Reviews/Osteoporosis Fracture Prevention/FRAX fracture risk assessment tool in osteoporosis
Fracture Prevention Services · ostfrac_18 · v2
This claim asks whether the FRAX risk tool helps doctors pick the right patients to treat for osteoporosis. The evidence here shows that selecting patients by their estimated fracture probability does point to higher-risk people than relying on a bone-density score alone, and several analyses support moving toward a probability-based approach [1][3][4]. But all of this comes from statistical modeling and registry data, not from a trial that actually gave FRAX-guided care to some patients and bone-density-guided care to others and then counted fractures. One large registry study found that the simple treat-or-not thresholds based on FRAX catch only a modest share of people who go on to fracture, especially younger women [2]. Another showed FRAX cannot track a patient's response to treatment over time [5], and a newer study found its ability to rank who will fracture is only moderate and is beaten by a competing tool [6]. The panel judged the direction of support genuine but indirect, drawn from overlapping data sources, and mixed within the corpus.
The corpus shows that FRAX-based intervention thresholds identify women at higher fracture probability than fixed T-score thresholds, particularly in the elderly [1], and that the FRAX model supports a shift from a DXA-based intervention strategy toward one based on fracture probability [3]. FRAX without BMD is highly sensitive (99.3% in women, 99.1% in men) for flagging those who would qualify for treatment under a 20% MOF threshold, allowing simple age-based testing rules [4]. This is risk-stratification and modeling evidence rather than outcome evidence. Against it, dichotomous FRAX treatment thresholds have low sensitivity and PPV, especially in women aged 40 to 49 [2], change in FRAX score did not independently predict incident fracture so it cannot serve as a treat-to-target monitor [5], and FRAX discrimination in men was only moderate (AUC 0.641) and bettered by μFRAC [6]. The contested point: The division is between FRAX as a continuous risk stratifier versus FRAX as a dichotomous treatment-decision threshold: the gradient of risk is strong [2], yet the cut-off strategies built on it have low sensitivity and PPV [2], and FRAX cannot track treatment response [5]. The panel also split on whether modeling and observational discrimination can establish 'improves targeting' absent any trial comparing FRAX-guided to BMD-guided care on fracture outcomes.
All 6 papers behind this claim with their DOIs and what each does to it, the panel’s arbitration, the evidence gap map and the open research questions. Free, and sent to your inbox.
We’ll also send you other claims from Osteoporosis Fracture Prevention, one at a time, and tell you when a release or a pivotal trial changes a grade. Unsubscribe by replying.
One reviewer's only job is to attack the finding. The challenges below were weighed before the grade was set.
All support is modeling and observational risk-discrimination evidence; no RCT in the corpus shows FRAX-guided treatment reduces fractures versus BMD-guided care, so 'improves targeting' rests on a surrogate.
Evidential independence is weak: [3], [4] and [5] all draw on the same Manitoba BMD registry / Leslie research group, inflating apparent concordance.
Where it remains uncertain. Dichotomous FRAX treatment thresholds have low sensitivity and PPV, especially in younger women, and the authors say they should be reassessed [2]; FRAX also cannot serve as a treat-to-target monitor [5] and shows only moderate discrimination (AUC 0.641) [6].
Aleatory
In women aged 40 to 49, who have the lowest incidence rates, threshold sensitivity for incident MOF ranged as low as 0.0%, making targeting inherently noisy where events are rare [2].
The load-bearing papers from a corpus of 6, each classified by what it does to the claim. Every DOI is taken from the corpus the claim was graded on.
Supports FRAX thresholds targeting high-risk individuals better than BMD-only thresholds.
“Intervention thresholds based on BMD alone do not optimally target women at higher fracture risk than those on age-matched individuals without clinical risk factors, particularly in the elderly.”
Panel auditInternal validity: medium · scope of inference: Men and women assessed under Kuwaiti guidelines, where FRAX has been calibrated; comparison of threshold strategies, not treatment outcomes.Key limitations: modelling study, not outcome-validated · single-country calibration · no head-to-head fracture reduction
Supports probability-based over DXA-based intervention targeting.
“The FRAX model supports a shift from a dual X-ray absorptiometry (DXA)-based intervention strategy, towards a strategy based on fracture probability for a major osteoporotic fracture.”
Panel auditInternal validity: medium · scope of inference: Canadian men and women aged ≥50; who gets designated high-risk under FRAX vs BMD criteria, not whether treatment targeting improves outcomes.Key limitations: descriptive proportions, no outcome comparison · conclusion is a supported shift, not demonstrated benefit · thresholds chosen a priori
Shows FRAX efficiently identifies treatment-eligible individuals.
“FRAX without BMD offers an effective strategy to identify individuals meeting the current Canadian treatment threshold based upon FRAX® with BMD (≥ 20%).”
Panel auditInternal validity: high · scope of inference: Canadian men and women age 50+ at average risk without prior fracture; FRAX-without-BMD as a triage tool to target BMD testing and treatment.Key limitations: targeting defined against FRAX-with-BMD, not fractures · high sensitivity but treatment benefit not measured · Canadian threshold-specific
Qualifies the claim: FRAX-threshold targeting performs poorly at dichotomous cut-offs, notably in younger women.
“The sensitivity and PPV of the strategies based on dichotomous cut-offs are low, especially among women aged 40 to 49 years (who have lowest incidence rates). Threshold-based approaches should be reassessed, particularly in younger women.”
Panel auditInternal validity: high · scope of inference: Women aged ≥40 in Manitoba, Canada; performance of dichotomous FRAX treatment thresholds, especially poor in younger women.Key limitations: dichotomous cut-offs perform poorly in younger women · no direct demonstration FRAX beats T-score targeting · continuous FRAX strong but threshold use weak
4 of 6 papers shown. The full review lists all of them, with the panel’s arbitration →
Claims are graded at each quarterly release, and re-graded between releases only when a pivotal trial lands. Every change is a new version of the claim.
Directional support for better risk-based targeting than BMD alone is genuine but indirect, source-correlated, and internally mixed, warranting an Uncertain verdict.
Release 2026-q4 re-graded every claim from its corpus under protocol v0.19. An earlier release's grade is shown for the record; it came from an earlier protocol and is not a movement.
Assumptions that carry weight but are not themselves established.
The panel records the gap that would most move the claim, and the study that would close it.
No fracture-outcome trial of FRAX-guided versus BMD-guided care. Every supporting paper is modeling or observational risk discrimination; 'improves targeting' cannot be confirmed as a clinical benefit without an outcome comparison [1][3][4]. The panel rates this the highest-importance open gap for the claim, and names the study design that would close it: A randomised or pragmatic trial allocating patients to FRAX-probability-threshold treatment versus BMD/T-score-threshold treatment, with incident fracture as the endpoint.
Open question: Does treating patients selected by FRAX probability thresholds reduce fractures more than treating those selected by BMD/T-score thresholds? Why it is still open: The corpus contains only modeling and observational risk-discrimination evidence [1][3][4]; no study measures fracture outcomes under the two targeting strategies.
Priors is a sceptical first reader, not peer review, and not medical advice.
Release 2026-q4 · graded 21 Oct 2026 · protocol v0.19