Article Text

Systematic review of dog bite prevention strategies
  1. Natasha Duncan-Sutherland1,
  2. Abbey C Lissaman2,
  3. Michael Shepherd3,4,
  4. Bridget Kool5
  1. 1Adult Emergency Department, Auckland Hospital, Auckland, New Zealand
  2. 2Department of Physiology, The University of Auckland, Auckland, New Zealand
  3. 3Starship Children's Emergency Department, Starship Foundation, Auckland, New Zealand
  4. 4Department of Paediatrics: Child and Youth Health, University of Auckland, Auckland, New Zealand
  5. 5Section of Epidemiology and Biostatistics, University of Auckland, Auckland, New Zealand
  1. Correspondence to Dr Natasha Duncan-Sutherland, Adult Emergency Department, Auckland Hospital, Auckland, New Zealand; quavernote{at}


Background The prevention of dog bites is an increasingly important public health topic, as the incidence of serious injury continues to rise.

Objectives To evaluate the effectiveness of interventions to prevent dog bites and aggression.

Methods Online databases were searched (PubMed, Cochrane Library, Embase and Google Scholar), using the search terms: dog/s, canine, canis, kuri, bite/s, bitten, aggression, attack, death, fatal, mortality, injury/ies, prevention, intervention, for studies between 1960 and 2021. All study designs were considered. Outcomes of interest were the incidence of dog bites or dog aggression. Non-English studies, and those without full-text access were excluded.

Results Forty-three studies met the review criteria, including 15 observational and 27 interventional studies. Fifteen studies investigating dog-control legislation, including leash laws, stray dog control and infringements indicated this can reduce dog bite rates. Breed-specific legislation had less of an effect. Six studies investigating sterilisation, showed while this may reduce dog bites through a reduction in the dog population, the effect on dog aggression was unclear. An alcohol reduction programme showed a significant reduction in dog bite rates in one study. Seven studies assessing educational approaches found that intensive adult-directed education may be effective, with one study showing child-directed education was not effective. Eight studies on dog training (two police-dog related), and six evaluating dog medication or diet were generally low quality and inconclusive.

Conclusions Multiple strategies including effective engagement with indigenous communities and organisations will be required to reduce dog-bites and other incidents involving dog aggression. This review provides some evidence that legislated dog control strategies reduce dog bite rates. Available evidence suggests greater restrictions should be made for all dogs, rather than based on breed alone. Due to a burden of child injury, protection of children should be a focus of legislation and further investigations. Prevention strategies in children require redirection away from a focus on child-directed education and future research should investigate the effectiveness of engineering barriers and reporting strategies.

  • Education
  • Legislation
  • Animal bites
  • Public Health
  • Engineering

Data availability statement

All data relevant to the study are included in the article. Not applicable.

This is an open access article distributed in accordance with the Creative Commons Attribution 4.0 Unported (CC BY 4.0) license, which permits others to copy, redistribute, remix, transform and build upon this work for any purpose, provided the original work is properly cited, a link to the licence is given, and indication of whether changes were made. See:

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.


The prevention of dog bites and other dog-related injuries is an increasingly important public health concern globally and in New Zealand (NZ), as the incidence of these events continues to rise, including during the current COVID-19 pandemic.1–3 As with other unintentional injuries, dog bites are not ‘accidents’, but preventable traumatic injuries. Annually in NZ around 10 951 people present to health professionals for a dog bite injury (242 per 100 000 people, 95% CI 240 to 245).1 Unacceptably high rates disproportionately affect our most vulnerable members of society: Māori (NZ’s Indigenous population), and those from low-socioeconomic areas.1 The prevalence of dog bites is also much greater as many cases do not seek medical attention.4 5 A 2015 UK cross-sectional survey reported an estimated annual dog bite incidence 1870 dog bites per 100 000 people (95% CI 1100 to 3180), with a quarter of respondents indicating they had been bitten by a dog at least once in their lifetime.5

Children and adults are equally as likely to sustain a dog bite that requires medical attention; however, children are more likely to suffer more serious injuries to the face, head or neck region, and have greater rates of hospitalisation than adults.6 Injuries to the hands are more common in adults and often occur when a person intervenes in a dog fight.4 Hand injuries can be serious with the potential for a significant loss of function and the sequelae associated with this.7

Substantial psychological consequences can follow an incident of dog aggression, with or without physical injury (including non-bite injuries), with several studies highlighting this as a significant traumatic event, with the potential for economic and social costs as well as the development of post-traumatic stress disorder.8–10 An NZ study of adult dog bite victims found that 87% of respondents described their injury as moderate to severe, with 72% describing psychological effects from the injury.11

Risk factors for dog bites include factors involving the dog, the physical environment, the owner or the victim, and occur in a variety of circumstances, within both private and public spaces, and urban and rural areas.11–13 There is a lack of robust evidence regarding dog factors such as the influence of breed or sex on dog aggression, due to a lack of appropriate studies with control groups, already existing restrictions on certain breeds and inaccuracy of breed identification.14–16 One recent Finnish study of pure-bred dogs showed differences in owner-reported aggression scores between breeds with Rough Collies, Miniature Poodles and Miniature Schnauzers being most aggressive. This is not reflected in published studies of dog bites or fatalities which depend in part on the proportion of popular breeds owned including a predominance of mixed breeds.11 17

Dogs that bite are most commonly owned by a neighbour, friend or relative (23%–83%), or an unknown person (20%–60%), and less commonly by the victim or family they live with (5%–30%).11–13 18–20 An NZ survey in 2002 of 535 adults who presented for medical attention following a dog bite, found that over one-third (36%) occurred in public places, 21% in the victims’ home and 43% in other private property; with 56% occurring in urban areas.11 Of note, 71% of dog-bite incidents were considered unprovoked. No NZ published data has investigated the geographical location of dog bite incidents among children.

Existing dog-bite interventions can be broadly categorised into education, engineering and enforcement, in line with the WHO’s framework for injury prevention,21 and will likely require input from multiple disciplines. Dog bite prevention in NZ has primarily focused on the 1996 Dog Control Act, which includes a set of national guidelines, directing predominantly locally governed legislation, and focusing on dog control in public spaces.22 The Act focuses on dog access legislation (leash laws), requirements for registration and microchipping, breed banning and restrictions for dogs considered to be dangerous based on either breed or behaviour. Restrictions for dangerous dogs can include infringements, disqualification of owners, sterilisation (neutering, spaying or chemical) with the perception that it reduces dog aggression, the use of muzzles and leashes in public, improved fencing, separate visitor access, a ban on re-homing from a shelter, signage on private property, owner-licensing, wearing of collars identifying them as high risk, or euthanasia.

Breed specific legislation (BSL) regarding restrictions for owners and dogs based on breed alone has been criticised as being ineffective and unfair for dogs and owners.23 24 The evidence for and against sterilisation has also been discussed.25–27

Improved reporting policies are likely to be an effective strategy, given a large number of bites are not reported to animal management services,11 13 28 29 and strong evidence suggesting that dogs who bite often have a history of aggression.11–13 30 31

Education has traditionally been a focus of dog bite prevention, and can be targeted toward dog-owners, children, parents, the general public or the dogs themselves (dog training). Messages vary broadly but are commonly based on the assumption that dog bites can be prevented by correctly interpreting a dog’s behaviour or not provoking a bite with an incorrect approach to a dog.32 Other educational messages and/or policies focus on the importance of puppy sourcing from appropriate breeders, early socialisation (the process of introducing a puppy to new experiences),33 or on the importance of exercising a dog, to prevent aggression and other behavioural issues.34 Dog training programmes also exist in NZ and are at times accompanied by the introduction of medication or a change in diet in dogs with behavioural issues including dog aggression. There are currently no standardised requirements for dog-training or the education of dog owners, children or parents in NZ.

In the home environment, engineering solutions such as baby gates, adequate fencing height, gate locks or separate dog areas are an emerging concept35 that have not been strongly promoted for the prevention of dog bites, despite their promotion in other areas of child unintentional injury.36 Neither is the use of home-care visits by well-child providers, which is proven to be an effective strategy for other unintentional trauma in children.37

Despite several published review articles that discuss dog bite prevention,32 35 38 39 no systematic reviews have evaluated the effectiveness of prevention strategies to reduce the incidence of dog bites or aggression. Therefore, the aim of this systematic review is to address this gap.


The methods of this systematic review were guided by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement.40 The methods for this review were developed after consultation with clinical and Indigenous leaders with a wide range of experience in the field.

Literature search

Four online databases were searched (PubMed, Cochrane Library, Embase and Google Scholar), using the search terms: “dog OR dogs OR canine OR canis OR kuri” AND “bite OR bites OR bitten OR aggression OR attack OR death OR fatal OR mortality OR injury” AND “prevention OR intervention”, for studies published between 1 January 1960 and 10 March 2021. Reference lists of located studies were reviewed for additional relevant studies, and several experts in the field were contacted for recommendations of further studies. The same method was applied to searches of the grey literature.

All study designs were included that investigated a dog-bite prevention strategy, including those that had a broader dog-related focus such as rabies prevention. There were no restrictions on dog-bite prevention interventions or dog environments (ie, free-roaming/stray, contained dogs). Outcomes of interest were the incidence of dog bites or dog aggression, or animal bites if a large proportion were caused by dogs. Non-English studies, and those without full-text access were excluded.

Screening and inclusion

The title and abstract of located studies were screened by a single researcher (ACL) to identify those for potential inclusion. Identified studies were evaluated in more detail by two researchers (ACL, ND-S), and exclusion criteria applied, with any discrepancies discussed and a collaborative decision made.


Information from the included studies was extracted and summarised, including authors, title, year of publication, country, study aims, design and length, participant demographics, description of the intervention, relevant outcomes measured, and key results.

The quality of included studies was critiqued by two researchers (ACL, ND-S) using the Critical Appraisal Skills Programme method of evaluating intervention studies.41 This included an appraisal of validity, reliability and concise reporting of results. Based on an analysis of strengths and limitations, the authors collaboratively assigned a study quality category (high, moderate or low), in line with current Cochrane recommendations.42

A meta-analysis was not possible due to the heterogeneity of study designs and outcomes.


There were 20 385 studies identified from the search criteria. Following a review of the title and abstract, the majority of studies (n=19 890) did not meet the inclusion criteria (figure 1). Many of these excluded articles did not study a specific intervention, often describing epidemiology only, or lacked outcomes of interest, including investigating rabies vaccination rates only. Four hundred and sixty-eight were further excluded, with 138 duplicates, 32 non-English, 28 with no full-text access and 270 that also did not investigate a specific intervention, or relevant outcomes.

Figure 1

Flow chart describing the process of inclusion and exclusion of studies.

Among the included studies, most were conducted in the Northern Hemisphere, with only three from Australia43–45 and one from NZ46 (online supplemental table 1). Study designs included 15 observational cohort studies46–59 and 27 interventional studies, 18 of which were predesign and postdesign with no control group,43 45 60–75 five non-randomised studies with a control group,44 76–79 three cross-over studies,56 61 80 and one randomised controlled trial (RCT).81 One study had inadequate information given to determine the study design.65 Two studies were located from the grey literature.50 69

Dog bites were measured in 27 studies, including four studies that reported animal bites (predominantly dog bites),70 75 77 82 a study reporting all dog attacks (including being rushed at by a dog),44 a study reporting bite ratio, a measure used by the police force54 and a study measuring bite attempts.76 Dog bite incidence as a proportion of the population was calculated in 14 studies.43–47 50 51 57 62 71 77 79 83 84 Dog bites were measured from a variety of data sources including hospital admissions, emergency department (ED) or primary care presentations, household surveys, or reports to animal management or public health services. Dog aggression was measured in 15 studies.55 56 59–61 63 66–68 72 73 78 80 81 85

Interventions identified focused on six main areas: dog-control legislation, sterilisation, alcohol reduction, education of people, dog training programmes, and medication or diet (online supplemental table 1). Fifteen studies investigated the effect of dog-control legislation, with a focus on either general dog control, or breed-specific strategies.44–52 57 65 74 75 83 84 Six studies investigated sterilisation.58 59 63 64 81 82 One study investigated a community alcohol reduction programme.43 Seven studies focused on education; with five on public education,69–71 77 79 and one each directed at children,62 and dog owners.78 Eight studies investigated predominantly dog-training,53–55 60 72 73 76 85 with a further six studies investigating predominantly medication or diet.56 61 66–68 80 Some studies investigated multiple interventions. In particular, legislation changes were often accompanied by education,57 74 75 75 84 and studies on dog training methods often included medication as part of the strategy.55 60 72 Studies were clustered by the predominant intervention. Five studies specifically mentioned implementing interventions from a One Health framework.45 57 58 74 77


Eleven studies44–47 50 52 57 65 74 75 84 investigating the effect of general dog control strategies, were considered moderate to high quality in all but three studies, providing some evidence that this decreases the rates of dog bites (online supplemental table 1).

A study of moderate quality conducted in Calgary, Alberta, showed a substantial (80%) reduction in the incidence of dog bites reported to animal management over a 30-year period from 99 per 100 000 people in 1984 to 20 per 100 000 in 2014.50 This time period included a change in legislation that focused on strict leash laws (including leash-length, and walking on the correct side of a path), increased ticketing, immediate return of stray dogs to owners, sterilisation of dogs that injure a person, reduced registration rates, restrictions including muzzling/caging requirements, and adjunctive public education about the laws.50 86 This was supported by a high-quality study by Clarke and Fraser demonstrating that ticketing for animal control violations and requiring licensing for domestic dogs in areas of Canada was associated with lower incidence of dog bites reported to animal management (p<0.01).84 However, the study found no association between higher budget or staffing rates, or public education, with reported dog bite incidence. An additional two small low-quality studies of intensive community dog control programmes in small indigenous communities in Canada, showed reduced dog bite rates from 6 to 10 per year to 1–2 per year.65 74

Two further similar small low to moderate quality studies conducted with high levels of community engagement within Indigenous communities in rural Australia provided free animal care and introduced dog control strategies.44 45 The study by Riley et al with a focus on animal welfare showed no change in dog bite presentations to healthcare clinics.45 The study by Ma et al focusing on dog control measures, including euthanasia or re-homing of dogs to other areas, had a small reduction in annual Council reported dog attacks (defined as ‘rushes at, attacks, bites, harasses or chases any person or animal, whether or not injury has occurred’) from 2.5 to 1.5 attacks per 1000 people in the preintervention year to <1 attacks per 1000 people postintervention (p=0.035), with a control community demonstrating an increase over a similar time-period from 4 to 8 attacks per 1000 people.44

In Spain, the introduction of dog control legislation in a moderate quality study of a rural/urban region resulted in a 38% significant reduction in dog bite hospitalisations over an 11-year period, from 1.80 per 100 000 (95% CI 1.47 to 2.13) prior to the legislation change, to 1.11 per 100 000 (95% CI 0.87 to 1.36) after it was introduced.51 Legislation included registrations, restrictions on ‘dangerous’ dogs by both breed and behaviour (such as a requirement to have a special licence, a psychological aptitude certificate and absence of criminal record of the owner), leash laws, muzzles in public and microchips.

The introduction of the NZ Dog Control Act in 1996, appeared to temporarily reduce dog bite hospitalisations from 7.5 per 100 000 people in 1996 to 5.5 per 100 000 in 1999 in a moderate quality study; however, 2 years later rates increased to 6.8 per 100 000.46

A historical study from Guam in the 1960s showed a 75% reduction in animal encounters (predominantly dog bites) from 995 in 1967 to 252 in 1969, after the mass euthanasia of >15 000 stray dogs in an effort to eradicate rabies.75 A study in Sri Lanka investigated a change from the euthanasia of stray dogs to a more comprehensive ‘One-Health’ rabies-prevention intervention.57 This involved the development of dog managed zones in public areas, targeted sterilisation, and the education of children and adults on bite prevention and rabies awareness.57 This moderate quality study showed a 34% non-significant decrease in the number of dog bites disclosed in household surveys over a 4-year period from 0.216 per person (n=23/1063) in 2007 to 0.0143 per person (n=8/559) in 2010 (p=0.31). The authors also noted a 9% increase in the number of people presenting to the hospital for bites from 131 in 2006 to 160 in 2011, which they postulated could have been due to an increase in reporting, which they were promoting.57

Only one study did not show a difference when dog control legislation was introduced.52 The moderate quality study from Scotland, compared a 3-month period the year dog control legislation was introduced, to a similar period 3 years later, showing number of people with a dog bite presenting to an ED was unchanged at 134.52 However, the 3-month data collection periods used were likely, not long enough to adequately investigate number of bites.52

The five moderate-high quality studies investigating BSL suggest there is possibly a small effect on dog bite rates.47–49 83 84

Raghavan et al83 and Clarke and Fraser84 both investigated dog control legislation by comparing the incidence of dog bites in Canadian jurisdictions with and without BSL policies in place. Clarke and Fraser described no significant difference in rates of dog bites reported to animal management between areas with (170 per 100 000) and areas without BSL (180 per 100 000).84 However, Raghavan et al showed that areas with BSL had 19% significantly less dog bite hospitalisations (2.92 per 100 000) than areas without (3.60 per 100 000, p<0.002). Contradicting this however, there was only a 9.6% non-significant reduction in the rate over time in areas with BSL following the introduction of the legislation.83

An Italian study of moderate quality showed an 18% reduction in the number of dog bites from various sources after BSL was introduced, which was stable in the long term (210 pre-BSL, 172 in short term, 174 in long term).49 Of note, BSL was removed in Italy in 2009, 5 years after the completion of this study. A moderate quality study in Denmark showed a 15% non-significant reduction in the average 6 monthly number of dog bites presenting to EDs over 13 years from 103 pre-BSL (95% CI 98 to 108) to 87 post-BSL (95% CI 82 to 93). There appeared to be a more marked, but again non-significant result, in private spaces with a 19% reduction from 75 (95% CI 71 to 79) to 61 (95% CI 56 to 66) compared with a 7% reduction in public spaces from 28 (95% CI 26 to 31) to 26 (95% CI 56 to 66).48 A Spanish study of moderate quality described no significant difference in the rates of dog bites reported to public health after BSL was introduced; however, data is missing in the publication to support this statement.47 This study also reports a non-significant 50% reduction in reported dog bites in highly populated areas, versus only a 2% reduction in low populated areas.

The main limitation of the studies on BSL legislation was that they did not compare dog bite rates in legislated breeds compared with a control group of non-legislated dogs, they often already had a decreasing trend prior to intervention implementation,49–51 and lacked robust statistical analysis.


Six studies investigated the effects of sterilisation on either dog bite rates or dog aggression58 59 63 64 81 82 (online supplemental table 1). One moderate-quality study evaluated the impact of stray dog sterilisation as a single intervention on the number of dog bites.64 The study undertaken in an Indian city demonstrated the chemical or surgical sterilisation (and then release) of free-roaming dogs in a city with a high rate of dog bites reduced the number of dog bites by 48%, from approximately 11 500 in 2003, to 6000 in 2011, likely due to a 28% decrease in the dog population.64 The authors suggest this also may in part have been due to female dogs protecting their young, as they noticed an increase in bites occurring 3 months after a peak in pregnancies.

Two further low-quality studies used stray dog sterilisation as a means of dog population control to prevent rabies,58 82 along with education regarding the importance of vaccination and post-exposure prophylaxis (PEP). One study in India had a nearly threefold increase in dog bites over a 9 year period (853 in 2005/2006 to 3314 in 2012/2013).58 A similar study in Thailand had an initial 66% increase in reported animal bites from potentially rabid animals (predominantly dogs) over the first 4 years (1692 in 1996 to 2816 in 2000), and then a drop the following year to preintervention levels.82 Both studies were limited by their data collection method, from either an unknown method (provided by Department of Health)58 or from potentially rabid (presumably unvaccinated) animals.82 Results from these rabies prevention programmes were also confounded by their promotion of presenting for PEP.

A further study on the sterilisation of stray dogs was conducted in Chile in 2016. This high-quality, blinded RCT demonstrated a significant increase in aggressive behaviour in 36 free-roaming dogs who were chemically sterilised, with no difference in 39 dogs who were surgically sterilised, or in 44 control dogs with no intervention.81 One small (n=57) low-quality study of domestic dogs in California in 1997, found a small percentage of owners (10%–25%) demonstrated a substantial (90%) reduction in aggressive behaviours toward following surgical sterilisation.63 A further similar study in 23 household dogs in the USA found a 26% reduction in aggression indoors and a 52% reduction outdoors.59 However, these studies were limited by small sample sizes, lack of a control group and subjective owner-reported measures of aggression which are likely subject to bias.


A 60% absolute reduction (ARR 0.6, 95% CI 0.4 to 0.9, p=0.024) in dog bites during a controversial strict alcohol reduction programme was shown in two Indigenous communities in the outback of Australia, aiming to decrease injury rates (online supplemental table 1).43 The communities with zero carriage of alcohol had a reduction in dog bites, of 61% in Community A from 12.4 per 1000 people in 2006/2008 to 4.8 in 2009/2011 (IRR 0.4, 95% CI 0.2 to 0.7, p=0.001), and 30% in Community C from 40.0 to 27.9 per 1000 people (IRR 0.7, 95% CI 0.5 to 1.0, p=0.033). The control community with restrictions limited to personal alcohol consumption had a 29% non-significant reduction, from 12.9 to 9.2 per 1000 people (p=0.317). Although there was no randomisation, this was a high-quality study, which measured bite incidence from primary care clinics, used large sample size (n=1684), and included a control group with less alcohol restrictions.


One high quality, large scale, preinterventional and postinterventional study in the Philippines, investigated the impact of educating 5764 school children (5–14 years) on rabies and bite prevention, on the incidence of dog bite presentations among same aged children (online supplemental table 1).62 There was only a minimal, non-significant reduction in child dog bite incidence captured in either the household surveys from 26.4 per 1000 (n=124/4700) in 2011, to 24.7 per 1000 (n=114/4700) in 2012 (p=0.46), or on hospitalisation rates from 8.6 per 1000 (n=79/9211) in 2011 to 7.5 per 1000 (n=69/9211) in 2012 (p=0.65). They did however report a reduction in the proportion of Category III (deeper) bites, (11% in 2011 to 3% in 2012, p<0.05). While there was no control group in this study, and the authors acknowledge the potential under-representation of children who were not enrolled in the traditional education system but may be at higher risk of dog bite injuries, this was a high-quality study demonstrating the limitations of providing education to children as an intervention strategy.

A small low-quality Canadian study (n=99) targeting dog owners found that giving behavioural advice at their puppy’s first veterinarian visit, regarding the importance of early socialisation and positive training techniques, resulted in less aggressive behaviour of the dogs toward unknown people and toward other dogs, compared with a control group, at 1-year follow-up (2% vs 16%, p<0.05).65 However, this study relied on non-validated owners’ interpretation of dog behaviour, with unknown adherence to the intervention, and a small sample size.

Five studies investigated the effects of providing intensive public education on rabies prevention, primarily aimed at local adult leaders (online supplemental table 1).69–71 77 79

One high-quality interventional study on rabies prevention from South India, compared three rural villages (n=1735) to three control villages (n=1080), resulting in a 30% significant reduction in animal bites (predominantly dogs) in the intervention villages from 2.7% (n=47/1735) to 1.9% (n=33/1735, p=0.04), and a 13% non-significant decrease in bite cases in the control villages from 2.8% (n=31/1080) to 2.5% (n=27/1080, p=0.55).77 The study also showed increased presentations of Category III dog bites, and attribute this to increased awareness of the need for medical intervention for more severe bites. This was a comprehensive community-based strategy, with control groups, and outcomes appropriately measured by a randomised household survey. However, it is possible that people may be more motivated to reduce dog bites if education is focused on rabies prevention.

A similar high-quality study in Northern Tanzania found that education on rabies prevention and dog vaccination had a 79% decrease in bites within intervention areas, and a 60% non-significant increase in control areas.79 However, this study only investigated the incidence of dog bites from potentially rabid (presumably unvaccinated) dogs, and thus has no benefit in estimating the effect on bite rates from all dogs. A further moderate-quality, preinterventional and postinterventional study in Southern Tanzania found that public education on rabies prevention resulted in an initial increase in bite incidence from 1.8 per 100 000 per quarter (n=1600), to 2.8 per 100 000 (n=2700) the following year, then a general decline to zero in the final quarter of 2016.71 However, this study had likely high variability in the data collection method, using dog bites reported to researchers by livestock field officers and healthcare workers. Likewise, a low-quality study on intensive adult directed rabies prevention education in Zanzibar had an unknown data collection method.69 A low-quality study with a more well-defined data collection method done in Philippines found that rabies prevention community education and dog vaccination resulted in an initial increase in animal bite presentations (83%–89% dogs) to eight animal bite treatment centres in the region, from 2015 in 2011 to a peak of 5908 in 2014, and then a fall to 5520 the following year.70 However, this was substantially confounded by a potential increase in presentations with increased awareness of the need for rabies PEP.

Dog training

Two studies investigated a change in police dog training methods from a ‘bite and hold’ method to a new ‘bark and hold’ method, which trains a dog to circle and bark during an arrest, but only to bite a suspect if the suspect moves or actively resists.53 54 A moderate-quality, Los Angeles study investigating the impact of the introduction of the ‘find and bark’ training method in 1992, resulted in a 90% reduction in the number of dog bites to incarcerated patients seen in the ED in 4 years before and 4 years after the policy was introduced from 639 to 66 bites.53 Patients also had fewer fractures, vascular complications, hospitalisations and multiple bites. In contrast to this, a low-quality Florida study surveying police dog handlers, compared 45 dogs who were trained using the new ‘bark and hold’ programme introduced in 2001, with 135 dogs who continued training in the standard ‘bite and hold’ method.54 This study found that the bite and hold method had significantly lower mean bites per arrest (bite ratio) than the new method (15.7 vs 22.4). However, this second study had a high (48%) non-response rate and thus small sample size. Both studies had unknown adherence to training and dogs were likely previously trained in the old method before introduction of the new method.

Six studies investigated dog training methods for dogs with a history of behavioural issues, two of which also used concurrent medication.55 60 72 73 76 85 Tortora in the high quality 1983 study of 36 household dogs with a history of aggression, showed reductions to near zero in a trainer-reported measure of dog aggression (using video footage of sessions with the dogs) over a 2.5-year intensive dog-training programme including the use of electric shock collars (p<0.001), with no improvement in the control group (p>0.05).76 Dodman et al, in two low-quality studies in the USA, found an 8-week non-confrontational behaviour modification programme reduced owner-reported dog aggression in 9/10 dogs with a history of aggression in one study (p<0.05),85 and 14/20 dogs in another study.73 A further similar study found that 10/24 dogs in the USA had a greater than 50% improvement in aggressive behaviours with dog training that was combined with sterilisation and progestin treatment.72 These studies were limited by their very small sample size with no control group. Knol conducted a low-quality study that found that in 133 dogs with behavioural problems in the Netherlands, owner satisfaction with a mixed positive and negative (collar) training programme (in conjunction with medication in only five dogs), was reported as ‘good/fair’ in 42% of cases, ‘moderate’ in 11% and ‘bad’ in 41%.60 However, this study used an indirect measure of dog aggression, with no control group, also included medications in some dogs, and had an unclear intervention with different methods for different behavioural problems. Likewise, Dinwoodie et al, in their low-quality retrospective cohort study of 963 dogs in the USA, where the owner employed one of 21 different behavioural techniques with or without medication, found that 82% of owners felt there was some improvement in aggression.55 This study was limited by asking owners retrospectively their view of a highly heterogenous group of interventions with no standardised outcome measure or control group for comparison. All located studies on dog-training were potentially influenced by financial gains from dog-behaviouralists undertaking studies of their own interventions.

Medication and diet

Three studies primarily investigated the effect of pharmaceutical intervention, in conjunction with behavioural modification.67 68 80 A further three studies investigated the use of medication only56 61 66 and one with diet alone (online supplemental table 1).56

The three studies investigating medication alone were low quality, and conclusions could not reliably be made.56 61 66 They were limited by small sample sizes, subjective and non-standardised owner-reported measures of aggression, and no control group. One used a cross-over design, however, had unreliable results with a sample size of nine.61 One study had an unknown period of medication given, and multiple medications used simultaneously,56 and one study used an inappropriate control group of non-aggressive dogs receiving the intervention.66 One moderate quality study looked at the effect of low or high protein diets, with or without tryptophan using an appropriate cross-over design with a 3-day washout period, exclusion of participants with recent medication use, and a well-defined outcome measure.56 They found no significant change in behaviour over a 4-week period with any of the groups. However, this study also had a small sample size, relied on owner-reports of aggression, excluded dogs with severe aggression or pregnancy, and appeared to have sponsorship from a pet food company. There did not appear to be sponsorship of studies by medication companies.

The three studies investigating medication in conjunction with behavioural therapy were likewise low-quality studies, with no conclusions able to be drawn.67 68 80 They were mostly limited by small sample sizes, lack of a control group, owner-reported measures of aggression, and no separation of dogs with or without behavioural therapy, a significant confounding factor.

Furthermore, with all studies on medication or diet, maintaining the recommended diet or dosage of medication would require highly motivated and likely high-income dog owners, and therefore these studies may have been subject to a significant response bias.


The aim of this review was to evaluate the efficacy of interventions to prevent dog bites and dog aggression, and used a systematic approach guided by the PRISMA statement.40

The strength of this review is that it provides a broad overview of the literature on a range of strategies to address dog bites and dog aggression, using an established public health framework. The wide search strategy predictably resulted in a large number of non-relevant studies, however, this also identified a number of relevant prevention strategies that would otherwise not have been considered. The outcome measure was clear and relevant, and did not rely on assumptions about how it might translate into injury prevention. The inclusion of grey literature reduced publication bias, and studies were systematically evaluated by two researchers, reducing individual biases.

This review should be considered in light of some limitations. Studies had a high degree of heterogeneity in both interventions and outcomes. A large number were excluded as they did not study an intervention or did not measure dog bites or aggression. Half of the studies were considered low-quality by the authors and were of limited value, primarily due to low sample size, the lack of a control group, limited statistical analysis, or bias in the outcome measure used, including subjective and heterogeneous owner-reported measures of dog aggression, dog bites reported to an authority, or hospitalisations which only represent a small proportion of dog bite injuries.1

In addition, study findings may not be generalisable between populations with socioeconomic or cultural differences. The authors also acknowledge that acceptability, ethical views or cultural appropriateness of interventions will vary, with this review touching on a number of controversial issues, including euthanasia, aversive dog-training techniques and BSL. A number of studies on rabies prevention (primarily addressed through vaccination and PEP), also employed dog bite prevention strategies.57 58 62 69–71 75 77 79 82 The main issue with these studies is that dog bite injuries measured through medical presentations may be confounded by an increase in presenting for PEP.


This review found moderate to high evidence that dog control strategies decreases dog bite injuries, particularly for those that substantially reduce the dog population through sterilisation, euthanasia or re-homing. Euthanasia of dogs is controversial with mass culling of dogs, a measure used historically for reducing the dog population,75 no longer considered acceptable in most countries. Likewise, euthanasia of dogs who pose a risk of serious injury, including those with a history of aggression or those relinquished to a shelter is controversial, with no well validated risk assessment tools available,87 and devastating health consequences for people when wrong decisions are made. This is of particular concern considering the results of a recent UK study showing the most common reasons for dog relinquishment to a shelter being aggressive behaviour between dogs in the home (20%) or aggression around children (19%).88 Sterilisation may be a more socially acceptable strategy. However, while there is evidence this decreases injuries through a reduction in the dog population, whether it reduces dog aggression in either contained or roaming dogs is still uncertain. Study findings are consistent with a previous systematic review of this topic in the context of rabies prevention.26

Further effective dog control strategies in located studies included dog management in public spaces (through microchipping, registrations, community patrols, property fencing requirements, infringements, establishment of dog shelters or leash laws). The use of leash laws is currently debated in many countries, with public calls for increased leash use as the population of dogs and people grow, with subsequent increased use of public spaces. There is also an increasing trend to establish dog-control methods to address roaming/stray dogs in low-socioeconomic areas with benefits for both the welfare of stray dogs and injury prevention.57 58 64 65 74 81 82 89 Of note, several included studies specifically adopted a ‘One Health’ approach.45 57 58 74 77 This multi-sectoral framework recognises the increasing interactions between humans and animals as the population of both grow, as a root cause for the spread of zoonoses (ie, rabies) or increasing injuries, and promotes prevention of these through humane dog control strategies, education and animal welfare. It is predominantly used within low socioeconomic regions, or indigenous cultures, however, can be applied to any group. Four included studies were conducted in small rural indigenous populations living within colonised countries (Canada and Australia), with high levels of community engagement.44 45 65 74 Of note, interventions within communities that employed dog control strategies65 74 89 had more impact than those focusing on dog health alone.45 One study also described that while these strategies can be effective, they require a high level of resource and can be difficult to implement with changes in community leadership or priorities over time.74

Restrictions on dogs by breed are also controversial world-wide. While there is a call to change this approach from the perspective of both dog-advocacy and injury prevention, an evaluation of the currently available evidence in this review has shown a small decrease in BSL. To address this, it seems a sensible approach to promote the widening of humane dog-control restrictions to be placed on all dogs, rather than limiting these to certain breeds. This is supported by studies showing that the most common breeds to bite are those that are the most popularly owned.11 24 90 91 An NZ study found the majority of dog bites (66%) were caused by mixed, undefined or unknown breeds, with bites from known breeds ranging from only 2% (Bull Terrier) to 8% (German Shepherd).11 A study in Ireland also found a lack of perceived threat from legislated versus non-legislated breeds, which were also less likely to be reported (27% vs 55%); a considerable concern, as people may underestimate the risk of the most popular non-legislated breeds.90

Legislative measures to reduce injury may also include promotion of reporting of issues either by members of the public or professional bodies. Internationally, while there are regional policies within the UK and USA, only Switzerland has a national legal requirement for the mandatory reporting of dog bites by health professionals or veterinarians for the purposes of dog-bite prevention, with a high rate of reporting after this strategy was introduced, and a 31% reduction in insurance claims for dog bites from approximately 3600 in 2005 to 2500 in 2007.92 93

Alcohol reduction as a general injury prevention strategy is an area of focus in NZ94 and globally, and may be linked through either decreased ability of victims to defend themselves or difficulties in providing adequate care for a dog. It has also been noted that fatal dog attacks have occurred in victims who are vulnerable such as those with disabilities, dementia or seizures, those with drug or alcohol compromise, the elderly, young children or infants.14


Among the seven studies included in this review that explored the impact of education on dog bite rates, five found intensive community based adult education reduced dog bite rates.69–71 77 79 A study conducted in the Philippines among children showed no significant change in dog bite rates.62 A recent low-quality Austrian study investigating an intensive child and parent-directed education programme on dog safety likewise showed no reduction in hospital presentations, with mean hospitalisation rates not reported.95 This is unsurprising, given child-directed education is not advocated for or commonly used in other areas of unintentional trauma in young children,36 96 and is consistent with research showing that dog bites in children and adults are frequently unprovoked, or occur with a minor interaction such as patting a dog.11 13 97 While there are a large number of studies investigating the effect of children’s education programmes on children’s knowledge of dog safety, including two systematic reviews,98 99 these did not investigate the impact of these programmes on either injuries or incidents involving dog aggression, and this should be considered in any future research on this topic.

The current review found no studies investigating the effect of educating caregivers on the importance of supervision of children on dog bite rates, despite this being a common educational message. Promoting this may not reduce injury, given supervision of children is a highly complex task that involves an understanding of a child’s developmental ability, along with an assessment of multiple hazards within different environments. Constant supervision is also simply not always possible, and depends on the ratio of children and adults, the environment they are in, socioeconomic factors, and the capability of the supervisor who may not be a parent.100

Given the predominance of attacks by dogs who are not owned by the victim or their immediate family, and dogs natural tendencies for resource guarding, territorial or predatory aggression,31 redirection toward owner-directed safety information is likely required. Non-aversive dog training is a strategy promoted by dog-advocates, and is further supported by studies demonstrating the negative impact of aversive training techniques on dog welfare.101 All studies located in the current review using positive behavioural techniques to reduce dog aggression were low-quality and inconclusive. Bias may also exist in studies undertaken by dog behaviouralists who have a financial interest in their own interventions. Furthermore, intensive dog training programmes are likely to be more effective for the highly motivated, higher-income owner. An example of this is given in a study that demonstrated low adherence to both internet-based or face-to-face dog training programmes in low socioeconomic groups, even when provided at no cost.102

Police dogs are a sub-population of dogs that are trained, and are often not considered in research on dog bites. A change in training strategy to ‘bark and hold’ in police dogs53 reduced the number of bites within a US prison, although the consequences to police or the public of not using the bite and hold method were not investigated. In NZ police dog bites make up less than 2% of bites presenting for medical attention,1 are highly regulated and used only in extreme situations.103

One study included owner-directed education regarding the importance of early socialisation,78 and no studies investigated education on dog walking to reduce dog aggressiveness. These may not be as important in reducing dog aggression as previously thought, as a recent large prospective cohort study in Finland comparing risk factors in 1791 aggressive to 7479 non-aggressive dogs, showed no differences in either early socialisation or number of hours walked between the groups.104

Medication and diet

Results from studies on the effect of medication on dog behaviour were inconclusive, and require larger RCTs to investigate this.

Engineering strategies (to modify the physical environment)

No studies were located that considered the effectiveness of dog-person barriers such as baby gates, adequate fencing height, gate locks, separate dog areas or the use of leashes as a single intervention, which are likely important given the role of physical barriers in other areas of unintentional injury in children.36 In-home barriers reduce the need for constant supervision of children, and do not rely on behaviour change in either dogs or people. Physical barriers such as pool fences, stair gates, playground safety standards, child-resistant packaging and restrictive window latches have been effective in other areas of unintentional child injury prevention.36 Likewise, no studies were located on the effect of home safety visits, with or without equipment provision, either at an early stage of dog-rearing, or following a dog bite, shown to be effective in other areas of unintentional injury in children.105 The implementation of fencing as a strategy may also be challenged by socioeconomic circumstances. For example, one article highlights the inability to modify shared or rented accommodation, as a barrier to improving child safety in more deprived areas.106

Future research

Future studies should employ interventional rather than observational study designs with an appropriate control group, ideally as RCTs, using adequate sample sizes and statistical analysis, and investigating specific well-defined interventions. Studies should also use the incidence of dog bites and other dog-related injuries as primary outcomes of interest (per 100 000 people), and use broad methods of data collection, including non-bite incidents of dog aggression from household surveys or presentations to primary care clinics, rather than more limited methods such as bites that are reported to animal management, or hospitalisation rates. Using standardised measures of owner-reported dog-aggression such as C-BARQ107 would aid in comparison between studies, despite the challenges with the inherent subjectivity that exists in these measures. Rigorous studies are specifically required for the use of positive dog training techniques, reporting strategies, and implementation of engineering barriers (fencing, baby gates, separate dog spaces or leashes) to protect children. Future research should also investigate macro-level strategies such as the distribution of funding provision of dog-control strategies, or access to services. Strong engagement with indigenous cultures should be prioritised in all future research on this issue, including exploration of indigenous approaches to dog control/safety.


Multiple strategies including effective engagement with indigenous communities and organisations will be required to reduce dog-bites and other incidents involving dog aggression. This review provides some evidence that legislated dog control strategies reduce dog bite rates. Available evidence suggests greater restrictions should be made for all dogs, rather than based on breed alone. Due to the burden of child injury, protection of children should be a focus of legislation and further investigations. Prevention strategies in children require redirection away from a focus on child-directed education and future research should investigate the effectiveness of engineering barriers and reporting strategies.

What is already known on the subject

  • Injury by dogs is an increasing and serious public health issue.

What this study adds

  • Dog control legislation, such as leash laws, stray dog control, infringements and restrictions or euthanasia of dogs with a history of aggression likely reduces dog bites, with less of an effect for breed specific legislation.

  • Sterilisation (spaying, orchidectomy or chemical) may reduce dog bites through a reduction in the dog population.

  • There is currently no evidence to support child education as a dog bite reduction strategy.

Data availability statement

All data relevant to the study are included in the article. Not applicable.

Ethics statements

Patient consent for publication

Ethics approval

Not applicable.


The contributions to this review by a number of people are greatly appreciated: Mareta Hunt (Ngāti Awa, Ngāi Tūhoe, Ngāti Maniapoto, Ngāti Kahungunu me Kai Tahu, Director, Safekids Aotearoa), Moses Alatini (Policy Analyst, Safekids Aotearoa), Melissa Wilson (Prior Director, Safekids Aotearoa); Dr Zachary Moaveni (FRACS (Pl Rec Surg), Middlemore Hospital); Dr Sylvia Boys (FACEM; Middlemore Hospital); Dr Inia Raumati (Ngāti Mutunga, Te Ātiawa, MBChB, Emergency Registrar, Auckland Hospital), Dr Inia Tomas (Te Rarawa, FACEM, Middlemore Hospital); Dr Eunicia Tan (FACEM, Middlemore Hospital); Dr Rebecca Hayman (FRACP, Middlemore Hospital); Denise Peters (Auckland Council Animal Management); Brent Lincoln (Tauranga Council Animal Management); Dr Lorelle Barrett (NZ Veterinary Association); Dr Lyndon Drake (Te Pīhopatanga o Te Tai Tokerau); Dr Jonathan McMillan (Paediatric Emergency Physician, Starship Hospital); Dr Elsa Flint (Dog behaviouralist); Inspector Todd Southall (National Coordinator, NZ Police Dogs); Dr Jessica Walker and Dr Alison Vaughan (NZ Society for Protection and Care of Animals). ‘Te piko o te mahuri tērā te tupu o te rākau’. A special acknowledgement is also given in reference to this whakataukī to Dr Helen Wihongi for her leadership and supervision supporting our Māori advisor, kia tupu tika, kia puawai i roto i ngā mahi rangahau. He puna mihi e kore e mimiti noa.


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.


  • Correction notice This article has been corrected since it was first published. The open access licence has been updated to CC BY.

  • Contributors The design of this review was developed by BK, ND-S and MS. The online database search was carried out by ACL. Screening, selection for inclusion and analysis of studies was undertaken by ACL and ND-S. The body of the article was composed by ACL and ND-S. ND-S is responsible for the overall content as guarantor.

  • Funding Starship Foundation. Registered Charity CC24272. Award number: A+ 9001. SF2137.

  • Competing interests None declared.

  • Patient and public involvement Patients and/or the public were involved in the design, or conduct, or reporting, or dissemination plans of this research. Refer to the Methods section for further details.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.