Leash, collar, harness — the underestimated pressure on the chest

In a Wiesbaden dog park on a Sunday three owners meet. One walks his Hovawart on a wide leather collar. The second has put a soft mesh Y-harness on her Husky, “because of the shoulders”. The third has fitted her Boxer with a front-clip model that has a ring between the front legs — the marketing promised that pullers turn themselves around in it “as if by themselves”. Three owners, three convictions, three very different mechanical loads on the animal. Which of them is evidence-based has been traceable in peer-reviewed literature for some two decades — but the rule of thumb “a harness is always better than a collar” falls strikingly short.

What is measurable at the collar

The first clinical impulse came in 2006 from a paper by Anita M. Pauli and colleagues in the Journal of the American Animal Hospital Association: collar pull reproducibly led in their experimental animals to a significant increase in intraocular pressure; the same manoeuvre on a harness was without effect.¹ These findings were confirmed and refined in 2025 in Veterinary Medicine and Science by Bailey and colleagues, in brachycephalic and dolichocephalic dogs: the intraocular-pressure effect was more pronounced in short-snouted than in long-snouted breeds — a biomechanically expected consequence of the different neck musculature.²

A direct force-measurement study by Carter and colleagues in the Veterinary Record (2020) tackled the question with an instrumented neck model: seven commercially available collars were tested at three pull intensities. Contact pressures ranged between 83 and 832 kilopascals, depending on collar and pull strength. The authors explicitly conclude that none of the tested models stayed below a pressure level already established as tissue-damaging in human-medical research, at any of the tested pull intensities.³

Thyroid, hyoid — and the missing stop reflex

Rarely addressed in general advice: the glandula thyroidea, the thyroid gland, sits directly in the typical collar-seating area. The chronic load of a collar that presses on glandular tissue at every pull is considered, in clinical veterinary medicine, a co-factor for inflammatory changes and subsequent autoimmune damage. Clinically the connection is most visible in breeds predisposed to hypothyroidism anyway — Labrador, Golden Retriever, Doberman, Rottweiler. The original Pauli pressure study documented the mechanical basis; the subsequent review literature extended the risk description.¹

A second, often overlooked structure is the os hyoideum, the hyoid bone. A retrospective CT study of 293 dogs identified hyoid fractures or luxations in 9 animals — 3.1 percent — several of which were explicitly linked in the case histories to collar-based training methods.⁴ A 2025 case report in Frontiers in Veterinary Science documented the extreme: complete separation of larynx and trachea from the hyoid apparatus, caused by a leash caught in an elevator.⁵

The point that often complicates the discussion: the dog does not self-regulate. Unlike humans, in most dogs tracheal pressure does not trigger a withdrawal reflex that ends pulling. As long as motivation or prey drive outweighs breathing resistance, the animal keeps pulling. How widespread this is was quantified by an online survey of 2,531 owners in the UK and Ireland, published in 2021: 82.7 percent (n = 2,092) of dogs pulled on the lead during walks.⁶ The animals, put bluntly, strangle themselves — and often do not notice until acute breathing distress.

What a veterinarian at the Royal Veterinary College measured

The scientific turn in the harness discussion is tied to a name: Pia Lafuente, MRCVS, PhD, Senior Lecturer at the Royal Veterinary College London. In 2018 she published in the Veterinary Record a controlled treadmill study of nine dogs that compared the two then common harness types directly — the classic restrictive chest-strap harness and the Y-harness marketed as non-restrictive. The counter-intuitive result: shoulder extension was 2.6 degrees less at walk and 4.4 degrees less at trot under the non-restrictive harness than under the restrictive one — the variant advertised as joint-friendly therefore restricted the shoulder measurably more, not less.⁷

In other words: what enters the market under the label joint-friendly also influences forelimb mechanics measurably — and sometimes does so more strongly than the model from which it differentiates itself. A 2024 study in the journal Reinvention (University of Warwick) with 30 dogs refined the picture further. It compared front-clip, back-clip and straight-front models and found: the front-clip harness produced the largest reduction in elbow and shoulder movement; straight-front models performed best.⁸

Front-clip: a legitimate training tool, not daily equipment

The front-clip argument deserves a differentiated answer here. Front-clip harnesses are widespread in training circles because they impose a soft turning motion on pullers, rather than blocking them hard against the neck region. The argument is not absurd: in the context of short-term training, during which a dog still has to learn self-control on the lead, a model that gives mechanical feedback is clearly preferable to a collar.⁹

What does not follow from this: that the same model is suitable for daily long-term use. The Reinvention study in 2024 measured range of motion during ordinary gait, not during a corrective moment. Anyone who makes the front-clip model the dog-walking standard fixes the forelimb mechanics for hours at a time — in young animals during growth the long-term consequences are unclear; in older animals with prior conditions such as elbow dysplasia they are foreseeably negative. The serious trainer consensus is accordingly: use the front-clip for targeted training; switch to a straight-front model for everyday use.

What the market says

The economic dimensions help to read the weight of the marketing. Published market estimates for 2024 vary considerably depending on market definition and methodology. Verified Market Research puts the global dog-harness market in 2024 at around USD 134 million, with a forecast of USD 168 million by 2032 (CAGR 3.6 percent); Cognitive Market Research and similar providers reach higher values in the low single-digit billions, but define the market more broadly (including leashes and collars).¹⁰ Across reports the United States dominates regionally (around 38 percent of revenue per Verified Market Research); Julius-K9, Hurtta, Ruffwear and EzyDog are routinely named as reach-strong brands across Europe. The advertising priorities of these manufacturers favour three messages that are often harder to back up than they sound: ergonomic, joint-friendly, Y-shape.

Concretely: the term “Y-harness” guarantees no biomechanical compatibility — only a cut form. The Lafuente study in 2018 identified exactly the Y-model as markedly shoulder-restrictive. Buying a model advertised on this logic is buying shape, not shoulder freedom.

What anatomy demands

In practical terms the matter is sharpened by the fact that there is no such thing as the dog. A chondrodystrophic breed with short, sturdy legs — Dachshund, Old English Bulldog, French Bulldog — brings a different chest-shoulder mechanic than a long-limbed sighthound or a German Shepherd. In short-legged breeds the sternum sits at a depth where even a slightly misfitting chest harness can press in the armpit; in long-limbed breeds the same cut may lie comparatively harmlessly. A proper fitting procedure — dog standing, dog walking, dog lifting the forepaw — is therefore not a luxury but a duty.

Anyone without the option to test in person should have the model sent by the manufacturer in several sizes, check free shoulder extension, and return the surplus pieces. The few weeks this costs are the smaller investment than an orthopaedically affected animal after three years of walks with a badly fitting harness.

Selection criteria

From the sum of the research and clinical practice, five points can be distilled:

1. Collars only as ID carriers. ID tag, emergency information, a GPS tracker — that is the documented indication. As a pull point on the walk, on current biomechanical evidence, it is barely defensible.
2. Straight-front before front-clip. Front-clip is a sensible training tool, not daily equipment. Daily walks belong on a harness whose leash attachment sits on the back and disturbs shoulder extension little.
3. The Y-cut is a shape, not a safety claim. Biomechanical measurements depend on the position of chest and armpit straps, not on the advertising message. Fitting is required.
4. For brachycephalic, chondrodystrophic, or already orthopaedically affected animals the choice is medical. Consultation with a veterinarian specialised in orthopaedics is not over-thinking here; it is appropriate.
5. Annual fit check. Dogs grow, muscle changes, harnesses warp. What fit in the first year does not automatically sit in the fifth.

Balance

The simple message “harness instead of collar” describes a correct but incomplete movement. The turn away from the neck is one answer — but the follow-up question where does the pressure go? is not settled by the purchase of any arbitrary harness model. The biomechanical studies of the last ten years — from Lafuente 2018 through the Reinvention paper 2024 to Bailey 2025 — agree on one point: there is no the harness recommendation. There is the individual choice for the individual animal — and it cannot be made without anatomical knowledge, without honest fitting, and without at least a glance at the peer-reviewed literature.

Anyone walking their dog on a leash daily for the next ten years needs more for this decision than the recommendation of the sales clerk in the pet store.

---