significantly associated with intake of metamizole and Vitamin D for at least 3 months (p = 0.013 and p = 0.045, respectively; Table 1). Of those factors, prolonged metamizole intake remained a significant predictor for a prevalent VF in multivariate analysis (p = 0.011). There was a tendency of more patients with VFs experiencing longer periods of immobility that did not reach statistical significance (p = 0.067). The association of patients without prevalent VFs taking bisphosphonates for at least 3 months did not reach statistical significance due to sparse data (p = 0.057). Clinical information was not available concerning immobility for n = 61, chemotherapy for n = 52, and drug intake for n = 75 patients.

Medical notes in the hospital database listed an established diagnosis of osteoporosis or low bone mass in 38 patients (5.3%) and ruled out such a diagnosis in 4 patients (0.6%), while in the majority of 676 patients, no information related to bone health could be found (94.1%; Table 3). Of all 219 fractured patients, CT reports mentioned an osteoporotic VF in only 54 patients (24.7%), used secondary terminology referring to a vertebral abnormality with or w/o mentioning deteriorated bone quality in 26 patients (11.9%), and only mentioned deteriorated bone quality in 11 patients (5%; Table 4).