2020 — Does Lumbar Spine Kinematics Contribute to Lumbar Bone Stress Injury?

Research theme

This is the test the 2018 abstract asked for. Having shown that upper (L1–L3) and lower (L3–L5) lumbar segments move differently, the team applied the same two-segment model to a group of bowlers who were MRI-screened for lumbar bone stress injury (LBSI). The question: do bowlers who get LBSI position their lumbar spine differently from bowlers who don’t?

Method: 45 elite male fast bowlers, each declared match-fit by a physiotherapist, bowled a minimum of six maximum-effort deliveries captured by an 18-camera VICON MX system at 300 Hz in an indoor cricket facility. Markers at the spinous processes of L5, L3 and L1 plus 5 cm bilaterally at L4 and L2 defined upper and lower lumbar segments; pelvis markers at the PSIS and ASIS. ZXY Euler decomposition. Fastest delivery per bowler analysed, filtered at 10 Hz. Every bowler received a 3.0T lumbar MRI, read by one radiologist, plus medical-record follow-up for injuries occurring afterwards. Groups compared by unpaired t-test (Mann-Whitney if normality violated), Cohen’s d for effect size.

What they measured

• Whether the bowler had lumbar bone stress injury (MRI at time of testing, plus subsequent medical records over two years).
• Upper lumbar (L1–L3) and lower lumbar (L3–L5) joint angles in three planes, each relative to the segment directly below it:
  • x-axis — arching back / bending forward (flexion-extension)
  • y-axis — leaning sideways (side flexion, contralateral vs ipsilateral to the bowling arm)
  • z-axis — twisting (rotation)
• Measured at back foot contact (BFC), front foot contact (FFC), ball release (BR), plus max flexion, max extension, max contralateral and max ipsilateral values.
• Convention throughout: anatomical position = 180°; below 180° means flexion, contralateral side flexion, or contralateral rotation.

Findings

1. Nearly half the elite bowlers had an LBSI. 21 of 45 (47%) were diagnosed. 16 on the same day as the biomechanical assessment; 5 in the subsequent two years (mean 305 days, range 52–675 days).

2. The groups were athletically indistinguishable. No significant difference in age (LBSI 18.96 ± 1.72 vs uninjured 19.94 ± 2.57, p = 0.26), height (1.87 ± 0.06 vs 1.89 ± 0.06 m, p = 0.24), body mass (82.61 ± 7.68 vs 83.08 ± 10.92 kg, p = 0.53) or bowling velocity (34.03 ± 1.76 vs 34.17 ± 2.26 m/s, p = 0.83). Note that last one: the injured bowlers were not faster.

3. The headline result is a null result. Not one of the 30 lumbar joint angles differed significantly between groups. Every p-value in the table was ≥ 0.15, and every effect size was small (d ≤ 0.43). Selected values (LBSI vs uninjured, degrees, mean (SD)):

4. The only signal — and it is explicitly non-significant — is upper lumbar extension. Injured bowlers were more extended (arched) in the upper lumbar segment at BFC (187° vs 185°), FFC (191° vs 189°) and max extension (193° vs 191°) — differences of about 2°, all with small effect sizes (d = 0.28–0.37) and p-values of 0.20–0.37. The authors flag it as warranting further investigation, nothing more.

5. The proposed mechanism (hypothetical, not demonstrated). Injured bowlers show more upper-lumbar extension combined with a flexed lower lumbar spine — i.e. greater lordosis at these instants. Increased lordosis is known to raise compressive and shear force on the vertebral neural arch (Shirazi-Adl & Parnianpour, 1999). At FFC, high vertical ground reaction forces plus high contralateral lower-lumbar side flexion compress the discs and displace them ipsilaterally, reducing the gap between vertebrae particularly on the contralateral side. Add upper-lumbar extension and the contralateral superior articular process of the vertebra below may repeatedly impact the contralateral inferior margin of the neural arch above — accelerating failure on that side. This would explain why LBSI is almost always unilateral and contralateral to the bowling arm. But it is reasoning from other people’s finite-element and cadaver work, not from data in this study.

6. Explicit conclusion: “lumbar spine joint angles alone may not contribute to LBSI, and greater contribution comes from other risk factors.”

Everything here is correlational / group-comparison, not causal. And the headline is that even the correlation is absent.

What a coach should look for on video

This paper supports no coaching cue, and that is its most useful message.

Read that plainly: 45 elite bowlers, MRI-confirmed injury status, a two-segment lumbar model that is more sensitive than anything used before — and the position of the lumbar spine at back foot contact, front foot contact and ball release did not distinguish the injured from the uninjured. If you are scrubbing video looking for “too much lumbar bend” as your injury screen, this study says that is not where the signal is.

Three things a coach should take away instead:

1. Stop screening on lumbar angle alone. No lumbar joint angle in this dataset separated injured from uninjured bowlers. The largest effect sizes (d = 0.43) were on max contralateral rotation and side flexion, and even those were nowhere near significance.

2. Ball speed is not the tell. Injured and uninjured bowlers released the ball at effectively identical speed (34.03 vs 34.17 m/s, p = 0.83). A coach cannot infer risk from pace.

3. If you must watch one thing from this paper, watch the arch — but hold it loosely.

• The cue: How arched the mid-to-upper low back is (the region between the bottom of the ribs and the middle of the low back).
• Camera view + frame: Side-on, scrubbed to back foot contact and again at front foot contact.
• What “good” looks like: No target is available. The uninjured group averaged 185° at BFC and 189° at FFC in the upper segment; the injured 187° and 191°. That is a two-degree difference and it is not statistically significant. You cannot see two degrees on a tablet.
• What the fault looks like: A visibly hollowed low back through the delivery stride, with the ribcage tipped back, while the region just above the belt line stays folded forward.
• Why it matters: Possibly injury risk via increased lordosis at the moment of peak ground force — but this paper did not demonstrate it. Treat as a hypothesis to watch, not a fault to correct.

Do not read this paper as “lumbar movement is safe.” It says lumbar joint angles at three discrete instants do not separate the groups. The follow-up work found the signal one joint lower down and in the pelvis.

Caveats and limits

• Sample: 45 elite male bowlers, mean age ~19. No women, no juniors, no club players.
• Injury classification is mixed retrospective/prospective. 16 of the 21 injured bowlers were diagnosed on the same day as the biomechanical assessment — so their technique may already be a response to an existing injury, not a cause of it. Only 5 were genuinely prospective. The authors flag this: bowlers “may have changed technique following undergoing biomechanical assessment and when being diagnosed with LBSI.”
• The uninjured group may not stay uninjured. The authors note many uninjured bowlers “are of an age where risk of LBSI is greater and may still sustain a LBSI.” The 2021 paper fixed this by requiring uninjured bowlers to be ≥23 with ≥150 professional match days.
• One trial per bowler (the fastest), justified by ICC = 0.98 whole-body repeatability from Felton et al. 2019.
• No static trial was used, deliberately, “to preserve the natural posture of the lumbar spine.” So the angles are inter-segmental, not referenced to a personal neutral.
• Discrete time points, not whole-movement analysis. Only BFC, FFC, BR and maxima were examined. Rates of change, timing, and coupling between planes were not.
• Absolute angles, not normalised to each bowler’s own range of motion. A bowler with limited lumbar ROM sitting at 175° may be at end-range while another at 175° has room to spare. The 2023 ROM paper in this cluster begins to address this.
• Underpowered for small effects. With n = 21 vs 24 and observed effect sizes of d ≈ 0.3–0.4, this study had limited power to detect the differences it was looking for. A null result here is not proof of no effect.
• Single radiologist read all MRIs.

Relationship to other Felton work

• Directly answers the call to action in the 2018 BASES abstract (2018 Alway — Upper vs Lower Lumbar Spine Kinematics in Fast Bowling), which ended: “Future research should compare upper and lower lumbar spine kinematics between stress fracture and non-injured fast bowlers.”
• Motivates the 2021 MSSE paper (2021 Alway — Cricket Fast Bowling Technique and Lumbar Bone Stress Injury). This paper’s own conclusion — “other joint rotations or other risk factors to LBSI may have a greater contribution” — is exactly what the 2021 study went looking for, and found: rear hip angle at BFC and lumbopelvic angle at FFC.
• Cites the same 3.0T MRI protocol and Loughborough/ECB cohort used across the cluster.

CONTRADICTION: The 2018 abstract’s implication — that separating the lumbar spine into upper and lower segments would illuminate LBSI aetiology — is not borne out. The two-segment model was applied, and no lumbar joint angle differed between injured and uninjured bowlers.

CONTRADICTION: This paper undercuts the widely-repeated claim (Ranson et al., 2008; Chosa et al., 2004 by extension) that excessive lumbar rotation and extension is the driver of LBSI in fast bowlers. Finite-element modelling says extension + rotation + compression loads the neural arch most; this study measured extension and rotation directly in injured and uninjured elite bowlers and found no significant difference in either.

CONTRADICTION (with the 2021 paper in this cluster): This 2020 study found no difference in lumbar side flexion between groups at any instant. The 2021 study, on a larger and cleaner prospective cohort using a lumbopelvic (rather than intra-lumbar) segment definition, found that injured bowlers had significantly less contralateral thoracolumbar side flexion at BR and (at p = 0.09, d = 0.57) more contralateral lumbopelvic side flexion. The two papers are measuring different joints — intra-lumbar (L1–L3 vs L3–L5) here, thoracolumbar-vs-lumbopelvic there — and the later result suggests the injury-relevant frontal-plane motion sits at the lumbopelvic junction, below where this paper was looking.