The route-(c) contraindication question for exercise in the osteoporotic/osteopenic stratum. The efficacy pages ask whether exercise prevents falls and fractures (Exercise for Preventing Falls in Older Adults); this page asks the inverted question a patient with fragile bone actually voices — will exercise break me, and which movements are dangerous? — a distinct decision (harm, not benefit) that gates whether the efficacy advice is even attempted. First source held on it: Kunutsor 2018, a gold systematic review pairing 14 observational studies (13 unique) on modality-specific harm with an update of three previously published interventional reviews (Howe bone-density and Giangregorio, both Cochrane; Sherrington’s BJSM falls meta-analysis). (Kunutsor et al., 2018)
Evidence-tier note (confidence: low). Single gold SR, but the underlying adverse-event data
are weak by construction: «none of the trials were powered to look at adverse events including
fractures, and these were poorly reported»
(Kunutsor et al., 2018).
The safety verdict is a reassurance drawn from absence of signal in trials not built to detect one
— so it is held low on total web support, and the under-reporting caveat below is the load-bearing
limit, not a footnote.
The bottom line: exercise is broadly safe, serious harm rare
The review’s headline verdict is reassuring: «Serious adverse events were rare and could not be attributed to the intervention… Patients with osteoporosis/osteopenia can safely participate in structured exercise programmes, whether at home or in supervised facilities.» (Kunutsor et al., 2018) Across the pooled interventional data (57 additional RCTs identified across the three updated interventional reviews) the excess was minor events only — muscle aches, transient musculoskeletal complaints — «what would normally be expected with physical exertion», not fractures. (Kunutsor et al., 2018)
- Sherrington falls-review update (107 trials, 23 288 participants). Of these, the 8 trials of community-dwelling participants that reported adverse-event counts for both arms gave a pooled intervention RR 1.08 (95% CI 1.04-1.13) — a real but small increase, driven by minor events. Falls were the highest-incidence adverse event (156/1000 person-years), musculoskeletal complaints next, and fractures the lowest (21/1000 person-years). (Kunutsor et al., 2018)
- Howe bone-density update (62 trials, 6607 participants). More minor adverse events in the exercise arms (RR 1.55, 95% CI 1.05-2.28, P=0.029) but no effect on falls overall (RR 1.04, 95% CI 0.79-1.37, P=0.779) — the minor-AE excess did not translate into more falls. (Kunutsor et al., 2018)
- Direction of harm is toward the trivial-and-transient, and «did not stop participation». (Kunutsor et al., 2018)
And a fracture-reduction hint — a trend, not a result
The same Howe update produced 31 fractures in 536 exercise participants (5.8%) vs 43 in 449 controls (9.6%), pooling to OR 0.60 (95% CI 0.36-1.01, P=0.053) — the authors call it «the first evidence to suggest that exercise interventions for bone health can reduce fractures.» (Kunutsor et al., 2018) State it as a trend: the CI crosses 1 and P=0.053 — this is a suggestive signal on an underpowered fracture endpoint, not a demonstrated reduction. It matters for the safety decision in one direction only: even the harm-focused review found no hint that exercise raises fracture risk in this stratum, and a weak hint it lowers it.
Which movements to avoid — flexion vs extension
The one route-(c) modality signal, from the observational arm plus mechanism:
- Repetitive/loaded spinal forward flexion is the flagged concern. Mechanistically «forward flexion increases the load on the front of the vertebral bodies, and is likely to increase the risk of vertebral fractures»; the observational Sinaki data showed a fracture rate of 89% in a flexion-exercise group vs 16% in an extension group (53% combined). (Kunutsor et al., 2018)
- The practical recommendation: «people with osteoporosis avoid specific and repetitive spinal forward flexion exercises, and instead focus on spinal extension.» (Kunutsor et al., 2018)
- But do not over-restrict — the important nuance. «Evidence that forward flexion exercise interventions cause vertebral fractures are limited» (Kunutsor et al., 2018), and crucially «there is no evidence that single forward flexion movements in daily activities or ‘roll down’ Pilates movements cause vertebral fractures - a common misconception»; the focus should be on safe lifting technique for everyday flexion, not on fearfully banning it. (Kunutsor et al., 2018)
- Other observational flags (weaker): brisk walking (raises fall risk vs normal-pace walking, 42 vs 26 falls/100 person-years), horseback riding, and golfing may associate with greater adverse-event risk — but the authors caution against strong avoidance conclusions from these. (Kunutsor et al., 2018)
So the modality rule is asymmetric and narrow: substitute extension for repetitive loaded flexion in a structured programme; do not police ordinary daily bending. This is a substitution frame (Exercise for Preventing Falls in Older Adults-style), not a prohibition.
The load-bearing limit: adverse events are systematically under-reported
(inferred from Kunutsor et al., 2018) — this is the streetlight instance on the harms side, and it caps every reassurance above. In the Sherrington update, 55 of 107 trials (51.4%) did not report on adverse events at all, and a further 26 (24.2%) reported zero occurrences — so three-quarters of trials contribute a blank or a bare zero. (Kunutsor et al., 2018) The review states the identification problem exactly: «it is unclear whether the many studies identified in this review that did not report any adverse events were doing so because none occurred, or because the data was not recorded.» (Kunutsor et al., 2018)
- This is the insufficient-evidence state, not no meaningful effect. A silent trial is not a safe trial; a reported zero from a trial not powered to detect fractures is not an observed null. The expectancy test fails here — if serious harm were happening we could not be confident these trials would have recorded it — so the correct reading is harm is unquantified, reframed by the reassurance into harm appears rare. Both are held; the gap between them is the finding.
- It is the downstream twin of Which Questions Get Studied: that concept is the upstream filter (which questions get funded and asked); this is the in-trial filter (a funded trial runs but does not record the harm). Same streetlight mechanism, different stage of the pipeline — cross-linked, not the same claim. Also an instance of Publication Bias and Selective Reporting operating within-study (selective outcome reporting of harms) rather than across studies.
Relationship to the falls-efficacy evidence — F/extension, not independent-E
provenance note guarding against laundered-E: Kunutsor 2018 is an update of the same review lineage as Exercise for Preventing Falls in Older Adults (it re-runs Sherrington’s BJSM falls meta-analysis; Julie Whitney co-authors both this review and the Sherrington work). The agreement that exercise is broadly safe is therefore a type-F refinement of a shared evidence base (same trials, re-examined for harms), not independent type-E corroboration — their convergence adds harm-specific granularity, not a second independent route to the safety claim. Marked so the confidence is not inflated by counting one review lineage twice.
Decision relevance
- Reassure, then prescribe. For an osteoporotic/osteopenic person afraid to move, the evidence supports structured exercise (home or supervised) as safe; serious harm is rare and unattributable, and inactivity’s harms are the larger rock. The fear itself is often the binding barrier.
- Substitute extension for repetitive loaded spinal flexion in programmed exercise; teach safe lifting for everyday bending rather than banning daily flexion (avoid the over-restriction misconception).
- Prefer normal-pace to brisk walking where fall risk is the concern.
- Hold the safety verdict as provisional — it rests on trials that did not measure harm well; the honest statement to a patient is no evidence it is dangerous, not proven safe, and the gap is a named Which Questions Get Studied-class hole ( for a purpose-designed adverse-event trial).