Background Previous studies have identified risk factors for dog bites in children, but use data from individual trauma centers, with limited generalizability. This study identifies a population risk profile for pediatric dog bites using the National Trauma Data Bank. We hypothesized that the population at risk was younger boys, that such bites occur at home, are moderately severe, and are on the face or neck.
Methods For this retrospective cross-sectional study, a sample of 7912 children 17 years old and younger with International Classification of Diseases (ICD)-9 event code E906.0, for dog bites, were identified. Datasets from 2007 to 2014 were used. Data included patient’s gender, age, ICD-9 primary and location E-codes, AIS body region and AIS severity.
Results Most children were 6–12 years old and female, but a similar number fell into the narrower range of 0–2 years old. Injuries in the younger group frequently occurred at home, on the face and head, and with minor severity. Age of the child predicts the location of incident (P<0.001), the severity of injury (P<0.001) and the body region of the injury (P<0.001). Body region of the injury predicted its severity (P<0.001).
Discussion Younger children are more likely to receive dog bites, and bites incurred are likely of greater severity. Children this young cannot yet be taught how to properly interact with a dog.
Conclusions Dog bites are a significant source of morbidity for children. Based on the population risk factors profile generated, this study recommends targeting live dog education towards the parents of young children.
- animal bites
- risk factor research
Statistics from Altmetric.com
The American Veterinary Medical Association estimates that 36.5% of households in the USA own a dog as a pet, amounting to about 70 million dogs.1 Given that dog ownership is so prevalent and that a dog has the potential to inflict a bite, dog bites are a fairly common form of trauma. The Centers for Disease Control and Prevention (CDC) estimates that about 5 million Americans per year are bitten by dogs, half of whom are children.2 Dog bites can be a major source of morbidity for children, given that they may inflict disability or disfigurement at a young age. Various educational programs exist to train both children and parents on how to behave around dogs and how to prevent bites, but previous research has shown that while education is beneficial for these groups, these prevention strategies can be enhanced by incorporating live dog training.3–5 Due to the resource intensity of live dog training components in educational programs, identifying those who are most at risk for dog bites is valuable because it allows for the provision of targeted interventions. Live dog training has been shown to be the most effective form of dog bite prevention for these groups.3
To date, public health research on dog bites lacks a national focus. No national surveillance system exists for dog bites, so most research has focused on small hospital systems, primarily collecting data from emergency departments,6 emergency departments of trauma centers7 and from single or small systems of trauma centers.8–11 These studies suggest risk factors for pediatric dog bites include being a boy under the age of 12,12 but the small sample size limits their generalizability. Demonstration of these risk factors as being valid at a population-based level, as well as discovering novel risk factors, will aid dog bite prevention efforts. We anticipate that our study will elucidate risk factors and assist us in creating a pediatric population risk profile by analyzing the National Trauma Data Bank (NTDB) from 2007 to 2014.
Our group examined the Research Data Sets of the NTDB,13 which catalogues select data from trauma cases presenting to participating trauma centers across the USA and US territories. Our group analyzed the 2007–2014 datasets including persons 17 years of age and younger (a child, as defined by the National Institutes of Health)14 with the ICD-9-CM E906.0 code, Dog Bite. Age was categorized to roughly follow the age ranges suggested by the National Institute of Child Health and Development.15 Data collected include gender, age, ICD-9 primary and location E-codes; AIS body region and AIS severity score. AIS body region is a set of denotations for the portion of the body involved in an injury. Possible descriptors include head, face, neck, thorax, abdomen, spine, upper extremity, lower extremity and unspecified. AIS severity score is a numerical value assigned to an injury to denote how severe it is, with possible values of 1 through 6, corresponding to descriptors of minor, moderate, serious, severe, critical and maximum. In some cases, a score may be not assigned. Our group analyzed these data through summary statistics, two sample t-tests and Fisher’s exact tests for gender differences, Wilcoxon rank-sum tests for location of incident (in the home vs outside of the home) as a predictor of age, Kruskal-Wallis rank-sum tests for body region of injury as related to injury severity and for body region of injury as a predictor of age, and Spearman’s rank correlations for age as related to severity of injury. RStudio, V.3.3.1,16 was used to analyze the data. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE)checklist for cross-sectional studies was followed in the preparation of this manuscript. This checklist is designed to standardize cross-sectional studies and ensure that all information necessary for proper study replication and analysis is contained within the published manuscript. The paper’s format and content were based on the recommendations found within.17
Our group identified 7912 cases with ICD-9 906.0 code and younger than 18 years of age in the NTDB 2007–2014 database. The plurality of children in the sample were 6–12 years old (34%); however, a similar percentage of children were in the narrower 0–2 age range (30%). The largest number of children of any age was 2 years old (n=1028) (table 1). The mean age was 5.4 years, while the median age was 4.0 years (table 1).
Given these indications of age being an important factor in dog bites, several subanalyses were performed. Children in the 6–12 age range were specifically found, via Fisher’s exact test, to have a significantly higher incidence of dog bites that occurred on the street (142 such incidences) versus all other ages (173 such incidences) (P<0.001).
The majority of children were female (57%) (table 1). Given this large gender difference, a one-sample test of proportion was run, under the assumption of 49% of children in the US being female18 to test for a difference in gender proportion in this sample, relative to the US distribution. This test showed a statistically significant difference between proportions (P<0.001) (table 1). The majority of cases (83.7%) occurred at home (table 2). Five hundred and eighty-seven cases where location was not assigned or unspecified were not included in this calculation, for a total of 7325 cases. Most of the injuries overall were minor (79.6%) (table 3) based on a mean injury severity score of 1.3 and a median severity score of 1.0. Ten cases where severity was unassigned were not included in this calculation, for a total of 7902 cases. The most common bites occurred on the face (54%) (table 4). Moreover, 1251 cases where body region of injury was unspecified were not included in this calculation, for a total of 6661 cases. Sixteen per cent of the cases involved unspecified body regions (table 4). Subgroup analyses were performed to investigate the difference in gender. There was no significant difference between the genders in the number of bites assigned as severe or greater severity (P=0.058). Sixty-six boys of 110 total children received severe or more severe injuries as defined by the AIS scoring system. Additionally, there was no significant difference in mean severity of injury found between the genders (1.31 in boys vs 1.28 in girls) (P=0.134). There was, however a small but significant difference in the mean age of girls (5.52 years) versus the mean age of boys (5.14 years) (P<0.001). Finally, a significant difference in body region of injury between genders was found (P<0.001). Five percent more boys were bitten on the face than girls, while 3.5% and 2.5% more girls than boys were bitten on the upper and lower extremities, respectively.
Age as a predictor of location of incident was analyzed. Children bitten in the home were younger than children bitten outside of the home. The difference in age distributions for these two types of bites were compared using a Wilcoxon rank-sum test (table 2). The difference was small (median age of 4 years in the home vs median age of 5 years outside of the home), but significant (P<0.001). Age was also analyzed as a predictor of severity of injury using Spearman’s rank correlation, rho. Cases where severity was unspecified were not included. Severity was found to be inversely associated with age (P<0.001), with a rho value of −0.06, meaning more severe cases tended to be younger (table 3). Severity of dog bites within the home (median severity of 1, or minor) was not found to differ from the severity of dog bites inflicted on the street (median severity of 1, or minor). To account for the significant number of empty cells and low counts of cases that were rated severe or higher, a second Spearman’s rank correlation rho was performed, with data on severe, critical and maximum severity cases being combined to minimize these effects. A similar P value was obtained (P<0.001), with a rho value of −0.06 (table 3). Severe injuries had the youngest median age (1 year) (table 3). In contrast, the majority of injuries were minor (79.4%), with a notably older median age of 5 years (table 3).
Age was analyzed as a predictor of body region of injury using a Kruskal-Wallis test. The distribution of age was found to significantly differ by body region injured (P<0.001, df=8) (table 4). The youngest children were bitten on the head, with a median age of 3 (table 4). The largest number of children (n=3651) were bitten on the face, with a median age of 4 (table 4).
Body region of injury was analyzed as a predictor of severity of injury using a Kruskal-Wallis test. The distribution of injury severity was found to differ between body regions (P<0.001, df=8) (table 5). Due to data sparsity, a sensitivity analysis was conducted by collapsing severe, critical and maximum injury severity categories into one group. A highly similar P value was obtained (P<0.001, df=4) (table 5). Injuries of the spine were found to be associated with the greatest median injury severity score, 2 (table 5).
The preceding analyses support our original hypothesis. We hypothesized that most children bitten by dogs and reported in the NTDB would be young. Disparities in age within the studied population are significant. Over 90% of the sample was 12 years of age or younger, and almost 60% of the sample was 5 years of age or younger. The data are right-skewed, indicating that the distribution of dog-bite victims shows a clear majority in the lower age ranges. This disparity is particularly striking, but corroborates previous studies.5 6 The NTDB shows more of a bimodal distribution for injuries of all types, with a clustering of injuries at the younger end, and a larger clustering at the older end of the pediatric age range.13 The trend toward younger children being bitten indicates a lapse in supervision as a likely source of pediatric dog bites. Parents typically determine the place and interactions of young children. Likewise, young children have challenges with understanding how to properly interact with a dog.19 Interventions designed to prevent dog bites should thus be targeted at parents of young children so as to confer the maximum overall benefit. The group of 6–12 years old would also benefit from similar education, but targeted directly to them, given that they suffered the most bites overall. Parents could possibly be educated on dog behavior and the importance of supervizing their children, even around known pets. In young children, dog bites are a significant source of morbidity. Dog bites are likely to leave scarring or other disabilities, including acute emotional distress, and such an incident occurring so early in life means that children injured by a dog bite are likely to experience significant morbidity.19
The results from this study suggest a profile for pediatric dog bites by indicating that age is likely to influence body region and severity of the bite, in addition to the location of the bite. Children bitten in the home and on the face and head were more likely to be younger. Given that younger children are known to travel head and face-first (ie, crawling) and are prone to explore through touch, it is likely that a dog responding to an annoyance by the child would bite at the closest part of the child’s anatomy: the face or head,19 which is consistent with previous studies.19–22
Age also predicted the severity of the injury, but the sample estimate indicates that there is only a weak inverse relationship between the two variables. As age increases, the severity of the injury decreases and vice versa. The result was significant, but the sample estimate is small enough to indicate that the relationship may be tenuous at best. Since most injuries were classified as minor, there was no significant difference in severity between age groups. In contrast, the severity of the injury differed significantly by body region. One particular finding was that younger children are most likely to be bitten on the head, resulting in a minor injury severity score. Additionally, the highest number of children were bitten on the face. Alternatively, bites in any other body region (not to the face or head) tended to be more severe. This profile indicates that younger children are most at risk for injuries that may not be particularly severe or life-threatening, but are still likely to cause a significant burden of morbidity to the child. However, our results still indicate that the younger a child is, the more severe their bite received is likely to be, even though the majority of bites do end up being minor.
Overall, the incident profile for dog bites confirmed our hypothesis: that the majority of incidents would occur at home and on the face of the victim. This profile also confirmed previous findings of preponderances of victims with facial bites incurred at home.6 12 These analyses reveal a pattern of injuries present in this population, specifically that younger children are being bitten on the face or the arm, in the home, and likely by a known dog. There were likewise noticeable numbers of injuries occurring on the head and the upper extremities of victims. Our finding that most injuries were of minor severity confirms the results of previous studies,10 but refuted our hypothesis that the majority of these injuries would be of moderate severity. We hypothesized this because data collected by the NTDB comes from trauma departments where more severe injuries are normally seen. Younger children suffering dog bites with a greater degree of severity was also consistent with previous studies.5
Due to the appearance of a bimodal distribution in age of bitten children, older children were further examined to characterize the setting of their injury. The difference in bites occurring on the street experienced by this group is potentially indicative of the fact that this age group is doubly vulnerable to such bites: they are unlikely to have learnt protective behaviors and they are also the most likely group to walk some distance between home and school (in addition to playing on the street, often unsupervized). Incident location was then examined to see if severity differed between home and street bites. It was hypothesized that dogs in the home may simply be inflicting ‘warning’ bites on children in response to a disturbance and that dogs on the street may be more aggressive or territorial, but a comparison of median severities did not validate such a difference.
The difference in gender represented by this analysis is initially striking. We hypothesized that boys were more likely to be bitten than girls, though our results do not support our initial hypothesis. Girls in the given sample were found to be the significant majority of pediatric dog bite victims. Previous studies have demonstrated at least slight skewing towards male victims.5 6 The NTDB also shows a preponderance of male pediatric victims across all trauma cases, making our finding of more female pediatric dog bite victims especially unique.13 The subgroup analyses performed reveal that these gender differences may be due to the body region of injury and the age of the child, though this age difference was very slight. While neither of these differences can explain the gender difference, they do offer some insights. Female children were more likely to be bitten on their extremities than male children. Male children were more likely to be bitten on the face. Given that this is a more traumatic area of injury, for reasons including noticeability, parents may be more inclined to bring facial or head injuries to trauma centers for treatment, which seemingly contradicts the gender difference that we found in the sample. Some differences may also exist between our sampling method (only including cases presenting to a trauma center) and those of previous studies (likely including cases presenting to a general hospital or family physician), possibly revealing a gender disparity in dog bite cases that present to a trauma center versus those that present to other facilities.
The results presented do have several limitations. First, the NTDB data are not generalizable to the entire nation. Rather, they are only applicable to the population studied: children who present to trauma centers participating in NTDB data collection. Selection bias may be present in this study as only those injuries that are deemed to be severe enough (whether this is done by parents or a medical professional) for presentation to a trauma center are represented here. Undoubtedly, many dog bites occur that do not present to trauma centers (even if they should). Given that so many minor face bites were present in the data, it may be possible that many of these cases are presenting to trauma centers for primarily cosmetic surgery. In addition, AIS body region codes do not take into account the possibility of multiple injuries in a patient, meaning that our analysis of body region may be slightly lacking. Finally, this sample likely does suffer from under-reporting due to issues with data coding, including ICD-9 primary and location E-codes, as well as injury severity score.
For those children ages 0–17 who present with a dog bite to trauma centers participating in the NTDB, we find that the child is most likely to be between the ages 6 and 12, female and have been bitten on the face at home, resulting in injuries classified as minor. The significantly narrower age range of 0 to 2 years is also of importance because it holds a similar number of children to the 6 to 12 range. Age was clearly shown to predict the location of the incident, as well as the body region of the bite, indicating a potential pattern of injuries involving known dogs in the home in this sample. These results may inform future preventive efforts, allowing them to target those that are most at risk for dog bites, ultimately improving outcomes in this population. The results generated here are indicative of a need for parents to be educated on the best way to prevent dog bites at home in their young children. Children age 2 and younger are generally unable to make decisions as to whether a given dog is displaying signs of aggression or not, and thus cannot extricate themselves from a dangerous situation.19 Therefore, the burden of preventing such injuries in young children falls ultimately on parents. School-age (ie, 6–12 years old) and older children will likely benefit from behavioral modification approaches as well as education on how to detect signs of aggression in dogs.23 Our study indicates that the majority of dog bites occur in the home, which likely means that the majority of dog bites are received from dogs that are known to the affected children, or at least to the parents. These types of injuries are preventable, likely through something as simple as live dog training, as outlined in previous research, for adults and children of appropriate age levels.3 Preventing such injuries would confer significant benefit to the health of the pediatric population as demonstrated by this study.
What is already known on this subject
2.5 million children are bitten by dogs per year in the USA, indicating that dog bites are a significant source of morbidity in this population.
Live dog training is a more effective prevention tool than education, but it is resource intensive.
Previous studies indicate that boys under age 12 are at risk, but most studies lack a national focus.
What this study adds
On a national scale, girls under age 2 are the most likely to be bitten and the most likely to receive the most severe dog bites.
Such bites are most likely to occur at home, where parents can supervise children and dogs.
Interventions, including live dog training, should focus on reaching the parents of the youngest children.
The authors would like to thank those people at the Trauma Program of the Children’s Hospital of Los Angeles who contributed their time and effort to make the Pediatric Injury Prevention Scholars program, and thus this research, a reality.
Contributors JF conceived this study, analysed the data retrieved and prepared the manuscript for publication. DB assisted in data analysis and helped to prepare the manuscript for publication. JSU helped to prepare the manuscript for publication. RVB assisted in study conception, data analysis and manuscript preparation. All authors read and approved the final manuscript.
Funding This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Ethics approval This study was approved by Children’s Hospital Los Angeles’ IRB (CHLA-16-00401). All data used in this study originate from a database, with subjects giving consent to have their data included.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement All data used by this study are publicly available for purchase via the American College of Surgeons.
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.