Dog Bites and Children: What the Risk Data Show and What Prevention Achieves

1. What the Numbers Actually Say

1.1 Hospital Admissions and Who Is in Them

Twenty years of English hospital episode data give the clearest picture available. Between 1998 and 2018, admissions for dog bites rose from 6.34 per 100,000 population (95% CI 6.12–6.56) to 14.99 per 100,000 (95% CI 14.67–15.31) — more than doubling (Tulloch et al., 2021).

The important finding is who drove that rise. It was adults. Children had among the highest age-specific admission rates across the whole period, but the child rate showed no obvious annual trend, averaging around 14.44 per 100,000. Boys were admitted more often than girls.

Two things follow. Children remain a high-risk age group for hospital admission, and the marked rise in admissions over time was driven by adults rather than children.

1.2 What Hospital Data Leaves Out

Admissions are the tip. In the same dataset, only 2.05% of emergency department attendances for dog bites resulted in admission (95% CI 0.93–3.17), so roughly fifty dogs bite someone badly enough for an emergency visit for every one that produces an admission.

Below the emergency department the data thin out almost completely. In a survey of 484 self-reported bite victims, 62.3% of injuries required no medical intervention at all (Oxley, Christley & Westgarth, 2018). Those bites exist only in the household's memory, and they are the ones a trainer is most likely to be told about.

This matters for interpretation. Severity data come from hospitals, frequency data come from surveys, and the two populations are not the same. Any statement of the form "most dog bites are X" needs to say which dataset it is drawn from.

1.3 Two Age Peaks, Two Different Problems

Pediatric bite presentations cluster at two ages: under two years, and nine to twelve years, with children under nine making up the majority of cases (Jakeman et al., 2020).

These are not the same phenomenon. A toddler bitten in the home at face height during an interaction it initiated, and an eleven-year-old bitten on the arm by a dog it approached outside the home, differ in circumstance, in prevention, and in which adult could have intervened. Treating "child bites" as one category loses the distinction that matters most.

2. How This Gets Studied, and What Each Design Can Show

Five designs produce almost everything that is known here, and each is blind in a different place. Knowing which one a claim comes from decides how much weight it carries.

2.1 Hospital Episode Statistics

National admission records cover a whole population with no sampling error and span decades, which makes them the only source that can show a trend. They record the injury and almost nothing else: not the dog, not the trigger, not what happened in the minute beforehand. They also capture only the severe end, so they describe surgery rather than behavior.

2.2 Victim Surveys

Self-report surveys reach the bites that never get to a hospital, which is most of them, and can ask about context. They pay for it with recall and selection: people who volunteer a bite story are not a random sample, and the account of a bite is reconstructed afterward by someone who was frightened at the time.

2.3 Referral Case Series

A behaviorist examining dogs after they have bitten a child produces the richest data available on the dog itself — guarding, anxiety, medical findings, prior history. The population is heavily selected: these are households that sought referral rather than rehoming or euthanasia, which is a minority. The figures describe referred biting dogs, not all biting dogs.

2.4 Video Coding

Coding recorded bites second by second is the only method that observes the event itself rather than reconstructing it. Nobody's memory is involved. But publicly posted videos are selected in ways that cannot be measured — filming was already happening, and the clip was worth uploading — and the sample is small once detailed coding is applied.

2.5 Trials and Systematic Reviews

Randomized trials of prevention programs can establish causation, and almost none exist. Where they do, they measure what is easy to measure — knowledge, or behavior toward a model dog — rather than injuries. Systematic reviews then inherit whatever the primary studies measured, which is why a review can be methodologically sound and still unable to answer whether anything prevents bites (how behavior gets operationalized).

2.6 Where They Agree

The agreement across designs is what makes the picture usable. Hospital data, victim surveys and referral series independently put the family dog in the home at the center. Video coding and signal-recognition studies independently converge on the quiet end of the behavioral range. A finding supported by two designs with different blind spots is worth more than a large number from one.

3. Where Children Get Bitten, and by Which Dog

3.1 The Family Dog, in the Home

Approximately 70% of bites to children involve a family-owned dog, and most occur in the home (Jakeman et al., 2020). In the broader victim survey, 66.1% of biting dogs were known to the person bitten and 40.5% of incidents happened on private property (Oxley, Christley & Westgarth, 2018).

The stray-dog image is close to useless for prevention. The dog that bites a child is usually the one the child lives with, in the room the child spends most time in, during an interaction nobody regarded as risky.

3.2 Body Regions by Age

Children under five sustain head and neck injuries most often; children of nine and over are bitten more on the extremities (Jakeman et al., 2020). The mechanism is height. A small child's face sits at the level of a standing dog's head, and a small child approaches face first.

The victim survey shows the same pattern from the other direction: head and neck accounted for 6.8% of all bites, but children under nineteen made up 78.6% of those — 15 of 22 head and neck incidents involved children aged nought to nine (Oxley, Christley & Westgarth, 2018). Upper extremities were the commonest site overall at 62.4%.

There is a severity gradient too. In video-recorded bites, infants were bitten more severely than older children (Owczarczak-Garstecka et al., 2018).

3.3 What the Child Was Doing

In the victim survey the largest single context, at 34.7%, was the person interacting with the dog — stroking, playing, handling, restraining. Walking past or near a dog accounted for 20.1%.

A separate analysis of 132 bites to the face found that 76% were preceded by the person bending over the dog, and 19% by putting their face close to the dog's face. More than two thirds of the victims were children, and none of the victims was an adult dog owner (Rezac, Rezac & Slama, 2015). Only adult dogs bit the face, and bites from large dogs required medical treatment more often than bites from small dogs.

That combination — bending over, face to face, by someone who is not the dog's owner — describes one of the clearest repeatedly identified high-risk interaction patterns in the literature, and it is also exactly how adults teach children to greet a dog.

4. The Triggers

4.1 Resource Guarding in the Youngest Children

A behavioral workup of 111 children bitten by 103 dogs, referred after the bite, separated the triggers by the child's age. In children under six, the commonest context was the dog guarding a resource: 44% of cases. Where the dog was familiar to the child, food guarding was the single commonest trigger at 42% (Reisner, Shofer & Nance, 2007).

Resource guarding is not aggression toward the child as such. It is a dog defending access to something, with a child happening to be the one approaching (resource guarding in dogs). The practical consequence is that the highest-risk minute of a toddler's day is often mealtime, and that the risk belongs to the arrangement rather than to the child's behavior.

4.2 Territory and the Older Child

In older children the commonest context shifted to territorial guarding, at 23% of cases. Where the dog was unfamiliar, territorial guarding was the commonest trigger at 53% (Reisner, Shofer & Nance, 2007).

That tracks what older children do: they leave the house, pass gates and fences, approach dogs they do not know, and do so without an adult nearby (territorial behavior in dogs). The prevention target is different from the toddler case — not household engineering but what the child does away from home.

4.3 The Resting Dog

A survey of 402 caregivers of children up to six years old recorded what the interactions actually look like. Hugging the dog was frequently observed by 46% of caregivers and kissing it by 27%. A quarter reported the child frequently taking its own toys back from the dog (Arhant, Beetz & Troxler, 2017).

Hugging is worth pausing on. It restricts the dog's movement, puts the child's face at the dog's head, and is read by most people as affection in both directions. For the dog it is a sustained restraint with no exit, which is the structure of an approach–avoidance situation rather than a social one (approach–avoidance conflict).

4.4 Discipline as a Trigger

In the referred cases, 59% of the dogs showed aggression when disciplinary measures were applied to them (Reisner, Shofer & Nance, 2007). Those measures were applied by adults, but their consequences land on whoever is next in the dog's space.

This is the point where the training method used in a household becomes a child-safety question rather than an ethical one. A dog that has learned that human approach can mean a physical correction has learned something that will be applied to small approaching humans too (punishment fallout, what aversive methods do neurologically).

4.5 Movement, Noise and Play

One mechanism is conspicuously absent from the trigger data and conspicuously present in practice: children move fast, at unpredictable angles, with high-pitched vocalization, at the height of a dog's head.

The referral series classified triggers as guarding, territorial, discipline-related and so on; a running, shrieking child does not fit those categories, and it is not clear whether such incidents were coded elsewhere or are under-represented because they are read as play gone wrong rather than aggression. The motivational systems involved are also different from defensive aggression — chase and grab sequences belong to a predatory rather than an agonistic repertoire (predatory behavior in dogs), and a high-arousal dog in a game has reduced behavioral control regardless of its intentions (arousal regulation).

This is stated as a gap rather than a finding. The practical handling is conservative: games between children and dogs that involve running, chasing or tugging get an adult-imposed structure and an end, not because the data demand it but because the data do not cover it.

5. The Biting Dog Was Usually Not a Known Biter

5.1 What the Behavioral Workup Found

The same referral series examined the dogs. Resource guarding was present in 61%. An anxiety abnormality was identified in 77%. And the figure that should change how these cases are discussed: 66% had never previously bitten a child, and 19% had never bitten any human (Reisner, Shofer & Nance, 2007).

Most of these dogs were also doing what responsible ownership prescribes. Ninety-three percent were neutered and 66% had attended obedience training.

5.2 Half of Them Had a Medical Problem

A potentially contributing medical condition was identified or suspected in 50% of the referred dogs. That is not a side finding. The study could not establish whether, or how much, those findings contributed to the bite. Pain and other illness can alter tolerance and defensive behavior, and dogs do not always display discomfort in the way people expect (why pain is missed in dogs, chronic pain and aggression).

For a household with children, this converts a routine veterinary check from good husbandry into a safety measure, and it makes any change in a dog's tolerance a reason to examine the dog rather than to correct it.

5.3 Why "He Never Did This Before" Is the Normal Case

Two thirds of these dogs genuinely had not bitten a child before. The sentence owners say after an incident is usually true, and it is usually useless as reassurance, because it describes the typical biting dog rather than an exception.

What it means for risk assessment is uncomfortable. Absence of a bite history is weak evidence of safety around children. What does carry information is the behavior these dogs showed beforehand — guarding, anxiety, reaction to being handled or corrected — none of which requires a bite to be visible (behavioral assessment in dogs).

6. "It Came Without Warning"

6.1 What the Video Study Found

Researchers coded 143 publicly posted videos of dog bites, with detailed second-by-second behavioral coding on 56 of them. Non-neutral codes for dog body posture, and some displacement and appeasement behaviors, increased from approximately 20 seconds before the bite; human tactile contact with the dog increased from about 21 seconds before (Owczarczak-Garstecka et al., 2018). The qualifier matters: the study reports no universal sequence, and not every behavior commonly described as an appeasement signal showed a clear temporal pattern.

So in this selected sample of recorded incidents, observable behavior changed well before the bite, and it changed on both sides: more touching by the person, more non-neutral posture and some appeasement from the dog. What the study does not establish is a fixed warning period or a reliable order of signals.

6.2 Why the Warning Was Not Seen

Twenty seconds is a long time to miss something. Three things explain it.

The signals that appear first are the quiet ones. Displacement and appeasement behavior — lip licking, nose licking, turning the head away, yawning, freezing — do not look like threat to an untrained observer, and are frequently read as the opposite (canine body language).

The behavior that matters is subtractive. A dog that stops wagging, stops moving, stops eating or stops looking at you has told you something, and absence is harder to notice than presence.

And the human half of the interaction was escalating at the same time. Someone leaning in and touching more is someone whose attention is on their own behavior rather than on the dog's.

6.3 What Adult Victims Say

Interviews with eight women bitten within the previous five years found a consistent pattern. Most knew, in principle, what dog aggression signals look like — and those who worked with dogs regularly tended to believe it would not happen to them. Blame was assigned to themselves or to the dog's owner, never to the dog. An attitude that the bite was "just one of those things" worked against prevention (Westgarth & Watkins, 2015).

The sample is small and qualitative, and it concerns adults rather than children. Its value is in naming the mechanism: knowing the signals and applying them to your own situation are different skills, and professional familiarity with dogs appears to weaken rather than strengthen the second one.

6.4 The Bites Nobody Records

In the victim survey, 62.3% of injuries required no medical intervention (Oxley, Christley & Westgarth, 2018). Those incidents leave no trace in any dataset, and they are the ones a trainer hears about first.

Two consequences. The household has usually already had the warning and classified it as a scratch, a nip or the dog being grumpy, so the history worth taking is not "has the dog ever bitten" but "has the dog ever made contact with teeth, and what was happening." And because the minor incidents outnumber the serious ones heavily, a household's own experience systematically understates what the dog is capable of — it has been sampling the low end of the distribution and concluding that is the whole of it.

The corollary is that the useful unit for risk assessment is not the bite but the event: an approach that was interrupted, a snap that missed, a growl over a bone. Those are observable, frequent, and not in anybody's statistics (why behavior is not the same as emotion).

7. Reading the Dog: What Children Can and Cannot Do

7.1 The Signal Hierarchy Children Get Wrong

A study trained 124 children aged three to five, and 40 parents, to identify dog signals from video, with 88 children completing the full longitudinal protocol (Meints, Brelsford & De Keuster, 2018).

Performance depended entirely on which kind of signal was being shown. Conflict-escalating signals — the loud ones — were recognized reasonably well and improved with training: five-year-olds went from 64% to 83% correct. Conflict-avoiding signals were poorly recognized throughout. Conflict-defusing signals were worst: three-year-olds scored 14 to 16% correct, before and after.

The two literatures use different classifications and should not be collapsed into one: Meints measured conflict-defusing signals, the video study coded posture and displacement behavior. What they jointly support is narrower and still decisive — the youngest children scarcely recognize subtle canine discomfort, and subtle canine discomfort is what changes before a bite. Relying on a three-year-old to notice it is unrealistic.

7.2 What Training Changed and What It Did Not

Three-year-olds improved from 47% to 50% on the escalating signals — which is to say, not meaningfully. Five-year-olds improved substantially on the same category. Neither age improved much on the quiet signals.

The honest reading is a developmental ceiling. A single-session video intervention moved five-year-olds on the easy category and moved three-year-olds almost nowhere. Whether a longer or repeated program would do better is untested, which is itself worth stating (developmental windows in dogs have an obvious human counterpart here that the literature has not pursued).

7.3 The Parents

The parents in the same study reached 100% correct on conflict-escalating signals after training. That is the most practically useful finding in this literature, with one limit stated plainly: it shows a substantial short-term improvement in signal recognition, not that a single session produces lasting household skill or fewer injuries.

It also relocates the target. If three-year-olds remain poor at recognizing quiet canine distress while adults improve quickly, prevention should not make the young child the primary safety system.

8. Supervision: The Word That Hides the Problem

8.1 What Caregivers Report Happening

"Never leave a child alone with a dog" is the most repeated advice in the field, and it is not wrong. It is also not what the data describe.

Among the 402 caregivers of children up to six, the reported interactions include hugging (46% frequently), kissing (27% frequently), and the child taking its own toys back from the dog (27% frequently) (Arhant, Beetz & Troxler, 2017). These are supervised interactions. An adult was present, watching, and did not intervene — because nothing visible was wrong.

That is the gap the word hides. Many bites to young children happen despite an adult being present. Supervision as usually practiced means being physically present and intervening when something looks obviously dangerous, and the literature above says the changes that precede a bite are not the obvious ones.

8.2 Attentiveness Goes Down as the Child Gets Older

The same survey found that caregivers' attentiveness decreased as the child got older, while the complexity of the child's interactions with the dog increased. The authors note that parental supervision does shape the child's behavior effectively — it is applied least where it would do most.

This fits the two-peak pattern in the admission data. The second peak, at nine to twelve, is the age at which supervision has largely been withdrawn and the child has started approaching dogs independently.

8.3 What Supervision Would Have to Mean

To do the job attributed to it, supervision would have to involve three things that ordinary presence does not.

Watching the dog rather than the child. The informative behavior is the dog's, and most adults watch the child.

Intervening on quiet signals rather than loud ones. A growl is late. The head turn, the lip lick and the freeze are the point of intervention, and the adult needs to know them — which, per the training data, takes one session.

Interrupting interactions that are going well. The bites in the video study were preceded by escalating contact, not by conflict. An interaction that has to be stopped while it still looks fine is counterintuitive, and it is what the data support.

9. Does Prevention Education Work?

9.1 The Cochrane Answer

A Cochrane review looked specifically for evidence that educating children and adolescents reduces dog bite injuries. Two randomized controlled trials met inclusion criteria, both in kindergarten and primary schools, both of moderate methodological quality (Duperrex et al., 2009).

Neither measured injuries. They measured inappropriate behavior around dogs and knowledge. One found reduced inappropriate behavior after a 30-minute lesson; the other found increased knowledge and caution.

The reviewers concluded that there is no direct evidence that educational programs reduce dog bite rates in children and adolescents — because the included trials did not measure injuries — and that education should not be the primary public health strategy. They called for studies using bite rates as the outcome. That was 2009.

9.2 Knowledge Versus Behavior

The gap was restated more recently in a pediatric review: educational schemes have the potential to increase knowledge of safe behavior around a dog in the short term, but do not appear sufficient to change children's behavior in the presence of a real dog, and education alone is unlikely to prevent bites (Jakeman et al., 2020).

Two mechanisms explain it. Knowing a rule and applying it under excitement are different, and children are usually excited around dogs (arousal and behavior applies to both species in the room). And the rules taught are mostly about what not to do with an unfamiliar dog, while most bites come from the family dog in a familiar room.

9.3 Why This Is Not an Argument Against Teaching

Two careful distinctions keep this from becoming nihilism.

Absence of evidence is not evidence of absence. Two moderate-quality trials that did not measure injuries cannot show that education fails to prevent injuries. They show that nobody has tested it properly.

And the target can be moved. What the evidence does not support is education of young children as the main line of defense. Adults improve their recognition of the critical signals after brief training and control the environment the child is in — though whether educating adults reduces injuries has not been tested either.

10. The Breed Question, Handled Honestly

10.1 What the Meta-Analysis Found

A systematic review with meta-analysis of facial dog bite injuries pooled 43 studies reporting more than 26,000 bites, alongside a 240-patient case series from two trauma centers. It found that injuries from pit bulls and mixed-breed dogs were both more frequent and more severe, and that breeds vary in rates of biting and in severity (Essig et al., 2019).

That is what the paper says, and pretending otherwise would be dishonest. It is also the paper most often cited by both sides of the breed argument, usually without its own caveats.

10.2 What Undermines It

Of the more than 26,000 bites, only 39.9% had a specific breed attributed at all. The remainder were unknown or recorded as mixed breed. The authors state the limitation themselves: breed reporting varied substantially in accuracy, and the preponderance of non-specific breed information constrained what could be concluded.

Three further problems apply to every study of this design. Breed in a hospital record is usually assigned by a bystander from appearance, which is unreliable. The denominator — how many dogs of each breed exist in the population — is almost never known, so frequency of appearance in a bite series cannot be converted into risk per dog. And severity confounds with size and jaw morphology independently of breed.

A pediatric review covering the same ground concluded that the evidence for specific breed, sex or age of dog as risk factors is both limited and conflicting (Jakeman et al., 2020) (breed versus behavior).

10.3 What Does Carry Information

A systematic assessment of fourteen papers on early experience and later human-directed aggression identified the source of the dog, the age at which it was rehomed, the reason for acquisition, and the owner's level of experience among the factors associated with risk, and proposed that early experiences limiting the development of effective coping strategies may matter particularly (Baslington-Davies et al., 2023). The authors are explicit that individual studies had methodological limitations and that they deliberately made best use of imperfect evidence rather than applying strict thresholds.

Taken together with the referral data — guarding in 61%, an anxiety abnormality in 77%, a medical condition in 50% — what carries information for an individual risk assessment is the dog's history, behavior, context and health rather than its appearance. That is less satisfying than a breed list and considerably more useful in a consultation.

11. What Actually Reduces Risk in a Household

11.1 Do Not Make the Child the Primary Safety System

The evidence points consistently in one direction. Children under about five cannot reliably read the signals that precede a bite. Adults can learn them quickly. Most bites happen in the home, during supervised interaction, with a dog that has never bitten.

What follows is a management principle rather than a tested protocol: let the environment carry as much of the safety burden as possible, instead of relying on anyone noticing risk in time. The arrangements below are derived from the observed trigger patterns; none of them has been tested in a trial using bite reduction as the outcome.

11.2 The Four Situations to Engineer Out

Food and possession. Resource guarding was the commonest trigger in the youngest children and was present in 61% of the referred dogs. The dog eats and chews behind a closed door, every time, regardless of how reliable it seems.

The resting dog. A sleeping or resting dog is approached by a child who cannot be expected to recognize that approach as intrusive. The dog's resting places are physically inaccessible to the child — not off limits by instruction.

Face-to-face greeting. Bending over the dog preceded 76% of facial bites. Children greet dogs from standing, at a distance, with the dog free to approach or not.

Restraint disguised as affection. Hugging, holding and kissing are reported as frequent by a quarter to a half of caregivers, and they are the structure that removes the dog's exit. This one is an instruction to the adults, because the child is doing what it has been shown.

11.3 Teach the Dog an Exit

The most underused intervention is on the dog's side. A dog that has a reliable, rewarded place to go, and has never been retrieved from it or corrected in it, has an alternative to escalating. Dogs that have learned that leaving does not work, because they were followed or corrected, escalate faster.

This is also where handling training earns its place as safety work rather than convenience: a dog accustomed to being touched, held and examined on its own terms has a wider margin before contact becomes threatening (cooperative care, desensitization and counterconditioning).

11.4 The Dog That Was There First

A household expecting a baby is the one case where the whole arrangement can be built before the risk exists, and it is routinely left until the child is mobile.

The evidence does not include a trial of pre-birth preparation, so this is reasoning from the trigger data rather than a tested protocol. What the trigger data say is that the four high-risk situations all involve space, possessions and rest. All four can be established as habits months before a child can crawl: the dog already eats behind a door, already has resting places a child will not be able to reach, already has a rewarded place to go when the household gets loud, and has already been through the equipment, the noise and the changed routine without anything bad happening (habituation and sensitization).

The alternative — imposing all of it in the week the child starts moving — takes everything away from the dog at once and correlates the loss with the child. That is a worse starting position than the same arrangement introduced neutrally a year earlier.

11.5 The Visiting Child

A dog that lives without children and meets them occasionally is in a different position from a family dog, and the data touch it only indirectly: territorial guarding was the commonest trigger where the dog was unfamiliar to the child, at 53% of those cases (Reisner, Shofer & Nance, 2007).

A visit combines several of the documented risk factors at once. The dog is on its own territory, the child is unfamiliar and has no established pattern with it, nobody present knows this particular dog's quiet signals, and the adults are talking to each other. Expecting the interaction to be read correctly under those conditions is unrealistic.

The conservative handling is separation by default for the duration of the visit, with any contact brief, structured and ended early — which is also the handling least likely to be adopted, because it looks unfriendly toward both the child and the dog (why a dog's experience does not generalize automatically).

11.6 What to Do With a Dog That Has Growled

A growl at a child in a household is the most useful information a family will ever get, and the standard response destroys it. Punishing a growl can suppress an observable warning without changing the motivation behind it, which leaves the household with less information before the next escalation (punishment fallout).

The correct response has three parts: separate immediately and without drama, get the dog examined for pain or illness, given that potentially contributing medical conditions were identified or suspected in half of the referred dogs, and identify the trigger precisely enough to engineer it out rather than training the dog to tolerate it (aggression in dogs, trigger stacking).

And the honest part: some combinations of dog and household are not safe to manage, and the families that most need to hear that are the ones being reassured that the dog never did this before. Two thirds of the dogs in the referral series had not.

12. Research Gaps

No trial has used bite rates as an outcome. The Cochrane reviewers asked for this in 2009 and it has not been delivered. Every claim that a prevention program works rests on knowledge or observed behavior, not on injuries.

The adult-education pathway is untested. Parents reach near-perfect recognition of escalating signals in one session. Whether that translates into different household arrangements, and whether those arrangements reduce bites, has not been studied.

Denominators are missing throughout. Without population-level data on how many dogs of each type live with children, bite series cannot be converted into risk, which is the single largest obstacle to answering the breed question.

The video findings need replication outside social media. A 20-second behavioral lead-in is the most actionable finding in the field, and it comes from publicly posted videos, which are a selected sample in unknown ways.

The medical contribution is barely characterized. A medical condition in half of referred biting dogs is a striking figure from one referral series. No study has quantified how often undiagnosed pain contributes to bites in a general population.

13. Conclusion

The evidence on children and dog bites is more coherent than the advice derived from it. Children are among the age groups most often hospitalized, but their rate has been flat for two decades while the overall rate more than doubled, driven by adults. The dog is usually the family's own, in the home, and usually has no bite history. The triggers are mundane: food, possessions, a resting dog, a child bending over its face. A potentially contributing medical condition was identified or suspected in half of the dogs examined after the event.

And there is a warning, with a caveat attached. In recorded bites, non-neutral posture and some appeasement behavior changed from approximately 20 seconds beforehand while the person touched the dog more — no fixed sequence, but a change on both sides well ahead of the event. That change sits at the quiet end of the signal range, which three-year-olds score 14 to 16% on and which many adults do not recognize as distress at all.

Put together, those two findings condemn the standard approach. Teaching young children to read dogs targets a skill they do not yet have, and supervising by watching for something obviously dangerous targets a signal that does not arrive until it is too late. The Cochrane verdict on child education — no direct evidence of reduced bite rates — has stood since 2009 and nobody has tested it properly since.

What the same literature supports is less popular and more boring. The adults learn the quiet signals, which takes one session — though whether that reaches real-world injury rates is untested. The household is arranged so that the four recurring high-risk situations cannot occur, rather than being watched for. The dog gets a place it is never disturbed in and an exit nobody punishes. And when a dog growls at a child, the growl is treated as information instead of being suppressed, because suppressing the announcement leaves the household with less warning, not less risk.

Key Insights

Children's bite admission rate has not risen. England's rate more than doubled over twenty years, driven by adults; the child rate showed no trend.

The dog is usually the family's own. About 70% of bites to children involve a family dog, mostly in the home.

Two thirds had never bitten a child before. "He never did this" describes the typical biting dog, not an exception.

Half of the referred biting dogs had a potentially contributing medical condition. A change in tolerance is a reason to examine the dog, not to correct it.

Bending over the dog preceded 76% of facial bites. It is also how adults teach children to greet dogs.

The change begins around 20 seconds before the bite. Posture and some appeasement behavior shift that far ahead — without a fixed sequence.

Three-year-olds score 14 to 16% on the signals that come first. Training did not move that figure.

Parents reached 100% on escalating signals after one session. The trainable party is the adult; whether it reaches injury rates is untested.

Adult presence alone does not reliably prevent bites. Supervision means active attention to the dog, not being in the room.

Attentiveness drops as the child gets older. Just as the child starts approaching dogs on its own.

No trial has measured whether education reduces bites. Cochrane asked for one in 2009.

Breed data cannot answer the risk question. Only 39.9% of 26,000 pooled bites had a breed attributed, and the denominator is unknown.

Punishing a growl can suppress the warning without changing the motivation. That leaves the household with less information, not less risk.

References

Arhant, C., Beetz, A. M., & Troxler, J. (2017). Caregiver reports of interactions between children up to 6 years and their family dog – implications for dog bite prevention. Frontiers in Veterinary Science, 4, 130. https://doi.org/10.3389/fvets.2017.00130

Baslington-Davies, A., Howell, H., Hogue, T. E., & Mills, D. S. (2023). An assessment of scientific evidence relating to the effect of early experience on the risk of human-directed aggression by adult dogs. Animals, 13(14), 2329. https://doi.org/10.3390/ani13142329

Duperrex, O., Blackhall, K., Burri, M., & Jeannot, E. (2009). Education of children and adolescents for the prevention of dog bite injuries. Cochrane Database of Systematic Reviews, 2, CD004726. https://doi.org/10.1002/14651858.CD004726.pub2

Essig, G. F., Sheehan, C., Rikhi, S., Elmaraghy, C. A., & Christophel, J. J. (2019). Dog bite injuries to the face: is there risk with breed ownership? A systematic review with meta-analysis. International Journal of Pediatric Otorhinolaryngology, 117, 182–188. https://doi.org/10.1016/j.ijporl.2018.11.028

Jakeman, M., Oxley, J. A., Owczarczak-Garstecka, S. C., & Westgarth, C. (2020). Pet dog bites in children: management and prevention. BMJ Paediatrics Open, 4(1), e000726. https://doi.org/10.1136/bmjpo-2020-000726

Meints, K., Brelsford, V., & De Keuster, T. (2018). Teaching children and parents to understand dog signaling. Frontiers in Veterinary Science, 5, 257. https://doi.org/10.3389/fvets.2018.00257

Owczarczak-Garstecka, S. C., Watkins, F., Christley, R., & Westgarth, C. (2018). Online videos indicate human and dog behaviour preceding dog bites and the context in which bites occur. Scientific Reports, 8, 7147. https://doi.org/10.1038/s41598-018-25671-7

Oxley, J. A., Christley, R. M., & Westgarth, C. (2018). Contexts and consequences of dog bite incidents. Journal of Veterinary Behavior, 23, 33–39. https://doi.org/10.1016/j.jveb.2017.10.005

Reisner, I. R., Shofer, F. S., & Nance, M. L. (2007). Behavioral assessment of child-directed canine aggression. Injury Prevention, 13(5), 348–351. https://doi.org/10.1136/ip.2007.015396

Rezac, P., Rezac, K., & Slama, P. (2015). Human behavior preceding dog bites to the face. The Veterinary Journal, 206(3), 284–288. https://doi.org/10.1016/j.tvjl.2015.10.021

Tulloch, J. S. P., Owczarczak-Garstecka, S. C., Fleming, K. M., Vivancos, R., & Westgarth, C. (2021). English hospital episode data analysis (1998–2018) reveal that the rise in dog bite hospital admissions is driven by adult cases. Scientific Reports, 11, 1767. https://doi.org/10.1038/s41598-021-81527-7

Westgarth, C., & Watkins, F. (2015). A qualitative investigation of the perceptions of female dog-bite victims and implications for the prevention of dog bites. Journal of Veterinary Behavior, 10(6), 479–488. https://doi.org/10.1016/j.jveb.2015.07.035