Discussion

This study provides much improved indicators of the true burden of dog bites on public health. One in four people had ever been bitten by a dog and when a bite occurred, only one-third sought any form of medical treatment. Only 1 of the 172 bites (0.6%) reported in more detail resulted in hospital admission; given the small sample size and self-selection of which bites to describe further, this finding must be interpreted with caution but may suggest that figures based on hospital admission records (eg, 6743 hospital admissions for ‘dog bites and strikes’ per year in England5 are a large underestimation of true incidence). Although in some senses this is reassuring that many dog bites do not cause significant physical injury, it is also known that even relatively minor bites can cause significant distress to the victim;16 thus they should not be considered unimportant to prevent. Perhaps multiple dog owners were more likely to complete the survey and inflate our dog bite incidences; however, the annual incidence of dog bites over the previous 12 months in our community was 18.7 per 1000 population which matches almost exactly population based studies in USA in the 1990s10 and 2000s.32 Our incidence of 1873 per 100 000 per year is nearly three times that of the 740 per 100 000 often cited but with no clear source.11 Regarding victim-level risk factors, our discovery that reporting being more emotionally stable is associated with reduced odds of having been bitten by a dog, is completely novel and unreported elsewhere. Males were also 1.6 times more likely to have been bitten than females and owners of multiple dogs were 3.3 times more likely to have experienced a bite than non-dog owners. Regarding our final aim, it was more common for participants to have been bitten by dogs completely unknown to them than familiar dogs.

This study has strengths and limitations. It only examined households in one geographical location and thus may not be generalisable to the wider population. It was part of a semirural small town, with a variety of housing types and socioeconomic factors.27 Within the survey population, 200 people (28.8%) reported currently owning a dog which is similar to estimations made by other studies conducted within UK.1 However, the respondents to the survey were proportionally more females (less likely to have been bitten) and older age groups (more likely to have been bitten). Furthermore, under 5 s, thought to be a high risk population, were not surveyed. In addition, the North-West of England experiences the highest rates of hospital admissions due to dog bites in the UK.5 Thus there are a number of reasons why our findings in terms of prevalence and incidence could have been both overestimated but also underestimated.

However, this study provides the first UK-based investigation of prevalence, incidence and risk factors for dog bites at a community population level and has significant advantages over hospital attendance data sources. One limitation of the data collected regarding the circumstances of the dog bite is that the respondent was asked to choose one bite event to describe if they had experienced multiple bites. However, a substantial number of bite victims within the study reported they had only been bitten once (57.6%), therefore it was possible to investigate the potential for bias by comparing relationships based on all dog bites with those based on individuals that had been bitten once and none was found. It is unknown whether the proportion of hospital admissions for dog bites, or even the actual rate of dog bites, has changed over the years, and as these data were collected retrospectively covering a number of years, it is unknown if it had an effect. While assessing bites retrospectively may lead to some details of the event being forgotten, it is unlikely that this would have affected the relatively simple questions regarding the bite required within this study. Recall problems are most likely when there is little time to retrieve information, the activity is common and timing is required to be recalled accurately.33 Dog bites are relatively uncommon over a lifetime, may be emotionally significant and we did not require specific timing of the event except for an indication of whether it had occurred in the past year.

Males were more likely to be bitten than females, which supports evidence from other studies at least in children.34–36 It may be that men receive more serious injuries and so are more likely to appear in hospital data; however, there was no evidence here that men were more likely to require medical treatment than women. It could be hypothesised that personality variation between genders has resulted in more males being subject to dog bites. Reporting being less emotionally stable was associated with an increased frequency of dog bites and so was being male. However, the multivariable analysis conducted adjust for the effects of these variables independently, and males were more likely to rate themselves as being more emotionally stable (data not shown). Negative indicators of emotional stability include descriptors such as insecurity, fear, instability and intrusiveness.37 Neuroticism (the other name for emotional stability) is well known to be linked to public health behaviours and outcomes;38 39 perhaps then this aspect of personality may be linked to performing behaviours that provoke dog bites. However, perhaps any association is not causal, in that people that have been bitten become more nervous or the association is being confounded by another factor such as socioeconomic status. Previous research has outlined mental disorders40 and attention deficit hyperactivity disorder in children41 as risk factors for dog bites. Another study found that less shy children took greater risks with a therapy dog.42 This adds strength to our finding that personality may be associated with dog bite incidence. There also is  evidence of an association between parent personality and child parenting style,43 which appears to echo in links between owner personality and dog personality44 and the nature of the dog–owner relationship.45 These studies suggest that nervous/anxious owners may have nervous/anxious dogs, which may be another explanation for increased bite risk. Much more research into the possible association with personality is now required, especially in order to understand if and how this knowledge could be used in dog bite prevention.

Many previous studies report that individuals are more commonly bitten by dogs that are familiar to them;15 23 46 however, this study has demonstrated that 54.7% of bites were by dogs the victim had never before seen. One hypothesis that could explain this difference is a possible bias towards people seeking medical treatment if bitten by a dog known to them,35 46 resulting in these types of dog bites appearing in hospital data. However, there was no evidence in our study that relationship to the dog varied with whether or not they received medical treatment. It may be that the association is present but not detectable given the study size.

Many of the historical bites reported occurred to individuals when they were young; however, only 3 of the 48 current children within the study (6.3%) had been bitten. This could indicate that childhood bites are becoming less common than they once were. Reduction of children playing outdoors in the modern era may be reducing the incidence of dog bites in children, in particular by strange dogs.47 The relatively high proportion of bites from unfamiliar dogs reported by participants in this study could be related to reporting of bites that occurred many years ago in childhood when free-roaming dogs were more commonplace. The relationships between risk of dog bite, age at bite and relationship to the dog require further investigation in a larger sample.

Multiple dog owners have previously been found to be at increased risk of dog bites,32 as here. It could be hypothesised that owners of multiple dogs may be a greater risk due to breaking up dog fights8 or simply having greater exposure to dogs and is a recommended area for further research, for example: are these bites from their own dog? are dog fights more likely? do these persons have increased exposure to other dogs in general? are multiple dog owners more likely to work in dog-related professions?

Our proportion of bites requiring hospital admission (0.6%) was very low and proportion requiring any medical treatment (33%) was higher rather than lower than the 19% reported in USA.32 Potential bias introduced into our estimate from those bitten multiple times is likely to be towards choosing to report to us the more serious injuries than less serious, meaning our proportion of bites described requiring hospital treatment may be underestimated. In fact, the proportions requiring medical treatment were similar even if someone had only been bitten once so we can be confident in our estimate in this respect. However, it must be noted that these statistical estimates is based on one community survey in one geographical area and thus it is questionable whether it can be generalised to the entire of England and certainly requires repeating to validate our proportions requiring medical treatment found.

Thus, future research should be conducted to corroborate our findings in other UK populations in order to assess the public health impact of dog bites and the significance of severity of injury. In particular, the causes and contexts of higher risk in owners of multiple dogs require further investigation and establishment of the repeatability and nature of the relationship between victim level of emotional stability/neuroticism and dog bite risk. Future studies should also investigate in more detail the contexts and situations regarding being bitten by familiar and unfamiliar dogs.48 Research is also required to estimate dog bite risk in under 5 s, which our study did not address. Our study also did not address any factors associated with the type of dog at risk of biting, such as sex, age and breed, which require more investigation as findings have been inconclusive.49

In conclusion, this study demonstrates that the most severe dog bites, of highest public health significance, are thankfully a small proportion of overall bites that occur. However, many more bites occur than present for treatment at a hospital and community level dog bite statistics such as those presented here are important to collect to validate our findings here. Unlike medical records-based studies, community survey data are likely to identify additional bites not requiring medical treatment and which are of unknown concern to the victim and whose public health significance is unclear. Further, this study has found that dog bite analysis based on hospital data does not always reflect the relationships seen when assessing population-based surveys. To better understand dog bites, future research should attempt to explore the circumstances of dog bites, the nature of the injury and victims’ perceptions and impacts on them. It is essential that previously assumed risk factors are reassessed as this study has revealed that prior beliefs, such as bites typically being from familiar dogs, are contested. Dog bite prevention initiatives may need to rebalance their target to include strange dogs with equal importance, as opposed to an emphasis on family pets in the home. Our study confirms that male victims are at increased risk and should be targeted for prevention. Dog bite prevention schemes may also need to target particular behaviours around dogs by different victim personality types. If UK or elsewhere hope to reduce dog bite incidence, it is essential that risk factors are accurately assessed in order to impose effective and well-informed dog policies in the future.