Introduction Preventing dog bites is an intractable problem given the complex dog bite injury environment. Desexing of dogs has the opportunity of creating a safer injury environment, given the potential links between desexing and behaviour change in dogs.
Methods A systematic review of the literature was conducted to examine the evidence for desexing of dogs to reduce dog bite risk within a population health paradigm. Medline and CAB Abstracts were searched for studies that reported data on the association of dog neuter status with the risk of dog bite. All definitions of dog bite were included and all empirical studies were included in the review, limited to those published in English. Quality appraisal and data extraction were based on the 2013 evidence-based practice and critical appraisal tool from the University of Auckland.
Results Five out of six observational studies, from four study populations found evidence that intact dogs were associated with an increased risk of dog bite compared with desexed dogs. The effect sizes ranged across the studies and given the heterogeneity of the studies no single effect size on the association between desexing and dog bite risk could be estimated.
Conclusions There is consistent evidence that desexing dogs is associated with a reduced risk of dog bite, although the studies reflect association and may not be causal. Although recent publications have suggested desexing is associated with health and behavioural costs in some breeds, population level evidence supports desexed dogs having a longer lifespan, and being less likely to wander with the added benefit of reducing unwanted litters. Thus, mandatory desexing presents a possible opportunity for prevention of dog bites expanding dog bite prevention beyond an education-only approach.
Statistics from Altmetric.com
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.
Controlling the risk of dog bites has proven to be an intractable problem worldwide. Estimates for rates of dog bite vary, but are in the order of at least 3–4 bites per 1000 population per annum.1 The events leading to dog bites are a complex interplay of innate and behavioural features of dogs, of the victims and aspects of the environment, in the classic injury triad.2 Factors such as dog temperament, sex and neuter status have been suggested to influence the risk of a bite, while victim age, gender, safety knowledge, risk perception and behaviour around dogs are thought to increase the risk of a dog bite.1 ,3 ,4 A dog's living conditions, including opportunity for social interaction, is also thought to affect the risk of dog bite. Many dog owners have passionate views about how dogs should be managed, which adds to the complexity of the injury environment and the consequent opportunities to implement effective interventions across the population.
Breed-specific legislation has been the legislative intervention of choice across a number of countries in an attempt to reduce the risk of serious injury or death from dog bites. This type of legislation attempts to control certain breeds of dogs in various ways, including breeding controls, and is based on disputed evidence that some breeds are more at risk of biting or of causing serious injury if they do bite than other breeds. These legislative approaches are thought to be inappropriate and ineffective for a number of reasons5–7 and in most studies have failed to reduce the risk of dog bite,8–11 with some exceptions.12 ,13 To date, there are no definitive tests of dog breed, and people are inaccurate when assigning dog breed based on appearance,14 making the implementation and enforcement of breed-specific legislation problematic. Available evidence on breed-specific dog attacks are limited in their validity as obtaining a reference population of dogs can be impossible, and breeds are often assigned incorrectly.5 Furthermore, all dogs have the propensity to bite given the right circumstances and so breed-specific legislation is likely to miss a substantial proportion of biting dogs.
There have been a number of attempts to use education programmes predominately targeted at younger children to reduce the risk of dog bite. Such programmes include school-based programmes using a variety of educational strategies (video, work book, dogs, websites, interactive software) to teach children about dog behaviour and how to behave around an aggressive dog.15–17 There are also programmes that target parents in settings such as antenatal care and both parents and children in emergency departments.18 ,19 While for some programmes there has been evidence of a change in knowledge and behaviour of children following the programme,15 ,20 there is no evidence that such programmes reduce the risk of dog bites.21 Training programmes for dogs (and their owners) have had mixed results depending on the style of training. A large cross-sectional survey in the UK demonstrated training methods using punishment were associated with increased interdog and dog-human aggression,22 ,23 which may result in a higher risk of dog bites.
Given the complexity of the dog bite injury environment, it is not surprising that education or training programmes alone have not made real changes to dog bite risk over time. Modifying the environment rather than attempting to modify the behaviour of people is a well-established principle in injury control and there are many historical examples of successful risk reduction employing this principle. Road accident prevention is a classic example of education campaigns (eg, do not speed) that are supported by legislation which is enforced (eg, speed cameras) and if an accident does occur better engineering of cars that reduces the risk of injury. Education programmes for dog bite prevention should be supported by effective public policy that is acceptable to the community and feasible to implement.
Desexing of dogs has the opportunity of creating a safer injury environment, given the possible links between desexing and behaviour change in dogs.1 ,24 An Australian state, South Australia recently introduced mandatory desexing of dogs (with exceptions) coupled with a proposed Breeding Code of Practice, which together aim to reduce the risk of dog bite among a range of other benefits for dogs25 and the wider community. In South Australia, breeding dogs will be exempt along with dogs that are exempted on medical grounds and some working dogs. The proposed Breeding Code of Practice prescribes principles for good breeding, including safe-guarding the health and welfare of dogs and also considering positive temperament of the litters. This new approach provides an opportunity to evaluate the effect of mandatory desexing of dogs across a population on the real-world risk of dog bites.
This paper aims to critically review the evidence for this novel intervention from a population health perspective. Studies that investigated a risk of dog bite for desexed dogs relative to entire dogs will be reviewed and consideration of the opportunity that mandatory desexing under legislation could provide to reduce the risk of dog bites across the community will be discussed.
Studies that reported data on the association of dog neuter status (desexed vs entire) with the risk of dog bite were included in the review. All definitions of dog bite were included. Studies reporting dog aggression as an outcome of desexing were not included in the review, given the concerns about the lack of a gold standard tool for measuring aggression in dogs.26 Given the paucity of published data, all empirical studies were included in the review, limited to those published in English. Case series were excluded as they did not report on a comparison group.
In December 2015, Ovid Medline was searched (from year 1946) to find eligible studies using the terms: ‘dogs’ (MeSH, using the Medical Subject Headings indexed term for dogs) AND ‘bites or stings’ (MeSH) OR ‘attack’ (keyword) AND NOT ‘infection or rabies’. CAB Abstracts from 1973 was searched using the terms: ‘bites or attack’ AND ‘dog or canine’ AND NOT ‘infection or rabies’. A further trace back search was conducted on the reference lists of all relevant reviews and included papers to find any additional papers, including any grey literature. The authors of included papers were contacted for further information if required.
One author (KD) selected eligible papers from the database search using either the title, abstract or full paper and all authors undertook data extraction and methodological appraisal of each paper selected for the review. Quality appraisal at the study level was based on the 2013 evidence-based practice and critical appraisal tool from the University of Auckland.27 This tool also facilitated the data extraction of the selected papers. Data items extracted included details of the study methodology, the dog bite definition and the severity of dog bite. Any discrepancy between the authors on the methodology appraisal or data extraction was discussed and resolved by consensus.
For ease of interpretation of results, where a single effect size for desexing on dog bite risk could be calculated from raw data this single value was presented. Furthermore, results from the study by Guy et al28 were reoriented from an OR of the effect of a desexed dog on dog bite risk to be an OR for the effect of an intact dog on dog bite risk to be consistent with the other studies and aid interpretation.
The search resulted in 1817 titles from Medline and 1286 from CAB Abstracts. There were 104 papers selected for full-text review and a reference list search found an additional six papers for full-text review. In total, seven papers met the inclusion criteria for the review with six being finally selected into the study (details in table 1 and figure 1).28–34 The paper by Drobatz was excluded as bite data from a dog or cat were combined, and dog only bite risk could not be sourced from the author. The 98 full-text papers were excluded on the basis of no desexing data being presented. All types of reported dog bite were included in the review, ranging from bites for which there was no visible injury through to fatal injuries from a dog bite(s). No meta-analysis was undertaken due to differences in the target populations, study design and dog bite definition across the papers (table 1).
There were no intervention studies examining the effect of neuter status on dog bite risk. This review found three retrospective cohort studies, two case-control studies, one cross-sectional study and five case series studies of fatal dog attacks. The latter were not formally selected to be reviewed given the non-inferential study design. Table 2 outlines the methodological appraisal of the reviewed papers.
Five out of the six studies found that desexing dogs was associated with a reduced risk of dog bite, although the magnitude of effect differed between studies and for the case-control study by Guy et al,28 the results did not meet traditional levels of significance (p≤0.05) (table 3). The two studies by Messam et al and the two by Guy et al were each conducted on the same population of dogs but using differing definitions of dog bites. Effect sizes ranged from a RR of 9.1 in the study by Shuler et al33 to near null in the study by Guy et al.28 The study by Messam et al32 had wide CIs for certain subgroup analyses given the small numbers involved (eg, zero neutered females in the bite during play group), in particular for the Jamaican group. Measurement error likely had a biasing effect on results of the cross-sectional study by Guy et al.34 This study collected data through the use of a survey completed by owners who had visited a participating veterinary practice, finding a negative association between desexing and dog bite risk. In particular, desexed male dogs had the greatest risk of biting when compared with other dogs, in contrast to the other studies in this review. When Guy et al undertook a subsequent case-control study using the same study population but instead asking further questions of owners by interview for a subset of the population they found that a number of the biting and non-biting dogs had been misclassified. Following re-classification of biting dogs, the association between desexed status and biting became positive (although not reaching traditional levels of significance).
Overall risk of bias
Selection and generalisability
Given the policy of mandatory dog desexing targets the entire population of dogs, studies that presented findings across the total population of dogs were deemed more relevant and most externally valid. Both Shuler et al33 and Gershman et al30 selected biting dog cases from the jurisdictional authority legislated reporting system, increasing the likelihood of all dog bites being reported, although both are likely to still be an underestimation of the true dog bite rate. Of note, the study by Shuler et al33 reported 19.8% of the attacks involved a family member, whereas in an earlier review it was suggested 25%–33% of bites were from the family dog1 and in a more recent study 28% of victims were bitten by their own dog,8 suggesting a modest undercount of these dog bites.
Missing data could have introduced selection bias in the study by Shuler et al,33 as desexing information was available on 71% of the dog bites reported, and in the study by Gershman et al30 only 18% of the total number of reported bites were analysed. Furthermore, the study by Gershman et al30 selected non-household-related bites only and so represents the minority of bites, given most dog bites occur in the dog's home (over 80%) and by the owned dog (over 60%).1 The two veterinary clinic-based studies selected cases from owned dogs at a veterinary practice and so likely represent the less severe end of the spectrum of dog bites.
The relatively small number of studies used a range of selection methods for participants, with selection from the general population of reported dog bites (2) or from a general veterinary caseload (4). It is likely that these different methods of selection have led to selection of differing levels of severity of dog bite and different populations of dog. Both the studies by Shuler et al and Gershman et al30 selected from reported dog bites and both are likely to have included a greater representation of severe bites, with the study by Gershman et al30 including only bites that led to injury requiring medical attention and Shuler et al33 reported 50.5% of included bites required medical attention. The veterinary clinic-based study by Guy et al28 reported 9.3% of included dog bites required medical attention and for Messam et al31 although the severity was not reported included bites that had no associated injury. There is some evidence from the included studies that studies with higher severity and broader range of dog bite scenarios included had bigger effect sizes (eg, Shuler) versus lower severity and more narrow inclusion criteria (eg, Guy). It is important to note, however, that in none of the studies was there a validated or systematic method for documenting bite severity.
None of the papers used data collection tools that were reported to be valid or reliable and so potentially misclassifying both the exposure (neuter status) and the outcome (dog bites) and so bias study findings. Furthermore, the lack of a well-accepted tool and different definition of dog bite means that the studies may not be comparable.
Studies by Messam et al31 ,32 relied on self-report from dog owners for both the exposure and outcomes and for the study by Gershman et al30 for the control group and so could have introduced measurement error. The case-control study by Guy et al28 also relied on self-report for both exposure and outcomes. The control group potentially included dogs that had bitten strangers, which would have the effect of biasing the neuter effect size towards the null. Furthermore, in the study by Shuler et al,33 non-licensed (non-registered) dogs that had bitten were included in the analysis, however, non-licensed dogs that had not had a reported bite were systematically excluded, which may have the effect of biasing the effect estimate away from the null.
All included studies except for the study by Shuler et al33 examined risk of dog bite using a multivariate model, which included a range of potential confounding factors such as dog sex, breed and weight, but none of the studies included age at desexing in the analysis. Age at desexing is potentially important as it is likely that the behavioural effects of desexing vary by the age at which a dog is desexed, although the effects are not well understood and requires further study. Given the observational study design and small number of variables included in the models, there is a possibility of residual confounding in the effect estimates because of unmeasured or poorly measured confounders. An example of a potential confounder is the dog care practices of owners that choose not to desex their dog versus those who do desex their dog, which may also lead to an increased risk of dog bites.
The studies found that desexed dogs were associated with a reduced risk of dog bite, possibly in favour of reducing the risk of serious bite. However, differential effects for male and female dogs were not able to be examined in detail due to the data available but may possibly be greater for females than males; this requires further investigation. Effect sizes for desexing and dog bite risk ranged from the largest in the study by Shuler et al (intact dogs RR 9.1 95% CI (7.3 to 10.5)) to near null in the case-control study by Guy et al (intact dogs OR 1.20 95% CI (0.53 to 2.7)). Of note, selection issues may have biased the Shuler et al effect sizes away from the null and for the study by Guy et al towards the null, suggesting an effect size somewhere in between the two studies. All studies had limitations in terms of potential selection, measurement and confounding bias, and only six published papers met the inclusion criteria for the review. The cross-sectional study by Guy et al was the only study to find a negative association with desexing and dog bites, however, it was a preliminary study with the reclassification of some biting and non-biting dogs following a subsequent interview for the case-control study (and reversal of the negative association). The study by Gershman et al was considered to be the most robust. It is important to note, however, that given the limitations of the papers reviewed that further research is required to determine if the association between desexing and dog bite risk is causal.
Five case series were identified in the review that reported on the circumstances for dog bites that led to (non-rabies) death.12 ,24 ,35–37 A study by Patronek et al24 examined the biggest series of cases (256 deaths over 10 years in the USA) that included a review of information from law enforcement agencies and an ascertainment check against national death data. In this series, 84.4% cases of dog attack that led to death involved entire dogs, which is consistent with findings from the other case series. While there is no denominator information with which to properly interpret these findings, a high proportion of entire dogs involved in severe dog attacks is consistent with the inferential studies discussed above.
Desexing of dogs has been associated with both costs and benefits. At a population level, the benefits, such as increased lifespan,38 outweigh the costs, although good quality evidence is lacking. Benefits to female dogs (bitches) include reduced risk of uterine infections (pyometra), which may occur in 23%–24% of entire bitches by 10 years of age.39 ,40 Although a reduced risk of mammary tumours has been thought to be associated with neutering,41 a systematic review reported the evidence is weak.42 In male dogs, benefits may include removing the risk of testicular disease, and other androgen-dependent diseases such as perineal hernias, perineal adenomas, prostatitis and benign prostatic hyperplasia.25 ,43 General costs in both males and females are associated with the risk of anaesthesia and surgery, plus in females urinary incontinence44 and in males increased risk of prostate cancer,45 and other types of cancer.38 ,46 Risks relating to desexing may vary according to the age of desexing and dog breed. In German Shepherd dogs, risk of cranial cruciate ligament tear increases in desexed males and females,47 while in another study the risk of cranial cruciate ligament tear increased in desexed male and female Golden Retrievers, and of hip dysplasia in desexed males.48 Some types of cancer (eg, mast cell cancer) are increased in desexed Golden Retrievers and Vizslas.48 ,49 However, well-designed studies to compare disease risk and longevity of neutered and entire dogs are lacking. In probably the highest quality evidence from 3961 deceased dogs attending first opinion veterinary practices in central and southeast England, entire females had the shortest lifespan, with an increase of 0.8 years in neutered bitches (95% CI 0.5 to 1.1) and both male entire and male neutered dogs living 0.4 years longer (95% CI 0.1 to 0.7 for both).38 On balance the overall effect is likely to be an increased lifespan, thus it is likely neutering in terms of population health has a positive impact.
Desexing is associated with a range of behavioural effects in dogs. Both androgens and oestrogens appear to facilitate aggressive behaviour, particularly during the mating season.50 In male dogs, desexing is likely to reduce the risk of objectionable sexual behaviour, roaming, intermale aggression and urinating in the house,51 although castration of aggressive dogs is not always an effective treatment.50 In female dogs, maternal aggression occurs when the nesting area or young are approached,50 which would be removed by desexing. However, in Vizslas desexed prior to 6 months of age, there may be an increased risk of a behavioural disorder.49 O'Farrell and Peachey suggested spayed bitches showed higher aggression than females. However, aggression was increased if the bitch already showed signs of aggression making the causal link questionable, and there were differences in the control and experimental (desexed) group at first interview in factors such as prior treatment with synthetic progestogen, making confounding likely.52 In a cross-sectional survey of English Cocker Spaniels, links between neutering and increased aggression were found in both males and females, but when the analysis was restricted only to dogs desexed prior to aggression being displayed, a significant effect remained only in neutered females for aggression towards children in the household, but not for the other 12 types of aggression.54 Although Azkona et al55 reported an increased risk of age-related cognitive impairment in desexed dogs, Katina et al56 found no significant impact of reproductive state on cognitive dysfunction. Existing evidence relies on cross-sectional data, and unknown confounders may significantly affect results. For example, if handling and pain relief are not optimal and desexing occurs during a critical fear period, this may result in a more fearful dog unrelated to the change in reproductive status. Confounders such as the age of desexing, environment during the first weeks of life and dog breed are likely to also be important, and well-controlled studies are needed to better understand factors at an individual level of the effect of desexing on the full scope of behaviour in dogs.
It is clear from the literature on the health effects of desexing dogs that there may be certain individual dogs who do not gain a health benefit from being desexed and who may be harmed by the procedure. Given the evidence for desexing is that on balance there is a positive effect of desexing across the population of dogs and the evidence detailed in this paper on the potential benefits through dog bite risk reduction, there is likely to be a net gain in health benefits for dogs and the community. There is also the potential for a positive interplay of factors resulting from mandatory desexing of dogs that would improve dog management and welfare in general. Figure 2 outlies the potential benefits of such a policy, in particular if accompanied by a breeding code of practice (or equivalent) which includes both health and welfare considerations for breeding dogs. Desexing dogs is associated with a reduced risk of wandering,57 which in turn may reduce risks of dog bite or dog injury (eg, from road accidents) and also of lost dogs being euthanised. Furthermore, breeding controls from mandatory desexing plus a breeding code of practice may reduce the risk of puppy farms that raise large numbers of puppies without regard to the health and welfare of dogs.
Why mandatory? While the majority of dog owners elect to have their dog desexed to realise the many benefits outlined above, there remains a proportion of dog owners that do not desex their dogs. In a recent survey of 645 Australian dog owners, 18% of dogs were not desexed,58 while in a study of 86 practices in central and southeast England in deceased dogs 21.3% of females and 28.9% of males were entire.50 Only 53% of people in an Australian survey of 1016 pet owners agreed desexing is good for a dog's health, while 62% agreed it is good for the dog's behaviour and 61% agreed it is expensive.59 In focus groups with Irish pet owners (dog and cat), negative perceptions of neutering included perceived financial cost, perceived adequacy of existing controls and negative perceptions of impact on pet health and welfare.60 Making desexing mandatory and enforcing the requirement would push those reluctant to desex their dogs increasing the proportion of desexed dogs in the population. The findings from this review suggest that this is in turn may reduce the risk of the dog injury environment.
A limitation to this review is that all studies were observational in design, and all had a slightly differing definition of a bite and source of study participants. Furthermore, only four study populations were identified. Together this has limited the ability to understand the causal effect of desexing on dog bite risk. Future research should examine this association in detail, ideally with an experimental study design in addition to careful evaluation of the mandatory desexing legislation in South Australia to examine the effectiveness of this policy at a population level in reduction of dog bite risk. Two major databases were used to find studies, which although broad may have missed important papers examining desexing risk on dog bites. This would have been limited by reviewing the reference list of all included studies. Furthermore, the search was limited to English language papers, which may also have missed relevant papers.
Mandatory desexing of dogs alongside comprehensive education and training programmes provides promise to effectively prevent some dog bites and is an important area for future research. Given the complexity of inter-relating factors resulting in a dog bite, altering one modifiable factor will not prevent all attacks; however, it is a critical step towards increasing the safety of the injury environments. There are multiple benefits of desexing dogs in addition to the potential benefits for dog bite risk, which will increase the feasibility of a mandatory desexing policy in terms of acceptability to a range of stakeholders.
What is already known on the subject?
Dog bites are a cause of serious injury in the community.
Controlling the risk of dog bites has proven to be an intractable problem.
What this study adds?
Desexing of dogs is associated with a reduced risk of dog bite.
Mandatory desexing of dogs alongside comprehensive education and training programmes provides promise to effectively prevent some dog bites.
We would like to acknowledge the work of The Dog and Cat Management Board of South Australia that was instrumental in progressing the adoption of mandatory desexing of dogs in South Australia.
Contributors All authors designed the study, KD searched the databases, all authors extracted data and conducted quality appraisal, KD drafted the manuscript and all authors reviewed and approved the final paper.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.