A vertebral fragility fracture occurs every 22 seconds worldwide in people over the age of 50. That is more than 1.4 million fractures every year globally, and approximately 750,000 in the United States. Vertebral compression fractures are the most common type of osteoporotic fracture, more frequent than hip or wrist fractures combined. Most of them go unrecognised.
Not because imaging is unavailable. Not because osteoporosis is a rare or poorly understood disease. But because of two distinct and compounding failures in the clinical pathway: fractures that are genuinely not detected, and fractures that are visible on existing imaging but are never reported, and therefore never treated.
This is the diagnostic gap at the heart of osteoporosis care. It falls disproportionately on women. Its consequences, in quality of life, subsequent fractures, and preventable deaths, are well documented and largely preventable.
A Disease of Scale and Invisibility
According to the International Osteoporosis Foundation, approximately 500 million people worldwide are currently living with osteoporosis or low bone mass. One in three women over the age of 50 will experience an osteoporotic fracture in their lifetime. One in five men will too.
Vertebral compression fractures are the leading edge of that burden. They are most common in postmenopausal women, whose bone density loss accelerates sharply after menopause due to oestrogen decline. Prevalence rises from approximately 3% in women under 60 to around 20% in women aged 70 and older. Across Europe, an estimated 12% of women aged 50 to 79 have evidence of at least one vertebral fracture. [1]
The scale is considerable. The visibility is not. Most vertebral fractures produce no dramatic event. There is no fall, no acute pain, no emergency presentation. Back pain that develops gradually, a slight loss of height, a change in posture: these are the signals, and they are easy to attribute to ageing. Patients do not present. Clinicians do not investigate. The fracture remains unrecorded. Only one in three vertebral fractures ever comes to clinical attention. [2]
Two Layers of the Same Problem
The diagnostic failure in vertebral fractures is not a single phenomenon. It operates at two distinct points in the patient pathway, and understanding both is essential to understanding why the gap persists.
The first layer is true non-detection
Many patients with vertebral compression fractures are never imaged for the relevant indication. DEXA (dual-energy X-ray absorptiometry) is the established pathway for osteoporosis screening, but it requires a clinical trigger: a referral, a risk score, a threshold reached. In primary care systems under pressure, a significant proportion of at-risk patients, particularly those without a prior fracture or strong family history, never reach that threshold. Their fracture exists, and no imaging ever captures it.
The second layer is detection without reporting
This is less intuitively obvious, but clinically at least as significant. Chest CT scans, abdominal CT scans, and thoracic CT scans acquired for entirely different indications routinely include the thoracic and lumbar vertebral column within their field of view. Those images sit in PACS systems every day. In a meaningful proportion of cases, the signs of vertebral fracture are present on those images.
They are not the reason the scan was ordered, and they are frequently not the reason it is reported.
A retrospective study of patients aged 60 and over found radiographic evidence of moderate or severe vertebral fractures in 22% of cases, yet only 55% of contemporaneous radiology reports mentioned those fractures [6]. The imaging existed. The fracture was identifiable. The clinical opportunity to intervene was not taken.
This is not a question of radiologist skill. CT scans are read in the context of the clinical question that generated them. The primary task determines the focus. Vertebral fracture assessment is not that task. What is needed is a systematic layer that applies to every relevant scan regardless of its indication.
What Happens When Fractures Go Untreated
The consequences of the diagnostic gap are not abstract.
An undetected vertebral fracture is one of the strongest available predictors of future fracture. Women with severe vertebral fractures carry a 12.6 times greater risk of a new vertebral fracture [3]. A prior vertebral fracture increases hip fracture risk 2.3-fold [7]. Over 55% of patients who present with a hip fracture have evidence of a prior vertebral fracture [8].
Hip fractures carry a 20 to 24% mortality rate in the first year [4]. Forty percent of patients who survive a hip fracture are unable to walk independently afterwards [4]. Thirty-three percent end up in long-term institutional care [4].
The vertebral fracture that goes undetected today is, in a significant proportion of cases, the precursor to the hip fracture that generates all of those outcomes.
Quality of life impacts from vertebral fractures themselves are also well established: chronic back pain, height loss, progressive spinal deformity, reduced mobility, and depression, all of which compound over time as fractures accumulate. The number of vertebral fractures is a direct determinant of quality-of-life deterioration in postmenopausal women.
Despite this, the treatment gap for osteoporosis in Europe stands at 71% [5]. Approximately 15 million eligible women in Europe remain untreated [5]. Fewer than half of patients identified with a vertebral fracture receive even one prescription for osteoporosis treatment [5].
The fractures are happening. The evidence is often present. The treatment is not being initiated.
The Fracture Liaison Service Model and Its Limits
The most established clinical response to cascade fracture risk is the Fracture Liaison Service (FLS). Capture the Fracture, the IOF’s global programme for secondary fracture prevention, now counts more than 1,300 FLS operating across 64 countries, identifying over one million patients annually. The evidence base for FLS is strong: coordinated, multidisciplinary post-fracture care reduces subsequent fracture rates and healthcare costs.
But FLS are downstream. They depend on an index fracture having been identified and a patient having entered the care pathway. For the two-thirds of vertebral fractures that never come to clinical attention, FLS cannot intervene. A patient whose vertebral fracture was present on a chest CT two years ago, but was never reported, never reaches the FLS.
Closing the upstream gap, the detection and reporting gap, is a prerequisite for the downstream prevention system to function at full capacity.
AI as a Systematic Detection Layer
This is where AI-assisted vertebral fracture detection addresses a structural problem rather than an individual case.
Tools such as CINA-VCF and CINA-VCF Quantix, developed by Avicenna.AI, operate on chest and abdominal CT scans acquired for any indication. They analyse the vertebral data that is already present in those acquisitions, systematically and on every relevant scan, regardless of the clinical question that generated it. They do not replace the radiologist’s primary reading task. They add a consistent, secondary assessment that would otherwise require the radiologist to step outside the scope of the original request.
CINA-VCF, FDA-cleared for the United States, detects suspected vertebral compression fractures and generates an alert within seconds. CINA-VCF Quantix, CE-marked for European deployment, adds quantitative analysis: automated vertebral labelling, height loss ratio calculations, Genant grading, and Hounsfield Unit values as a proxy for bone mineral density. When a height loss ratio exceeds 20 to 25%, consistent with a moderate to severe VCF, a passive notification is generated for radiologist review. Sensitivity: 92.3% (95%CI: 81.5%–97.9%). Specificity: 91.7% (95%CI: 80.0%–97.7%) for CINA-VCF Quantix; 95.2% sensitivity and 92.9% specificity for CINA-VCF.
A peer-reviewed performance evaluation of CINA-VCF Quantix on opportunistic CT scans confirmed its detection performance on real-world data across CT acquisitions performed for unrelated indications. A separate multi-centre validation study of CINA-VCF, conducted across 474 CT scans from France and the United States, confirmed equivalent detection performance in a blinded, multinational cohort.
Every flagged finding is reviewed by a radiologist before any clinical action is taken. The AI does not generate a clinical report. It creates the condition for the radiologist to include a finding that the primary clinical task might not have surfaced.
Regulatory Recognition: NICE HTG760
In October 2025, NICE published Early Value Assessment guidance HTG760, evaluating AI technologies for opportunistic vertebral fracture detection. The guidance recommends five AI tools for NHS use during a three-year evidence generation period. CINA-VCF Quantix is one of the five recommended systems.
NICE’s assessment reflects the clinical rationale directly: these tools address the diagnostic gap for patients who are already being imaged, whose imaging already contains the relevant vertebral data, and who are currently leaving radiology departments without a fracture assessment having been made.
Four AI tools for vertebral fracture detection were endorsed by NICE for NHS use, with the assessment noting that opportunistic deployment could identify fractures in thousands of patients who would otherwise go undiagnosed. The evidence generation plan accompanying HTG760 will build the real-world data needed to quantify those outcomes across NHS sites.
The Broader Context: Prevention at Scale
Avicenna.AI is a member of PRECCO, the initiative established under European Society for Clinical and Economic Aspects of Osteoporosis and Metabolic Bone Diseases (ESCEO) and the IOF to strengthen osteoporosis management and fracture prevention globally.
The clinical argument for opportunistic vertebral fracture detection is straightforward. The imaging infrastructure already exists. The at-risk population is already being scanned for other reasons. The vertebral data is already in the system. What has been missing is a consistent, scalable mechanism to extract clinical value from that data and connect it to the prevention pathway.
That connection matters most for the patients who are furthest from it: women in their 60s and 70s undergoing chest or abdominal CT for an unrelated condition, whose vertebral fracture is present on the scan, and who will leave without a bone health assessment unless the imaging is examined with that purpose in mind.
FAQ
What is opportunistic vertebral fracture detection?
Opportunistic detection means identifying vertebral fractures on CT scans ordered for an unrelated clinical indication. Chest, abdominal, and thoracic CT scans routinely include the thoracic and lumbar spine within the imaged volume. AI tools assess vertebral data systematically, flagging suspected fractures that would not otherwise be part of the reading task.
Why are so many vertebral fractures detected but not reported?
Radiology reports are structured around the clinical question that generated the scan. A chest CT is read for pulmonary, cardiac, or mediastinal findings. Vertebral assessment is not within the scope of that task unless specifically requested. Without a structured mechanism to prompt that assessment, findings that are visible in the imaging data often go unrecorded.
How does this fit with existing DEXA screening pathways?
DEXA measures bone mineral density and informs osteoporosis treatment decisions. It requires a clinical referral and is not routinely performed on patients who have not yet been identified as at risk. Opportunistic CT-based VCF detection operates upstream: it can identify a patient who already has a fracture and who should be referred for DEXA and treatment, but who has not yet triggered the pathway. The two approaches are complementary, not alternatives.
References
[1] O’Neill TW et al. (EVOS Study Group). *The prevalence of vertebral deformity in european men and women: the European Vertebral Osteoporosis Study.* Journal of Bone and Mineral Research. 1996;11(7):1010–1018. doi: 10.1002/jbmr.5650110719
[2] Felsenberg D et al. *Incidence of vertebral fracture in Europe: results from the European Prospective Osteoporosis Study (EPOS).* Journal of Bone and Mineral Research. 2002;17(4):716–724. doi: 10.1359/jbmr.2002.17.4.716
[3] Lindsay R et al. *Risk of new vertebral fracture in the year following a fracture.* JAMA. 2001;285(3):320–323. doi: 10.1001/jama.285.3.320
[4] Cooper C. *The crippling consequences of fractures and their impact on quality of life.* Osteoporosis International. 1999;9(Suppl 2):S2–8. doi: 10.1007/PL00022774
[5] Hernlund E et al. *Osteoporosis in the European Union: medical management, epidemiology and economic burden.* Archives of Osteoporosis. 2013;8(1–2):136. doi: 10.1007/s11657-013-0136-1
[6] Williams AL et al. *Vertebral fractures: how large is the silent epidemic?* AJR Am J Roentgenol. 2004;182(2):297–302. https://doi.org/10.2214/ajr.182.2.1820297
[7] Melton LJ et al. *Long-term fracture prediction by bone mineral assessed at different skeletal sites.* Osteoporosis International. 1999;9(4):310–315. PMID: 10525713
[8] Gonnelli S et al. (BREAK Study). *The risk of re-fracture and its determinants in subjects with vertebral fractures.* Osteoporosis International. 2013;24(4):1151–9. PMID: 23011681. doi: 10.1007/s00198-012-2029-3