Injury is the leading cause of death in children in most developed countries.1 2 Boys are known to have a higher rate of injury incidence and mortality than girls.3^–9 The gender difference in childhood injury increases with age3 7 8 and is already noticeable in infancy for some types of injury, for example, unintentional head injury.4 9 10 Boys experience injury with both greater frequency and severity than girls.4 9 These gender differences may have multiple explanations and may reflect more than varying exposure11; physical and cognitive development, spatial abilities and motor coordination may be important.12 13 Boys who incur injuries tend to display higher activity levels5 14 and appear to take more risks than girls.15 Injury-prone behaviour by boys is reinforced by expectations of the male role from parents16 and society, and promoted through the media and advertising.17 Conversely, children perceive girls to have a higher risk of injury.18

There appears to be a large potential scope for reducing boys’ injury mortality rates. A study of the Organisation for Economic Cooperation and Development countries found that if boys had the same rate as girls, 5000 lives could be saved every year.8 The same study reported wide variation in m:f injury mortality ratios between different countries ranging from 1.34:1 in Sweden to 2.3:1 in Ireland.

Although some researchers have investigated the reasons and associated risks behind gender differences in injury incidence and mortality,15 19^–22 gender variation in injury mortality risk over time has seldom been explored. This study aimed to investigate the nature and consistency of gender variation in child injury mortality in Scotland over a 25-year period.

METHODS

Injury (external cause) mortality data for children aged 0–14 years, resident in Scotland at the time of death, were obtained from the General Register Office for Scotland (GROS). The study period was 1982–2006 inclusive. The data were selected using primary cause of death codes (E800-999) from the Ninth Edition of the International Classification of Diseases (ICD-9) for the period 1982–1999, and codes V01-Y98 from the Tenth Edition (ICD-10) for 2000–2006. These include deaths from all forms of injury and poisoning, whether intentional or unintentional. In 2000, the coding of deaths in Scotland changed from ICD-9 to ICD-10. A bridge coding exercise reported high overall and gender compatibility in cause of death coding between ICD-9 and ICD-10 for the major injury classifications used in this study.23

The records obtained were anonymous and contained the variables: age, gender, year, and cause of death. Annual population estimates were obtained from GROS and used as the denominator in the calculation of mortality rates (per 100 000 person years). Mortality data were divided into age groups (0, 1–4, 5–9 and 10–14 years), aggregated into 5-year periods and analysed in terms of mortality rates and m:f ratios (with 95% confidence intervals where appropriate). The proportional changes in the mean rates between the first 5 years and the last 5 years of the study period were also calculated. To determine the nature and statistical significance of trends, methods of simple linear regression were employed using data for each of the 25 years individually.

RESULTS

Gender differences, all ages (0–14 years)

Over the 25-year study period, there were 2140 child injury fatalities, of which 1371 (64%) were males. The m:f ratio of all-cause injury mortality was 1.7, a significant male excess (males 11.29 and females 6.64 per 100 000 person-years) (table 1). The 5-yearly m:f ratio was relatively stable over time, ranging from 1.48 to 1.87. All-cause injury mortality rates declined in both genders over the study period but in males fell especially sharply with a trend gradient of −0.68 (p<0.001), almost double that of females (−0.36, p<0.001) (table 2). The proportional reductions between the first and last 5-year time periods were slightly higher in males than females (−74% vs −67%).

Gender differences by age

Over the whole study period, both male and female injury mortality rates were highest in infancy (age 0) and lowest in 5–9 year olds (table 1). In all age groups except infancy, males were significantly more likely to die from injuries than females (table 1). In infancy, both male and female injury fatality rates fell markedly over the study period (from 35.2 to 7.3 and 26.3 to 6.9 per 100 000 population, respectively), but remained higher than other age groups (table 1).

Over time, there was a significant decline in all-cause injury mortality rates in both males and females in each age group but the declines were steeper in males (table 3). This was particularly noticeable in the 5–9 years age group where the trend gradient in males was almost three times that of females and in the 10–14 years age group where it was more than double. In each age group the proportional decline in mortality rates was greater in males than in females (table 3) so that by the end of the study period, the male excess had disappeared in the two youngest age groups and narrowed substantially in the oldest two (table 1). Only the 10–14 years age group showed a significant male excess of deaths from injury at the end of the study period.

Gender differences by cause

Some causes of injury death were more strongly patterned by gender than others (tables 2 and 4). Fatalities due to fires showed a weak male excess with only one of the five time periods being significant. By contrast, fatal drowning and suffocation injuries (including choking/asphyxia) showed a male excess in all five time periods (mean m:f ratio 2.09, table 2) but this became non-significant in the latter two. Although rates were small, poisoning (including drug deaths) was the most highly gender related of all causes with an overall m:f ratio of 3.2. This was most pronounced in the 1–4 years age group (m:f ratio 8.6) but was also significant in the 10–14 years age group (m:f ratio 3.1) (table 4).

In the case of both types of road traffic injuries (non-pedestrian and pedestrian), males had significantly higher mortality rates than females over the 25-year period (table 2). Older boys (5–9 years and 10–14 years), were significantly more likely to be fatally injured than girls (table 4).

Male and female mortality rates for non-pedestrian road traffic injuries decreased steadily over time (table 2). The reduction was greater in females than in males (overall −78% vs −65%) so that the significant male excess observed at the start of the study period, whilst remaining relatively stable over the first three periods (m:f ratio 1.5), increased over the latter two (m:f ratio 2.0 in 1997–2001, 2.4 in 2002–2006).

The male pedestrian fatality rate fell sharply over the study period (table 2), by the end of which males and females had similar risks.

Mortality rates for falls in females fluctuated over time although in males there was a significant linear decline between 1982 and 2006 (p = 0.001) (table 2). Fatal falls were few but were significantly more common in males at the start, middle and end of the study period (table 2); over the study period as a whole; and significantly more likely in males aged 10–14 years (table 4).

Fire was less gender related than other injury causes. There was a significant decline in both male and female rates between the start and end periods of study, and the relative difference between the genders fell from a significant male excess in 1997–2001 to a (non-significant) female excess at the end (table 2).

Drowning and suffocation were divided into two categories to explore more closely two different causes of death (table 4). Infants, particularly males, were the main victims of suffocation with very high 25-year mortality rates (males 7.9 and females 4.8 per 100 000 person-years,), giving a significant m:f ratio of 1.66. However, fatality rates among infants fell strikingly over time in both genders (90% in males, and 93% in females). Males aged 10–14 years were also significantly more likely to die from suffocation than their female counterparts although mortality rates were very low (table 4) and remained largely stable over time.

In the case of drowning, all age categories (except infants, where numbers were very small) had a significantly higher gender ratio and the difference increased with age (from m:f ratio 1.85 for age group 1–4 years to 4.66 in age group 10–14 years; table 4). While female drowning rates decreased with age, male rates remained similar, hence the increasing gender gap.

Compared with females, males (m:f ratio 1.35), particularly those aged 10–14 years (m:f ratio 1.58), were significantly more likely to die from intentional injuries (table 4). When considering suicide separately, males aged 10–14 years (the only age group with suicides) were again significantly more likely to die than females (m:f ratio 2.19). Intentional injuries were the only causal categories where higher mortality rates were observed at the end of the study period than at the start (table 2). These trends were not statistically significant, however, and the number of suicides was very small. The only period with a significant gender gap was 1992–1996 when the number of male suicides increased substantially (from three fatalities in each of the previous two time periods, to 17, mortality rate 0.69 per 100 000 population); at the same time, the female rate fell to its lowest (0.04). This resulted in a significant m:f ratio of 16.3 although the confidence interval was wide.

Infant males were not significantly more likely to die of any injury cause compared with infant females other than suffocation, but some gender ratios were of borderline significance, including intentional injuries and falls (table 4). The number of fatalities from falls was, however, small.

DISCUSSION

Some epidemiological studies have specifically investigated the gender patterning of childhood injuries,19^–21 although few have focused on time trends. Our study sought to help remedy that deficiency using an epidemiologically robust dataset.

We found a considerable gender variation in child injury mortality in Scotland, with a male excess that is just discernible in infancy and becoming more pronounced with increasing age. The gender pattern is, however, complex and changing over time to the point where the previous male excess has disappeared in some age and cause categories. This is a departure from the conventional view of a consistent male excess of injury risk across causes and time, but concurs with a recent Swedish study which also found a narrowing gender gap.6 Nevertheless, in our study, if the injury fatality rate of boys in 2002–2006 had been equal to that of girls, about 36 boys’ lives would have been saved in that period.

Males experienced a steeper reduction in injury mortality than females; this could be due to their higher rates at the start of the study period. The proportional reduction between 1982–1986 and 2002–2006 was, however, similar (about 70%), suggesting that injury prevention efforts may have impacted equitably. The continuing male excess in all-cause injury mortality may, therefore, be no more than a historical legacy which, following recent trends, will disappear in the near future. The overall trend of decreasing excess injury mortality risk to males, however, obscures a fluctuating or increasing gender variation in some individual cause categories.

Poisoning fatalities were highly gender patterned in our study but the rates were low. A French study of emergency admissions found that boys aged under 4 years were more likely to suffer acute poisoning, while girls were at highest risk aged between 12 and 15 years.24 By contrast, although the gender gap was widest in 1–4 year olds (with a significant m:f ratio of 8.6), our study revealed a significant male excess in 10–14 year olds and that both male and female rates peaked in this age group. These contradictory findings may reflect differences between morbidity and mortality; alternatively, there may be real differences in exposure or behaviour between the study populations.

Our data indicate that drowning was more common in boys with an m:f ratio of almost 3:1. A previous study found that three-quarters of children who drowned were boys25 but another, including near-drowning, showed that boys were only slightly more at risk.26 Male drowning rates have been shown to peak at age 1–2 years then decline to age 10 years before increasing during the late teens; female rates peak by 1 year and then decline to a steady low rate.3 Our results showed a corresponding steady decline in female drowning rates from infancy but that male rates peaked in the 5–9 years age group resulting in an increasing gender gap with age. This could reflect a greater tendency for boys to wander27 and explore their environment, and to variable parental supervision between boys and girls.28 29

Excessive male involvement in road traffic accidents has long been recognised, but previous studies have either concentrated on gender differences in adults3 30 31 or reported them incidentally.12 32 Our findings concur with other studies suggesting that the rate of pedestrian injury for boys is about twice that of girls.12 32 Analysis over time, however, revealed a consistent downward trend in pedestrian injury mortality in both genders resulting in equal risks by 2002–2006. Declining pedestrian injury fatalities may be partly due to the implementation of road safety measures and the provision of safer play areas, but changes in levels of exposure through increasing car use33 and more leisure activity taking place indoors, may have provided relatively more protection to male pedestrians thereby contributing to the disappearing gender gap.

Our study revealed significant improvements in mortality rates from non-pedestrian road traffic injuries in both genders over time, perhaps due to improvements in road safety and vehicle design. This was, however, accompanied by an increasing and unexplained gender disparity. The excess male risk in the oldest two age groups could be a result of variation in the use of restraints between boys and girls. One study reported that inappropriate restraint was associated with the perceived size of the child by parents and that only 16.5% of children who required a booster cushion were properly restrained.34

Falls are the most common cause of childhood injury,9 35 but account for only a small proportion of injury deaths.8 A significant gender difference in unintentional head trauma through falls has been observed even in infants aged younger than 3 months, which may reflect differing parenting practices for male and female infants.10 Although our study revealed a significant difference between male and female fatal falls only in 10–14 year olds, there were more male than female fatalities in each age group.

Fire fatalities were not gender related in our study, perhaps because most fatalities are due to residential fires,3 36 many of which occur at night or early morning when most residents are asleep3 37 and children of both genders are at equal risk.3

No significant changes in intentional injury mortality rates were observed over the study period. Nevertheless, we found a significant male excess of deaths from intentional injuries (all causes), and suicide in particular. Although numbers were small, the proportionately large increase in male suicides in 1992–1996 is likely to have influenced these results. Mortality data for intentional injuries in children, however, can be subject to misclassification.38^–41 Although one study reported slightly higher ascertainment of maltreatment deaths in girls in the USA,38 we found no evidence to suggest that misclassification of intentional injuries occurs disproportionately by gender in the UK. Furthermore, underascertainment in boys would mean that the gender ratios observed in our study were underestimates.

To minimise misclassification bias, in 10–14 year olds we analysed suicides together with deaths from injuries of undetermined intent but it is still possible that some suicides in this age group were misreported as deaths from unintentional injuries.42 We do not know whether any misclassification or misreporting in suicide deaths varied by gender. The number of suicides in our study was small and boys were significantly more likely to die than girls from unintentional injuries of the type that might have been misreported suicides (eg, poisoning, drowning, suffocation). We conclude that misclassification or misreporting may have impacted to a small extent on overall mortality rates, but that the gender disparities in intentional injuries and suicide are likely to be real.

In conclusion, we have demonstrated a striking variation in the gender patterning of child injury mortality in Scotland over time. Moreover, we have shown that the gender gap in deaths from all-cause injury in children aged 0–14 years has steadily declined over a 25-year period to the point where the male excess might disappear within the next few years. A similar pattern was evident in some specific age and cause categories where gender differences were either converging or had disappeared entirely. These findings are largely unexplained. Future research should focus on the generation and testing of specific hypotheses regarding the varying gender pattern in mortality risk from specific causes of injury. Meanwhile, policy makers should resist pressures to target preventive measures exclusively or predominantly at boys without carefully reviewing the gender patterning of injury risk in their populations.