Effects of Neighbourhood, Family, and Child Behaviour on Childhood Injury in Canada

1. Introduction

Theoretical and empirical evidence from the injury literature suggest three identifiable sets of influences on childhood injury: the child, the family, and the neighbourhood. At the individual level, the risk of injury is linked to child age, gender, and behaviour (Davidson, 1987; Hillier & Morrongiello, 1998). At the family level, a high frequency of injuries among young children in the home indicates the importance of understanding how family can affect childhood injury occurrence (Matheny, 1988). For example, research suggests that differences in parenting practices may explain differences in injury rates among families (Zettle & Hayes, 1983; Aagran, Winn, Anderson, & Del Valle, 1998). At the neighbourhood level, studies suggest that neighbourhood and community characteristics have important influences on childhood injuries and other health-related issues (Bronfenbrenner, 1986; Jencks & Mayer, 1990; Kupersmidt, Griesler, DeRosier, Patterson & Davis, 1995; Malmstrom, Sundquist, & Johansson, 1999). Children who live in disorganized environments are known to be at increased risk for injury, as are children who live in low-income neighbourhoods (Jolly, Moller, & Volkmer, 1993; Matheny, 1986, 1987; Nersesian, Petit, Shaper, Lemieux & Naor, 1985; Valsiner & Lightfoot, 1987). Despite evidence of the importance of contextual variables, characteristics of neighbourhoods and families have been studied less frequently than individual characteristics. Consequently, little is known about the neighbourhood and family processes that influence childhood injuries (Gallagher, Hunter & Guyer, 1985; Hu, Wesson, & Kenney, 1993; Peterson & Stern, 1997). In the following sections, we will review some of the evidence that links characteristics of the child, the family and the neighbourhood to childhood injury.

1.1 Child Characteristics and Childhood Injury

Injury risk is linked to a child's age. Young children tend to identify fewer risk factors, and do so more slowly than older children. In a study of 120 children aged 6-10 years, 6-year-old children identified fewer risk factors, and did so more slowly than 10-year-old children. The 6-year-old children also had more difficulty than the older children did in identifying how to prevent injuries (Hillier & Morrongiello, 1998).

Injury risk is also linked to gender. From birth to 24 years of age, boys are more likely than girls to sustain injuries (Soubhi, Raina et al., 1999). Boys are also more likely than girls to suffer the most severe forms of injury and be hospitalized (Vital Statistics Agency, 1996). These gender differences tend to be more pronounced in older children (Baker, O'Neill, & Karpf, 1984; Canadian Institute of Child Health, 1994; Matheny, 1988; Rivara & Mueller, 1987).

The causes of increased risk of injury among boys are not well known. Using data from 197,516 consumer product-related injuries, Rivara, Bergman, LoGerfo, and Weiss (1982) found that differences in exposure to risks only partly explained gender differences in injury rates. Gender differences in injury rates may be related to differences in behaviour or to differences in risk perception.

According to parental reports, boys are more active than girls and are more likely to sustain injuries (Bijur, Stewart-Brown, & Butler, 1986; Langley, McGee, Silva & Williams, 1983; Kohen, Soubhi, & Raina, 2000). In experimental studies, boys were observed to be more active, disruptive, less manageable, and to have more contact with hazards than girls (Matheny, 1986; Matheny, 1988; Cataldo et al., 1992). Manheimer & Mellinger (1997) found an association between the frequency of injury occurrence and maternal reports of child activity levels. The association was not significant when child activity ratings were reported by teachers. However, in Manheimer and Mellinger's study, injuries were retrospectively reported and the analyses were not adjusted for socioeconomic factors.

Boys tend to be more aggressive than girls (Bijur, Stewart-Brown, & Butler, 1986). There seems to be general agreement that aggressiveness is a risk factor for injury (Davidson, 1987). Studies show that early infant temperament characterized as "difficult" is associated with increased risk of injury during the preschool years (Bijur, Golding, Haslum, and Kurzon, 1988). In a review of the literature, Wazana (1997) found aggressive behaviour to be consistently related to general injuries but not to pedestrian injuries.

Aggressive behaviour is often highly correlated with hyperactivity. However, the link between hyperactivity, aggressive behaviour and injury is not clear, and in most studies the level of risk of injury due to hyperactivity is low (Bijur, Stewart-Brown & Butler, 1986; Davidson, 1987). In a critical review of the literature, Davidson (1987) reported that with all prospective designs hyperactivity did not predict the occurrence of injury while aggressive behaviour always did. Using a sample from the British Births Survey that included 11,966 children from a representative birth cohort, Bijur, Stewart-Brown, and Butler (1986) found that aggressive behaviour was more strongly associated with injuries than was hyperactivity. The authors also reported an interaction between hyperactivity and aggressive behaviour. They concluded that aggressive behaviour might increase risk-taking and impulsiveness among overactive children.

Another study by Davidson, Taylor, Sandberg, and Thorley (1992) used a prospective cohort (16-month follow-up) to examine hyperactivity as a risk factor for subsequent injury. The cohort included 1,740 boys from age six to eight years, attending school in a borough of London, England. The rate of injury occurrence among the boys was assessed from the records of five emergency departments in the borough. The behaviour of the boys was measured by three sources: the parents, the teachers, and by direct observation by the investigators. The study accounted for the socioeconomic characteristics of the boys' parents that included income, education, and occupation. The study also accounted for whether the parents would allow the same degree of independence in their boys if they were hyperactive as they would if they were non-hyperactive. Although the length of follow-up may not have been long enough to detect any changes in the boys' behaviour, the study did not find any relationship between hyperactivity and injury, regardless of the source that measured the behaviour.

Gender differences in injury occurrence also seem to be related to differences in risk perception. In addition to displaying higher activity levels than girls, boys tend to underestimate risks and engage in more risk-taking behaviours than girls do (Alexander, Somerfield, Ensminger, Kim, & Johnson, 1995). Boys are more likely to repeat behaviours that led to previous injuries (Coppens & Gentry, 1991; Ginsburg & Miller, 1982; Hillier & Morrongiello, 1998). Boys are also more likely to attribute injuries to bad luck, while girls are more likely to attribute injuries to their own behaviour (Morongiello, 1997).

In summary, there is a consensus that injury risk among children varies by child age and gender. Differences in risk perception and risk management as well as behavioural differences seem to be related to gender differences in injury rates. Younger children tend to identify fewer risk factors, and do so more slowly than older children. Finally, the link between hyperactivity, aggressive behaviour, and injury remains unclear. The elevation of risk of injury due to hyperactivity is small in most studies while aggressive behaviour in children is more frequently linked to an increased risk of injury.

1.2 Family Characteristics and Childhood Injury

Childhood injuries are linked to factors such as a child's age, gender and behavioural traits. They are also linked to home and family characteristics (Cataldo et al., 1992; Ciastko, 1997; Davidson, 1987; Matheny, 1987). Schor (1987) examined 693 two-parent families with one, two, and three children aged from birth to ten years that were enrolled in the Columbia Medical Plan from 1974 to 1979. The study found a clustering of individuals within families that exhibited similar unintentional injury patterns. These patterns were stable over time for the individuals and their families. Boys were at greater risk for injury than girls were and children aged 6 to 14 years were the most likely to be injured. Schor (1987) could not determine whether the clustering of injuries was due to similar behavioural risk factors, similar physical and emotional environments, or to familial patterns of health care utilization. However, Schor's study supports the hypothesis that a child's injury experience can be influenced by home and family characteristics.

Dershewitz & Christophersen (1984) indicate that most injury related deaths of children younger than 5 years of age occur in the home. Gallagher, Hunter and Guyer (1985) reported that the percentage of injuries that occurred in the home was approximately two thirds of all childhood injury occurrences, and 91% of these injuries occurred to children under the age of 5 years. Hu, Wesson and Kenney (1993) conducted a study using injury surveillance data gathered from the emergency department of the Hospital for Sick Children in Toronto from 1990 to 1991. During this one-year period, 66 percent of injuries among children two years old or younger occurred in the home. The authors found that 35% of home injuries were head injuries, 17% of these were severe enough to require hospital admission. Two-thirds of all home injuries occurred in the living room or bedroom. Falls accounted for the greatest percentage of home injuries among 1 to 4-year olds (55%), followed by being struck with an object (18%), while cutting, piercing, and poisoning injuries accounted for 6%. A recent analysis of data gathered from 1990 to 1996 by the emergency department of British Columbia's Children's Hospital found that 49% of all injuries occurred in the home, and 18% occurred at school or in a public building. Forty-seven percent of injuries among boys occurred in the home, and approximately 19% occurred at school or in a public building. Fifty-one percent of injuries among girls occurred in the home, and 17% occurred at school or in a public building (Soubhi, Raina et al., 1999).

The influence of family characteristics on childhood injury seems to vary according to child age. Using cohorts of children from the longitudinal Louisville Twin Study, Matheny (1987) surveyed two groups of children for the occurrence of injuries. The children in the first group were monitored for the first three years of their lives (n = 96), while those in the second group were monitored from age 6 to 9 years (n = 76). Independent variables included parental temperament, home injury hazards, family functioning, and family socioeconomic status. In the younger cohort, there was a high risk of injury for children with irregular sleeping and eating habits, and noise and confusion in the home. A high risk was also observed for children whose mother was less active and less emotionally stable, and whose father was impulsive and less sociable. In the older cohort, parental characteristics were less likely to influence the occurrence of injury. Instead, child variables showed a stronger association with injury occurrence: active boys with irregular sleeping and eating habits sustained more injuries that required medical attention.

The high frequency of injury occurrence in the home is an indication of the importance of understanding family characteristics and how they influence childhood injury (Matheny, 1988). Matheny (1987) suggested that cohesive families, families governed by rules, families that stressed active involvement in family activities, and families that supported child autonomy had fewer childhood injuries. Schor (1987) suggested that patterns of frequent injury occurrence should be regarded as possible evidence of poor family functioning. However, few studies have considered childhood injury in combination with both parental and family characteristics (Matheny, 1987). Consequently, little is known about how family functioning relates to parenting and childhood injury (Aagran, Winn, Anderson, & Del Valle, 1998; Gable & Peterson, 1998; Gallagher, Hunter & Guyer, 1985; Hu, Wesson, & Kenney, 1993; Peterson & Saldana, 1996; Peterson & Stern, 1997).

1.2.1 Family Functioning, Maternal Health and Childhood Injury

Family dysfunction can influence child behaviour. A study by Campbell, March, Pierce, Ewing, and Szumowski (1991) highlights the potential roles that family dysfunction and negative maternal control may have in the occurrence of behavioural problems in young children. Campbell et al. (1991) compared preschool boys who were identified by their teachers as being active, inattentive, and impulsive (N = 42) with matched classroom controls (N = 43) and with parent-identified problem-boys (N = 27) on measures of family functioning. The teacher-identified problem-boys and the parent-identified problem-boys did not differ on measures of their family adversity. Both groups came from less well-functioning families than the comparison boys. In addition, mothers who were feeling more depressed and overwhelmed were likely to report more health and behaviour problems in their children (Campbell et al., 1991). In a compliance test, the mothers of the problem-boys were observed to be more negative and controlling toward their sons than the mothers of the comparison boys were to their sons. A follow-up on the problem-boys and their families a year later verified the predicted behavioural problems in the boys. The adverse behaviours developed in response to the presence of maternal depression and negative maternal control within the family unit. The authors concluded that maternal depression, family change and instability may place children at risk of behavioural problems because these stresses make parents less available to meet their young children's developmental needs (Campbell et al., 1991).

Impairments in family functioning are seldom due to only one factor. A combination of factors such as maternal depression, poverty, inadequate housing, and single parenthood can upset the balance of family functioning and, as a result, can hinder child development and increase the risk of childhood injury (Zayas, 1995). A study by Backett and Johnston (1997) showed that maternal illness and stress strongly influenced the rate of childhood injury occurrence. An example of the influence of parental depression on childhood injury can be shown with a study by Weissman et al. (1986). Using a sample from the Yale Family Study of Major Depression, Weissman et al. (1986) studied sixty-five couples and their children (N=153). One or both parents had been treated for major depression. A control group with comparable sociodemographic backgrounds consisted of 26 couples and their children (N=67). The data from both groups were collected through blinded interviews. The final sample of children included 105 boys and 115 girls, aged 6 to 23 years. The analyses showed that children from families with at least one depressed parent suffered more head injuries, adverse perinatal events, retarded development, convulsions, surgical operations, suicide attempts and depressions than children who had healthy parents (Weissman et al., 1986). The results of this study may be a reflection of stress and dysfunction in ill families. They may also reflect inadequate parenting skills among ill mothers, and may not be specific to maternal depression.

1.2.2 Parenting and Childhood Injury

Parenting skills involve the creation and application (via praise and punishment) of verbally based parental rules, which have only a gradual influence on child's behaviour over time (Kendall & Wilcox, 1979; Peterson, Mori, & Scissors, 1986; Zettle & Hayes, 1983). Research indicates that parents do not necessarily have a correct appreciation of their child's competence and ability to judge risks (Klein, 1980). Level of education is likely to influence the parent's perception of risk as well as parenting behaviours (Glik, Kronenfeld, & Jackson, 1993). Parents often overestimate their child's knowledge and ability to make decisions about safety (Dunne, Asher & Rivara, 1992; Yarmey & Rosenstein, 1988). For example, parents of children over two years of age tend to be less vigilant about preventing their children from exposure to hazards, even though these children are still at risk for making poor safety judgments. Furthermore, parents may confuse their child's verbal ability with cognitive development and expect the child to understand verbal commands to avoid risks (Christoffel, 1993). In general, the literature suggests that families with poor parenting skills are likely to have higher rates of injury occurrence among their children. Some studies suggest that parents may be less likely to supervise boys than girls when they are playing, resulting in boys reporting more injuries (for which an adult was not present) than girls (Block, 1983; Saegart & Hart, 1976).

Studies indicate that parenting can be disrupted by economic hardship and that parents of low SES are more apt to use harsh, authoritarian parenting practices (Simons, Johnson, Beaman, Conger, & Whitbeck, 1996). Focusing on low-income, inner-city families, Zayas (1995) reviewed the literature on the influence of culture on parenting behaviours towards young children. He also reviewed the impact of poverty, urban stresses, parental psychopathology, and family social support networks on parental behaviour and early childhood development. Zayas's review suggests that poverty is associated with negative parenting practices such as the use of physical punishment; the issuance of commands without explanation or consultation with the child; and the neglect to reward the child for desirable behaviours.

1.2.3 Family Socioeconomic Status and Childhood Injury

There is evidence that children living in low-income families are more likely to die from injury than children from higher income families are (Nersesian, et al., 1985). Alwash and McCarthy (1988) found that social disadvantage increased the risk of childhood injury in the home. Characteristics associated with low socioeconomic status (SES) such as single marital status, poor maternal health, inadequate education, and poverty have been linked to the occurrence of childhood injuries (Parker et al., 1991; Nersesian, et al., 1985). A review of studies associated with child-pedestrian injuries revealed that the mothers of injured children were more likely to be young, poorly educated, and single (Wazana, Krueger, Raina et al., 1997).

Poor maternal physical and mental health, low social class and marital discord have been identified as risk factors for injuries in young children. These family characteristics have also been determined to be risk factors for aggressiveness and hyperactivity in the child (Bijur, Stewart-Brown, & Butler, 1986). In a study by Bijur, Golding, Haslum, and Kurzon (1988), 10,394 children from a birth cohort were interviewed on their injury experiences at their fifth and tenth birthdays. Significant linear trends indicated higher levels of aggressiveness and hyperactivity in children of low-income families living in crowded and deteriorated housing. Similar results were observed for children whose mothers were distressed and unhappy, and for children whose families moved frequently. The association of aggressiveness and the occurrence of injuries remained even after control for socioeconomic factors. Hyperactivity continued to be significant in boys, but not girls. However, the results of this study do not apply to children from ethnic and racial minorities in Great Britain.

It is likely that SES and maternal perceptions of risk, stress, coping, and parenting behaviours influence the presence of hazards in a home. Safety practices such as knowing how to perform the Heimlich manoeuver, covering electrical outlets with safety plugs, and using car seats are more prevalent in high SES households where the parents are also more likely to be well-educated (Kramer, Allen, & Gergen, 1995). Low SES can also contribute to the occurrence of injuries through its influence on the cognitive development of the child. Results from a United States national cohort of 2,531 children of ages 6 to 16 years show that low SES is inversely related to a child's cognitive development. Thus, children from low-SES families are more likely to have difficulty with perceiving and managing injury hazards (Kramer, Allen, & Gergen, 1995).

In summary, research suggests that family can influence the occurrence of childhood injury, and that family's influence varies according to child's age. For toddlers, the pattern includes the influence of parental, home, and child temperament and behaviour, while among older children, child's behaviour seem to predominate. Risk factors for childhood injuries in the home include the developmental status of young children and their inability to perceive risk, impairments in family functioning and the quality of parenting. The family measures most associated with increased risk of injury appear to be indexes of poverty, social disadvantage, family stress, and family dysfunction. In turn, family dysfunction is rarely due to only one factor. Factors such as parental depression and ill health, single parenthood, poverty and inadequate housing upset the balance in family functioning and consequently may affect parenting behaviours, children's behaviour and increase the risk of injury.

1.3 Neighbourhood Characteristics and Childhood Injury

Research suggests that neighbourhood and community characteristics have important influences on child health and childhood injury (Bronfenbrenner, 1986; Diez-Roux, 1998; Jencks & Mayer, 1990; Kupersmidt, Griesler, DeRosier, Patterson & Davis, 1995; Malmstrom, Sundquist, & Johansson, 1999; Wazana, Kreuger, Raina et al., 1997). In general, studies show that children who live in disorganized environments or low socioeconomic neighbourhoods are at increased risk of injury (Jolly, Moller, & Volkmer, 1993; Matheny, 1986, 1987; Nersesian, et al., 1985; Valsiner & Lighfoot, 1987). Children living in low socioeconomic environments are more likely to die from a motor vehicle crash, from drowning or from fire (Dowswell et al., 1996; Nersesian, et al., 1985). Child pedestrian injuries for example, have been associated with living in communities characterized by household and neighbourhood crowding in which numerous families live below the poverty level (Durkin, Davidson, Kuhn, O'Connor, & Barlow, 1994; King & Palmissano, 1992; Rivara & Barber, 1985). Adverse environmental and demographic conditions of crowded urban neighbourhoods can interact with social and personal factors to increase the risk of pedestrian injuries among children from poor families (Bagley, 1992). Child pedestrian injuries have also been attributed to living in poor neighbourhoods with houses in close proximity to busy streets and restricted access to play space. Pless, Verreault, Aresenault, Frappier, and Stulginkas (1987) showed that in Montreal, children of all ages and of both genders from low-income areas had higher rates of traffic and pedestrian injuries compared to children from middle- and upper income areas of the city.

1.3.1 Neighbourhood Disadvantage and Childhood Injury

Reading, Langford, Haynes, and Lovett (1999) found an independent effect of socioeconomic deprivation at the area level, with much higher injury rates in deprived urban neighbourhoods than in affluent areas. A multi-level analysis showed that variations in the occurrence of childhood injury could be explained by factors such as the gender of the child, the age of the mother, and whether the child had older siblings. The characteristics of disadvantaged neighbourhoods had a small, but noticeable effect on the rate of childhood injury: a higher number of severe injuries were reported from the most disadvantaged neighbourhoods. However, socioeconomic differences at the neighbourhood level are not limited to the most severe types of injury, and the reasons for the associations between neighbourhood socioeconomic status and injury remain unclear (Reading et al., 1999). Studies suggest that not only the absolute amount of income is important for health, but also the relative disparity of income distribution in a population (Kaplan, Pamuk, Lynch, Cohen, & Balfour, 1996; Kennedy, Kawachi, Prothrow-Stith, 1996; Wilkinson, 1996). A study conducted by Lynch et al. (1998) showed that high-income inequality was associated with a higher frequency of injury occurrence and with all-cause mortality rates. In studies that relate neighbourhood socioeconomic status (a combination of level of income, education and occupation) to variations in injury rates, the level of income and the inequalities of income appear to be the most strongly related to injury. The main explanations focus on lack of material resources and inability to protect children from injury (Black, Morris, Smith, & Townsend, 1982).

A Canadian population study of pedestrians and bicyclists involved in motor vehicle collisions found higher rates of death among children from census tracts that had the highest percentage of low-income families (Dougherty, Pless & Wilkins, 1990). Dougherty, Pless and Wilkins (1990) examined motor vehicle traffic deaths and injuries to pedestrians and bicyclists aged birth to 14 years by their income quintile of residence. The injury rate of children living in the poorest neighbourhoods was four times greater than that of children living in the richer neighbourhoods. In census tracts of Northern Manhattan, Durkin et al. (1994) investigated the relationship between several socioeconomic factors and the occurrence of severe childhood injury. Their report included injury data from motor vehicle collisions, pedestrian injuries, falls, gunshot wounds, and burns. The census tract percentage of low-income households was the single most important predictor for the occurrence of all types of injuries. Children living in areas with predominantly low-income households were twice as likely to be injured from all causes than were children living in areas with few low-income households. Similarly, two studies conducted in Australia (Jolly, Moller, & Volkmer, 1993) and the United States (Nersesian, et al., 1985; Mierley & Baker, 1983) examined patterns of fatal and non-fatal injury among children and adolescents. These studies found significant relationships between injury rates and low-income neighbourhoods. Both of these studies showed that the risk of injury for low-income neighbourhoods was nearly three times greater than that of the highest income neighbourhoods.

Occupational structure, from blue collar to professional positions, can also explain variations in health and injury (Lantz, House, Lepkowski, et al., 1998; Lynch, et al., 1998; Durkin, et al., 1994; Emerick, Foster, Campbell, 1986; Beautrais, Fergusson, & Shannon, 1982). Communities with higher levels of occupations and smaller differences between occupational classes are likely to have more resources for promoting healthy lifestyles and lower levels of stressful and alienating social conditions (Sclar, 1980). Studies show large differences in mortality by occupational class for many causes of death (Hertzman, Frank, & Evans, 1994), and among children, unintentional injury shows the steepest gradient of social class disadvantage (Black, Morris, Smith, & Townsend, 1982).

Although empirical evidence is sparse, neighbourhood disadvantage may have an effect on childhood injuries through its impact on children's behaviour. Child pedestrian injuries significantly co-vary with child's behavioural problems such as over-activity, conduct disorder, and delinquency (Roberts, 1994; Roberts, Norton, & Jackson, 1995). In turn, delinquency reflects a variety of neighbourhood and ecological factors such as family poverty and dysfunction (Bagley, 1992). Limited available data suggest that the adverse effects of socioeconomic disadvantage on child behaviour are mostly attributable to family level measures (Boyle & Lipman, 1998). A study of 673 five- to six-year olds from the United States examined the influence of selected neighbourhood characteristics on a child's problem behaviour. The impact of neighbourhood on child I.Q. results and reading scores was high, yet it had very little impact on child problem behaviour (Chase-Lansdale & Gordon, 1996). In a representative sample of Canadian children aged 4 to 11 years, Boyle and Lipman (1998) tested a multi-level model of the interrelationships between neighbourhood, family, and child problem behaviour. This model could predict approximately 25 percent of the variance in child problem behaviour. Six to seven percent of the variance was due to differences between enumeration areas, and the remaining 18 percent were due to family socioeconomic status. However, in Boyle and Lipman's study, the nature of the influence of neighbourhood was not entirely clear as there were no measures of the social influence of neighbourhood such as level of neighbourhood cohesion or prevalence of neighbourhood problems.

1.3.2 Social Influence of Neighbourhood on Childhood Injury

The concept of social influence on health and injury includes the influence of social factors such as social cohesion and community stress. Either jointly or independently, these factors can influence injury rates among children and youth (Corin, 1994). The term "community stress" refers to the proportion of the population that is considered at social risk and includes people on social assistance, unemployment insurance, or in shelters for the homeless or abused (FCM, 1998). Evidence suggests that people who live in communities that have high proportions of individuals who are at social risk, have more stressful and less healthy lives (Rosengren, Orth-Gomer, Wedel, & Wilhelmsen, 1993; Rubin, 1976). Social cohesion on the other hand is defined as a sense of social unity and cooperation among community members built on egalitarian standards, and aimed at promoting some common good (Wilkinson, 1996).

A number of population-based studies have examined social cohesion as a factor that is related to a population's health (Fullilove, 1998; Kawachi, Kennedy, Lochner, & Prothrow-Stith, 1997; Putnam, 1994; Wilkinson, 1996). Putnam (1994) demonstrated a high correlation between a stronger social cohesion, a lower infant death rate, and a longer life expectancy in females. Putnam (1994) also noted a highly significant correlation of a narrower income distribution with an index including the percentage of voters in referenda, newspaper readership, and number of associations for voluntary activities per capita. However, few studies have examined the relationship between the occurrence of injuries and social cohesion. Sampson, Raudenbush, & Earls (1997) conducted a survey on injury occurrence in 343 neighbourhoods in Illinois. The authors tested the hypothesis that cohesion among neighbours, combined with their willingness to intervene on behalf of the common good, is linked to reduced neighbourhood violence. Neighbourhood cohesion yielded high between-neighbourhood reliability and was negatively associated with violence.

In general, the translation of social and neighbourhood factors into processes that can explain the occurrence of injuries is complex. This translation may involve mechanisms that may act differently for various types of injuries and have different effects at distinct stages in life (Goodman, 1999). For example, studies suggest that neighbourhood conditions may have a stronger impact on adolescent behaviour than on child behaviour. This may be due to increased independence during adolescence accompanied with the lessening of family influence and the ascendancy of peer influences (Boyle & Lipman, 1998; Williams, Currie, Wright, Elton, & Beattie, 1996). Another mechanism is family disruption such as separation or divorce, which lead to single parenthood. Such disruptions increase a child's anxiety level, which in turn increases the occurrence of injuries (Manheimer and Melinger, 1997). Children in single parent families are generally disadvantaged with respect to housing and income as compared to children whose parents are both present. Platt and Pharoah (1996) conducted a review of statistics on child health in the UK. In 1993, 23 percent of all families with dependent children were lone parents, and 21 percent of them were lone mothers. Lone parent families with dependent children were more likely to live in overcrowded, rented accommodation without central heating. Only 53 percent of lone mothers with children over five years old were employed, while 74 percent of married mothers were employed. In the United States, neighbourhoods with a higher than average proportion of single parents, disadvantaged ethnic minorities, household crowding, and low income have significantly higher rates of injury involving young pedestrians (Rivara & Barber, 1985).

In summary, children who live in disorganized environments and low-income neighbourhoods are at increased risk for injury. There is an increasing social disadvantage gradient in mortality for childhood injuries, but socioeconomic differences at the neighbourhood level are not limited to the most severe forms of injury. Neighbourhood disadvantage may have an effect on childhood injuries through its impact on the behaviour of the child. However, studies suggest that neighbourhood conditions may have a stronger impact on adolescent behaviour than on child behaviour, and the adverse effects of socioeconomic disadvantage on child behaviour seem mostly attributable to family level measures. Family disruption that leads to single parenthood may also be associated with increased risk of childhood injuries. Children in lone parent families are generally disadvantaged with respect to housing and income as compared with children from families with two parents. Neighbourhoods with a higher than average proportion of single parents tend to have significantly higher rates of childhood injuries.

1.4 Critical Remarks on the Literature

In the studies that we have reviewed, the consistency of the results seems to vary according to the type of injuries considered and to the type of research design, i.e. prospective versus cross-sectional. For example, in cross-sectional studies, hyperactivity in children appears to be related to injury while in prospective studies, hyperactivity does not predict the occurrence of injury. In contrast, aggressive behaviour remains a strong correlate of injury in both types of designs. Other studies show that aggressive behaviour is consistently related to the occurrence of injuries in general, but not to pedestrian injuries.

Study results also seem to vary according to the source of measurement. Researchers often report that there is an association between the frequency of childhood injury occurrence and maternal reports of child activity levels. However, in studies where the teachers report the activity levels of the children, the association does not remain significant. Finally, not all studies account for socioeconomic factors, and some studies cannot be generalized because they were conducted with small samples or in experimental settings (e.g. observation in simulated settings).

Few studies on the occurrence of injuries have included measures of social influence of neighbourhood such as level of neighbourhood problems or cohesion. Consequently, the translation of social and neighbourhood factors into processes that can explain the occurrence of childhood injury remains complex and poorly documented. The literature suggests an evolving pattern of influence of family and neighbourhood on injury that varies according to child age. Furthermore, multiple mechanisms may act differently for various types of injury outcomes and have different effects at distinct stages during childhood. Few studies have considered these environmental influences across different age groups that cover the span of childhood.

The literature we have reviewed suggests that a complex model of relationships between child, family and neighbourhood factors may account for the variations in the occurrence of childhood injuries. In general, the factors that seem to be most associated with increased risk of childhood injury include social disadvantage at the neighbourhood or family level, parental illness and depression, and family stress and dysfunction. The neighbourhood and the social environments of the family can interact in various ways to increase (or decrease) the impact of stressful events on the health of family members, disrupt (or reinforce) family functioning and positive parenting and increase (or decrease) the occurrence of childhood injuries. Family functioning can affect the risk of injuries either directly or through child behaviour or parenting. In turn parenting can influence childhood injuries through verbal rules, which favor the development of safe behaviours in children. The literature also suggests that parenting can be disrupted by economic hardship. Low SES families tend to use fewer positive interactions with their children and more harsh and punitive parenting practices. Family dysfunction and economic hardship can also affect maternal health and depression, which in turn can affect parenting and/or child behaviour and ultimately increase the risk of injuries. It is also possible that family SES modifies the effect of parenting and/or child behaviour on the occurrence of injuries. At yet another level, neighbourhood disadvantage can affect childhood injuries either directly or through its influence on child behaviour. Furthermore, the influence of neighbourhood can also be modified by family SES or by social cohesion. In turn social cohesion can modify the effect of child problem behaviour and reduce the risk of injuries. Thus, there are a number of plausible pathways of influence of neighbourhood and family factors on childhood injury. However, the value of a model is not necessarily its complexity or completeness, but whether it suggests testable hypotheses for major factors that influence injuries and can be modified to prevent them or reduce their severity (Robertson, 1998). The next section presents the specific research questions that were addressed in the present study.

Contents

Effects of Neighbourhood, Family, and Child Behaviour on Childhood Injury in Canada - March 2001