What was the actual absolute risk reduction in HORIZON-PFT, not just the relative 70%?

The absolute risk reduction for morphometric vertebral fractures was 7.6 percentage points over three years (10.9% placebo vs. 3.3% zoledronic acid). This translates to a number needed to treat of approximately 13.

Were non-white women adequately represented in HORIZON-PFT?

No. The trial population was predominantly white. Representation of Black, Asian, and other ethnic groups was limited, which constrains how confidently the results can be generalized to those populations.

Did the acute-phase reaction compromise blinding in the trial?

It may have. Roughly 32% of zoledronic acid recipients experienced fever, myalgia, or headache after the first infusion compared to about 7% in the placebo group. No formal blinding assessment was reported.

Was HORIZON-PFT funded by the drug manufacturer?

Yes. Novartis Pharmaceuticals funded the trial and participated in study design, data collection, analysis, and manuscript preparation. The FDA later independently confirmed the efficacy findings.

Did the atrial fibrillation signal in HORIZON-PFT hold up in later studies?

Not consistently. The serious atrial fibrillation rate was higher with zoledronic acid (1.3% vs. 0.5%), but subsequent analyses and the companion HORIZON-RFT trial did not replicate the finding. The FDA considers the causal relationship unestablished.

Why does the morphometric vertebral fracture endpoint matter for interpretation?

Morphometric fractures are detected on scheduled X-rays and are often asymptomatic. They are a valid surrogate, but clinically symptomatic fractures and hip fractures (both also reduced in the trial) may be more meaningful to patients.

Was HORIZON-PFT compared head-to-head against oral alendronate?

No. HORIZON-PFT was placebo-controlled. No direct comparison with alendronate or any other oral bisphosphonate was included. Indirect cross-trial comparisons are unreliable for drawing superiority conclusions.

How much missing data was there in HORIZON-PFT?

Approximately 24% of participants lacked evaluable spine radiographs at 36 months. Sensitivity analyses supported the primary finding, but this level of missing data warrants caution.

Does HORIZON-PFT tell us how long to treat with zoledronic acid?

Not directly. The trial lasted three years. Extension studies out to six years and separate drug-holiday analyses have since addressed duration, but the key trial itself does not answer this question.

Honest Criticisms and Limitations of the HORIZON-PFT Trial

Q: Is yearly zoledronic acid safe in patients with reduced kidney function?

HORIZON-PFT excluded patients with creatinine clearance below 30 mL/min. Post-marketing reports of acute renal failure led to updated labeling. Prescribers should ensure adequate hydration and check renal function before each infusion.

By HealthRX Medical Team

Published May 25, 2026Updated May 25, 2026Last reviewed May 25, 2026

At a glance

| Parameter | Detail | |---|---| | N | 7,765 postmenopausal women | | Intervention | Zoledronic acid 5 mg IV once yearly | | Comparator | Placebo IV infusion | | Duration | 3 years | | Primary endpoint | New morphometric vertebral fracture | | Key result | 70% relative risk reduction in vertebral fractures (3.3% vs 10.9%; p <0.001) | | Registration | NCT00049829 |

Why a Critical Appraisal Matters

When a trial reports a 70% relative risk reduction, the number commands attention. The HORIZON-PFT results published in the New England Journal of Medicine became a cornerstone of osteoporosis treatment guidelines within months. Zoledronic acid (marketed as Reclast) earned FDA approval for postmenopausal osteoporosis partly on the strength of these data. But headline numbers tell you what happened inside the trial. They do not automatically tell you what will happen in your clinic. This page dissects the gaps.

Enrollment Biases and Population Selection

The trial enrolled postmenopausal women aged 65 to 89 with a femoral neck T-score of −1.5 or worse (with existing vertebral fracture) or −2.5 or worse (without fracture). That two-pathway entry criterion matters for interpretation.

High-risk enrichment. Roughly 63% of participants had at least one prevalent vertebral fracture at baseline. This population sits at considerably higher fracture risk than the average woman referred for a DEXA scan. Absolute risk reductions in a lower-risk population would likely be smaller, even if relative reductions held steady.

Age floor of 65. Women in their late 50s and early 60s, a group frequently started on bisphosphonates in clinical practice, were excluded. Whether the same magnitude of benefit applies to younger postmenopausal women cannot be confirmed by HORIZON-PFT alone.

Ethnic and geographic homogeneity. The cohort was predominantly white and recruited across sites in Europe, North America, Latin America, and Oceania. Representation of Black, Asian, and Middle Eastern women was limited. Given known differences in bone mineral density distribution and fracture epidemiology across ethnic groups, generalizability is constrained. The National Osteoporosis Foundation guidelines later acknowledged that most key bisphosphonate trials underrepresent non-white populations.

Concomitant calcium and vitamin D. All participants received calcium (1,000 to 1 to 500 mg/day) and vitamin D (400 to 1 to 200 IU/day). The fracture reductions attributed to zoledronic acid occurred on top of this supplementation. Patients who are non-adherent to calcium and vitamin D in real-world settings may not reproduce the same benefit.

Statistical Considerations Worth Flagging

The primary efficacy analysis was sound: intention-to-treat, hierarchical testing to control for multiple endpoints, and pre-specified subgroup analyses. Still, several statistical features deserve scrutiny.

Relative vs. absolute risk. The 70% relative risk reduction (RRR) corresponds to an absolute risk reduction (ARR) of 7.6 percentage points for morphometric vertebral fractures over three years. The number needed to treat (NNT) is approximately 13. That NNT is genuinely impressive for an osteoporosis drug, but the 70% headline figure overstates the clinical impact if read without context.

Morphometric vertebral fractures as the primary endpoint. These are fractures identified by scheduled spinal radiographs, not by patient symptoms. Many morphometric fractures are clinically silent. The trial also reported clinical vertebral fractures (symptomatic), which showed a 77% reduction, and hip fractures, which showed a 41% reduction (p = 0.002). The hip fracture result is arguably more clinically meaningful, but it was a secondary endpoint with a smaller event count.

HealthRX Limitation Severity Framework. We grade each limitation on a three-tier scale: (1) minor, meaning unlikely to change clinical conclusions; (2) moderate, meaning the limitation could meaningfully alter the expected benefit in specific populations; (3) major, meaning prescribers should actively account for this gap when making decisions. Enrollment bias rates moderate because the high-risk enrichment inflates absolute benefit estimates for average-risk patients. The morphometric primary endpoint rates minor because secondary clinical endpoints were also positive. The ethnic homogeneity rates moderate for non-white patient populations.

Multiplicity of secondary endpoints. The trial tested vertebral, hip, non-vertebral, clinical, and all-clinical-fracture endpoints. A hierarchical gatekeeping procedure controlled the family-wise error rate. However, some subgroup analyses (e.g., fracture reduction by age tertile, by baseline T-score stratum) were exploratory. Post-publication summaries sometimes cite these exploratory subgroups as though they carry the same evidentiary weight as the primary result.

Dropout and missing radiographs. Approximately 24% of randomized participants did not have evaluable spine radiographs at 36 months. The investigators used multiple imputation and last-observation-carried-forward sensitivity analyses, both of which supported the primary finding. But a 24% radiograph-missing rate is substantial. If dropout was differentially related to fracture occurrence (e.g., sicker patients leaving the study), residual bias is possible.

Acute-Phase Reactions and the Blinding Question

Within three days of the first infusion, roughly 32% of zoledronic acid recipients experienced an acute-phase reaction: fever, myalgia, arthralgia, headache. This rate dropped to about 7% after the second infusion and 3% after the third. In the placebo arm, the rate was around 7% at each time point.

This discrepancy raises a practical concern about blinding integrity. A participant who develops fever and body aches 24 hours after infusion may reasonably guess she received active drug. If unblinding influences subsequent health-seeking behavior, fracture reporting, or adherence to calcium and vitamin D, it introduces detection bias. The HORIZON-PFT publication did not report a formal assessment of blinding success (e.g., asking participants or investigators to guess allocation).

The trial was funded by Novartis Pharmaceuticals, the manufacturer of Reclast. Several lead authors disclosed consulting fees, lecture honoraria, or grant support from Novartis. The sponsor participated in study design, data collection, data analysis, and manuscript preparation, as stated in the original publication.

This does not automatically invalidate the findings. Large osteoporosis trials require resources that typically only pharmaceutical companies can provide. However, sponsor-conducted data analysis, without independent statistical verification publicly disclosed at the time of publication, is a known source of interpretive caution in evidence-based medicine. The subsequent FDA medical review for Reclast provided an independent re-analysis that broadly confirmed the efficacy findings, which strengthens confidence.

Safety Signals That Emerged After Publication

Atrial fibrillation. HORIZON-PFT reported a statistically significant increase in serious atrial fibrillation events (1.3% vs 0.5%; p <0.001). This signal was unexpected. Subsequent analyses, including pooled data and the HORIZON Recurrent Fracture Trial, did not consistently replicate the finding. The FDA ultimately concluded that a causal relationship was not established, but the signal remains noted in the Reclast prescribing information.

Osteonecrosis of the jaw (ONJ). One case of ONJ was reported in the zoledronic acid group during the trial. The low incidence in an osteoporosis-dose trial is consistent with later epidemiologic data suggesting ONJ risk is primarily dose-dependent and much higher with oncology-dose IV bisphosphonates. Still, the trial was not powered to detect rare events occurring at rates below 1 in 1,000.

Renal safety. Transient increases in serum creatinine were more common with zoledronic acid. Patients with creatinine clearance below 30 mL/min were excluded. In real-world use, patients with borderline renal function are sometimes infused, a population for which HORIZON-PFT provides no safety data. Post-marketing reports of acute renal failure led to updated labeling recommending adequate hydration and monitoring of creatinine.

What Post-Publication Commentary Surfaced

Letters to the editor following the 2007 NEJM publication raised several points.

Clinicians questioned whether the acute-phase reaction rate would reduce real-world adherence. Oral bisphosphonate non-adherence is well-documented (roughly 50% discontinuation at one year). One theoretical advantage of yearly IV dosing is bypassing daily or weekly pill burden. But if patients refuse a second infusion after a flu-like reaction, the adherence advantage is partially lost. The HORIZON extension studies later showed that the acute-phase reaction did decrease with subsequent doses, and three-year treatment completion in the extension was reasonable.

Others noted the absence of a head-to-head comparison with oral alendronate, the dominant standard of care at the time. HORIZON-PFT was placebo-controlled. Clinicians wanting to know whether IV zoledronic acid is more effective than weekly alendronate cannot answer that question from this trial. Indirect comparisons using separate trials (HORIZON-PFT vs. the Fracture Intervention Trial for alendronate) are hypothesis-generating at best.

Duration of Follow-Up and Long-Term Questions

Three years is standard for a registration trial in osteoporosis. But bisphosphonates accumulate in bone and continue to exert effects for years after discontinuation. HORIZON-PFT, by design, cannot address:

Optimal duration of treatment before a drug holiday.
Whether fracture protection persists after stopping at three years.
Long-term risk of atypical femoral fractures, which typically emerge after 5 to 10 years of bisphosphonate exposure.

Extension data from HORIZON (6 years total) and the FLEX trial for alendronate have partially addressed these questions, but the original key trial is silent on them.

Bottom Line for Prescribers

HORIZON-PFT remains a well-conducted, large, key trial with clinically meaningful fracture reductions across multiple skeletal sites. Its limitations are real but graded. The enrollment bias toward high-risk, predominantly white women means prescribers should temper absolute benefit expectations in lower-risk or non-white patients. The morphometric primary endpoint, the acute-phase reaction blinding concern, the sponsor's role in data analysis, and the atrial fibrillation signal are all worth knowing. None of them erase the efficacy finding, but together they form the honest fine print that a prescriber should carry alongside the 70% headline.

Frequently asked questions

›

References

Black DM, Delmas PD, Eastell R, et al. Once-yearly zoledronic acid for treatment of postmenopausal osteoporosis. N Engl J Med. 2007;356(18):1809-1822. PubMed
Lyles KW, Colón-Emeric CS, Magaziner JS, et al. Zoledronic acid and clinical fractures and mortality after hip fracture (HORIZON-RFT). N Engl J Med. 2007;357(18):1799-1809. PubMed
Black DM, Reid IR, Boonen S, et al. The effect of 3 versus 6 years of zoledronic acid treatment of osteoporosis: a randomized extension to the HORIZON-PFT. J Bone Miner Res. 2012;27(2):243-254. PubMed
Reclast (zoledronic acid) prescribing information. Novartis Pharmaceuticals. FDA Label
Cosman F, de Beur SJ, LeBoff MS, et al. Clinician's guide to prevention and treatment of osteoporosis. Osteoporos Int. 2014;25(10):2359-2381. PubMed
Black DM, Cummings SR, Karpf DB, et al. Randomised trial of effect of alendronate on risk of fracture in women with existing vertebral fractures (FIT). Lancet. 1996;348(9041):1535-1541. PubMed