Inside the FIT (Fracture Intervention Trial) Methodology: What Most Summaries Skip

At a glance

| Detail | Value | |---|---| | Trial name | Fracture Intervention Trial (FIT), Vertebral Fracture Arm | | N | 4,432 | | Intervention | Alendronate 5 mg/day for 2 years, then 10 mg/day | | Comparator | Matching placebo | | Duration | 36 months (median follow-up) | | Primary endpoint | New morphometric vertebral fractures | | Key result | 47% relative risk reduction in vertebral fractures (RR 0.53 to 95% CI 0.41, 0.68) | | Publication | JAMA 1998; Cummings et al. |

The Two-Part Structure Most Summaries Collapse

FIT was not one trial. It was two parallel, pre-specified arms with distinct enrollment criteria. The Vertebral Fracture Arm (FIT-1) enrolled 2,027 women who already had at least one prevalent vertebral fracture at baseline. The Clinical Fracture Arm (FIT-2) enrolled 4,432 women with low femoral neck BMD (T-score ≤ −1.6) but no baseline vertebral fracture. Most secondary literature cites only FIT-1 when reporting the 47% figure. FIT-2, published separately in 1998, showed a smaller and non-significant reduction in clinical fractures among women without prevalent fractures, unless BMD was below a T-score of −2.5.

This distinction matters clinically. The headline result applies most directly to women with existing vertebral deformity, not to the broader population of postmenopausal women with mildly reduced bone density.

Randomization and Allocation Concealment

Randomization used a computer-generated sequence with permuted blocks, stratified by clinical center. Allocation was concealed through a centralized telephone system, meaning neither site investigators nor participants could predict or influence group assignment. This is a stronger method than sealed envelopes, which can be subverted.

All participants received calcium (500 mg/day) and vitamin D (250 IU/day) as background therapy. This was a deliberate choice: investigators wanted to isolate the pharmacologic effect of alendronate from simple nutritional repletion. It also means the trial tested alendronate on top of baseline supplementation, not as a standalone intervention.

Blinding: Harder Than It Sounds for Bisphosphonates

Both participants and investigators were blinded. The placebo tablets were matched in appearance, and the dosing instructions (take on an empty stomach with a full glass of water, remain upright for 30 minutes) applied equally to both arms. This is worth noting because the distinctive dosing ritual of oral bisphosphonates can function as a practical unblinding cue in open-label or single-blind designs.

GI side effects were tracked and occurred at similar rates in both groups, which argues against differential unblinding through adverse-event patterns. The FDA label for alendronate later reflected this GI profile. No formal unblinding assessment was reported, a limitation common to trials of this era but one that weakens certainty about masking integrity.

Inclusion and Exclusion Criteria: Who Was Actually Studied

FIT-1 enrolled postmenopausal women aged 55 to 81 with femoral neck BMD ≤ 0.68 g/cm² (roughly T-score ≤ −1.6 by the Hologic reference) and at least one radiographically confirmed vertebral fracture. Key exclusions: recent bisphosphonate use, conditions affecting bone metabolism (hyperparathyroidism, Paget disease, osteomalacia), and major GI disease.

Three features of this population shape how the result generalizes:

Age range was narrow. Women under 55 and over 81 were excluded. The mean age was 71. Extrapolating to younger postmenopausal women or the very elderly requires caution.

Prevalent fracture was required. This enriched the sample for women at high fracture risk, increasing statistical power but also meaning the NNT applies to a high-risk subgroup. The absolute risk reduction was 7.0 percentage points over three years (placebo 15.0% vs. alendronate 8.0%), translating to an NNT of roughly 14. In a lower-risk group, the NNT would be substantially larger.

Ethnic and geographic homogeneity. Participants were predominantly White women recruited from 11 U.S. clinical centers. The trial does not inform bisphosphonate efficacy in men, premenopausal women, or non-White populations with different fracture epidemiology.

The Primary Endpoint: Morphometric Vertebral Fractures

The choice of morphometric (radiographic) vertebral fracture as the primary endpoint deserves close attention. Morphometric fractures are identified by serial lateral spine radiographs, with fracture defined as a ≥20% reduction in vertebral body height (anterior, middle, or posterior) compared to baseline.

This approach captures fractures that are clinically silent. Roughly two-thirds of vertebral fractures produce no acute symptoms. Using morphometric assessment increases event capture and statistical power, but it also means the primary endpoint includes events that may not have been noticed by the patient.

Clinical vertebral fractures (those producing symptoms leading to medical attention) were a secondary endpoint. Alendronate reduced clinical vertebral fractures by 55%, a larger relative effect, though with wider confidence intervals due to fewer events.

Radiographic reading was performed centrally by trained readers blinded to treatment assignment and temporal sequence (baseline vs. follow-up films were read in random order). This centralized, blinded adjudication is a significant methodological strength. Site-level reading with knowledge of treatment assignment would introduce measurement bias. The original FIT publication details this adjudication process.

The Mid-Trial Dose Escalation

Participants received alendronate 5 mg/day for the first two years, then 10 mg/day for the third year. This dose change was implemented based on emerging phase II dose-finding data suggesting that 10 mg produced greater BMD gains at the spine and hip than 5 mg.

Protocol amendments mid-trial introduce interpretive complexity. The 47% vertebral fracture reduction reflects a blended exposure: two years at 5 mg followed by one year at 10 mg. It is not possible to attribute the effect cleanly to either dose. The approved dose of alendronate for osteoporosis treatment is 10 mg daily (or 70 mg weekly), partly informed by this trial and subsequent data. Whether the full three-year benefit would have been observed at 5 mg throughout remains unknown.

Statistical Framework and Power

The primary analysis used an intention-to-treat framework: all randomized participants were included regardless of adherence or discontinuation. The pre-specified significance level was α = 0.05 (two-sided), and the trial was powered to detect a 40% relative risk reduction in morphometric vertebral fractures with 90% power.

Survival analysis (Kaplan-Meier and Cox proportional hazards) was used for time-to-first-fracture endpoints. The morphometric vertebral fracture endpoint used a life-table approach with interval censoring (radiographs were obtained at baseline, 24 months, and 36 months, so the exact fracture date was unknown).

Interval censoring is a subtle but important feature. Because vertebral fractures were assessed at fixed timepoints rather than continuously, the analysis captured whether a fracture occurred between visits, not when it occurred. This limits the ability to estimate time-dependent hazard patterns or assess early vs. late treatment effects with precision.

What the Estimand Actually Tells You

The trial estimated the effect of being assigned to alendronate (ITT), not the effect of taking alendronate consistently. Adherence was approximately 82% over three years, with similar discontinuation rates between groups. The ITT estimate is therefore conservative: it dilutes the true pharmacologic effect by including non-adherent participants.

A per-protocol analysis was also reported and showed a slightly larger treatment effect, consistent with this expectation. The per-protocol estimate is more susceptible to bias (adherent patients differ systematically from non-adherent ones), so the ITT result is generally preferred for causal inference about the treatment policy.

No formal estimand framework (as defined in ICH E9(R1)) was applied, as that guidance postdates the trial by two decades. Under the current framework, the primary estimand would likely be classified as a "treatment policy" strategy for intercurrent events like discontinuation.

Limitations the Authors Acknowledged

The original publication and subsequent commentary noted several limitations:

• Duration. Three years does not address long-term safety or sustained fracture reduction beyond that window. Later extension studies (FLEX trial, 10-year data) partially addressed this, showing attenuation of benefit after discontinuation.
• Non-vertebral fractures. Hip fracture reduction (51%) was a secondary endpoint and the trial was not independently powered for it. The hip fracture finding is suggestive but not confirmatory from this trial alone.
• Generalizability. The population was restricted to postmenopausal women with very low BMD and prevalent fractures. Results should not be assumed to apply to men, premenopausal women, or glucocorticoid-induced osteoporosis.
• GI tolerability. While upper GI events were similar between groups, the controlled trial setting (with strict dosing instructions and exclusion of patients with active GI disease) may not reflect real-world tolerability.

How Design Choices Shape Interpretation

The convergence of population enrichment (prevalent fractures), sensitive endpoint capture (morphometric assessment), centralized blinding, and adequate power produced a clean, internally valid result. The 47% relative risk reduction is well-supported within the studied population.

The practical clinical question, whether to prescribe alendronate for a specific patient, requires translating this result across differences in baseline risk, age, comorbidities, and willingness to adhere to dosing requirements. The 2020 AACE/ACE guidelines and 2022 ACP recommendations both cite FIT as foundational evidence but apply it within broader treatment algorithms that incorporate FRAX scoring, BMD thresholds, and competing options like denosumab or romosozumab.

FIT established that oral bisphosphonate therapy meaningfully reduces vertebral fractures in high-risk postmenopausal women. The methodology, particularly the centralized radiographic adjudication and enriched population, is what made that conclusion defensible. Understanding where the methodology is strong (internal validity, blinding, power) and where it is limited (generalizability, duration, dose change) is what separates reading the abstract from reading the trial.

Frequently asked questions

›

›

›

›

›

›

›

›

›

›

References

• Cummings SR, Black DM, Thompson DE, et al. Effect of alendronate on risk of fracture in women with low bone density but without vertebral fractures: results from the Fracture Intervention Trial. JAMA. 1998;280(24):2077-2082. 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. Lancet. 1996;348(9041):1535-1541. PubMed
• Black DM, Schwartz AV, Ensrud KE, et al. Effects of continuing or stopping alendronate after 5 years of treatment: the Fracture Intervention Trial Long-term Extension (FLEX). JAMA. 2006;296(24):2927-2938. PubMed
• Alendronate sodium prescribing information. U.S. Food and Drug Administration. FDA Label
• Camacho PM, Petak SM, Binkley N, et al. American Association of Clinical Endocrinologists/American College of Endocrinology clinical practice guidelines for the diagnosis and treatment of postmenopausal osteoporosis, 2020 update. Endocr Pract. 2020;26(Suppl 1):1-46. PubMed
• Qaseem A, Hicks LA, Etxeandia-Ikobaltzeta I, et al. Pharmacologic treatment of primary osteoporosis or low bone mass to prevent fractures in adults: a living clinical guideline from the American College of Physicians. Ann Intern Med. 2023;178(1):79-88. PubMed