Osteoporosis: a silent disease diagnosed by a line drawn on a smooth curve

Topic: osteoporosis: bone density as a continuous risk gradient, absolute-risk assessment, antiresorptive and anabolic therapy, and the treatment gap · Since 1990 · Grounded citations only · Published 2026-08-30

Start here: what osteoporosis is

Bone is not inert scaffolding. It is living tissue that is continuously torn down and rebuilt — remodelled — throughout life, by two opposed cell types working in sequence: osteoclasts, which resorb old bone, and osteoblasts, which lay down new [1] [2] [3]. In youth the two are balanced or building. From middle age onward, and abruptly after the menopause, resorption outpaces formation, and the skeleton loses both density and structural quality — the internal trabecular architecture thins and connections are lost.

The result is bone that breaks under loads it should tolerate: a fall from standing height, sometimes a cough. That is a fragility fracture, and the three that matter most are hip, spine and wrist.

It is silent until it isn't. Losing bone produces no symptoms. There is no ache that warns you, and the first sign in most people is the fracture itself. Vertebral fractures are stranger still — many are never clinically diagnosed at all, occurring without a memorable event and showing up later as height loss, stooping or an incidental finding on an X-ray taken for something else [4]. This is why osteoporosis is described as a silent disease, and why it is so heavily underdiagnosed: in a study of 200,160 postmenopausal women with no previous osteoporosis diagnosis, almost half had previously undetected low bone density, including 7% who met the definition of osteoporosis [5].

Why it matters. A hip fracture is not a broken bone in the way a wrist fracture is a broken bone. It is an event that frequently ends independent living and carries substantial mortality. The worldwide prevalence and disability associated with osteoporotic fractures have been estimated formally [6], hip fracture incidence and lifetime probability mapped across countries [7], and the global burden of fractures quantified in the Global Burden of Disease framework [8], within the broader burden of musculoskeletal conditions [9].

Who. Postmenopausal women above all — oestrogen withdrawal accelerates bone loss sharply — and older adults of both sexes. There are also secondary causes, of which long-term glucocorticoid therapy is the most important and most preventable [10].

And it is treatable. That is the part worth insisting on, because a silent, symptomless condition with effective therapy is exactly the kind of disease that goes untreated.

Three pillars follow — measurements, medicines, and progress — with a model between the first two that explains why the number used to diagnose this disease identifies only a minority of the people who will break a bone.

Pillar 1: measurements and diagnosis

DXA and the T-score

Bone mineral density is measured by dual-energy X-ray absorptiometry (DXA), a low-dose scan of the hip and lumbar spine. The result is expressed as a T-score: how many standard deviations the patient's density sits above or below the mean of a healthy young adult.

The WHO definitions are thresholds on that scale: osteoporosis is a T-score of −2.5 or below, and osteopenia — low bone mass — is between −1.0 and −2.49 [5]. Those two numbers determine who gets a diagnosis, and, in many health systems, who gets treatment.

They are also, as the model below shows, conventions rather than biological boundaries.

FRAX, and why absolute risk is the better question

A T-score answers "how dense is this bone?" The clinically useful question is "what is the chance this person breaks something in the next ten years?" — and density is only one input to that. Age matters enormously, as do prior fracture, parental hip fracture, smoking, glucocorticoid use, rheumatoid arthritis, alcohol intake and body-mass index.

FRAX combines those clinical risk factors, with or without a BMD measurement, into a 10-year probability of major osteoporotic fracture and of hip fracture, calibrated to each country's own epidemiology [11]. European guidance for diagnosis and management is built around this absolute-risk framing [12] [13].

The shift from T-score to absolute risk is the single most important conceptual change in this field, and the figure below is why it happened.

Bone turnover markers, and finding the fractures already there

Blood and urine markers of bone resorption and formation reflect how fast remodelling is running. They change within weeks of starting therapy — far faster than density does — so their main clinical uses are checking that a treatment is working and that the patient is actually taking it.

Vertebral imaging matters because the fractures that most often go unnoticed are the ones that most strongly predict the next one. Beyond dedicated imaging, abdominal CT scans performed for entirely unrelated reasons contain usable bone-density information, and opportunistic screening of those existing scans has been shown to identify osteoporosis in people who were never referred for a DXA [14]. The National Bone Health Alliance has argued for a clinical diagnosis of osteoporosis that does not depend on the T-score alone [15].

Centerpiece: a simple simulatable model of density and fracture risk

The question this model answers is whether the diagnostic threshold does the job people assume it does.

Start with the relationship between density and risk. A meta-analysis of eleven prospective cohorts, about 90,000 person-years and over 2,000 fractures reported the relative risk of fracture per one standard deviation decrease in bone mineral density: 1.5 (95% CI 1.4–1.6) at any measurement site generally, 2.3 (1.9–2.8) for spine density predicting vertebral fracture, and 2.6 (2.0–3.5) for hip density predicting hip fracture [16].

A constant relative risk per standard deviation is an exponential statement. So the curve is the source's own model, written out:

RR(T) = k^(−T), where T is the T-score and k the published per-SD relative risk.

−4 −3 −2.5 −2 −1 0 bone mineral density, T-score (SD from young-adult mean) 0.5 1 2 5 10 20 40 relative fracture risk (log scale) the threshold is a line drawn on a smooth curve: T = −2.6 carries just 21% more hip-fracture risk than T = −2.4 Fracture risk is continuous in bone density any site, any fracture ×1.5 per SD spine BMD, vertebral fracture ×2.3 per SD hip BMD, hip fracture ×2.6 per SD osteoporosis T ≤ −2.5 osteopenia normal normal T > −1.0 osteopenia −1.0 to −2.49 osteoporosis T ≤ −2.5 0 10 20 30 40 50 60 70 % of women / % of all fractures 53.2% 35% 39.6% 46% 7.2% 19% pale = share of women solid = share of fractures 81% of fractures occur in women who are NOT osteoporotic by T-score So most fractures happen outside the diagnosis 1× 1.8× 4.03×
Computed relationship between bone density and fracture risk, and its consequence for who actually fractures. LEFT PANEL: relative fracture risk on a logarithmic axis against T-score, under RR(T) = k^(-T). This is not an imposed shape - "a relative risk of k per one standard deviation decrease" IS an exponential in standard deviations, so the curve is the source's own model written out. The three values of k are published with their intervals: 1.5 (95% CI 1.4-1.6) for any site and any fracture, 2.3 (1.9-2.8) for spine density predicting vertebral fracture, and 2.6 (2.0-3.5) for hip density predicting hip fracture, from a meta-analysis of 11 prospective cohorts, about 90,000 person-years and over 2,000 fractures (Marshall et al., BMJ 1996 [W2036125852]). Shaded bands are those intervals; shaded vertical regions are the WHO categories. The script asserts that each curve returns exactly its published relative risk at T = -1 and 1.0 at the mean, that every point estimate lies inside its interval, that risk rises monotonically as density falls, and that site-matched measurement dominates generic measurement at every low density. The annotated consequence: someone at T = -2.6, just inside the diagnosis, carries only 21% more hip-fracture relative risk than someone at T = -2.4, just outside it - the threshold is a line drawn on a smooth curve, not a cliff. RIGHT PANEL: no model and no fitted parameters. Six published numbers from 200,160 postmenopausal women with no previous osteoporosis diagnosis (Siris et al., NORA, JAMA 2001 [W2153579902]) - prevalences of 53.2% normal, 39.6% osteopenia and 7.2% osteoporosis, and fracture rate ratios of 1.00, 1.80 (95% CI 1.49-2.18) and 4.03 (3.59-4.53) - are multiplied and normalised to give each category's share of all fractures: 35%, 46% and 19%. The script asserts that osteoporosis carries the highest risk per person yet contributes the smallest share of fractures, that osteopenia contributes more fractures than osteoporosis does, and that 81% of fractures occur in women who are NOT osteoporotic by T-score. ILLUSTRATIVE and flagged: extending the per-SD relative risk as a smooth exponential across the whole range assumes it stays constant, which the source reports as an average across cohorts rather than verifying at every density. These are RELATIVE risks against an age-adjusted mean, so the left panel says nothing about ABSOLUTE risk, which rises steeply with age at any T-score - precisely why absolute-risk tools exist and why this curve alone cannot decide treatment. The two sources measured at different skeletal sites by different methods, so the panels are two views of one phenomenon rather than one continuous calculation.

What the model explains. Four things.

First, that the diagnostic threshold is arbitrary in the precise sense. Risk climbs smoothly and steeply as density falls, with no discontinuity anywhere. Two women either side of −2.5 differ in hip-fracture relative risk by about 21% — while one receives a disease label and often treatment, and the other receives neither. The line is useful for standardising research and reimbursement; it is not a fact about bone.

Second, why most fractures happen to people without the diagnosis. This falls out of arithmetic rather than argument. Osteoporotic women have four times the fracture rate — but they are only 7.2% of the population, while osteopenic and normal-density women together are 92.8%. Multiply prevalence by rate and 81% of all fractures occur outside the osteoporotic range, with osteopenia alone contributing more fractures (46%) than osteoporosis (19%) [5]. A treatment policy keyed to T-score ≤ −2.5 is aimed at the fifth of the problem where risk per person is highest, not at where most fractures are.

Third, why FRAX exists. If density is a strong but partial predictor, the way to find high-risk people is to combine it with everything else that predicts fracture — which is exactly what absolute-risk tools do [11]. The model shows the gap that FRAX was invented to fill.

Fourth, the source's own conclusion, which is stronger than its numbers suggest. Marshall and colleagues wrote that bone density measurements "can predict fracture risk but cannot identify individuals who will have a fracture", and on that basis explicitly did not recommend population screening of menopausal women by bone densitometry [16]. A relative risk of 2.6 per SD is a strong epidemiological association and still a weak individual test. Those are different things, and conflating them is the commonest error in reading this literature.

What the model deliberately does not do. It is entirely about relative risk, and relative risk against an age-adjusted mean at that. A 50-year-old and an 80-year-old with identical T-scores have very different absolute risks, and it is absolute risk that determines whether treatment is worthwhile — so nothing here should be read as a treatment rule. It assumes the per-SD risk is constant across the whole density range. And bone density is not bone strength: architecture, turnover rate and material quality all contribute to fragility and none is captured by a DXA number.

Pillar 2: medicines — slowing the loss, or rebuilding

Treatment has a base and two pharmacological strategies.

The base: calcium, vitamin D, and not falling over

Adequate calcium and vitamin D are the substrate for mineralisation, and supplementation in men and women aged 65 and over has been trialled for its effect on bone density [17]; hypovitaminosis D is common enough among medical inpatients to be worth actively looking for [18]. These are a foundation, not a treatment: they do not substitute for the drugs below in someone at high risk.

Falls prevention deserves equal billing and rarely gets it. Almost every hip fracture is a fall plus a fragile bone, and the classic prospective analysis of hip-fracture risk factors in older women found that many of the strongest predictors were about falling — not about bone at all [19]. Removing either term prevents the fracture, and exercise interventions in older adults address strength and balance directly. Consensus statements have long placed these alongside the drugs rather than beneath them [20].

Antiresorptives: stop the demolition

Bisphosphonates are first-line. They bind avidly to bone mineral, are taken up by osteoclasts during resorption, and disable them — nitrogen-containing bisphosphonates by inhibiting farnesyl pyrophosphate synthase in the mevalonate pathway, which blocks the protein prenylation osteoclasts need to function [1]. The fracture evidence is old and solid: alendronate reduced fracture risk in women with existing vertebral fractures [21] and was tested separately in women with low bone density but no prior vertebral fracture [22].

Denosumab is a fully human monoclonal antibody against RANKL, the signal that drives osteoclast development and activity. In FREEDOM, 7,868 women aged 60–90 with a T-score below −2.5 but not below −4.0 received denosumab or placebo six-monthly for 36 months. New radiographic vertebral fracture fell from 7.2% to 2.3% — a 68% relative reduction (risk ratio 0.32, 95% CI 0.26–0.41); hip fracture fell 40% (HR 0.60, 0.37–0.97); non-vertebral fracture fell 20% (HR 0.80, 0.67–0.95) [23]. Note the gradient across fracture types: the vertebral effect is much larger than the non-vertebral one, which is typical of this whole drug class.

Denosumab has a specific and important catch. Its effect does not persist after stopping. Discontinuation is followed by a rebound in bone turnover and a documented risk of multiple vertebral fractures — described in a systematic review of rebound-associated cases [24] [25], confirmed in a post hoc analysis of the FREEDOM trial and its extension [26], and now the subject of a formal position statement on how to stop the drug safely [27] [28]. So it is not a drug that can simply be discontinued, and transition to another antiresorptive is required. This is a genuine difference from bisphosphonates, which persist in bone for years after the last dose — and, unlike denosumab, teriparatide's fracture-risk reduction was sustained after withdrawal [29].

Anabolics: rebuild

Antiresorptives slow loss. Anabolics add bone.

Teriparatide is a fragment of parathyroid hormone acting at the PTH/PTHrP receptor. Continuous PTH exposure resorbs bone; intermittent daily injection stimulates formation instead — a genuinely counterintuitive pharmacology [30]. It outperformed alendronate in glucocorticoid-induced osteoporosis [10], and abaloparatide, a related peptide, reduced new vertebral fractures against placebo [31].

Romosozumab blocks sclerostin, an inhibitor of Wnt signalling in bone, and is unusual in doing two things at once: increasing formation and decreasing resorption. In FRAME it was tested against placebo [32]; in ARCH, 4,093 postmenopausal women with osteoporosis and a fragility fracture received romosozumab or alendronate for 12 months, then open-label alendronate. At 24 months new vertebral fractures were 48% lower (6.2% vs 11.9%), clinical fractures 27% lower, non-vertebral 19% lower and hip fracture 38% lower [33]. The trial also reported the safety signal that shapes its use: during year 1, adjudicated serious cardiovascular adverse events occurred more often with romosozumab than alendronate (2.5% vs 1.9%) [33].

Sequence matters

ARCH is not only a superiority trial; it is a sequencing trial — anabolic first, then antiresorptive to hold the gain. That order is not interchangeable: anabolic agents build bone that an antiresorptive then preserves, whereas starting with an antiresorptive blunts the subsequent anabolic response. Comparative efficacy across the whole class has been assembled in network meta-analysis [34], in living systematic review [35], in a broad fracture-reduction and safety synthesis [36], and in society guidance [37] [38] [39].

The harms, stated plainly

Two rare adverse effects dominate public perception of these drugs, and both are real and both are uncommon. Osteonecrosis of the jaw has an international consensus definition and management pathway [40]. Atypical femoral fracture is associated with long-term bisphosphonate use, and the mechanism is thought to be severely suppressed bone turnover — a complication described early in the alendronate era [41]. These are the reason for drug holidays after several years of bisphosphonate therapy in lower-risk patients, a question addressed directly by an ASBMR task force report on managing long-term bisphosphonate treatment [42]. They are not a reason to leave a high-risk patient untreated: the fractures prevented vastly outnumber the fractures caused [35].

Pillar 3: progress

Closing the treatment gap

The largest available gain in this field is not a new molecule. It is treating the people who have already fractured — a group whose risk is unambiguous and who are nonetheless frequently discharged from a fracture clinic with no bone assessment at all. Fracture liaison services, which systematically identify and investigate every patient presenting with a fragility fracture, exist precisely to close that gap — and the scale of the gap they address has been quantified across Europe as a matter of burden, management and missed opportunity [43]. Given that the first fracture is the strongest predictor of the next, this is the clearest case in the review of a known problem with a known solution and an implementation failure.

Risk-stratified and sequenced therapy

The model's implication — treat by absolute risk rather than by T-score — is now guideline practice [12] [13], and the arrival of anabolic agents has added a second axis: not only whether to treat but in which order, with anabolic-first reserved for those at highest imminent risk [38] [37].

Better use of scans already taken

Opportunistic CT screening turns imaging performed for other reasons into osteoporosis case-finding at essentially no marginal cost [14] — an attractive answer to a disease whose central problem is that nobody knows they have it.

Dig deeper in lmmol

Osteoporosis intersects with several conditions covered elsewhere here:

  • Chronic kidney disease — CKD disturbs calcium, phosphate, vitamin D and parathyroid hormone handling, producing a mineral and bone disorder in which standard osteoporosis thresholds and treatments cannot simply be applied.
  • Rheumatoid arthritis — a double hit on bone: the inflammation itself causes loss, and glucocorticoids used to treat it are the leading secondary cause of osteoporosis [10].
  • Obesity — higher body-mass index reduced the likelihood of low bone density in the study behind this review's figure [5], one of the few places where it is protective.
  • Depression — common after a hip fracture and a barrier to the rehabilitation that determines whether independence is regained.
  • The health reviews index collects the rest of the series.

Then move down into lmmol's graph, to the molecules the drugs act on:

  • Farnesyl pyrophosphate synthase — the mevalonate-pathway enzyme that nitrogen-containing bisphosphonates inhibit, disabling the osteoclast's protein prenylation [1].
  • Cathepsin K — the protease osteoclasts use to digest bone collagen, and the target of a drug class that was developed and then abandoned [44] [39].
  • PTH/PTHrP receptor — where teriparatide acts, and where continuous versus intermittent exposure produces opposite effects on bone [30].
  • Estrogen receptor — whose withdrawal at menopause drives the accelerated bone loss that makes this predominantly a disease of women.
  • Vitamin D receptor — the endpoint of the supplementation that forms the base of treatment [17].
  • For entities without a linked static page here, use the graph index, all proteins, or all diseases rather than guessing an entity URL.

Key papers

  1. W2124157902: Bone remodelling at a glance (cited 499×)
  2. W1970144926: Cellular and Molecular Mechanisms of Bone Remodeling (cited 1,387×)
  3. W1965696823: Cellular mechanisms of bone remodeling (cited 824×)
  4. W2033401946: Incidence of clinically diagnosed vertebral fractures: A population-based study in rochester, minnesota, 1985-1989 (cited 1,543×)
  5. W2153579902: Identification and Fracture Outcomes of Undiagnosed Low Bone Mineral Density in Postmenopausal Women (cited 1,226×)
  6. W2110396808: An estimate of the worldwide prevalence and disability associated with osteoporotic fractures (cited 4,741×)
  7. W2025881993: A systematic review of hip fracture incidence and probability of fracture worldwide (cited 1,385×)
  8. W3195195170: Global, regional, and national burden of bone fractures in 204 countries and territories, 1990–2019: a systematic analysis from the Global Burden of Disease Study 2019 (cited 1,276×)
  9. W2154461664: Burden of major musculoskeletal conditions. (cited 3,536×)
  10. W2042007405: Teriparatide or Alendronate in Glucocorticoid-Induced Osteoporosis (cited 912×)
  11. W2110768118: FRAX™ and the assessment of fracture probability in men and women from the UK (cited 2,461×)
  12. W2107490395: European guidance for the diagnosis and management of osteoporosis in postmenopausal women (cited 2,953×)
  13. W4253165656: European guidance for the diagnosis and management of osteoporosis in postmenopausal women (cited 1,874×)
  14. W2027387089: Opportunistic Screening for Osteoporosis Using Abdominal Computed Tomography Scans Obtained for Other Indications (cited 942×)
  15. W2115587655: The clinical diagnosis of osteoporosis: a position statement from the National Bone Health Alliance Working Group (cited 675×)
  16. W2036125852: Meta-analysis of how well measures of bone mineral density predict occurrence of osteoporotic fractures (cited 3,726×)
  17. W2314892802: Effect of Calcium and Vitamin D Supplementation on Bone Density in Men and Women 65 Years of Age or Older (cited 2,296×)
  18. W2325704874: Hypovitaminosis D in Medical Inpatients (cited 1,438×)
  19. W2323830513: Risk Factors for Hip Fracture in White Women (cited 3,593×)
  20. W175660244: Osteoporosis Prevention, Diagnosis, and Therapy (cited 5,532×)
  21. W2153009489: Randomised trial of effect of alendronate on risk of fracture in women with existing vertebral fractures (cited 3,857×)
  22. W2161837912: Effect of Alendronate on Risk of Fracture in Women With Low Bone Density but Without Vertebral Fractures Results From the Fracture Intervention Trial (cited 2,290×)
  23. W2160005424: Denosumab for Prevention of Fractures in Postmenopausal Women with Osteoporosis (cited 3,485×)
  24. W2594420699: Clinical Features of 24 Patients With Rebound‐Associated Vertebral Fractures After Denosumab Discontinuation: Systematic Review and Additional Cases (cited 369×)
  25. W2220606005: Rebound-associated vertebral fractures after discontinuation of denosumab—from clinic and biomechanics (cited 159×)
  26. W2766957057: Vertebral Fractures After Discontinuation of Denosumab: A Post Hoc Analysis of the Randomized Placebo-Controlled FREEDOM Trial and Its Extension (cited 722×)
  27. W3094494322: Fracture Risk and Management of Discontinuation of Denosumab Therapy: A Systematic Review and Position Statement by ECTS (cited 367×)
  28. W3120383256: Denosumab Discontinuation and the Rebound Phenomenon: A Narrative Review (cited 193×)
  29. W2015350578: Sustained Vertebral Fracture Risk Reduction After Withdrawal of Teriparatide in Postmenopausal Women With Osteoporosis (cited 330×)
  30. W2076064293: Parathyroid Hormone and Teriparatide for the Treatment of Osteoporosis: A Review of the Evidence and Suggested Guidelines for Its Use (cited 723×)
  31. W2516614130: Effect of Abaloparatide vs Placebo on New Vertebral Fractures in Postmenopausal Women With Osteoporosis (cited 863×)
  32. W2520995084: Romosozumab Treatment in Postmenopausal Women with Osteoporosis (cited 1,780×)
  33. W2754720807: Romosozumab or Alendronate for Fracture Prevention in Women with Osteoporosis (cited 1,509×)
  34. W2924633197: Efficacy of Pharmacological Therapies for the Prevention of Fractures in Postmenopausal Women: A Network Meta-Analysis (cited 330×)
  35. W4313455245: Effectiveness and Safety of Treatments to Prevent Fractures in People With Low Bone Mass or Primary Osteoporosis: A Living Systematic Review and Network Meta-analysis for the American College of Physicians (cited 272×)
  36. W4367681339: Fracture risk reduction and safety by osteoporosis treatment compared with placebo or active comparator in postmenopausal women: systematic review, network meta-analysis, and meta-regression analysis of randomised clinical trials (cited 211×)
  37. W3006754541: Pharmacological Management of Osteoporosis in Postmenopausal Women: An Endocrine Society Guideline Update (cited 426×)
  38. W4309368148: Anabolic therapy for osteoporosis: update on efficacy and safety (cited 54×)
  39. W2587417535: New horizons in treatment of osteoporosis (cited 168×)
  40. W1957467225: Diagnosis and Management of Osteonecrosis of the Jaw: A Systematic Review and International Consensus (cited 1,364×)
  41. W2022094899: Severely Suppressed Bone Turnover: A Potential Complication of Alendronate Therapy (cited 1,430×)
  42. W1777836280: Managing Osteoporosis in Patients on Long-Term Bisphosphonate Treatment: Report of a Task Force of the American Society for Bone and Mineral Research (cited 666×)
  43. W3016358164: Fragility fractures in Europe: burden, management and opportunities (cited 804×)
  44. W1964031097: Osteoclastogenesis, Bone Resorption, and Osteoclast-Based Therapeutics (cited 178×)