Early Risk Behaviors and Adolescent Injury in 25 European and North American Countries

Abstract

Injury is a leading cause of mortality and morbidity among adolescents in developed countries. Jessor and Jessor’s Problem Behavior Theory suggests an association between risk behaviors (e.g., smoking, drunkenness, cannabis use, and sexual intercourse) and adolescent injury. The present study examined whether early engagement in risk behaviors would predict injury at age 15. It also examined whether such associations were consistent in strength across countries. Based on the data from the 2005-2006 Health Behaviour in School-aged Children (HBSC) survey, a multigroup logistic regression analysis was conducted. Our findings demonstrate a cross-national consistent association (with relative odds of injury rising to 1.85; 95% CI: 1.70-2.02). Based on the study findings, early engagement in risk behaviors was considered a marker for a trajectory that places adolescents at higher risk for physical injury, independent of their national context.

Keywords

early substance use early sexual debut injury cross-national analysis

Introduction

In developed countries, injury is a leading cause of mortality and morbidity among young people (Chiolero & Schmid, 2000; Currie, Nic Gabhainn et al., 2008; Danseco, Miller, & Spicer, 2000; Lescohier & Scavo Gallagher, 1996). In addition to negative physical and psychological effects on individuals, pediatric injury exacts a significant toll on society in terms of direct and indirect costs associated with treatment, rehabilitation, and long-term care, as well as loss of income and productivity due to disability (Krug, Sharma, & Lozano, 2000). The etiology of adolescent injury is, therefore, an obvious health research priority.

Researchers from different countries have related substance use with higher risk of adolescent injury (Jessor, 1998; Koven, McColl, Ellis, & Pickett, 2005; Pickett et al., 2006; Pickett, Garner, Boyce, & King, 2002; Pickett et al., 2005; Pickett, Schmid et al., 2002). Whereas some studies have suggested that substance use (e.g., alcohol misuse) is a direct cause of injury (Pickett et al., 2005), another plausible theory is that substance use is related to injuries in an indirect manner. Jessor and Jessor (1977), for example, proposed a theoretical framework, the problem behavior theory, which stated that a large variety of risk behaviors in adolescence are interrelated as they are all expressions of an underlying propensity for problem behavior. Whether or not a youth has this propensity depends on various factors, including the adolescent’s personality and his or her environment (e.g., peer influence, parental support, school system, neighborhood). Youth with a propensity for problem behavior are more likely to adhere to a high-risk lifestyle, which fosters both engagement in risk behaviors and the experience of adverse health outcomes such as injuries.

Sexual intercourse is rarely examined as a risk behavior in etiological studies of adolescent injury, as it is obviously not a direct risk factor. However, early sexual intercourse has a prominent place in problem behavior theory. Moreover, recent studies (Bellis et al., 2008; Willoughby, Chalmers, & Busseri, 2004) demonstrated that early sexual activity is part of a special cluster within problem behavior theory, which also includes alcohol misuse, smoking tobacco, and illicit drug use. This cluster may be perceived as an expression of the underlying propensity of problem behavior, which is, according to the theory, predictive of adolescent injury.

Whereas Jessor and Jessor’s (1977) theoretical framework focused on risk behaviors of adolescents in general, the present work focused on risk behavior specifically in early adolescence. This choice was based on the assumption that risk behaviors are not, as a rule, problem behaviors. From a developmental perspective, adopting adult-like behaviors is part of the normal growing-up process (Moffitt, 2006). Since most adults use or have used substances and are sexually active, it is not surprising that young people would adopt these behaviors (Luijpers, 2000). For them, such behaviors might fulfill important social functions and could be an essential aspect of psychosocial development (Moffitt, 2006). It is, therefore, important not to problematize risk behaviors. For early adolescents, however, engagement in smoking, drinking, cannabis use, and sex is deviant; most early adolescents are not (yet) interested in these kinds of behaviors. The behaviors, therefore, do not fulfill the same social function they do for older youth. Indeed, previous research has shown that early onset of risk behaviors can be considered problematic. A famous study by Moffitt (1993) demonstrated that youth with an early onset of risk behaviors were more likely to become involved in problematic levels of substance use and engage in risky and unsafe sexual activities later in life compared to youth with a later onset of risk behaviors (Moffitt, 1993, 2006). Although for late onset youth engagement in risk behaviors during adolescence was perceived as a temporary phenomenon, it was considered the beginning of lifelong engagement in risk and other unhealthy behaviors for early onset youth. A variety of other studies, in different countries, confirmed that early onset risk behaviors are a marker for later adherence to a high-risk lifestyle (Basen, Edmundson, & Parcel, 1996; Bellis et al., 2008; DuRant, Smith, Kreiter, & Krowchuk, 1999; Gullone, Moore, Moss, & Boyd, 2000; Petridou et al., 1997).

To the authors’ knowledge, no study to date has investigated the relationship between early onset of risk behaviors and later injury in a cross-national sample. As policies and attitudes toward risk behaviors, as well as prevalence rates of early risk behaviors and injuries, differ considerably cross-nationally, identification of a cross-national consistent association would be quite profound. Furthermore, there is neither scientific evidence nor a logical rationale to assume that early risk behaviors in some countries would be more strongly related to the occurrence of injury than in others. To date, only two studies have conducted cross-national analyses on the association between adolescent risk behavior and injuries. However, neither tested whether associations were consistent across countries. Moreover, these studies focused on concurrent risk behaviors (Pickett et al., 2005) or only one risk behavior, rather than the clustered risk behavior concept (i.e., alcohol misuse; Borges et al., 2006).

The present study investigated the association between injury and early engagement in alcohol, tobacco, and cannabis use and sexual intercourse in 25 European and North American countries. We first described the prevalence of medically attended injuries and early risk taking in each country, then evaluated the association between injury at age 15 and the number of early risk behaviors youth had engaged in. We hypothesized that the more early risk behaviors youth had engaged in, the higher their likelihood of being injured at age 15. On the basis of our assumption, that early risk taking might be a phenomenon with the same meaning in different populations of early adolescents, we expected our findings to be similar in strength across different countries and cultures.

Method

This study was based on international records from the 2005-2006 Health Behavior in School-aged Children (HBSC) survey. The HBSC study is a World Health Organization (WHO) collaborative cross-national study on health and related behaviors and the social contexts of young people’s health (Currie, Nic Gabhainn et al., 2008). Students aged 11, 13, and 15 years were surveyed across Europe and North America. In the present study, we used reports from 15-year-old students. Data were collected using a questionnaire that was completed in the classroom. Participating countries and regions obtained institutional ethics approval.

Countries

In 2005-2006, 38 countries participated in the HBSC study. Countries were excluded from these analyses if they had not asked students about the age of onset of at least one of the four risk behaviors of interest (seven countries: Ireland, Norway, Poland, Spain, Sweden, Turkey, and the United States) or if they had at least 50% missing values on variables that were included in the analyses (six countries: Bulgaria, Greece, Greenland, Malta, Israel, and Macedonia). Included countries were Austria, Belgium, Canada, Croatia, the Czech Republic, Denmark, Estonia, Finland, France, Germany, Iceland, Luxembourg, the Netherlands, the Russian Federation, Switzerland, Ukraine, and the United Kingdom (England, Scotland, Wales). In addition, eight countries (Hungary, Italy, Latvia, Lithuania, Portugal, Romania, Slovakia, and Slovenia), which had very few students who reported engagement in all four early risk behaviors, were included in the analysis but with a limited range of early risk indicators (at most three). National sample sizes for 15-year-olds ranged from 1,191 to 5,006.

Sampling

The HBSC sampling methods are described extensively in the international protocol (see Currie, Samdal, Boyce, & Smith, 2001; Roberts et al., 2009). Cluster sampling was conducted in accordance with the structure of national education systems within each country. Sampling was stratified by region or school type, as appropriate. The primary sampling unit was the school class, or the whole school where a sample frame of classes was not available.

Measures

Injury

Occurrence of injury was measured via a single question that asked how often students experienced an injury event in the past 12 months that required medical attention from a doctor or a nurse. The question was preceded by the following statement: “Many young people get hurt or injured from activities such as playing sports or fighting with others at different places such as on the street or at home. Injuries can include being poisoned or burned. Injury does not include illnesses such as Measles or the Flu.” Response options were: I was not injured in the past 12 months, one time, two times, three times, four or more times. The outcome was dichotomized into not injured or injured at least once as per existing precedents (Pickett, Garner et al., 2002; Pickett, Schmid et al., 2002).

Early risk behaviors

Participants were asked to report if and at what age they first engaged in (a) smoking a cigarette, (b) being drunk, and (c) having sexual intercourse. Although drunkenness is a subjective term, we focused on drunkenness rather than the age of the participants’ first time drinking alcohol because drunkenness indicates alcohol misuse. Participants were also asked whether they had ever used cannabis. Of note, asking youth to report the age at which they first engaged in cannabis use was considered a sensitive topic in many countries; therefore, it was not included as a question in the mandatory international questionnaire. Early initiation was defined as having smoked a cigarette before age 13, having been drunk before age 14, having had sexual intercourse before age 15, and having used cannabis by age 15. We selected these cutoffs based on literature-based precedents (see Baumeister & Tossmann, 2005; Currie, Nic Gabhainn et al., 2008; Zimmer & Helfand, 2008) and on the average age of onset of these behaviors in the 25 countries (see Table 1). The average age of onset per risk behavior across countries provided an indication of the age of onset that would be considered early for each risk behavior. Finally, a summary score was created by adding the four dummy variables for early engagement in the risk behaviors (values ranged from 0 to 4).

Table 1.

Ages of Onset of Risk Behaviors Among 47,106 Children in 25 Countries^a

		Smoking		Drunkenness		Sexual intercourse
Country	N	M	SD	M	SD	M	SD
Austria	1,495	12.58	1.28	13.55	1.02	13.83	1.03
Belgium	3,030	13.06	1.29	13.92	1.06	14.38	.87
Canada	2,299	13.21	1.32	13.84	1.08	14.09	1.02
Croatia	1,630	13.20	1.42	13.86	1.16	14.18	1.17
Czech Rep.	1,666	12.65	1.30	13.71	1.13	14.50	.75
Denmark	1,556	13.38	1.25	13.92	0.96	13.28	.92
Estonia	1,589	12.43	1.37	13.61	1.24	14.41	.86
Finland	1,686	13.06	1.34	13.72	1.08	14.27	.91
France	2,223	13.10	1.32	14.06	0.98	14.15	.98
Germany	2,554	12.84	1.26	13.97	0.95	14.06	.97
Hungary	1,191	13.09	1.27	14.07	0.96	14.40	.83
Iceland	1,886	13.69	1.27	14.21	0.97	14.21	.91
Italy	1,340	13.66	1.23	14.45	0.87	14.31	.97
Latvia	1,330	12.84	1.44	13.83	1.20	14.55	.83
Lithuania	1,861	13.03	1.43	13.89	1.14	14.37	1.05
Luxembourg	1,513	12.93	1.24	13.93	1.08	13.90	1.15
Netherlands	1,365	12.86	1.28	13.88	1.04	14.16	.94
Portugal	1,383	13.15	1.32	14.07	1.08	14.08	1.10
Romania	1,605	13.51	1.45	14.01	1.29	14.29	1.07
Russian Fed.	2,754	13.05	1.42	13.82	1.25	14.38	.96
Slovakia	1,252	12.89	1.32	13.76	1.09	14.29	1.03
Slovenia	1,561	13.25	1.34	13.99	1.05	14.45	.84
Switzerland	1,502	12.98	1.30	13.82	1.02	14.16	.94
Ukraine	1,829	13.14	1.49	14.17	1.12	14.35	1.03
United Kingdom	5,006	13.12	1.30	13.85	1.09	14.23	.97
Mean	1,884	13.10	1.33	13.80	1.08	14.20	.96

Note: Cannabis was excluded from this table as no data on exact age of onset for cannabis use were available. Means and standard deviations are pooled estimates of the six imputed files (Schafer, 1997).

Means are restricted to those youth who reported engagement in the behaviors.

Covariates

Additional variables included in the HBSC database and selected as potential confounders were gender, physical exercise frequency, and socioeconomic status as measured by the Family Affluence Scale (FAS; Currie, Molcho et al., 2008; Williams, Currie, Wright, Elton, & Beattie, 1997). Existing studies have demonstrated that boys are more likely to be injured than girls (Koven et al., 2005) and that high family affluence is related to an increased risk for injury, mainly via participation in organized sports (Mazur, Scheidt, Overpeck, Harel, & Molcho, 2001; Pickett, Garner et al., 2002). Although physical activity is part of a healthy lifestyle, it also increases the risk for injury through an increased exposure to injury-causing activities (King, Pickett, & King, 1998; Riley et al., 1996).

Imputation of Missing Values

As our sample was very large and the number of missing values was rather small (2.9% to 37.5% across countries; mean 25.9%) missing values were imputed using multiple imputation in Amos (version 17: Arbuckle, 2008). The Bayesian imputation method was selected, which is considered an appropriate approach with large samples (Schafer, 1997). With multiple imputation, missing values are replaced by several simulated values. In this study, six simulated values were used, which resulted in six complete data sets. Based on the results of these six analyses, pooled model parameter estimates were computed following Schafer’s guidelines.

Imputed values for missing data are generated within a so-called imputation model that contains all the variables from the final research model. When possible, variables that are assumed to be related to missingness are added to the model. In the present study, several variables (i.e., gender, academic achievement, self-rated health, and involvement in physical fights) were added to the imputation model for this purpose. To ensure an adequate model fit as well as ensure that the data were consistent with observed values, the imputation model was defined as a saturated multivariate regression model with risk behaviors as dependent variables and the other variables as independent variables. Furthermore, this model was applied to the original data set of each country. Subsequently, the data sets of the countries were merged to allow for multigroup analyses across countries.

Statistical Analyses

Descriptive statistics from each country were presented to estimate the average age of onset of each of the four early risk behaviors (based on youth reporting the behaviors). Next, percentages of children who reported early risk behaviors were presented, as were country-specific rates of injury.

Subsequently, we performed a series of logistic regression analyses to predict injury at age 15 based on the additive risk behavior score. We also analyzed a multigroup model with country as a grouping variable, using the software package Mplus (version 4.0: Muthén & Muthén, 2006). We included all 25 countries and examined the number of early risk behaviors (0, 1, 2, and 3 risk behaviors) as a predictor of injury. To test whether associations were similar across countries, the model was analyzed initially with all regression paths constrained. We subsequently allowed regression paths to vary across countries and compared the fit of the two models. Because the sample size was very large and as the chi-square statistic is sensitive to sample size, we focused on the Tucker–Lewis Index (TLI) and Root Mean Square Error of Approximation (RMSEA) values as evidence of model fit (Chen, 2007). The TLI is related to the total variance accounted for in the model and corrects for model complexity; values larger than 0.90 are desired (Kline, 2010). In addition, the RMSEA is related to the residual variance; values less than 0.05 indicate a good fit; (Kline, 2010). Comparison of fit for the current analysis was based on Chen’s guidelines (i.e., the fit of the two models differs significantly if ΔTLI > 0.010 and ΔRMSEA > 0.005). If the fit of the freely estimated model was significantly better than the fit of the constrained model, then the freely estimated model was preferred, which would indicate that associations between early risk behaviors and injury differed in strength across countries. If this was not the case, then the constrained model was preferred, which would indicate that the associations were similar across countries.

Next, we performed a multigroup analysis with those countries (N = 17) that had sufficient students who reported engagement in four early risk behaviors, following the same procedures as described earlier. On the basis of our findings, we defined a final model for presentation.

Finally, as an additional analysis, we performed a multigroup analysis for boys and girls to determine whether the associations differed by gender.

Since the type of cluster sampling varied from one country to another, commonly applied sandwich estimator methods were not possible. A decomposition of the variances revealed that school and class levels accounted for only 1.3% and 0% of the variance in injury, respectively. To be conservative in our analyses, we therefore applied the 0.1% alpha-error threshold rather than the usual 5% threshold.

Results

In total, 47,106 adolescents were included in our analyses. Each country had a large, nationally representative sample, 48.5% of whom were boys. The rate of engagement in physical activity each day differed substantially across countries, ranging from 9.2% (Iceland) to 26.9% (Russian Federation). Table 1 presents the mean age of onset of risk behaviors in the 25 countries included in this study (based on youth reporting the behaviors). Smoking tobacco yielded the lowest mean age of initiation (13.1), followed by drunkenness (13.8), and sexual intercourse (14.2). The percentage of children reporting engagement in early risk behaviors and injuries within countries differed substantially cross-nationally (see Table 2). Overall, 40.7% of the adolescent population in the 25 countries engaged in at least one early risk behavior; 18.2% engaged in multiple early risk behaviors.

Table 2

Percentage of Adolescents Reporting Early Risk Behaviors and Injuries

Country	Early smoking (age < 13)	Early drunkenness (age < 14)	Early cannabis use (age ≤ 15)	Early sexual intercourse (age < 15)	At least one early onset risk behavior	Multiple early onset risk behaviors (2+)	At least one medically attended injury
Austria	33.2	24.0	14.1	21.1	50.9	25.8	44.7
Belgium	15.4	12.5	22.7	13.5	38.7	16.3	44.2
Canada	10.1	16.3	36.2	14.7	44.4	20.5	45.0
Croatia	20.1	18.2	14.5	9.6	38.5	15.5	41.1
Czech Rep.	27.8	18.2	25.8	8.2	44.8	22.3	49.5
Denmark	12.0	20.2	17.9	35.4	48.6	23.4	48.5
Estonia	40.2	28.0	25.3	9.6	55.6	30.0	36.3
Finland	19.1	22.8	8.6	13.8	36.9	18.0	40.6
France	16.7	8.4	28.1	16.3	43.6	17.2	41.4
Germany	22.2	11.1	18.1	15.2	39.7	17.2	47.2
Hungary	20.4	11.9	14.2	10.7	36.3	13.6	42.3
Iceland	6.4	9.3	10.3	17.3	25.9	10.3	54.2
Italy	9.8	4.2	23.4	11.9	33.0	12.0	47.0
Latvia	30.3	21.5	23.1	6.1	49.3	21.6	44.9
Lithuania	27.8	24.1	15.2	6.8	46.3	19.5	49.6
Luxembourg	21.0	11.4	24.1	19.3	42.8	20.2	42.9
Netherlands	19.2	11.8	24.2	14.5	38.4	19.2	36.5
Portugal	14.3	8.5	9.9	12.8	29.7	10.9	41.1
Romania	12.1	12.6	3.3	10.0	26.6	8.5	30.0
Russian Fed.	23.0	19.7	18.4	12.7	42.9	20.1	37.1
Slovakia	21.8	18.6	18.5	4.9	38.3	17.0	42.5
Slovenia	15.9	14.4	18.6	8.5	35.7	14.8	34.3
Switzerland	20.7	12.3	33.9	12.3	45.7	20.7	50.0
Ukraine	23.4	14.5	16.8	10.8	39.8	17.4	33.4
United Kingdom	15.7	22.8	28.7	15.9	45.0	23.6	45.8
Mean	19.9	15.9	19.2	13.3	40.7	18.2	42.8

Note: Percentages are pooled estimates of the six imputed files (Schafer, 1997).

A multigroup model that examined the number of early behaviors (0, 1, 2, 3), as a predictor of injury in the 25 countries, had a good model fit when all regression paths were constrained (see Table 3). Model fit did not improve when regression paths were allowed to vary across countries (ΔTLI = 0.003 and ΔRMSEA = 0.001), which indicated that the cumulative association between early risk behaviors and injury was consistent across countries. Odds ratios were 1.17 (95% CI = 1.13-1.21), 1.34 (95% CI = 1.29-1.40), and 1.49 (95% CI = 1.41-1.57) for early engagement in one, two, and three risk behaviors, respectively.

Table 3.

Model Fit Statistics

Predictors	Model	Model specification	χ^2c	df	TLI	RMSEA	ΔTLI	ΔRMSEA
Early onset of 1, 2, or 3 risk behaviors^a	1	Risk behaviors 1-3 fixed across countries	280.59-288.63	171	0.930	0.018	0.003	0.001
	2	Risk behaviors 1-3 free	171.78-185.73	106	0.927	0.019	—	—
Early onset of 1, 2, 3 or 4 risk behaviors^b	1	Risk behaviors 1-4 fixed across countries	199.33-216.42	126	0.946	0.018	0.008	0.002
	2	Risk behaviors 1-4 free	104.53-114.55	69	0.954	0.016	—	—

Note: RMSEA = root mean square error of approximations; TLI = Tucker–Lewis index. Δ statistics indicate the difference with Model 1.

Countries included: Austria, Belgium, Canada, Croatia, Czech Republic, Denmark, Estonia, Finland, France, Germany, Iceland, Italy, Latvia, Lithuania, Luxembourg, Netherlands, Portugal, Romania, Russian Federation, Slovakia, Slovenia, Switzerland, Ukraine, and United Kingdom.

Countries included: All countries under (a) except for Hungary, Italy, Latvia, Lithuania, Portugal, Romania, Slovakia, and Slovenia.

As Mplus does not provide pooled estimates of the χ² value, the range of values of the individual imputed files was reported (p < .001 for all files and models).

The second multigroup analysis included only those countries that had sufficient students reporting engagement in four early risk behaviors (N = 17). This model had a good fit with all paths constrained. Allowing the paths to vary across countries did not improve the fit (ΔTLI = 0.008 and ΔRMSEA = 0.002), which supported the consistency of our findings across countries. Adolescents who engaged in multiple early risk behaviors were up to 1.85 fold more likely to report medically attended injuries compared to adolescents who did not engage in early risk behaviors (see Table 4). As the odds ratios of engagement in one, two, and three early risk behaviors were comparable to those in the previous model (N = 25), we assumed that the odds ratio of engagement in four early risk behaviors in the final model (N = 17) was also applicable to the eight countries that did not have sufficient students engaging in four early risk behaviors. The final model also revealed that (a) boys were more likely to be injured than were girls, (b) increasing frequency of physical activity was associated with increasing chances of becoming injured, and (c) adolescents with higher family affluence were more likely to be injured.

Table 4

The Cumulative Effect of Early Risk Behaviors on Medically Treated Injuries at Age 15: Model Estimates

Predictor	B*	SE	OR (95% CI)
Early engagement in 1 risk behavior	0.19	0.02	1.21 (1.17-1.26)
Early engagement in 2 risk behaviors	0.32	0.02	1.38 (1.32-1.48)
Early engagement in 3 risk behaviors	0.44	0.03	1.56 (1.46-1.66)
Early engagement in 4 risk behaviors	0.62	0.04	1.85 (1.70-2.02)
Gender (being a boy)	0.17	0.01	1.19 (1.15-1.22)
Physical exercise each day^a	0.45	0.02	1.57 (1.50-1.64)
Physical exercise 4-6 days a week^a	0.40	0.02	1.49 (1.44-1.55)
Physical exercise 2-3 days a week^a	0.21	0.02	1.23 (1.19-1.28)
High FAS^b	0.21	0.02	1.23 (1.17-1.29)
Medium FAS^b	0.11	0.02	1.11 (1.07-1.16)

Note: FAS = Family Affluence Scale.

Reference category: physical exercise 1 day per week or less.

Reference category: low FAS.

p < .001.

Finally, to test whether the current findings were similar in strength for boys and girls, we defined and tested a multigroup model with two groups defined by gender. Both the model where regression paths were constrained across gender and the model where they were freely estimated had a perfect fit (RMSEA = 0.000, TLI = 1.00 and RMSEA = 0.001, TLI = 0.999, respectively). As there was no significant difference between the fit of the models (ΔTLI = 0.001 and ΔRMSEA = 0.001), we chose the most parsimonious model, which was the model with constrained paths. This indicates that there were no observed gender differences in the association between early risk behaviors and injury.

Discussion

This study resulted in two major findings. First, it profiled the prevalence of early risk taking in 25 countries in terms of smoking, drunkenness, cannabis use, and sexual intercourse. Second, it demonstrated a cross-national consistent and cumulative association between the number of early risk behaviors that adolescents engaged in and the occurrence of injury at age 15. The fact that we found an association is remarkable, as much time had passed between the engagement in some risk behaviors (e.g., smoking before age 13) and the experience of injury at age 15. Because of the cross-sectional design of the current study, we cannot make inferences regarding causality. However, a causal pathway between early risk behaviors and injury has been questioned in previous studies that had found a high correlation between psychosocial characteristics, stress, risk taking, and injuries (Macdonald, Erickson, Wells, Hathaway, & Pakula, 2008). It seems likely that a third variable (i.e., a high-risk lifestyle) explains both adolescent early risk behavior and injury at age 15. This conclusion is consistent with Moffitt’s (1993, 2006) findings in that early onset of risk behavior is related to unhealthy behaviors and negative outcomes later in life.

It should be noted that, due to the cross-sectional nature of the current data, the temporal sequence between some reported risk behaviors and injury remains unclear. To illustrate, the 15-year-old youth were asked whether they had experienced injuries in the year before this study; however, participants were also asked whether they had ever used cannabis. Thus, participants may have experienced an injury prior to using cannabis. However, this was not viewed as a major issue for the interpretation of our findings as we placed both behaviors in the context of a high-risk lifestyle.

The cross-national relationship as demonstrated by our modeling is important. Specifically, the consistency of our findings is remarkable in that there are many factors that could potentially eliminate the associations between early risk behaviors and injury. For example, there are large variations in access to medical care for the treatment of injuries across countries (Schoen & Doty, 2004). Furthermore, cultural factors are likely to influence individual adolescents’ decision to seek medical assistance when injured. Existing studies have demonstrated that cultural dimensions play an important role in medicine use (Deschepper et al., 2008), medical communication (Meeuwesen, van den Brink-Muinen, & Hofstede, 2009), and medical compliance (Larsen, Stovring, Kragstrup, & Hansen, 2009). All these factors may cause an injured adolescent in one country to visit a doctor and be treated, whereas an adolescent in a different country with a similar injury might not be treated or might not even seek medical assistance. Finally, countries differ considerably in terms of prevalence rates and policies with respect to young people’s substance use and sexual activity. Sex education strategies for youth in European and North American countries differ noticeably concerning educational curricula and approach taken in the classroom (e.g., Lewis & Knijn, 2003). Similar differences in approach can be found with respect to substance use policies. For example, though alcohol is prohibited for youth up to age 20 in Iceland, youth from age 16 onwards are legally allowed to purchase and drink alcoholic beverages in Italy and Portugal (International Center for Alcohol Policies, January 2011). Because so many differences exist concerning laws on substance use, it is noteworthy that we observed a cross-national consistent relationship between early risk behavior and injury.

Past efforts have focused on the detection of cross-national differences in risk behaviors and their relationship to health outcomes. Therefore, it is significant that the current results yielded similarities in associations across countries. They suggest that the relationship between early risk behaviors and physical injuries may not be significantly influenced by cultural factors. The possibility remains, however, that mediating effects vary between countries; future research is encouraged to investigate these mechanisms in more detail.

A secondary finding that deserves attention in this discussion is that youth from more affluent families were more likely to be injured, independent of their engagement in early risk behaviors, physical exercise, and gender. A possible explanation for this is that youth from more affluent families have better access to medical facilities compared to youth from less affluent families. This may especially be the case in countries with large socioeconomic inequities. In addition, youth from more affluent families are more likely to engage in injury-producing sports frequently than are youth from less affluent families (Gorely, Atkin, Biddle, & Marshall, 2009; Seabra, Mendonça, Thomis, Peters, & Maia, 2008; Walters, Barr, Wall, & Neumark, 2009).

Limitations of this study include the self-reported nature of our data, which are subject to social desirability and recall bias (Aday & Cornelius, 2006). Although far from perfect, self-administered surveys are virtually the only practical method that can be used to simultaneously obtain a broad range of health indicators for a study of this size. Our questionnaire items were developed iteratively over many years and were supported by validation efforts (Currie, 2000; Currie, Nic Gabhainn et al., 2008; Currie et al., 2002; Roberts et al., 2009). Furthermore, the effect estimates in our final model were modest, which may lead one to conclude that the associations are not important. However, the estimates should be interpreted with caution as they were conservative; specifically (a) the association tested was indirect, and (b) our outcome variable of injury was broadly defined. Rather than focusing on severe injuries, we included any type of injury an adolescent had experienced for which they received medical attention. This broad definition of injury likely biased our results toward no effect. A final limitation of the current study was the cross-sectional nature of the study design, and no causal inferences could be made. Therefore, longitudinal confirmation of the observed relationships is encouraged.

The current study is unique in several respects. First, our examination of the cumulative versus individual effects of the four risk behaviors, though consistent with problem behavior theory (Jessor & Jessor, 1977), has rarely been the focus of past etiological studies of injury in adolescent samples. Second, the current focus on early risk behaviors was unique in comparison to other research that has predominantly focused on concurrent risk behaviors as well as a more direct link between risk behaviors and the occurrence of injury. Third, the quality of our analyses was considered high due to our application of modern imputation techniques, which minimized the potential for bias from missing data. Finally, a major strength of this study was its international scope; it included populations of young people from diverse cultures. Our confirmation of a cross-national consistent pattern that links early risk behaviors with later occurrence of injury is noteworthy.

Our findings have implications for prevention as they provide confirmation that early engagement in risk behaviors may be related to a lifestyle that includes a high risk for injury. Public health interventions may need to consider the multiple risk behavior phenomena during early adolescence as an important determinant of health and focus on this clustered risk behavior phenomenon, as opposed to focusing on individual risk behaviors in isolation. At a societal level, this may affect the content of health promotion campaigns. At an individual level, children who engage in a number of early risk behaviors may benefit from individual, focused interventions that promote a healthier lifestyle in general. Our list of risk behaviors provides a basis for a useful administrative tool that may be used in the early screening for at-risk adolescents.

Footnotes

Acknowledgements

HBSC is a WHO collaborative study. International coordinator of the 2005-2006 study: Candace Currie, University of Edinburgh, Scotland. Data Bank manager: Oddrun Samdal, University of Bergen, Norway. This publication reports on data from the following countries (principal investigators at that time): Austria (Wolfgang Dür), Belgium (Flemisch: Le Maes, French: Danielle Piette), Canada (William Boyce), Croatia (Marina Kuzman), Czech Republic (Ladislav Csémy), Denmark (Pernille Due), Estonia (Katrin Aasvee), Finland (Jorma Tynjälä), France (Emmanuelle Godeau), Germany (Ulrike Ravens-Sieberer), Hungary (Ágnes Németh), Iceland (Thoroddur Bjarnason), Italy (Franco Cavallo), Latvia (Iveta Pudule), Lithuania (Apolinaras Zaborskis), Luxembourg (Yolande Wagener), Netherlands (Wilma Vollebergh), Portugal (Margarida Gaspar de Matos), Romania (Adriana Baban), Russian Federation (Alexander Komkov), Slovakia (Elena Morvicova), Slovenia (Helena Jeriček), Switzerland (Emmanuel Kuntsche), Ukraine (Olga Balakireva), and the United Kingdom (England: Antony Morgan, Scotland: Candace Currie, Wales: Chris Roberts). We are especially thankful to Margarida Gaspar de Matos and Katrin Aasvee for their feedback on the manuscript. For the purpose of this study, Ms. Margaretha de Looze had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The data collection in each country was funded at the national level. We are grateful for the financial support offered by the various government ministries, research foundations, and other funding bodies in the participating countries and regions.

References

Aday

L. A.

Cornelius

L. J.

(2006). Designing and conducting health surveys: A comprehensive guide. San Francisco, CA: Jossey-Bass.

Arbuckle

J. L.

(2008). Amos™ 17.0 User’s Guide. Crawfordville, FL: Amos Development Corporation.

Basen

K.-E.

Edmundson

E. W.

Parcel

G. S.

(1996). Structure of health risk behavior among high school students. Journal of Consulting and Clinical Psychology, 64(4), 764-775.

Baumeister

S. E.

Tossmann

(2005). Association between early onset of cigarette, alcohol and cannabis use and later drug use patterns: An analysis of a survey in European metropolises. European Addiction Research, 11(2), 92-98. doi: 10.1159/00008303810.1159/000083038

Bellis

M. A.

Hughes

Calafat

Juan

Ramon

Rodriguez

J. A.

. . . Phillips

P.-H.

(2008). Sexual uses of alcohol and drugs and the associated health risks: A cross-sectional study of young people in nine European cities. BMC Public Health, 8, 155-166. doi:10.1186/1471-2458-8-15510.1186/1471-2458-8-155

Borges

Cherpitel

Orozco

Bond

Macdonald

. . . Poznyak

(2006). Multicenter study of acute alcohol use and non-fatal injuries: Data from the WHO collaborative study on alcohol and injuries. Bulletin of the World Health Organization, 84, 453-460.

Chen

F. F.

(2007). Sensitivity of goodness of fit indexes to lack of measurement invariance. Structural Equation Modeling: A Multidisciplinary Journal, 14(3), 464-504. doi: 10.1080/1070551070130183410.1080/10705510701301834

Chiolero

Schmid

(2000). Morbidité accidentelle à l’adolescence: une étude représentative auprès des écoliers de 12 à 15 ans en Suisse. [Accidental morbidity in adolescents: A retrospective study in 12–15-year-old schoolchildren in Switzerland]. Schweiz Med Wochenschr, 130(37), 1285-1290.

Currie

(2000). The international HBSC study: Rationale, history and description. In Currie

Hurrelman

Settertobulte

Smith

Todd

(Eds.), Health and health behaviour among young people. Health behaviour in school-aged children: A WHO cross-national study (HBSC) international report (pp. 8-10). Copenhagen, Denmark: WHO Regional Office for Europe.

10.

Currie

Molcho

Boyce

Holstein

Torsheim

Richter

(2008). Researching health inequalities in adolescents: The development of the Health Behaviour in School-Aged Children (HBSC) Family Affluence Scale. Social Science & Medicine, 66(6), 1429-1436.

11.

Currie

Nic Gabhainn

Godeau

Roberts

Smith

Currie

. . . Barnekow

(2008). Inequalities in young people’s health: Health behaviour in school-aged children (HBSC). International report from the 2005/2006 survey. Copenhagen, Denmark: WHO Regional Office for Europe.

12.

Currie

Roberts

Morgan

Smith

Settertobulte

Samdal

Rasmussen

V. B.

(2002). Young people’s health in context. Health Behaviour in School-Aged Children (HBSC) study: International report from the 2001/2002 survey. Copenhagen, Denmark: WHO Regional Office for Europe.

13.

Currie

Samdal

Boyce

Smith

(2001). Health behaviour in school-aged children: A WHO cross-national study (HBSC), research protocol for the 2001/2002 survey. Edinburgh, Scotland: Child and Adolescent Health Research Unit (CAHRU), University of Edinburgh.

14.

Danseco

E. R.

Miller

T. R.

Spicer

R. S.

(2000). Incidence and costs of 1987-1994 childhood injuries: Demographic breakdowns. Pediatrics, 105(2), 27. doi:10.1542/peds.105.2.e27

15.

Deschepper

Grigoryan

Lundborg

C. S.

Hofstede

Cohen

Kelen

G. V.

. . . Haaijer-Ruskamp

F. M.

(2008). Are cultural dimensions relevant for explaining cross-national differences in antibiotic use in Europe? BMC Health Services Research, 8, 123-132. doi:10.1186/1472-6963-8-12310.1186/1472-6963-8-123

16.

DuRant

R. H.

Smith

J. A.

Kreiter

S. R.

Krowchuk

D. P.

(1999). The relationship between early age of onset of initial substance use and engaging in multiple health risk behaviors among young adolescents. Archives of Pediatrics and Adolescent Medicine, 153(3), 286-291.

17.

Gorely

Atkin

A. J.

Biddle

S. J.

Marshall

S. J.

(2009). Family circumstance, sedentary behaviour and physical activity in adolescents living in England: Project STIL. The International Journal of Behavioral Nutrition and Physical Activity, 6, 33-40. doi:10.1186/1479-5868-6-3310.1186/1479-5868-6-33

18.

Gullone

Moore

Moss

Boyd

(2000). The adolescent risk-taking questionnaire: Development and psychometric evaluation. Journal of Adolescent Research, 15(2), 231-250. doi:10.1177/074355840015200310.1177/0743558400152003

19.

International Center for Alcohol Policies. (2011). Minimum age limits worldwide. Retrieved from http://www.icap.org

20.

Jessor

(1998). New perspectives on adolescent risk behavior. Cambridge, UK: Cambridge University Press.

21.

Jessor

S. L.

(1977). Problem behavior and psychosocial development: A longitudinal study of youth. San Diego, CA: Academic Press.

22.

King

M. A.

Pickett

King

A. J. C.

(1998). Injury in Canadian youth: A secondary analysis of the 1993-94 health behaviour in school-aged children survey. Canadian Journal of Public Health, 89(6), 397-401.

23.

Kline

R. B.

(2010). Principles and practice of structural equation modeling (3rd ed.). New York, NY: Guilford.

24.

Koven

McColl

M. A.

Ellis

Pickett

(2005). Multiple risk behaviour and its association with head and neck injuries: A national analysis of young Canadians. Preventive Medicine, 41, 240-246. doi:10.1016/j.ypmed.2004.11.00910.1016/j.ypmed.2004.11.009

25.

Krug

E. G.

Sharma

G. K.

Lozano

(2000). The global burden of injuries. American Journal of Public Health, 90(4), 523-526.

26.

Larsen

Stovring

Kragstrup

Hansen

D. G.

(2009). Can differences in medical drug compliance between European countries be explained by social factors: Analyses based on data from the European Social Survey, round 2. BMC Public Health, 9, 145-156. doi:10.1186/1471-2458-9-14510.1186/1471-2458-9-145

27.

Lescohier

Scavo Gallagher

(1996). Unintentional injury. In DiClemente

R. J.

Hansen

W. B.

Ponton

L. E.

(Eds.), Handbook of adolescent health risk behavior (pp. 225-258). New York, NY: Plenum.

28.

Lewis

Knijn

(2003). Sex education materials in the Netherlands and in England and Wales: A comparison of content, use and teaching practice. Oxford Review of Education, 29(1), 113-132.

29.

Luijpers

(2000). Intentie tot exploratie, sociale binding en delinquent gedrag van Nederlandse jongeren [Intention to explore, social bonds and delinquent behavior of Dutch adolescents]. Doctoral dissertation, Utrecht University, the Netherlands.

30.

Macdonald

Erickson

Wells

Hathaway

Pakula

(2008). Predicting violence among cocaine, cannabis, and alcohol treatment clients. Addictive Behaviors, 33, 201-205. doi:10.1016/j.addbeh.2007.07.00210.1016/j.addbeh.2007.07.002

31.

Mazur

Scheidt

P. C.

Overpeck

M. D.

Harel

Molcho

(2001). Adolescent injuries in relation to socioeconomic status: An international perspective. Journal of Injury Control and Safety Promotion, 8(3), 179-182.

32.

Meeuwesen

van den Brink-Muinen

Hofstede

(2009). Can dimensions of national culture predict cross-national differences in medical communication? Patient Education and Counseling, 75, 58-66. doi:10.1016/j.pec.2008.09.01510.1016/j.pec.2008.09.015

33.

Moffitt

T. E.

(1993). Adolescence-limited and life-course-persistent antisocial behavior: A developmental taxonomy. Psychological Review, 100, 674-701.

34.

Moffitt

T. E.

(2006). A review of research on the taxonomy of life-course persistent versus adolescence-limited antisocial behavior. In Cullen

F. T.

Wright

J. P.

Blevins

K. R.

(Eds.), Taking stock: The status of criminological theory (pp. 277-312). New Brunswick, NJ: Transaction Publishers.

35.

Muthén

L. K.

Muthén

B. O.

(2006). MPlus user’s guide (4th ed.). Los Angeles: Muthén & Muthén.

36.

Petridou

Zavitsanos

Dessypris

Frangakis

Mandyla

Doxiadis

Trichopoulos

(1997). Adolescents in high-risk trajectory: Clustering of risky behavior and the origins of socioeconomic health differentials. Preventive Medicine, 26(2), 215-219. doi:10.1006/pmed.1996.013010.1006/pmed.1996.0130

37.

Pickett

Dostaler

Craig

Janssen

Simpson

Shelley

S. D.

Boyce

W. F.

(2006). Associations between risk behavior and injury and the protective roles of social environments: An analysis of 7235 Canadian school children. Injury Prevention, 12(2), 87-92. doi:10.1136/ip.2005.01110610.1136/ip.2005.011106

38.

Pickett

Garner

M. J.

Boyce

W. F.

King

M. A.

(2002). Gradients in risk for youth injury associated with multiple-risk behaviours: A study of 11,329 Canadian adolescents. Social Science & Medicine, 55(6), 1055-1068.

39.

Pickett

Molcho

Simpson

Janssen

Kuntsche

Mazur

. . . Boyce

W. F.

(2005). Cross-national study of injury and social determinants in adolescents. Injury Prevention, 11(4), 213-218. doi:10.1136/ip.2004.00702110.1136/ip.2004.007021

40.

Pickett

Schmid

Boyce

W. F.

Simpson

Scheidt

P. C.

Mazur

. . . Harel

(2002). Multiple risk behavior and injury: An international analysis of young people. Archives of Pediatrics and Adolescent Medicine, 156(8), 786-793.

41.

Riley

A. W.

Harris

S. K.

Ensminger

M. E.

Ryan

Alexander

Green

Starfield

(1996). Behavior and injury in urban and rural adolescents. Injury Prevention, 2(4), 266-273. doi:10.1136/ip.2.4.26610.1136/ip.2.4.266

42.

Roberts

Freeman

Samdal

Schnohr

C. W.

de Looze

M. E.

Nic Gabhainn

. . . Rasmussen

(2009). The Health Behaviour in School-Aged Children (HBSC) study: Methodological developments and current tensions. International Journal of Public Health, 54, 140-150. doi:10.1007/s00038-009-5405-910.1007/s00038-009-5405-9

43.

Schafer

J. L.

(1997). Analysis of incomplete multivariate data. London: Chapman & Hall/CRC.

44.

Schoen

Doty

M. M.

(2004). Inequities in access to medical care in five countries: Findings from the 2001 Commonwealth Fund International Health Policy Survey. Health Policy, 67(3), 309-322.

45.

Seabra

A. F.

Mendonça

D. M.

Thomis

M. A.

Peters

T. J.

Maia

J. A.

(2008). Associations between sport participation, demographic and socio-cultural factors in Portuguese children and adolescents. European Journal of Public Health, 18, 25-30. doi:10.1093/eurpub/ckm04910.1093/eurpub/ckm049

46.

Walters

Barr

D. J.-A.

Wall

Neumark

D.-S.

(2009). Does participation in organized sports predict future physical activity for adolescents from diverse economic backgrounds? Journal of Adolescent Health, 44(3), 268-274. doi:10.1016/j.jadohealth.2008.08.01110.1016/j.jadohealth.2008.08.011

47.

Williams

J. M.

Currie

C. E.

Wright

Elton

R. A.

Beattie

T. F.

(1997). Socioeconomic status and adolescent injuries. Social Science & Medicine, 44(12), 1881-1891. doi:10.1016/S0277-9536(96)00297-310.1016/S0277-9536(96)00297-3

48.

Willoughby

Chalmers

Busseri

M. A.

(2004). Where is the syndrome? Examining co-occurrence among multiple problem behaviors in adolescence. Journal of Consulting and Clinical Psychology, 72(6), 1022-1037.

49.

Zimmer

M. J.-G.

Helfand

(2008). Ten years of longitudinal research on US adolescent sexual behavior: Developmental correlates of sexual intercourse, and the importance of age, gender and ethnic background. Developmental Review, 28(2), 153-224. doi:10.1016/j.dr.2007.06.00110.1016/j.dr.2007.06.001