Patterns of youth injury: a comparison across the northern territories and other parts of Canada

Background Injury is the leading cause of death for young people in Canada. For those living in the northern territories (Yukon, Nunavut, and the Northwest Territories), injury represents an even greater problem, with higher rates of injury for people of all ages in northern areas compared with the rest of Canada; however, no such comparative studies have focussed specifically on non-fatal injury in youth. Objectives To profile and examine injuries and their potential causes among youth in the northern territories as compared with other parts of Canada. Design Cross-sectional data from the 2009/2010 Health Behaviour in School-aged Children survey (youth aged 11–15 years) were examined for the Canadian northern territories and the provinces (n=26,078). Individual survey records were linked to community-level data to profile injuries and then study possible determinants via multilevel regression modelling. Results The prevalence of injury reported by youth was similar in northern populations and other parts of Canada. There were some minimal differences by injury type: northern youth experienced a greater percentage of neighbourhood (p<0.001) and fighting (p=0.02) injuries; youth in the Canadian provinces had a greater proportion of sport-related injuries (p=0.01). Among northern youth, female sex (RR=0.87, 95% CI 0.81–0.94), average (RR=0.88, 95% CI 0.80–0.97) or above-average affluence (RR=0.84, 95% CI 0.76–0.91), not being drunk in the past 12 months (RR=0.77, 95% CI 0.69–0.85), not riding an all-terrain vehicle (RR=0.81, 95% CI 0.68–0.97) and not having permanent road access (RR=0.89, 95% CI 0.80–0.98) were protective against injury; sport participation increased risk (RR=1.45, 95% CI 1.33–1.59). Conclusions Patterns of injury were similar across youth from the North and other parts of Canada. Given previous research, this was unexpected. When implementing injury prevention initiatives, individual and community-level risk factors are essential to understand; however, specific positive safety assets that might exist in different community contexts must also be considered.

an adolescent (1) and having low socio-economic status are all associated with increased injury risks (13,14). Although there are many behavioural risk factors associated with injury, excessive alcohol consumption and the use of motorized vehicles are two notable concerns in youth populations throughout Canada (15Á18).
Previous studies have determined that the injury experiences of residents of the northern territories differ from those of Canadians as a whole. In addition to higher rates of hospitalizations and mortality, northerners experience a greater proportion of injuries due to selfinflicted harm, off-road vehicle use, being a pedestrian, fighting, assaults and maltreatment compared with southern regions of Canada (19). Few previous studies have focussed their efforts on individual and community exposures and injury experiences specific to northern youth populations, and very few have focussed on the positive assets that might exist in communities for injury prevention. The purpose of this study is to examine injury occurrence among northern youth and to compare injury patterns with those of youth from other parts of Canada. We profile both individual and community-level risk or protective factors for injury. This analysis provides novel evidence to support the development and targeting of specific injury prevention efforts.

Methods
Description of data set and study population To examine injury occurrence, data were taken for secondary analysis from the 2009/2010 (Cycle 6) Health Behaviour in School-aged Children (HBSC) study in Canada. The HBSC survey provides information on youth health behaviours and outcomes, and in 2009Á2010, 26,078 youth aged 11Á15 years participated (Table I). Of these, 3,942 students attended one of 80 schools located in the three northern territories, and were classified as ''northern youth'' for this study. The experiences of these students were compared with those from participating provinces. The data set provides information on demographic characteristics, risk behaviours and information for evaluation of injury experiences, including information on injury frequency, location, activity and severity. Community demographic characteristics such as population size and population demographic were determined from the census subdivision (CSD) level data in the 2006 census

Outcome variables
The main outcome for this study was a student's experience with injury. Study participants were asked how many times they had been injured and treated by a doctor or nurse during the past 12 months. Response options were ''I was not injured in the past 12 months,'' or a range from ''1'' to ''4 times or more.'' For the most serious injury reported, participants were asked about injury location (''Where were you when this one most serious injury happened?''); activity at time of injury (''What were you doing when this one most serious injury happened''); and injury severity (''did your injury required medical treatment such as the placement of a cast, stitches, surgery or staying in a hospital overnight?'' and/or ''missed at least one full day from school or other usual activities''). Injuries were deemed severe if they required any of the mentioned treatments or resulted in one or more days of missed activities (5).

Exposure variables
The HBSC survey asks students to identify their sex and school grade. Relative family affluence was determined by the question: ''How well off do you think your family is?'' (5 options: ''Very well off'' through ''Not at all well off''). Ethnicity was reported using a number of standard categories including Métis, Inuit or First Nation (20). Students of mixed or multi-ethnic backgrounds could choose up to three options for ethnicity. Impaired driving behaviours were evaluated using the question: ''During the past 30 days, how many times did you drive (or ride in) a car or other vehicle while under the influence, or being driven by someone under the influence of alcohol, using marijuana or other illegal drugs?'' [''Never,'' ''1Á3 times'' (sometimes) and ''4 or more times'' (often)]. Off-road vehicle use was determined by ''During the past 12 months, how often did you wear a helmet when riding other vehicles (e.g. snowmobile, ATV, dirt bike)?''; participants were categorized as ''Yes/ riders'' if they selected ''Never,'' ''Sometimes,'' ''Most of the time'' or ''Always,'' and ''No'' if they selected ''I did not ride other vehicles.'' Helmet use was inferred directly from this question. Participants were asked, ''On how many occasions (if any) have you done the following things? In the past 12 months: been drunk, smoked cigarettes. Responses included ''Never,'' 1Á2, 3Á5, 6Á9, 10Á19, 20Á39 and 40 or more times; for study purposes, these were categorized into ''Never,'' ''Sometimes'' (1Á5) and ''Often'' (6 or more times). Finally, sports participation was determined by ''Do you participate in a sports club or team?'' (yes/no).
Population size was categorized as small ( B500), medium (500Á3,000) or large (!3,000). Communities were classified into groups ranging 0Á20%, 21Á60%, and !60% of the community with Aboriginal ethnicity. The percentage of dwellings requiring repair in a community was categorized into three groups: B5%, 5Á10% and !10%. These cut-points represented approximate tertiles in the distribution of each variable in the northern sample of communities and were determined for the neighbourhood within a 1 km buffer around each school where the survey was administered. Permanent road access, defined as any permanent, year-round access road linking the school community to other communities, was determined using GIS/Google maps (Google Maps and Earth Imagery, 2014). Communities with only seasonal roads or year-round non-road access were not classified under the designation ''permanent road access.'' Since the 1970s, local option provisions in provincial and territorial Liquor Control Acts allow communities to decide, through plebiscite, their preferred approach to community alcohol control (21). By hand searching the regulatory database we were able to classify each community as ''open/wet'' (no community regulations beyond what is found in the Territory's Liquor Act), ''restricted/ damp'' (restrictions to limit the quantity of alcohol that may be imported, consumed, purchased or transported), or ''prohibited/dry'' (complete ban on the sale, purchase, transport or consumption of alcohol) in 2008.

Statistical analysis
All analyses were conducted in SAS 9.4 (SAS Institute, Cary, NC). Youth from the northern territories (Yukon, Nunavut, and the Northwest Territories) and other parts of Canada (all provinces except New Brunswick and Prince Edward Island where data were not available) were first described by their individual and home community characteristics. For the children in the Canadian provinces, standardized population weights were applied to ensure representativeness at the provincial level. This was not required for the northern children as a near census was conducted. The Rao-Scott chi-square test, which adjusts for clustering at the school level, was used to assess whether differences between the two regions were statistically significant (p B0.05). Next, the injury experiences of children from the two regions were profiled. Rate ratios, and corresponding 95% confidence intervals, were calculated using the SAS procedure PROC GENMOD.
The final analysis of injury risk factors focussed exclusively on northern children. A multilevel, multivariable, log binomial regression model was used to examine the association between individual and community characteristics, and ''any injury.'' A priori, the decision was made to keep grade, sex and level of affluence in the final model because of the previously identified strong relationships with injury (1,5,14). A backwards elimination approach was used to assess all others. Ethics Ethics approval was granted by the Queen's University General Research Ethics Board, including specific approval for the use of the northern data with student ethnicity. All data were de-identified. Consent was obtained by active, passive or other community-based approaches as per local protocol. The HBSC 2009Á10 study was reviewed and approved by the Aurora Research Institute (Northwest Territories), the Nunavut Research Institute and the Heritage Branch of the Yukon Territory Department of Tourism and Culture. The survey is administered in partnership with territorial and provincial government representatives from health and education in each jurisdiction through the Joint Consortium for School Health.

Results
Individual and home community context Table II describes the populations of youth from both the northern territories and the provinces of Canada. Both populations had similar sex and grade distributions; about half of each sample identified as male, and 60% of children from both locations fit into the grade 6Á8 category. Over half (53.2%) of northern children reported Aboriginal ethnicity, compared with 6.1% of children residing in other areas of Canada. Self-reported levels of affluence were marginally different between the two populations, with a slightly higher percentage of northern youth falling into the average or not well-off category. Impaired driving, or driving in a vehicle with an impaired driver, in the past 30 days was reported ''sometimes'' by 8.8 and 5.9% of northern and provincial youth, and ''often'' by 5.9% of northern youth and 4.2% of provincial youth. All-terrain vehicle (ATV) and motorized vehicle use differed considerably between the two populations; 76.2% of northern youth reported use in the past year, compared with only 52.1% of young people from other parts of Canada. Youth from both populations had similar drinking habits, with about 75.0% reporting having never been drunk in the last 12 months, and approximately 11.0 and 14.0% reporting this behaviour occurring sometimes or often.
The home community characteristics varied considerably between the two populations (Table III). Children living in the North more often lived in small communities; over 70% (14.4'57.7%) of northern youth lived in communities of less than 3,000 people, compared with about 35% (5.6'30.9%) of youth from the rest of Canada. Northern communities had a higher percentage of Aboriginal people and had more homes in need of major repairs. Over a quarter (26.2%) of northern children lived in communities without permanent road access, and 51.4% (47'4.2%) lived in communities with prohibitive or restrictive alcohol policies.

Injuries
Reports of injury experiences were available for 3,682 northern youth and a weighted sample of 25,413 young people from other parts of Canada. No difference in the occurrence of overall injury was observed between the two populations: 41.8% of northern and 41.5% of provincial youth sustained injury within the past year (Table IV). There were also no significant differences found in injury severity or frequency. For both populations, about 17% of children reported a severe injury in the past year, and approximately 21% reported being injured two or more times in the past year. When injury occurrences were profiled by location and activity, the distribution of injuries by location or type differed slightly between populations. Northern youth experienced a slightly higher proportion of injuries located in the neighbourhood (4.7% compared with 2.9%, p B0.001) and sustained in fighting (3.0% compared with 2.2%, p 00.01), whereas youth in the provincial subsample experienced slightly higher proportion of sport-related injuries (22.4% compared with 20.1%, p 00.02). No other significant differences were found by injury location or activity. For both populations, a sports facility or field was the most common location for injuries to occur. Sports were the primary activity leading to injury, accounting for over 20% of injuries in both populations.

Injuries and northern youth
Several individual characteristics and community exposures were shown to be protective against risk of injury among northern youth, after adjustment for potential confounding factors (Table V). Being female and of at least ''average'' affluence were protective. Certain individual behaviours, such as not getting drunk (RR00.77, 95% CI 0.69Á0.85) or not riding a motorized vehicle (RR00.81, 95% CI 0.68Á0.97) in the past year, were also protective against injury. Organized sport participation was determined to be a risk factor for injury (RR 01.45, 95% CI 1.33Á1.59). There appears to be a trend towards injury protection in restricted and dry communities. Higher rates of injury were reported in restricted and open communities compared with prohibited communities, but differences were not statistically significant. Lack of permanent road access was found to be protective against injury (RR 00.88, 95% CI 0.80Á0.98).

Discussion
This study examined the sociodemographic context and patterns of injury occurrence among Canadian children grades 6Á10, with a comparison of injury rates between youth in the northern territories and other parts of Canada. Unexpectedly, both populations had similar overall injury prevalence; however, we did identify some differences between northern and other Canadian youth in terms of patterns of injury; and this confirms previous findings (19). Northern youth experienced higher proportions of neighbourhood and fighting injuries and lower proportions of sport-related injury. The slightly elevated risks for neighbourhood injury may be attributable to where and how youth spend their non-school time. If northern youth spend different amounts of time and engage in different activities in neighbourhoods than other Canadian youth, this could have an influence on their injury risk and experiences in these environments. More study on time and space use patterns among youth would also be useful. Time-use patterns have been shown to be different between urban and rural youth (22) with implications for substance use for example.
The finding that fighting injuries occur 36% relatively more often among northern children is sobering, yet consistent with other findings with respect to violence (19) and violent crime (23). In contrast, we noted that youth living in the northern territories had lower prevalence of sport-related injury. As with other context specific injuries, this may be attributable to exposure, as smaller and remote communities may have fewer opportunities for organized sport and recreational activities (24). Despite differences in proportions of sport-related injury, it is still the most prevalent activity associated with injury in both populations, and is an important area of focus for future injury prevention initiatives. Factors and characteristics found to be protective against injury included being female; of average or high affluence; not being drunk in the past year; not riding an ATV, snowmobile or dirt bike; and living in a community without permanent road access. Although many of these exposures are known protective factors, off-road motorized vehicle use and community road access warrant comment. Many northern youth use ATVs, snowmobiles or dirt bikes (76.2%); helmet use is variable. Northern populations experience a proportion of injuries associated with the use of these kinds of vehicles that is higher than the Canadian average (19). Although not riding is protective, it may not be reasonable to suggest ATV or snowmobile disuse, as they are a principal mode of transportation in Canada's north (25). Instead, injury prevention initiatives should focus on risk management, for example, education and enforcement of safe use of motorized vehicles, or discouraging recreational ATV or snowmobile use by children (25). Additionally, impaired driving among youth and those using ATVs is an important concern (26) and essential to focus upon. In this study, wearing a helmet was found to increase risk of injury; however, this anomaly may be attributed to the fact that helmet use likely indicates greater frequency of motorized vehicle use and more time or opportunity for risk.
Our finding of equivalent prevalence of injury in northern as compared with other Canadian youth was unexpected given past literature describing the numerous risk factors for injury in rural, remote and northern communities, as well as above average rates of injury and injury mortality previously reported in northern populations (3,27Á30). There might be a number of explanations for our disparate results. First, past studies have focussed more on severe injuries or injury-related fatalities (3,4) and included adults. Injury patterns may differ when considering injuries of different levels of severity, injury mortality, or among population of varied ages. Second, injury incidence may indeed be higher in northern environments, but we are capturing fewer injuries among northern youth because of health care access issues, as the HBSC measures injuries ''treated by a doctor or nurse.'' While many northern communities have nurses and nursing stations, fewer have doctors or other health care staff on a permanent basis. If there were differential access to, or use of, health care services for injuries, this would impact our ability to compare the two groups. A third explanation for the unexpected parity in northern and provincial youth injury could relate to possible protective factors that may exist in northern families and communities that may mediate the effects of hazardous exposures if they are indeed greater in northern settings. For example, while populations in rural and remote communities have been identified as being at increased risk of injury and other health disadvantages, dry communities in northern regions have reported overall fewer alcohol-related hospitalizations and lower rates of violent crime than other communities (31,32). We have found in ongoing analysis with HBSC data that youth in dry communities consume less alcohol on average than their peers in wet or damp communities (paper in preparation). In our study of injury in youth, there appears to be an encouraging trend for communities with alcohol prohibition and restriction to have lower injury rates (see unadjusted and adjusted estimates in Table V). Presence of a community alcohol policy was not statistically significantly associated with risk for injury after adjustment for other factors; however, in analytic assessment of this, we determined that it was the addition of the variable ''permanent road access'' in the fully adjusted model that accounted for the reduction in the alcohol policy influence. Knowing the northern community context, this is not surprising. While not completely collinear, community policy option does tend to distribute roughly by community remoteness, where fly-in communities represent most of the dry communities (i.e. only one of the dry communities had permanent road access). Not having permanent road access, a proxy measure for remoteness, is associated with social and physical environmental factors that may influence youth injury. These factors include, for example, the number and type of vehicles in a community; existence and condition of community infrastructure like playgrounds or sports fields; and the diversity or overall number of adults available to take on roles of oversight, care or mentorship. These characteristics may represent important factors in the prediction of youth injury. No permanent road access may also reduce injuries associated with riding or driving in vehicles such as cars or trucks. By adjusting for both alcohol policy option and permanent road access, some of the influence of these factors may have been hidden. However, these analyses point towards important potential safety assets at a community level. Future analyses could explore these and additional community factors as unique predictors to further establish potential aetiologic pathways.
Strengths of this study include the size and representative nature of the sample, which permitted subanalyses by specific injury characteristics. The comparison between the northern territories and youth in other parts of Canada gives researchers and policy-makers a frame of reference when making decisions regarding where injury prevention efforts need to be focussed. Another strength of this study lies in our use of advanced multilevel models, accounting for clustering that occurs at the school or community-level, to simultaneously examine associations between individual and community characteristics, and injury. This study was also one of the first to examine nonfatal injury among a population based sample of northern youth; many previous studies focussed only on injury mortality or injuries presenting in a hospital setting.
This study also has limitations. Weaknesses stemming from the HBSC survey include its self-reported nature and the fact that it is cross-sectional and does not allow definitive tests of causation. Some analyses yielded wide confidence intervals; this could be improved by larger sample sizes or a different analytical approach. Northern communities in the HBSC sample were identified as those in any of the three northern territories; we recognize there are other definitions of ''northern'' where areas of the provinces might also be included. In addition to lacking information on fatal injury outcomes and the physical nature of injuries, the HBSC survey does not differentiate unintentional injury from intentional injury. Intentional injuries are a leading cause of hospitalizations and death throughout Canada (1,2), particularly within the North (27,28), and it could be beneficial for future studies to examine intentional injuries uniquely.
Public health efforts centred on injury prevention can benefit from the findings of this research, which encourages acknowledgement of the similarities and differences between northern youth and those living in other parts of Canada. It also encourages emphasis on positive safety assets that might exist within communities. In some northern communities, restrictions on alcohol availability or lack of road access (with its implications for vehicle use) both appear protective for example. Targeting safe sport participation is important. Variations in the distribution of injuries by location and activity, although minimal, reveal areas where prevention efforts could also be focussed. In any youth population, careful consideration of the context in which injuries occur and knowledge of community-specific exposures is essential to create effective and successful injury prevention programmes. Our findings indicate that community and socio-economic characteristics influence youth injury risk. Prevention efforts should focus in these areas in addition to individual behavioural interventions.

Disclosure
The authors have no financial relationships relative to this article to disclose. CIHR Team in Youth Injury Prevention (grant number MOP97961) and the CIHR and the Heart and Stroke Foundation of Canada (grant number PCR 101415).