Research theme
Lumbar stress fracture is the most costly injury in professional cricket, and everyone assumes it is caused by the hammering the body takes at front foot contact. But nobody had measured how much of that hammering actually reaches the lower back. This study strapped accelerometers up the body and measured the impact shock at every level, at all three ground impacts of the action. Method: 11 healthy male medium–fast bowlers (19.3 ± 2.3 y, 80.5 ± 9.8 kg, 1.86 ± 0.06 m), mix of county cricketers and Loughborough 1st XI, injury-free for 2 months. Nine Vicon Blue Trident IMUs at 1600 Hz taped over: both ankles (5 cm above lateral malleolus), both knees (5 cm above lateral femoral condyle), both hips (greater trochanter), and the L5, L1 and C7 vertebrae. Six maximum-velocity deliveries each, front foot on the popping crease; the three fastest legal trials averaged. Four Vicon Bonita cameras at 250 Hz identified the impact instants; a Stalker II radar gun measured pace. This is direct measurement, not simulation — but of surface (skin) accelerations, which are not the same as bone load.
What they measured
- The size of the impact shock at each level of the body (peak resultant acceleration, in g) at the ankle, knee, hip, lower back (L5), mid back (L1) and base of neck (C7).
- How long after ground contact the shock arrived at each level (time to peak resultant acceleration, ms).
- All three of these at each of the three ground impacts of the action: back foot contact (BFC), front foot contact (FFC) and the follow-through (FT) landing.
Findings
Front foot contact is by far the biggest hit, and the ankle takes ~150 g. Peak resultant acceleration, mean (SD), in g:
Impact Ankle Knee Hip L5 (low back) L1 (mid back) C7 (neck) Back foot contact 55.92 (27.57) 30.44 (15.34) 10.96 (4.58) 5.82 (1.85) 4.39 (1.63) 3.70 (1.06) Front foot contact 150.17 (49.67) 90.75 (36.03) 53.69 (30.49) 28.76 (13.32) 19.27 (5.93) 9.54 (3.26) Follow-through 110.21 (41.54) 87.70 (36.70) 46.84 (17.50) 18.20 (8.63) 12.77 (12.78) 5.21 (1.61) Over 90% of the shock is gone before it reaches the head/neck. Percentage reduction relative to the ankle sensor, mean ± SD:
Impact Knee Hip L5 L1 C7 BFC 45 ± 17% 77 ± 11% 88 ± 3% 91 ± 2% 93 ± 2% FFC 38 ± 15% 63 ± 14% 80 ± 10% 86 ± 5% 93 ± 4% FT 18 ± 20% 55 ± 11% 82 ± 8% 88 ± 4% 95 ± 3% Sensor position had a highly significant effect on both magnitude and timing (both p < 0.01), with every adjacent pair significantly different (Bonferroni post-hoc, p < 0.01). The body attenuates at every step — ankle→knee, knee→hip, hip→L5, L5→L1, L1→C7. There is no single “shock absorber”; it is a chain.
The shock arrives later the higher up you go — a progressive temporal delay at every step (p < 0.01). This delay is the attenuation mechanism: joint rotations and tissue deformation spread the impulse over more time, which lowers the peak. Bowling phase had no effect on the delay (p = 0.38).
The knee is the weakest link at follow-through. Knee attenuation was only 18 ± 20% at the follow-through versus 45% at BFC and 38% at FFC, and follow-through knee acceleration (87.70 g) was almost as high as at front foot contact (90.75 g). The follow-through landing is not a benign phase for the knee.
Attenuation continues past L5, all the way to C7. The authors note this contradicts McErlain-Naylor, King & Allen (2021), who found no attenuation beyond L5 in drop-jump landings, and attribute the difference to fast bowling generating much larger ankle accelerations in the first place.
Authors’ headline conclusion: even when the incoming shock is largest, the body dissipates more than 90% of it before it reaches the vital organs — so the attenuating structures (joint rotations, joint compression, soft-tissue displacement, intervertebral discs) matter a great deal, and existing cricket simulation models that use rigid pin joints will get segment accelerations — and therefore calculated spinal loads — wrong.
What a coach should look for on video
This paper is an instrumentation study. It supports no direct technique cue on video — it does not relate any accelerometer number to any joint angle, and it does not compare good and bad techniques. Saying otherwise would be inventing a cue. What it does support:
Cue: which impact to scrub to when you care about loading.
- Camera view + frame: Side-on, slow motion. Scrub to front foot contact — the frame the front foot first strikes the ground — and step through the next ~40 ms.
- What “good” looks like: nothing measured here. But this is the frame that matters: front foot contact produced ~2.7× the ankle shock of back foot contact (150 g vs 56 g) and ~5× the L5 shock (28.8 g vs 5.8 g).
- What the fault looks like: n/a from this paper.
- Why it matters: it tells a coach where in the action to spend their attention when the concern is load rather than pace.
Cue: do not ignore the follow-through landing.
- Camera view + frame: Side-on, scrub past ball release to the first follow-through ground contact.
- What “good” looks like: not established by this paper.
- What the fault looks like: not established.
- Why it matters: follow-through knee acceleration (87.7 g) was statistically indistinguishable from front foot contact, and the knee attenuated only 18% there. The follow-through is a real loading event, not a wind-down, and it is routinely un-filmed.
Practical takeaway for workload monitoring (from the numbers, not a video cue): a single tibia/ankle-mounted IMU sees the largest signal by far and is the practical wearable site. An L5-mounted sensor sees only ~20% of it, with the attenuation fraction varying between bowlers (SD ±10% at FFC) — so an ankle sensor is a good proxy for the impact the body absorbs, but a poor proxy for what any individual’s lower back actually receives. See the caveats.
Caveats and limits
- n = 11, all male, mean age 19.3 — young, healthy, injury-free by design. No injured bowlers and no women.
- Skin-mounted accelerometers measure surface motion, not bone load. The companion 2023 ISB paper makes the point explicitly: surface accelerations disregard muscle forces and do not represent internal bone loads. You cannot read “28.76 g at L5” as a stress-fracture risk score.
- Laboratory, six maximum-effort deliveries, best three averaged. Nothing about fatigue, spell length, or accumulated workload — the exact things workload monitoring cares about.
- Soft-tissue artefact: even with adhesive tape, a sensor over the greater trochanter or L5 moves relative to bone. Reported magnitudes should be read as approximate.
- Between-bowler variability is large and unexplained. Ankle FFC acceleration was 150 ± 50 g — a third of the mean. The study does not ask which bowlers ran high or why.
- An internal reporting error: the Results state “significant differences between BFC and FT (p < 0.01) and FFC and FT (p < 0.01), however there were no significant differences observed between FFC and the FT phase.” The second clause contradicts the first. The companion 2023 ISB paper resolves it: BFC accelerations were significantly less than both FFC and FT, and FFC vs FT did not differ. Read the 2022 sentence as a typo.
- Purely descriptive: no link to injury outcome, to technique, or to ball speed.
Relationship to other Felton work
- The first output of Matthew Lamb’s PhD, supervised by King, McErlain-Naylor, Felton, Brooke-Wavell and Peirce (funded by the ECB and an EPSRC DTP).
- Extended by Lamb et al. 2023, ISB/JSB Kyoto, which adds frequency-domain analysis, and by Lamb et al. 2023, 19th ISCSB, which builds a compliance simulation model — both in this folder.
- Directly criticises the Felton simulation line. It names Felton, Yeadon & King (2020), “Optimising the front foot contact phase of the cricket fast bowling action” as an example of a whole-body cricket model that “focused solely on performance and [has] not yet explored the relationship between performance and injury”, and cites Allen, King & Yeadon (2012) — that models with rigid pin joints are fine for simulating performance but unsuitable for simulating internal loading. It argues future cricket simulations must include joint compliance.
TENSION: this paper’s mechanism — attenuation comes from joint rotations and tissue deformation, i.e. from the lower limb being compliant — points in the opposite direction to the performance finding running through Felton’s other work (Felton, Yeadon & King 2020; Worthington, King & Ranson 2013) that a straighter, braced front leg produces more ball speed. A stiffer, straighter front leg is a worse shock absorber. This paper does not test the trade-off, but it sets it up; Lamb et al. (2023, ISCSB) tests it directly and finds stiffness does raise transmitted acceleration. See that file and the folder README.