This is one of the highest-stakes decisions in adult medicine: a broken hip in your seventies carries roughly the one-year death rate of a heart attack, and the right treatment cuts that risk by something like half. The evidence base is one of the strongest in bone medicine — three decades of large trials counting actual fractures, not just bone scans. The catch is cost and complexity. The right drug for a very-high-risk patient can be a $20 000 course of monthly injections, followed by another year of locking-in therapy. Cheaper paths exist and work for the majority — but the choice has to be made deliberately, not by clinic default.
Bone isn't a static structure. It's constantly being torn down and rebuilt by two opposed cell types — osteoclasts dig out old bone matrix, osteoblasts lay down new. In a healthy adult the two are matched. In osteoporosis, especially after menopause, resorption pulls ahead. Net bone is lost faster than it's added back, and the accumulated deficit shows up as a low T-score on a DXA scan and eventually as a fracture from a fall that wouldn't have broken a healthier bone.
Drug treatment intervenes at one of the two ends of that loop. Bone-preservers — the bisphosphonates (alendronate, risedronate, zoledronic acid) and denosumab — turn down the osteoclasts. Bisphosphonates stick to bone and poison any osteoclast that tries to chew through them; denosumab is an antibody that blocks the signal osteoclasts need to mature in the first place (Cummings 2009). Bone turnover slows, repair fills in faster than damage opens up, and density creeps up over years.
Bone-builders work the other end. Teriparatide is a fragment of parathyroid hormone; abaloparatide is a closely related synthetic. Given as a daily injection, both push osteoblasts into overdrive and make new bone faster than usual (Neer 2001). Romosozumab is different again — an antibody against a molecule called sclerostin that bones use to brake their own building program. Block it and formation jumps, resorption falls at the same time, for a one-shot window of about a year (Cosman 2016).
That mechanical difference is why the order matters. Bone-builders create a window where the skeleton is actively laying down fresh matrix. Closing that window with a bone-preserver locks the new bone in. Closing it with nothing — or never opening it because you started with a bone-preserver instead — leaves the gain on the table.
How well it works, in actual broken bones
Bone medicine has the unusual luxury of measuring its endpoint directly: did the patient break a bone or not. Three decades of large trials, with thousands of women each, give us reasonably clean numbers.
For the older bone-preservers: weekly alendronate cuts new spine fractures by about half in women who already had one (Black 1996). A once-yearly zoledronic acid infusion does better — about 70% off spine fractures and 41% off hip fractures over three years (Black 2007). Denosumab, given every six months, lands in the same range: 68% spine, 40% hip (Cummings 2009).
The bone-builders pulled ahead of the bone-preservers in the late 2010s for the highest-risk patients. The trial that pinned it down was ARCH: women with osteoporosis and a prior fragility fracture given a year of monthly romosozumab and then a year of alendronate had about half the new spine fractures, a quarter fewer of all clinical fractures, and 38% fewer hip fractures than women given two straight years of alendronate (Saag 2017).
Teriparatide showed the same pattern against the older bone-preserver risedronate in VERO: 56% fewer new vertebral fractures in women with severe osteoporosis over two years (Kendler 2018). And the STRUCTURE trial — women already on a bisphosphonate, switched to either romosozumab or teriparatide — showed romosozumab actually gained hip density while teriparatide lost a touch of it (Langdahl 2017). The clinical signal is consistent: in the patients at highest risk, opening with a bone-builder beats opening with a bone-preserver.
The major clinical bodies have caught up. The 2020 update from the American Association of Clinical Endocrinologists put the recommendation in plain text: bone-builder first for very-high-risk patients (Camacho 2020). The Endocrine Society (Eastell 2019) and the Bone Health and Osteoporosis Foundation (LeBoff 2022) agree.
What untreated looks like in real life
Osteoporosis is silent. No pain, no swelling, no morning stiffness — the disease announces itself with a fracture. For most readers picturing this, the typical case is a 65-to-75-year-old whose first hint of trouble was a wrist broken in a slip on ice, or a vertebra that quietly compressed during a normal day.
Year one untreated. Likely nothing felt at all. Bone is being lost faster than it's replaced, and the next DXA in two years will show another quarter to half a T-score point gone. The first fracture, when it comes, roughly doubles the chance of the next one.
Five years untreated. The vertebral fractures most people don't notice — sneeze fractures, lift-a-grandchild fractures — start to add up. Friends comment that you've gotten shorter. Bras that fit don't anymore; trouser legs sit differently. The chronic mid-back ache that you assumed was a desk job is sometimes a settled compression fracture pressing on a nerve root.
Ten to fifteen years untreated, post-menopausal. The hip fracture is the dominant event. In US registry data, roughly one in four adults over 65 who breaks a hip is dead within a year — comparable to a serious heart attack (Brauer 2009). About a third of survivors don't go home; they go to assisted living. The fracture itself isn't the killer. The pneumonia after surgery, the clot from immobility, the cascade of new medications and lost independence — that's the killer.
This is what the 90-day decision is buying you out of. Not a feeling that lifts; a future event that doesn't happen.
What the 90 days actually look like
The window splits cleanly into three roughly equal pieces: confirm, stratify, choose.
Days 0–30: confirm and look for a cause
The diagnosis is usually a DXA T-score of −2.5 or below at the spine, hip, or femoral neck, or a low-trauma fracture in someone over 50. Either is enough; you don't need both. About one in four to one in three cases — more in men, more in premenopausal women — has a treatable cause sitting underneath: low vitamin D, an underactive thyroid, undiagnosed coeliac disease, a slow-burning bone marrow problem, low testosterone, hidden steroid exposure. The labs that catch most of these are inexpensive and standard (Camacho 2020): calcium, vitamin D, parathyroid hormone, thyroid, kidney function, blood count, and a 24-hour urine for calcium. Conditional adds depending on the picture: tests for myeloma, coeliac, low testosterone in men, high cortisol if Cushing's is on the table. Two contributors the labs won't flag are worth a look here too: years on a daily acid blocker, and a heavy preformed-vitamin-A intake from supplements or liver — both can chip away at bone over time, and both are easy to miss.
Imaging that's easy to miss: a vertebral fracture assessment on the same DXA machine, or a lateral spine X-ray. About one in five women over 70 has a silent vertebral fracture nobody's noticed (LeBoff 2022). Finding one shifts a patient from osteoporosis to very-high-risk, which changes the treatment recommendation.
Days 30–60: size up the actual risk
The T-score alone isn't the decision. A free online calculator called FRAX combines T-score with age, prior fracture, family history of hip fracture, smoking, alcohol, steroid use, rheumatoid arthritis, and other conditions to produce a ten-year probability of a major fracture and a separate ten-year probability of a hip fracture (Kanis 2008).
The cutoffs that matter for treatment choice (Camacho 2020):
- Very-high-risk — a T-score of −3.0 or below, a fracture in the last year (especially hip or spine), more than one fracture ever, a fracture that happened while already on a bone-preserver, or a FRAX ten-year hip fracture probability above 4.5% / major-fracture probability above 30%. Bone-builder first.
- High-risk — meets the diagnostic criteria but doesn't cross the very-high-risk threshold. Bone-preserver first.
Days 60–90: start, and schedule the follow-up
When the default plan doesn't apply
Each drug class has its own list of people it isn't right for. The clinician walks the list; the patient should know the highlights.
Teriparatide and abaloparatide are avoided in anyone with a history of bone radiation, Paget's disease, unexplained elevations in alkaline phosphatase, or active high calcium. The osteosarcoma warning these drugs carried for years came from a rat study at very high doses; the human signal has not materialised over two decades of use, and the warning has been softened in updated labelling.
Bisphosphonates are avoided when kidney function is poor (eGFR below roughly 30–35), in active oesophageal disease for the oral forms, and in people with low blood calcium that hasn't been corrected. The rare complications people worry about — jawbone necrosis after dental procedures, atypical thigh-bone fractures after many years of use — are real but uncommon at typical osteoporosis doses (jaw issue: roughly 1 in 10 000 to 1 in 100 000 patient-years; atypical fracture: roughly 1 in 1000 after eight or more years of continuous use) (Khosla 2017). The fracture risk being treated is several orders of magnitude larger.
Denosumab requires a hard commitment to the every-six-months schedule or to a planned exit ramp via zoledronic acid. Anyone who can't reliably make the dosing schedule — frail elderly with unreliable transport, patients who travel for long stretches — is better served by an annual IV infusion or oral bisphosphonate.
Where this goes wrong in practice
The denosumab cliff. Denosumab works as long as you keep getting the injection every six months. Miss a dose by more than a few weeks and the bone gain not only reverses but rebounds — within 8 to 18 months of the missed dose, a substantial fraction of patients sustain multiple new vertebral fractures, often in people who'd been doing fine on the drug for years (Cummings 2018). The European Calcified Tissue Society now treats this as a hard rule rather than a preference: if denosumab is being stopped for any reason, a single dose of zoledronic acid six months later locks the gain in and prevents the rebound (Tsourdi 2021). Anyone going on denosumab needs to understand that they're signing up for either indefinite continuation or a planned exit ramp.
Starting with the wrong drug. The mechanical penalty for starting with a bone-preserver and then switching to a bone-builder is real and measurable. Bone that's been quieted by bisphosphonates responds less to the subsequent bone-builder; the spine and hip gains are smaller than they would have been if the order were reversed (Langdahl 2017). This is why the very-high-risk identification in the first 90 days matters — once a patient is on alendronate for a year, that decision has partially been made.
The bone-builder without a follow-on. Stopping teriparatide, abaloparatide, or romosozumab without immediately starting a bone-preserver lets the new bone resorb out within six to twelve months (Cosman 2017). The pattern that works: bone-builder for its allotted window, then antiresorptive consolidation starting within the month.
Not treating after a hip fracture. The largest implementation failure isn't choosing the wrong drug — it's choosing none at all. Across US Medicare data, the share of patients started on osteoporosis therapy after a hip fracture fell from 40% to 21% between 2002 and 2011, exactly the period during which the evidence for treatment got stronger (Solomon 2014). A fragility fracture in someone over 50 is itself a diagnosis. If the post-fracture conversation about treatment never happens, the next fracture is the conversation that does.
What most clinics get wrong
"Bisphosphonate is always the right first move." This was true in 2005. Alendronate was the only well-trialled drug; everyone got alendronate. Twenty years of newer trials changed the picture for the very-high-risk subset, and the guidelines have moved (Camacho 2020). The reason clinic practice is slower to move is partly cost (anabolics require prior authorisation), partly habit, partly that the bone-builder window is only opened well if you identify the very-high-risk patient up front.
"The T-score tells me what to do." A T-score is one input. A 75-year-old who broke a hip last year sits at very-high-risk even if her T-score is −2.3; her FRAX hip risk will be well above the threshold and her treatment plan is the same as someone at −3.5 (Kanis 2008). The reverse also holds: a 55-year-old with a T-score of −2.6 and no other risk factors usually doesn't need an injection course.
"Calcium and vitamin D are the treatment." They aren't. Calcium and vitamin D are baseline for everyone with low bone density, but trials of calcium and D alone in established osteoporosis show modest density change and inconsistent fracture reduction. They support the drug; they don't replace it (LeBoff 2022).
"I'll be on this forever." Bisphosphonates are typically taken for 3–5 years and then paused — they keep working for a while after stopping, because the drug stays bound to bone. Denosumab and the bone-builders are different: they require either continuation or a planned exit ramp. The duration question is part of the original treatment decision, not an afterthought.
"This is a women's disease." About one in three hip fractures in adults over 50 happens in men, and men with osteoporosis are diagnosed later and treated less often — scanning men's bone density on time is its own underserved problem. Secondary causes are also more common in men — low testosterone, chronic steroids, heavy alcohol — and the workup matters more (LeBoff 2022).
What changes if you start
Most osteoporosis treatments don't feel like anything. There's no symptom to relieve. The payoff is a future event that doesn't happen, which is harder to picture than a felt change. So picture the timeline.
Weeks 2–6. Blood markers of bone turnover shift — formation markers rise on a bone-builder, resorption markers fall on a bone-preserver. Nothing the patient feels. The infusion or first injection might leave a couple of days of flu-like aching. Then nothing.
Months 6–12. The fracture-risk curves separate from untreated in the large trials around month six and become clearly different by month twelve (Cosman 2016). The patient still feels nothing — there's no symptom to relieve — but the underlying odds have already shifted. A repeat DXA at 12 months on a bone-builder shows a roughly 13–15% gain at the spine and several percent at the hip (Cosman 2016).
Years 1–3. Continued therapy compounds. By the end of year three a patient who started in the osteoporotic range often crosses into the osteopenic range on T-score — still below average, but no longer below the diagnostic line. Fracture rates in the cohorts continue to track below untreated.
Decade-scale. The visible part is what didn't happen. The spine that didn't quietly collapse — height held, the dowager's kyphosis that runs in the family didn't develop. The hip that didn't break in the bathroom at 82. The Christmas you spent at home instead of in a rehab facility. None of these are events; they're absences. That's what bone medicine sells, and it's harder to feel grateful for than a symptom that left.
Related ground
This entry stops at the 90-day decision. Adjacent topics worth their own reading: the longer-horizon question of when to stop or pause a bisphosphonate (the "drug holiday" question, distinct from the initial choice); the prevention end of the spectrum (DXA screening cadence, weight-bearing exercise programmes, falls prevention in the home); the secondary-cause workups that overlap with this one (vitamin D status, thyroid management, hypogonadism in men, glucocorticoid-induced osteoporosis); and the specific case of post-menopausal hormone therapy, which addresses bone density alongside vasomotor symptoms and sits on a different risk/benefit ledger.
- — Fixing fall risks like positional vertigo matters most when weak bone makes any fall a fracture.
- — Smoking speeds bone loss and slows fracture healing. Quitting is part of any serious bone plan.
- — Years on a daily acid blocker quietly raise fracture risk — worth flagging while you size up bone loss.
- — Too much preformed vitamin A (retinol from supplements and liver) weakens bone over time — worth auditing if you're managing osteoporosis.
- — The estrogen drop at menopause is a major bone-loss driver; HRT started in time protects the skeleton.
- — Progressive resistance training is the exercise side of protecting bone after an osteoporosis diagnosis.
- — Resistance training is the exercise prescription for protecting bone density, especially through menopause.
- — Correcting a vitamin D shortfall is one of the basics underneath any osteoporosis plan.
- — Carrying heavy loads loads the skeleton too, and that mechanical stress is part of keeping bone from thinning further.
- — Paired with resistance training, daily creatine slows the bone loss this protocol is fighting.
- — Weight-bearing loading like rucking is one of the bone-stimulating habits that supports bone density.
- — Vitamin K2 is one of the quiet, decade-scale supports for keeping bone density up.
- — Diagnosis starts with a DEXA bone scan — the same machine reads your muscle and fat too.
- — The T-score that starts this whole decision comes from a DEXA scan — men get told to get one far too late.
- — Undiagnosed celiac is a fixable cause of early bone loss; malabsorption is worth ruling out.
- — The fastest bone loss of a woman's life is the menopause transition — which is why the scan and the symptom timeline belong together.
- — When heavy lifting risks fragile bones, restriction training is a lighter way to hold onto muscle.
- — Rapid weight loss on GLP-1 drugs strips bone along with fat, so fragile bones need the protein-and-lifting plan even more.
Substance and claimed effects
The substance is the first 90 days after an osteoporosis diagnosis — the bounded clinical window in which a patient and clinician (1) confirm the diagnosis and rule out secondary causes, (2) stratify fracture risk, and (3) select and initiate pharmacological therapy. The decision that anchors this window is the sequencing call: anabolic-first (teriparatide, abaloparatide, or romosozumab, followed by an antiresorptive consolidator) versus antiresorptive-first (oral or IV bisphosphonate, or denosumab). Claimed consequences within the catalogue's scoring dimensions: large reduction in fracture risk and mortality (Black 2007, Saag 2017, Cummings 2009); meaningful gain in bone mineral density measured by DXA (Cosman 2016, Neer 2001); preservation of vertebral height and posture over the long run; modest day-to-day burden of injections, oral dosing, and follow-up DXA. Felt experience is otherwise muted — the disease is silent until a fracture, so the payoff of treatment is risk averted rather than symptom relieved.
Evidence by addressing question
Mechanism
Bone is constantly remodelled by a coupled pair of cell lineages: osteoclasts resorb old matrix, osteoblasts lay down new matrix. In osteoporosis, resorption outpaces formation, especially after the oestrogen withdrawal of menopause. The two therapy classes intervene at opposite ends of this loop. Antiresorptives — bisphosphonates (alendronate, risedronate, zoledronic acid) and denosumab (anti-RANKL monoclonal antibody) — suppress osteoclast activity. Bisphosphonates bind hydroxyapatite and poison osteoclasts that ingest them; denosumab binds RANKL and prevents osteoclast differentiation (Cummings 2009, Khosla 2017). Bone turnover slows, the resorption pit closes faster than the formation cycle, and BMD rises modestly over years.
Anabolic agents act differently. Teriparatide (recombinant PTH 1–34) and abaloparatide (PTHrP analogue) exploit a paradox: continuous PTH exposure resorbs bone, but pulsatile daily SC injection stimulates osteoblast precursors and increases bone formation more than resorption (Neer 2001, Miller 2016, Tabacco 2019). Romosozumab is a monoclonal antibody against sclerostin, an osteocyte-derived Wnt-pathway inhibitor; blocking it both stimulates formation and suppresses resorption, a brief dual-action window of about 12 months before the effect attenuates (Cosman 2016, Saag 2017).
The sequencing logic falls out of this mechanism. Anabolics open a remodelling window — newly formed bone matrix that has not yet mineralised. If that window closes with a strong antiresorptive (typically zoledronic acid or denosumab), the gain is locked in and BMD continues to rise (Cosman 2017, Leder 2015). If the window closes with nothing, BMD reverts toward baseline within months. Conversely, starting with a bisphosphonate and then switching to an anabolic blunts the anabolic response — bisphosphonate-conditioned bone has a smaller formation-response capacity, demonstrated head-to-head in the STRUCTURE trial (Langdahl 2017). Mechanism, not just trial outcome, drives the order.
Evidence
Antiresorptive pivotal trials. The Fracture Intervention Trial (Black 1996) randomised 2027 women with low BMD and prior vertebral fracture to alendronate or placebo over 3 years: ~47% reduction in new clinical vertebral fractures, ~51% reduction in hip fractures. The HORIZON-PFT trial (Black 2007) tested annual IV zoledronic acid in 7765 postmenopausal women: 70% reduction in vertebral fractures, 41% reduction in hip fractures, 25% reduction in any clinical fracture over 3 years. FREEDOM (Cummings 2009) tested denosumab in 7868 postmenopausal women: 68% vertebral, 40% hip, 20% non-vertebral reduction over 3 years.
Anabolic pivotal trials. The teriparatide pivotal (Neer 2001) randomised 1637 postmenopausal women with prior vertebral fracture: 65% reduction in new vertebral fractures, 53% reduction in non-vertebral fractures over a median 19 months. The trial was stopped early because of an osteosarcoma signal in lifetime-exposed rats, never observed in humans but driving the 2-year cumulative cap. The abaloparatide ACTIVE trial (Miller 2016) showed an 86% relative reduction in new vertebral fractures and 43% reduction in non-vertebral fractures versus placebo at 18 months. FRAME (Cosman 2016) randomised 7180 postmenopausal women to 12 months of romosozumab versus placebo, then 12 months of denosumab in both arms: 73% reduction in vertebral fractures at 12 months and persistent advantage at 24 months.
Head-to-head sequencing. The ARCH trial (Saag 2017) randomised 4093 postmenopausal women with prior fragility fracture to 12 months of romosozumab followed by alendronate versus 24 months of alendronate. At 24 months: 48% lower risk of new vertebral fracture, 27% lower risk of clinical fracture, 38% lower risk of hip fracture for the romosozumab-first arm. The STRUCTURE trial (Langdahl 2017) compared romosozumab to teriparatide in women already on bisphosphonates: romosozumab produced superior total hip BMD gain (+2.6% vs −0.6% at 12 months). VERO (Kendler 2018) randomised 1360 postmenopausal women with severe osteoporosis to teriparatide or risedronate for 24 months: 56% lower vertebral fracture rate on teriparatide.
BMD outcomes by class. Aggregated across trials: oral bisphosphonates produce ~5–6% lumbar spine BMD gain over 3 years; zoledronic acid ~6%; denosumab ~9% at 3 years and ~21% at 10 years on continued therapy; teriparatide ~9–13% over 18–24 months at the lumbar spine but only modest hip gain; abaloparatide ~10–11% at the spine; romosozumab ~13–15% at the spine and ~6–7% at the total hip over 12 months — the steepest known BMD response (Bouxsein 2019, Khosla 2017).
Practice / clinical consensus. The 2020 AACE/ACE guideline (Camacho 2020) introduced the very-high-risk category and explicitly recommended anabolic-first therapy (teriparatide, abaloparatide, or romosozumab) for this group, followed by antiresorptive consolidation. The Endocrine Society 2019/2020 update (Eastell 2019, Shoback 2020) followed suit. The 2022 Bone Health and Osteoporosis Foundation clinician's guide (LeBoff 2022) endorses the same logic. USPSTF 2018 (USPSTF 2018) covers screening, not treatment, but anchors the diagnostic gate.
Protocol
Days 0–30: confirm and characterise. Diagnosis is usually a DXA T-score ≤ −2.5 at the lumbar spine, total hip, or femoral neck, or a fragility fracture (low-trauma fracture in someone over 50). A 90-day workup confirms the diagnosis, rules out secondary causes (responsible for ~20–30% of cases in men and premenopausal women), and stratifies fracture risk. Standard baseline labs (Camacho 2020, LeBoff 2022): serum calcium, albumin, creatinine, 25-hydroxyvitamin D, intact PTH, TSH, CBC, alkaline phosphatase, 24-hour urinary calcium and creatinine. Conditional adds: serum protein electrophoresis and free light chains (rule out multiple myeloma) for unexplained osteoporosis especially with anaemia; coeliac serology; morning testosterone in men; 24-hour urinary free cortisol if Cushing features. Imaging: vertebral fracture assessment (VFA) on DXA or lateral spine X-ray to detect silent vertebral fractures — present in ~20% of women over 70 even without back pain.
Days 30–60: stratify and choose. FRAX (Kanis 2008) generates 10-year probabilities of major osteoporotic fracture (MOF) and hip fracture using femoral neck BMD and clinical risk factors. The very-high-risk thresholds (Camacho 2020): T-score ≤ −3.0, recent prior fracture (especially hip or vertebral within the last 12 months), multiple fractures, fracture while on antiresorptive therapy, or MOF probability > 30% / hip probability > 4.5%. Very-high-risk patients get anabolic-first; moderate-risk get bisphosphonate-first.
Days 60–90: initiate and schedule follow-up. Anabolic regimens: teriparatide 20 µg SC daily for up to 24 months; abaloparatide 80 µg SC daily for up to 24 months; romosozumab 210 mg SC monthly for 12 months. Each is followed by an antiresorptive consolidator — typically zoledronic acid 5 mg IV annually or denosumab 60 mg SC every 6 months — initiated within 1 month of the last anabolic dose to lock in the gain (Leder 2015, Cosman 2017). Antiresorptive-first regimens: alendronate 70 mg oral weekly, risedronate 35 mg oral weekly, ibandronate, or zoledronic acid 5 mg IV annually; denosumab 60 mg SC q6 months as the injectable alternative. Calcium 1000–1200 mg/day and vitamin D 800–1000 IU/day are baseline adjuncts. Reassess BMD at 1–2 years.
Contraindications
Per-agent contraindications are central to the choice. Romosozumab carries an FDA boxed warning for cardiovascular events: in ARCH (Saag 2017) there was a numerically higher rate of serious cardiovascular adverse events on romosozumab versus alendronate (2.5% vs 1.9%). Avoid in patients with MI or stroke in the past 12 months. Teriparatide and abaloparatide carried a boxed osteosarcoma warning from rat data (removed for teriparatide in 2020 after 15+ years of post-marketing data showed no human signal); both still avoided in patients with prior radiation to bone, Paget's disease, unexplained alkaline phosphatase elevation, or hypercalcaemia. Bisphosphonates are avoided in eGFR < 30–35 mL/min, severe oesophageal disease (oral forms), or hypocalcaemia. Rare complications with cumulative exposure: osteonecrosis of the jaw (~1 in 10 000–100 000 person-years) and atypical femur fracture (~1 in 1000 after 8+ years) (Khosla 2017). Denosumab requires uninterrupted dosing — see failure-modes.
Misconceptions
"Bisphosphonate is always first." This is the inheritance of two decades of practice when alendronate was the only well-trialled option. ARCH and VERO change the picture for very-high-risk patients (Saag 2017, Kendler 2018): in those subgroups, anabolic-first reduces fractures more than antiresorptive-first, and the order is not interchangeable. Mechanism also matters — STRUCTURE (Langdahl 2017) shows pre-treatment with a bisphosphonate blunts the subsequent anabolic response.
"T-score is the decision." The T-score is one input. FRAX integrates age, prior fracture, parental hip fracture, smoking, glucocorticoid use, secondary osteoporosis, alcohol, and rheumatoid arthritis. A 75-year-old with a T-score of −2.3 and a recent hip fracture is very-high-risk despite not meeting the −2.5 DXA threshold; treatment is fully indicated (Camacho 2020).
"Treatment fixes the bone, then I can stop." Bisphosphonates have a long bone half-life and tolerate drug holidays (typically after 5 years oral or 3 years IV in lower-risk patients). Denosumab and anabolics do not — stopping without a bridge produces rebound bone loss and, for denosumab, a documented surge in multiple vertebral fractures (Cummings 2018).
"Calcium and vitamin D are the treatment." They are baseline adjuncts, not therapy. Trials of calcium + vitamin D alone in established osteoporosis show modest BMD effects and inconsistent fracture reduction; they do not substitute for pharmacological therapy in someone diagnosed by DXA T-score ≤ −2.5 or fragility fracture (LeBoff 2022).
Failure-modes
Denosumab discontinuation without consolidation. Stopping denosumab without an immediate transition to a bisphosphonate produces a rapid loss of all accrued BMD and a documented rebound surge of multiple vertebral fractures within 8–18 months of the missed dose. A post hoc FREEDOM analysis (Cummings 2018) showed the annualised rate of vertebral fractures after discontinuation was 7.1%, with the majority experiencing multiple vertebral fractures. The European Calcified Tissue Society position (Tsourdi 2021) explicitly recommends a single dose of zoledronic acid 6 months after the last denosumab injection — this is a hard rule, not a preference.
Bisphosphonate-then-anabolic order. Prior bisphosphonate exposure attenuates the BMD response to subsequent teriparatide and romosozumab. STRUCTURE (Langdahl 2017) and DATA-Switch (Leder 2015) document this directly. The clinical implication: when a patient is identified as very-high-risk at diagnosis, starting them on a bisphosphonate "to be safe while we figure things out" forecloses options.
Anabolic without consolidation. Stopping teriparatide, abaloparatide, or romosozumab without transitioning to an antiresorptive produces rapid BMD loss within 6–12 months. The anabolic creates new bone that has not yet matured; without the antiresorptive to suppress remodelling, that bone is preferentially resorbed (Cosman 2017).
Under-treatment after a hip fracture. The largest implementation failure is that most patients hospitalised with a hip fracture never receive osteoporosis therapy. A large US study (Solomon 2014) of Medicare hip-fracture patients between 2002–2011 found osteoporosis medication initiation declined from 40% to 21% over the decade — the opposite of what the trial evidence supports. The first 90 days after a fragility fracture is the window in which this gap closes or doesn't.
Alternatives
Lifestyle interventions — weight-bearing and resistance exercise, fall prevention, smoking cessation, alcohol moderation, adequate calcium and vitamin D — reduce fracture risk modestly and are recommended as adjuncts for everyone (LeBoff 2022). They do not replace pharmacotherapy at established osteoporosis. Hormone replacement therapy (HRT) reduces fracture risk in postmenopausal women but is no longer first-line for osteoporosis given the WHI safety findings; it remains reasonable in women within 10 years of menopause being treated for vasomotor symptoms with osteoporosis as a co-indication. Raloxifene (a SERM) reduces vertebral fracture risk modestly with secondary benefit on breast cancer risk reduction; weaker on hip fractures, so reserved for women whose dominant risk is vertebral and who can't tolerate first-line agents (Eastell 2019). Calcitonin is essentially historical — modest effect, withdrawn in Europe over a malignancy signal.
Stakes
Untreated osteoporosis is silent until a fracture. The first fragility fracture, often a wrist or a vertebral compression, roughly doubles the risk of the next. A hip fracture in someone over 65 carries a 12-month all-cause mortality of approximately 20–25% in US registries (Brauer 2009); roughly a third of survivors lose independence and require long-term assisted care. Vertebral fractures, often subclinical, accumulate and produce the kyphotic posture, height loss, and chronic mid-back pain associated with the disease — many are detected only on incidental imaging. The clinical evidence shows the 90-day decision window is consequential: starting on the right therapy at diagnosis reduces fracture incidence by ~50–70% across the major trials, and reduces hip fracture mortality in observational follow-up; under-treatment of the post-fracture population is the dominant real-world failure (Solomon 2014).
Payoff
BMD response is detectable on DXA at 12 months, often earlier on bone turnover markers (CTX falls within weeks on antiresorptives; P1NP rises within weeks on anabolics). Fracture incidence starts to diverge from placebo within 6 months in FRAME and FREEDOM, with statistically significant separation by 12 months. For the right candidate on anabolic-first therapy, expect ~13–15% lumbar spine BMD gain by month 12 on romosozumab (Cosman 2016); after antiresorptive consolidation, BMD continues to rise modestly. Over 5+ years on continued therapy, a substantial fraction of patients move from osteoporotic to osteopenic T-score range — they remain at elevated absolute risk versus a never-osteoporotic peer, but the trajectory of accumulating fractures is bent. The felt experience is muted: no symptom relief, just risk averted. Posture preservation, retained height, and the avoided hospital admission are the long-shadow consequences.
Practicalities
Cost. Oral bisphosphonates are generic and cost roughly $100–300/year out-of-pocket without insurance, often $0–50 with Medicare Part D. Annual IV zoledronic acid is ~$200–1000 depending on infusion-centre billing. Denosumab is ~$1500–2500/dose at twice yearly (~$3000–5000/year list). Teriparatide and abaloparatide list at ~$3000–4000/month (course total ~$70 000+). Romosozumab lists at ~$1900/month (12-month course ~$23 000). Insurance coverage for anabolics typically requires documented T-score and prior fracture or failed antiresorptive trial; prior authorisation is the dominant administrative friction.
Logistics. Daily SC self-injection (teriparatide, abaloparatide) requires refrigeration and a once-daily routine; the pen is straightforward. Romosozumab is two SC injections monthly, in clinic. Denosumab is one SC injection every 6 months, in clinic. Zoledronic acid is a single annual IV infusion, in clinic, ~15 minutes; acute-phase reaction (flu-like symptoms) is common after the first dose. Oral bisphosphonates require empty stomach with water and upright posture for 30–60 minutes after — the adherence cost that drove the shift to IV and SC options.
Credibility range
The optimist case
The 2017–2020 wave of trial evidence — ARCH, FRAME, ACTIVE, VERO, STRUCTURE — produced a clean, mechanism-consistent answer for the highest-risk osteoporosis patients: anabolic-first sequencing reduces fractures more than antiresorptive-first, and the magnitude of effect is large (38% lower hip fracture rate in ARCH over 24 months). Romosozumab in particular is the steepest BMD response ever observed in a Phase 3 osteoporosis trial. The mechanism underwrites the trial result: anabolic agents build new bone, antiresorptives prevent loss; building first, then preserving, is the rational order. The major guideline bodies have aligned (Camacho 2020, Eastell 2019, LeBoff 2022). The first 90 days after diagnosis is the lever — the decision made here echoes for years.
The skeptic case
Anabolics are expensive, parenterally administered, and short-course; bisphosphonates are generic, oral, and well-trialled for decades. The cardiovascular safety signal for romosozumab is unresolved — ARCH showed a numerical excess of cardiac events versus alendronate, and the FDA boxed warning is not pro forma. The osteosarcoma signal that drove the teriparatide warning was rat-only but has shaped 20 years of cautious use. Most patients diagnosed with osteoporosis are not very-high-risk; for them, the ARCH advantage is unproven and the cost difference is substantial. Adherence to weekly oral bisphosphonate is mediocre, but adherence to monthly clinic injections is also imperfect, and dropping denosumab mid-course is dangerous. The "first 90 days" framing can be over-medicalised: a 70-year-old with a T-score of −2.6 and no other risks may be reasonably served by an oral bisphosphonate and lifestyle optimisation, not a $23 000 course.
The author's call
The sequencing question is real and the evidence is settled enough to act on: very-high-risk patients should get anabolic-first therapy, and that decision needs to be made in the first 90 days because the order is mechanistically locked in (starting with a bisphosphonate compromises the anabolic later). Most diagnosed patients are not very-high-risk and a bisphosphonate-first plan remains correct for them. The implementation gap — most post-hip-fracture patients getting no treatment at all — is a larger problem than the sequencing debate. Evidence quality is high (5); controversy is moderate (3): the field broadly agrees on the framework, but the romosozumab cardiovascular signal, the cost-effectiveness of anabolics outside very-high-risk, and the duration of antiresorptive therapy ("drug holiday" timing) remain debated.
Stakeholder and incentive map
- Commercial. Amgen (romosozumab Evenity, denosumab Prolia), Lilly (teriparatide Forteo, biosimilars now), Radius Health (abaloparatide Tymlos), generic-bisphosphonate manufacturers. The new-agent makers have a strong financial interest in the very-high-risk anabolic-first framing; generic-bisphosphonate cost is low so the counter-pressure is weaker than in other drug classes.
- Professional. AACE, the Endocrine Society, the Bone Health and Osteoporosis Foundation (formerly NOF), the International Osteoporosis Foundation. Endocrinologists and rheumatologists are the natural specialists; primary care manages most diagnoses. Specialty guideline bodies have moved earlier toward anabolic-first than community primary care.
- Payer. Medicare and private insurers gate anabolic agents tightly; prior authorisation typically requires documented T-score, prior fracture, or failed antiresorptive. The gating is the dominant friction in actually implementing anabolic-first.
- Skeptic / counter. Cardiology raised the romosozumab MI/stroke flag during FDA review. Geriatricians emphasise polypharmacy and the implementation gap (post-fracture under-treatment) over the sequencing debate. Patient communities sometimes resist bisphosphonates due to ONJ and AFF anxieties — usually disproportionate to the actual risk magnitudes.
Population variability
- Postmenopausal women are the modal population; almost all trial evidence is here. The very-high-risk framing applies cleanly.
- Men account for ~30% of hip fractures and are systematically under-diagnosed and under-treated. Secondary causes (hypogonadism, glucocorticoid use, alcohol, hypercalciuria) are more common; the 90-day workup should include morning testosterone. Trial evidence is thinner — most agents have small bridging trials in men with BMD endpoints rather than fracture endpoints.
- Glucocorticoid-induced osteoporosis develops fast (within months of starting prednisone-equivalent doses ≥ 7.5 mg/day) at any age. Treatment thresholds are lower; teriparatide and zoledronic acid have specific indications. Premenopausal women on chronic steroids are an important under-recognised group.
- Premenopausal women with low BMD are most often secondary — coeliac disease, anorexia, early menopause, glucocorticoids. The workup matters more than the choice of agent.
- Ethnic variation. Hip fracture risk is highest in white and Asian populations and lower in Black populations at equivalent BMD; FRAX uses ethnicity-specific calibration where available. Genetic admixture and population-specific bone geometry contribute; the literature is sampled overwhelmingly from white postmenopausal cohorts.
- Renal impairment. eGFR < 30–35 mL/min excludes most bisphosphonates; denosumab is RANKL-targeted and not renally cleared but carries severe hypocalcaemia risk in low-eGFR patients.
Knowledge gaps
- Long-term sequencing. Trials run 1–3 years. Optimal sequencing over a 10–20-year horizon — including when to cycle back to an anabolic — is largely extrapolated, not trialled.
- Romosozumab cardiovascular signal. The ARCH excess was modest and the mechanism unclear; post-marketing surveillance continues. The signal was not seen in FRAME (vs placebo); whether the comparator (alendronate, which may have a small cardioprotective effect) drives the difference is unresolved.
- Real-world adherence. Trial adherence and effect sizes overstate community results. Adherence to oral bisphosphonates beyond 12 months is poor; switching costs to injectable agents are largely unmeasured.
- Drug-holiday duration. When to stop a bisphosphonate, for how long, and how to re-initiate is empiric. The FLEX (Black 2006) and HORIZON-extension (Black 2012) trials provide some guidance, but personalised re-initiation triggers remain debated.
- Men, premenopausal women, glucocorticoid-induced osteoporosis. Pivotal-quality fracture-endpoint trials are largely absent; BMD-endpoint bridging trials are the basis for treating these groups.
Scope vs brief. The brief named the workup, treatment selection, BMD effects, fracture risk, and anabolic-vs-antiresorptive sequencing. All five are covered end-to-end. No narrowing.
Voice translation choices. "Anabolic" and "antiresorptive" are translated to "bone-builder" and "bone-preserver" throughout reader prose. The technical terms still appear once or twice (in lists of agent names) because the reader will encounter them in clinic; introducing the friendly term while keeping the medical one searchable is the compromise.
Hard call on stance. The article lands on the AACE 2020 / Endocrine Society / BHOF anabolic-first framing for very-high-risk patients rather than the older bisphosphonate-default. This is the current guideline consensus but practice is uneven. Considered a more hedged "both are defensible" framing and rejected it — the trial signal (ARCH, VERO, STRUCTURE) is large enough that hedging would mislead the reader more than the residual debate.
Rating difficulty: cost_burden. Wide range by treatment path. Generic oral bisphosphonate is essentially free; a 24-month abaloparatide course lists at $70 000+. Landed on 3 reflecting the substantial cost of the very-high-risk anabolic path most readers consulting this entry will face. A defensible 2 if weighted toward the modal bisphosphonate path.
Rating difficulty: controversy. Landed on 3. The sequencing framework is broadly agreed on by major guideline bodies, but real-world practice lags and the romosozumab cardiovascular signal is genuinely unresolved. A 2 would understate the cardiac question; a 4 would overstate the framework-level agreement.
Future-link candidates (entries that should cross-link once they exist):
- Bisphosphonate drug holiday (when to stop, for how long, how to re-initiate)
- DXA screening cadence (USPSTF thresholds, who and when)
- Vitamin D status and supplementation
- Hypogonadism in men (overlapping secondary cause)
- Glucocorticoid-induced osteoporosis (different workup, lower thresholds)
- Post-menopausal hormone replacement therapy (alternative bone-density agent in a narrow subset)
- Falls prevention in the home
- Weight-bearing and resistance exercise for bone health
Separate-entry candidates surfaced during writing:
- Glucocorticoid-induced osteoporosis warrants its own entry — different thresholds, different age range, different first-line agents.
- Osteoporosis in men is sufficiently distinct (workup-heavy, under-treated, different evidence base) that a dedicated entry would be more useful than expanding this one.
Deliberately excluded. The drug-holiday decision (when to stop a bisphosphonate after 3–5 years) is out of scope — it's a year-3+ decision, not a first-90-days decision, and the FLEX/HORIZON-extension evidence base deserves its own treatment.
Osteoporosis: The First 90 Days
Multiple large pivotal RCTs with hard fracture endpoints (FIT, HORIZON, FREEDOM, FRAME, ARCH, ACTIVE, VERO, STRUCTURE) plus aligned AACE/ACE 2020, Endocrine Society 2019/2020, and BHOF 2022 clinician guidelines (Black 1996, Black 2007, Cummings 2009, Saag 2017, Cosman 2016, Miller 2016, Kendler 2018, Langdahl 2017, Camacho 2020, Eastell 2019, LeBoff 2022).
Hip fracture in adults over 65 carries 20-25% 12-month all-cause mortality (Brauer 2009); pivotal trials show 40-70% relative reductions in hip and vertebral fracture across antiresorptive and anabolic classes, with anabolic-first sequencing delivering 38% lower hip fracture risk vs alendronate over 24 months (Black 2007, Cummings 2009, Saag 2017).
Daily SC self-injection (teriparatide, abaloparatide), monthly in-clinic injections (romosozumab), or annual IV infusion / 6-monthly SC injection (zoledronic acid, denosumab). Add: baseline lab panel, DXA at 1-2 years, ongoing calcium/vitamin D. Minor lifestyle integration, not life-dominating.
Wide range by path: generic oral bisphosphonates ~$0-300/year; denosumab ~$3000-5000/year; anabolic courses (teriparatide, abaloparatide) ~$70 000+ over 24 months; romosozumab ~$23 000 over 12 months. Insurance typically covers anabolics with prior authorisation. Score reflects the substantial cost typical of the very-high-risk anabolic-first path (Camacho 2020).
Vertebral fracture prevention preserves height and forestalls kyphotic posture over decades; the aesthetic component is real but secondary to the medical effect (Black 2007, Cosman 2016).
The 90-day workup commonly surfaces correctable secondary contributors (vitamin D deficiency, hypothyroidism, hypogonadism, coeliac, hypercalciuria) responsible for ~20-30% of cases; the disease itself is silent so the treatment delivers no felt-experience lift in the first 90 days (Camacho 2020, LeBoff 2022).