2022 — Lumbar Bone Mineral Adaptation in Adolescent Cricketers

Research theme

This paper flips the injury question around. Instead of asking “what technique breaks the spine?”, it asks “what technique builds it?” Elite fast bowlers are known to have unusually dense lumbar spines, and unusually lopsided ones — more bone on the side opposite the bowling arm. Nobody had checked whether that lopsidedness is already present in teenagers, or which parts of the bowling action drive it.

Method: 39 adolescent male fast bowlers (age 15.6 ± 1.1 years; height 1.79 ± 0.07 m; mass 68.7 ± 10.7 kg; fat free mass 56.5 ± 8.3 kg), recruited from professional academies and strong school/club programmes, aged 14–17 with ≥2 years of high-level cricket. Each bowled at least 12 maximal-effort good-length deliveries (landing 4–7 m from the batter’s stumps) captured by an 18-camera Vicon system at 300 Hz with two Kistler force plates at 1800 Hz under the front foot. 47 reflective markers. On the same day each received a total-body and antero-posterior lumbar (L1–L4) DXA scan, with a custom analysis measuring bone mineral density in the lateral third of the contralateral and ipsilateral sides of L3 and L4 separately (spinous process omitted).

Then: partial correlations (controlling for fat-free mass) between 47 kinematic + 6 kinetic parameters and whole-lumbar BMD; bivariate correlations against L3/L4 asymmetry; significant ones fed into forward stepwise linear regression.

What they measured

• How dense the low back bones are (L1–L4 BMD, g/cm²; and Z-score vs age-matched population).
• How lopsided that density is (percent difference between the contralateral and ipsilateral lateral thirds of L3 and L4).
• How lean the bowler is (fat-free mass from total-body DXA — used as a covariate because it drives both BMD and force).
• Hip angles (back hip at BFC, front hip at FFC).
• How the pelvis is tipped, dropped and turned (pelvis orientation — tilt, drop, twist — at BFC, FFC, BR, plus min and max between BFC and BR).
• The hinge between pelvis and low back (lumbopelvic angle: flexion/extension, side flexion, rotation — same five instants).
• The hinge between ribcage and low back (thoracolumbar angle: flexion/extension, side flexion, rotation — same five instants).
• How hard the front foot hits (peak vertical GRF, vertical loading rate, vertical impulse — in kN and bodyweights).
• Convention: anatomical position = 180°; below 180° = flexion, contralateral side flexion, contralateral rotation.

Findings

1. Teenage fast bowlers already have dense, lopsided lumbar spines.

• L1–L4 BMD 1.214 ± 0.199 g/cm²; Z-score +1.0 ± 1.2 (95% CI 0.7–1.4) — a full standard deviation above the population norm for their age.
• Contralateral-to-ipsilateral BMD asymmetry: L3 = 9.0 ± 9.6% (95% CI 5.8–12.1%), L4 = 8.2 ± 10.1% (95% CI 4.9–11.5%). Both significantly different from zero (p < 0.001).
• The asymmetry is established by mid-adolescence. Compared to elite adults (Alway et al. 2019): whole-lumbar BMD lower (1.21 vs 1.56 g/cm², p < 0.001), L3 asymmetry the same (9.0% vs 8.9%, p = 0.972), but L4 asymmetry smaller (8.2% vs 14.6%, p < 0.001) — so the L4 lopsidedness keeps growing into adulthood.

2. Twisting, not thumping, predicts bone density. Best single technique predictor of L1–L4 BMD (with fat-free mass controlled): maximum contralateral thoracolumbar rotation between BFC and BR, explaining 56.7% of variance (fat-free mass alone: 47.1%). Adding maximum ipsilateral lumbopelvic rotation → 61.5%. Adding contralateral pelvic drop at FFC → 65.2%.

Regression model (a) for L1–L4 BMD, final:

Direction: more contralateral thoracolumbar rotation, less ipsilateral lumbopelvic rotation, and more contralateral pelvic drop at FFC all go with higher whole-lumbar BMD.

Key partial correlations (covariate: fat-free mass) with L1–L4 BMD:

3. The rotation that matters happens BEFORE front foot contact. Maximum contralateral thoracolumbar rotation between BFC and BR was correlated at r > 0.80 with thoracolumbar rotation at BFC. Thoracolumbar rotation at FFC and at BR, and maximum ipsilateral thoracolumbar rotation, were not predictors of BMD. The authors state plainly: “the thoracolumbar rotation which is correlated with lumbar BMD occurs prior to FFC.”

4. Different technique drives the lopsidedness. Best predictors of BMD asymmetry:

Bivariate correlations behind these: lumbopelvic rotation at FFC vs L3 asymmetry r = 0.411, p = 0.009; max lumbopelvic rotation (BFC–BR) vs L3 r = 0.365, p = 0.022; lumbopelvic rotation at BR vs L4 asymmetry r = 0.460, p = 0.003; thoracolumbar side flexion at BR vs L4 r = 0.390, p = 0.014; pelvic drop at FFC vs L3 r = −0.342, p = 0.033.

Direction: more ipsilateral lumbopelvic rotation → more lopsided bone at L3 and L4. Less contralateral thoracolumbar side flexion at BR → more lopsided bone at L3.

5. There is an internal tension the authors flag themselves. Smaller maximum ipsilateral lumbopelvic rotation goes with higher whole-lumbar BMD, while larger ipsilateral lumbopelvic rotation goes with greater asymmetry. Their explanation: larger lumbopelvic ipsilateral rotations shunt the load further towards the contralateral side and the lower vertebrae (L3–L4), producing lopsidedness; smaller rotations spread the strain across the whole L1–L4 region, raising overall density.

6. Ground reaction force explained NOTHING once body size was accounted for. Peak vertical GRF 3.5 ± 1.2 kN (5.2 ± 1.3 BW); vertical loading rate 131 ± 90 kN/s; vertical impulse 85 ± 48 N.s. Not one kinetic parameter correlated with L1–L4 BMD (partial, FFM-controlled) or with L3/L4 asymmetry (bivariate) — in absolute or bodyweight-normalised terms. Peak VGRF was raw-correlated with BMD, but that association vanished once fat-free mass was controlled (FFM vs peak VGRF: r = 0.673, p < 0.001).

7. Mechanism proposed: torsional loading from muscle, not impact from the ground. The authors argue that the opposing rotations — thoracolumbar rotating ipsilaterally while lumbopelvic rotates contralaterally between FFC and BR — generate torsion on the less mobile inferior vertebrae. Resisting that requires eccentric contraction of contralateral multifidus, erector spinae, external obliques and contralateral internal obliques. Torsional loading is known to strain bone more than axial loading (Rubin et al., 1996), and eccentric muscle action has the greatest osteogenic effect. This aligns with Frost’s mechanostat theory: muscle, not gravity, is the main source of bone strain.

Status of evidence: correlational and cross-sectional. Bone was measured on the same day as bowling. There is no longitudinal tracking of BMD change, no intervention, and the direction of causality is assumed, not shown.

What a coach should look for on video

Be careful with this one. This paper is about bone adaptation, not injury. High lumbar BMD is generally a good thing — it is bone getting stronger in response to load. But the same rotational mechanics that build bone are the mechanics implicated in asymmetric adaptation, and asymmetric loading is what fractures the contralateral neural arch. The paper does not tell you whether more BMD or less asymmetry is the better outcome for a given bowler. It ends by calling for exactly that research.

With that framing, two cues are genuinely supported:

Cue 1 — Trunk rotation at back foot contact (the bone-building one)

• The cue: How much the ribcage is turned away from the bowling arm relative to the pelvis as the back foot lands.
• Camera view + frame: Behind-the-arm or high overhead, scrubbed to back foot contact. Overhead is far better if you can get it — rotation is what you are judging.
• What “good” looks like: Genuine contralateral rotation of the ribcage on the low back at BFC. In this cohort thoracolumbar rotation at BFC averaged 178 ± 4° (below 180° = contralateral), and the bowlers further below that number had higher lumbar BMD (r = −0.430, p = 0.007).
• What the fault looks like: A trunk that stays square through the back foot landing — the ribcage and pelvis turning together as a block.
• Why it matters: Bone strength, explaining 56.7% of variance in lumbar BMD once lean mass is accounted for. This is a positive adaptation cue, not an injury cue. Important caveat: the 2021 Alway paper found that more contralateral thoracolumbar rotation at BFC was one of the significant differences in bowlers who went on to be injured (injured 177 ± 5° vs uninjured 182 ± 4°, large effect). So the same motion that appears to build bone appears in the injured profile. See the contradiction section. Do not coach this as a straightforward “do more of it.”

Cue 2 — Rotation at the waistband through the delivery (the lopsidedness one)

• The cue: How much the pelvis turns towards the bowling arm relative to the low back, from front foot contact through to release.
• Camera view + frame: High overhead or behind-the-arm, scrub from front foot contact to ball release, watching the belt line relative to the ribs.
• What “good” looks like: No target exists. In this cohort lumbopelvic rotation was 203 ± 9° at FFC, 187 ± 6° at BR, max 207 ± 10° (above 180° = ipsilateral rotation).
• What the fault looks like: Large, late ipsilateral turning of the pelvis under a relatively fixed low back.
• Why it matters: This is the best predictor of side-to-side bone asymmetry — max lumbopelvic rotation explains 11% of L3 asymmetry alone, 24.6% with thoracolumbar side flexion at BR added; lumbopelvic rotation at BR explains 19% of L4 asymmetry. Asymmetric loading is the pattern that concentrates strain on the contralateral neural arch where LBSI occurs. This is an inference — the paper measured bone, not injury.

What NOT to look for: a heavy front foot

Front foot impact was irrelevant to bone in this cohort. Peak vertical GRF, loading rate and impulse showed no relationship with lumbar BMD or asymmetry once fat-free mass was controlled. Whatever you think a “heavy” or “soft” front foot does, it is not driving lumbar bone adaptation in these 39 teenagers.

The genuinely actionable, non-video takeaway

Fat-free mass alone explained 47.1% of lumbar BMD variance — more than any single technique parameter added on top of it. For an adolescent bowler, building lean mass is the largest lever on lumbar bone density in this dataset. That is a strength-and-conditioning intervention, not a technique cue.

Caveats and limits

• Sample: 39 adolescent male bowlers, aged 14–17, mean 15.6. Nothing here applies to women or to adults directly. The authors explicitly call for female work.
• Cross-sectional, correlational. Bone density measured once. No tracking of change over time, no intervention. The causal claim (technique → bone) is assumed from plausibility and mechanostat theory, not demonstrated.
• Adolescents mid-puberty. The authors name this as a weakness: substantial hormonal and growth adaptation, and different historical bowling workloads, “has the potential to skew the findings.”
• Workload was NOT included in the analysis — the authors say it was too difficult to assess accurately prior to inclusion. So the biggest known driver of bone adaptation (how much they’ve bowled) is missing from the model entirely.
• History of LBSI was not an exclusion criterion, and injury status is not reported per participant. Some of these bowlers may already have had bone stress injury.
• 2D DXA, not 3D. DXA measures areal density from one projection and cannot see bone architecture or the posterior elements where fractures happen. The authors note 3D methods would be better but carry radiation cost in growing children.
• No multiple-comparison correction. 47 kinematic + 6 kinetic parameters × 3 outcomes, all at α = 0.05, deliberately unadjusted to avoid Type 2 errors. The authors state the results “should be treated cautiously as an increased risk of Type 1 errors occurring remains.”
• Forward stepwise regression carries bias in parameter estimation and reliance on a single best model; mitigated by computing all same-size models for comparison.
• Discrete instants and absolute angles, not continuous curves or ROM-normalised values — the authors list both as limitations.
• The asymmetry models are weak. 11.0%, 19.0% and 24.6% of variance explained. These are not strong predictions.
• One trial per bowler (fastest, minimal marker loss, front foot on the plate).

Relationship to other Felton work

• Directly builds on Alway et al. 2019 (Bone), “Lumbar bone mineral asymmetry in elite cricket fast bowlers,” which established the adult pattern (L1–L4 BMD 1.56 ± 0.16 g/cm², Z-score 2.45 ± 1.24, L4 contralateral 14.6% denser). This paper is the adolescent equivalent.
• Uses the 2021 Alway MSSE paper (2021 Alway — Cricket Fast Bowling Technique and Lumbar Bone Stress Injury) as its injury reference point throughout, and explicitly tests whether that paper’s injury predictors also predict bone. They do not: “No association was found between these technique variables [increased hip flexion at BFC, increased lumbopelvic extension at FFC] and L1-L4 BMD or BMD asymmetry.”
• Agrees with the 2021 paper on ground reaction force. Both found GRF irrelevant — this one to bone adaptation, that one to injury. The authors note the concurrence directly.
• Companion to the other 2022 Keylock paper (2022 Keylock — Lumbar Bone Stress Injuries and Risk Factors in Adolescent Fast Bowlers), same research programme, overlapping adolescent cohort, but that one tracks injury prospectively while this one measures bone cross-sectionally.
• Shares the Worthington et al. (2013) marker set and processing pipeline with the whole cluster.

CONTRADICTION: Impact loading does not build the fast bowler’s lumbar spine — muscle torsion does. The received view, and the reason gymnastics and basketball are cited as osteogenic sports, is that high vertical impacts drive bone adaptation. In these bowlers, no kinetic parameter — peak vertical GRF, loading rate, or impulse, absolute or normalised — was associated with lumbar BMD or with L3/L4 asymmetry once fat-free mass was controlled. The conclusion: “muscular forces, through the initiation and control of thoracic and lumbar rotation, are the predominant contributor.” For a coach this reframes what “loading the back” means: it is what the trunk muscles do resisting rotation, not what the front foot does hitting the ground.

TENSION (important, and the authors half-acknowledge it): The rotation that appears to build bone is also in the injured bowlers’ profile. This paper’s single best BMD predictor is greater contralateral thoracolumbar rotation around BFC. The 2021 Alway paper found injured bowlers had significantly more contralateral thoracolumbar rotation at BFC than uninjured bowlers (177 ± 5° vs 182 ± 4°, large effect size) — one of only two large-effect BFC differences in that study. Keylock et al. spot this and read it as promising rather than contradictory: “larger contralateral thoracolumbar rotation angles at BFC… have been associated with LBSI. This may provide evidence of a link between kinematic parameters, lumbar bone adaptation and LBSI.” But it leaves a genuine open problem for coaching: the same motion may simultaneously strengthen the bone and be part of the injury pattern. No one should be coached toward or away from it on the strength of this paper.

TENSION (internal to this paper): Smaller maximum ipsilateral lumbopelvic rotation predicts higher whole-lumbar BMD, while larger ipsilateral lumbopelvic rotation predicts greater L3/L4 asymmetry. The authors call these “seemingly conflicting associations” and explain them as load being either spread across L1–L4 or shunted to the contralateral side of L3–L4. It is a plausible reading, but it means the same variable points in opposite directions depending on which bone outcome you care about.

TENSION (with the LBSI literature generally): Adolescent fast bowlers have lumbar BMD a full standard deviation above population norms (Z-score +1.0), yet lumbar bone stress injury is the most prevalent injury in the sport. High bone density is clearly not protective on its own — the companion Keylock adolescent risk-factor paper found injured bowlers had non-significantly greater contralateral BMD than uninjured ones (g ≥ 0.812).