Are Traumatic Brain Injuries Associated With Criminality After Taking Account of Childhood Family Social Status and Disruptive Behaviors?

In the United States, medical records indicate that more than 1.3 million persons sustain a TBI each year.¹ Meta-analyses^2–5 and successive studies^6–14 report that the prevalence of TBIs is higher among juvenile and adult criminal offenders than in age- and sex-matched general population samples. Evidence is lacking about the age at which injuries were sustained. Some studies^{6,9–11,13,14} have determined that TBIs preceded arrest or conviction. For example, in a large population sample of Swedes, a TBI (excluding concussion) recorded in health records was associated with a threefold increase in the risk of a subsequent conviction for a violent crime.⁶ In a similar study of a large Australian cohort, among males, having sustained a TBI was associated with an increase of 1.58 times (95% CI=1.46–1.72) in the risk of a subsequent conviction for any type of crime and with a 1.65 times (95% CI=1.42–1.92) increase in the risk of a subsequent conviction for a violent crime.¹³

The association between TBIs and criminality, however, may be confounded by antecedents and correlates of criminality. For instance, the Swedish study did not adjust analyses for prior offending, the strongest predictor of offending. It found that comparisons with siblings who showed similar risks for offending reduced the strength of association of TBIs with subsequent offending.⁶ In the Australian study, compared with nonadjusted models, the risk of subsequent offending in males who sustained TBIs was lower after adjusting for drug and alcohol abuse, mental illness, Aboriginality, socioeconomic disadvantage, and year of birth.¹³ A U.S. study using data from the Traumatic Brain Injury Model System National Database concluded that having sustained a TBI might not have been associated with an increased risk of arrest among men presenting known antecedents of criminality.¹⁵ Similarly, a study of U.S. veterans showed that TBIs were not associated with subsequent offending when analyses were adjusted for predictors of offending, such as having witnessed parents fighting, previous arrests, substance misuse, irritability, and posttraumatic stress disorder (PTSD).¹⁶

A recent study conducted in the United Kingdom⁷ that used parent- and self-reports found that the association of TBIs with offending was robust to adjustment for mother’s age, education, alcohol and nicotine use, social class, gender, life events, and parenting style, but detected no difference in risks of offending for participants who sustained TBIs and those who sustained orthopedic injuries. A prospective study of a New Zealand birth cohort using official health and criminal records reported that after adjusting for gender, family socioeconomic status, and parent-rated behavior problems up to age 5, TBIs sustained at different ages and of differing severity were associated with an increased risk of arrest for any type of crime, property crimes, and violent crimes.¹⁰ Age at TBI and severity of TBI may further modify the association of TBIs with criminality.^7,10

Taken together, the extant literature suggests that the association between TBIs and offending is weakened after taking account of known predictors of criminality. Robust evidence shows that childhood behavior problems predict criminal offending^17–21 and that most violent crimes are committed by men with a history of childhood behavior problems.^22–25 Thus, although an association between TBIs and crimes has been established, it is presently not known whether this association would persist after taking account of childhood predictors of criminality and the age at which the TBI is sustained.

In the present study, we examined a sample of 724 males from Quebec, Canada, followed to age 24. Our aims were to 1) document the prevalence of TBIs among individuals with offender status and nonoffender status in childhood, adolescence, and early adulthood; 2) determine whether TBIs preceded or followed criminal conviction; 3) compare childhood characteristics of individuals with offender status who sustained TBIs before or after conviction with individuals with nonoffender status with no TBI; and 4) determine whether experiencing a TBI at different developmental stages predicted offending after taking account of known childhood predictors of offending. Data were collected prospectively from age 6 to age 24, allowing us to determine temporal associations of childhood behaviors and TBIs with criminal convictions. Well-known predictors of criminality included social status of the participants’ family when they were age 6 years and teacher ratings of disruptive behaviors when participants were ages 6, 10, and 12 years. Information on TBIs was extracted from health files. Information on juvenile and adult criminal convictions for any type of crime and for violent crimes was extracted from official records.

Methods

Participants

Participants were males drawn from two cohorts recruited when they entered elementary school^26,27 and followed to age 24. From this sample of 2,631 males, 371 who were charged with a criminal offense from age 18 to age 24 and a random sample of 371 without a criminal charge from age 18 to age 24 were selected. Complete data for the present study were available for 724 of these 742 men.

Measures

TBIs.

In Quebec, there is one universal health system, and each inhabitant has one digitalized health file from birth to death. The Régie de l’Assurance Maladie provided digital health records for each participant, which contained ICD-9 codes and dates for every diagnosis. TBIs were defined as ICD-9 codes: 800.0–800.9 fractures of vault of skull; 801.0–801.9 fractures of base of skull; 802.0–802.9 fracture of face bones; 803.0–803.9 other and unqualified skull fractures; 850.0–850.9 concussion; 851.0–851.9 cerebral laceration and contusion; 852.0–852.9 subarachnoid, subdural, and extradural hemorrhage, following injury; 853.0–853.9 other and unspecified intracranial hemorrhage following injury; 854.0–854.9 intracranial injury of other and unspecified nature; 959.0 head injury unspecified. Previous studies defined TBIs by using these same ICD-9 codes.^9,13,14 Since a diagnosis is noted in the file each time a physician sees a patient, all TBI diagnoses recorded within 30 days of each other were counted as one TBI.

Criminal convictions.

Criminal records were available from age 12 to age 24. Criminal convictions were coded as any crime in the criminal code or violent crime (homicide, assault, sexual offenses, offenses with arms, burglary, harassment, and other crimes that physically hurt people), according to the Correctional Services of Canada classification.²⁸

Childhood behaviors.

When participants were ages 6, 10, and 12, their classroom teachers rated conduct problems (CP), hurtful and uncaring behaviors, and inattention hyperactivity (IH) using the Social Behavior Questionnaire.²⁷ Items for each rating are described in Supplementary Material.

Family social status (FSS).

This variable included parents’ highest level of education, prestige of parental employment, age of the mother and father at participant’s birth, and whether or not the participant lived with both biological parents. Elevated scores indicated lower FSS.²⁹

Ethics Approval

Initially, parents provided consent for participants’ teachers to rate their child’s behavior and also consented to their own participation in the study. Once participants were 18 years old, they provided consent. The Commission d’Accès à l’Information de Québec approved the use of data from health files and criminal files. The study was approved by ethics committees at the Université de Montréal, Centre Hospitalier Universitaire Sainte-Justine, and the Institut Philippe-Pinel de Montréal.

Statistical Analyses

Categorical variables were compared using chi-square tests and continuous variables using Mann-Whitney U tests or analysis of variance, depending on distributions of values. To determine whether TBIs predicted criminal convictions when accounting for childhood behaviors and FSS, logistic regression models were computed to predict at least one conviction for any type of crime and for a violent crime, from age 12 to age 17 and from age 18 to age 24. Among participants who had no juvenile convictions, similar logistic regression models were computed to predict being convicted for the first time from age 18 to age 24. In all models predicting at least one conviction for a violent crime, participants with a nonviolent conviction during the same period were excluded from analyses. Since we found no association between the number of TBIs and the likelihood of conviction, presence or absence of TBI was entered into regression analyses as a predictor. For each dependent variable, three initial models (ages 6, 10, 12) tested associations of childhood behaviors and previous TBIs with convictions. The significant predictors from these initial models were included in final model along with FSS. In models predicting adult convictions for any crime and for a violent crime, TBIs from age 13 to age 17 and having a juvenile conviction were added as predictors. Significant results of models are presented as odds ratios: each increase of 1 in a score for a childhood behavior increases the risk of conviction by one odds ratio; odds ratios for FSS indicate the risk for a participant with the lowest score as compared with one with the highest score.

Results

By age 24, 142 participants (19.6%) had sustained at least one TBI: 114 (15.7%) sustained a single TBI, 22 (3%) sustained two, and six (0.83%) sustained three or more TBIs. Sixty-one participants sustained at least one TBI up to age 12 (24 up to age 6), 37 sustained at least one TBI from age 13 to age 17, and 56 from age 18 to age 24. By age 24, 355 (49.0%) participants had acquired at least one conviction for any type of crime and 132 (18.2%) for a violent crime.

Are Criminal Offenders More Likely to Sustain a TBI Than Nonoffenders?

As presented in Table 1, similar proportions of offenders (22.0%) and nonoffenders (17.3%) sustained a TBI by age 24. While the proportions of offenders and nonoffenders who sustained TBIs up to age 12 and from 13 to 17 years were similar, from 18 to 24 years significantly more of the offenders (11.0%) than nonoffenders (4.6%) sustained a TBI. Similar proportions of violent offenders and nonoffenders sustained TBIs up to age 12 and from age 13 to age 17, while significantly more violent offenders (13.6%) than nonoffenders (4.6%) sustained a TBI at age 18 or older.

Table 2 presents comparisons of the median number of TBIs for offenders and violent offenders as compared with nonoffenders. No differences were detected in childhood or adolescence, but offenders and violent offenders sustained significantly more TBIs from age 18 to age 24 than nonoffenders.

As shown in Tables 3, 4, and 5, the proportions of participants with a conviction from age 12 to age 17, a conviction for a violent crime from age 12 to age 17, a conviction from age 18 to age 24, a conviction for a violent crime from age 18 to age 24, a first conviction from age 18 to age 24 for any crime, or a first conviction for a violent crime from age 18 to age 24 did not vary as a function of the number of prior TBIs.

Do TBIs Precede or Follow First Criminal Convictions?

Among the 78 offenders with TBIs, 46 (59.0%) sustained their first TBI before their first conviction, and 32 (41.0%) after their first conviction ([N=78] χ²=2.513, df=1, p=0.113). Of the 34 violent offenders with TBIs, 19 (55.9%) sustained their first TBI before their first conviction for a violent offense, and 15 (44.1%) after ([N=34] χ²=0.471, df=1, p=0.493). The mean length of time from first TBI to first crime was 95.4 months (SD=71.28) (range, 1.2–225.36 months). Among the offenders who sustained a TBI prior to their first crime, the average number of TBIs was 1.13 (SD=0.34).

Do Offenders Who Sustained a TBI Before a First Conviction Differ From Nonoffenders as to Childhood Predictors of Crime?

As presented in Table 6, offenders who sustained a TBI before being convicted of a crime were raised in families of lower social status and obtained higher scores at age 6 for CP, hurtful and uncaring behaviors, and IH; at age 10 for CP, hurtful behavior, and IH; and at age 12 for hurtful behavior and CP than the nonoffenders with no TBI.

Do Offenders Who Were Convicted of a Crime Before Sustaining Their First TBI Differ From Nonoffenders as to Childhood Predictors of Crime?

Also presented in Table 6 are results showing that the 32 men who were convicted of a criminal offense before sustaining a TBI were raised in families with lower social status than the nonoffenders with no TBIs. They obtained higher scores at age 6 for CP; at age 10 for CP, hurtful and uncaring behaviors, and IH; and at age 12 for hurtful behavior, CP, and IH.

Does a TBI Predict Criminal Convictions When Taking Account of Well-Known Childhood Predictors of Criminal Offending?

Juvenile convictions.

Results of initial models revealed that having sustained a TBI prior to age 6, 10, or 12 was not associated with a juvenile conviction for any type of crime. The final model, presented in Table 7, included behavior scores that were significant in the initial models and FSS. The risk of a juvenile conviction for any type of crime was increased 1.19 times (95% CI=1.07–1.32) by age 10 hurtful behavior, 1.1 times (95% CI=1.02–1.19) by age 10 IH, 1.3 times (95% CI=1.11–1.48) by age 12 uncaring behavior, and 6.1 times (2.59–14.39) by FSS.

A similar series of regression models was computed to predict a juvenile conviction for violence. In the age 6, 10, and 12 models, having sustained a prior TBI was not associated with a juvenile conviction for violence. In the final model, childhood behavior scores that were significant in the initial models, and FSS were entered as predictors. The results, presented in Table 7, show that the risk of a violent conviction was increased 1.16 times (95% CI=1.04–1.29) by age 10 IH, 1.52 times (95% CI=1.20–1.93) by age 12 uncaring behavior, 1.36 times (95% CI=1.17–1.57) by age 12 hurtful behavior, and 22.29 times (95% CI=5.52–90.01) by FSS.

Convictions from age 18 to age 24.

Initial models detected no association of TBIs prior to age 6, 10, or 12 with convictions from age 18 to age 24 for any type of crime. The final model included as predictors having sustained a TBI from age 13 to age 17, scores for childhood behaviors that were significant in the initial models, FSS, and a juvenile conviction. Results are presented in Table 8. The risk of a conviction in adulthood for any type of crime was increased 1.22 times (95% CI=1.11–1.35) by age 6 hurtful behavior, 1.27 times (95% CI=1.15–1.41) by age 12 hurtful behavior, 7.20 times (95% CI=3.44–15.07) by FSS, and 4.26 times (95% CI=2.62–6.93) by a juvenile conviction.

A similar series of analyses was computed to determine the association of having sustained a TBI prior to age 18 and a conviction for violence from age 18 to age 24, after taking account of FSS and childhood behavior scores. Again, the initial models of predictors at ages 6, 10, and 12 indicated no association of prior TBIs with an adult conviction for violence. In the final model, presented in Table 8, predicting a conviction for a violent crime in adulthood, a TBI from age 13 to age 17, FSS, the behavior scores that were significant in the initial models, and presence or absence of a juvenile conviction were entered as predictors. The risk of a violent conviction in adulthood was increased 1.20 times (95% CI=1.06–1.35) by age 10 IH, 1.40 times (95% CI=1.16–1.66) by age 12 hurtful behavior, 16.34 times (95% CI=3.75–71.19) by FSS, and 12.63 times (95% CI=5.82–27.43) by a juvenile conviction.

Convictions for a first crime from age 18 to age 24.

A similar series of models was computed to determine whether having sustained a prior TBI was associated with a first conviction for any type of crime from age 18 to age 24 or a first conviction for violence from age 18 to age 24. Only participants without a juvenile conviction were included in these models. Initial models showed that having sustained a TBI prior to age 6, 10, or 12 was not associated with an increased risk of a first conviction for any type of crime or for a violent crime in adulthood. The final models included presence or absence of a TBI from age 13 to age 17, childhood behavior scores that were significant in the initial models, and FSS. As shown in Table 9, prior TBIs were not associated with an increased risk of a first conviction for any type of crime or for a violent crime from age 18 to age 24, while childhood behavior problems and FSS were associated with increased risks.

Discussion

In a sample of 724 males, proportionately more of those who had acquired a criminal conviction by age 24 sustained a TBI from age 18 to age 24, but not at younger ages. While previous studies reported a higher prevalence of TBIs among offenders than nonoffenders,^2–5 in the present study the higher prevalence of TBIs among offenders emerged only at age 18 or later. We further extended knowledge by showing that similar proportions of offenders had sustained a first TBI before and after their first conviction for any type of crime or for a violent crime. Offenders who sustained at least one TBI prior to conviction and those who were convicted before sustaining a TBI were similar in childhood, having been raised in families of low social status and presenting high levels of disruptive behaviors. Taken together, these results suggest that it was men who had been raised by single parents who were young and poorly educated with low prestige jobs, and who had displayed conduct problems, behaviors that hurt other children, inattention, and hyperactivity through elementary school who were at increased risk to sustain a TBI in adulthood.

Most offenders have a history of childhood disruptive behaviors.^17–25 As children presenting disruptive behaviors transition to adolescence, they begin to misuse substances³⁰ and persistently engage in reckless behaviors that become more dangerous as they age. The same childhood behaviors that predict crime have been reported to predict accidents^31,32 and death in early adulthood.²³ Thus, engagement in risky behaviors such as driving motor vehicles at high speed, particularly while intoxicated, may account for the increased number of TBIs sustained by the offenders in early adulthood. The results of our study add to previous findings showing that adolescents and adults involved in the criminal justice system constitute a population at risk for TBIs. Interventions aimed at preventing, assessing, and treating TBIs among adult offenders are needed, especially since repeated TBIs among individuals with a history of aggressive behavior have been reported to lead to increased aggressive behavior.³³ Our results additionally suggest that reducing disruptive behaviors in childhood so as to limit subsequent substance misuse and reckless behavior could potentially prevent TBIs.

We used prospectively collected data and found no evidence that sustaining a TBI prior to age 18 increased the risk of offending after taking account of well-known childhood predictors of criminality. TBIs were not associated with juvenile convictions for any type of crime or for a violent crime, for convictions for any type of crime or a violent crime from age 18 to age 24, or for any type of first crime or a first violent crime from age 18 to age 24 when FSS and disruptive childhood behaviors were included in the prediction models. In the initial logistic regression models, sustaining a TBI prior to age 6, prior to age 10, and prior to age 12 were entered as predictors, and in the models predicting convictions in adulthood TBIs sustained from age 13 to age 17 were additionally entered. These results add further evidence¹⁶ and support to the hypotheses^15,33 that TBIs do not predict criminal convictions after taking account of characteristics of the family of origin and previous aggressive behavior and criminality. These results require replication in cohorts with different demographics to assess generalizability.

One previous study¹⁰ reported that TBIs sustained by age 5 were associated with offending up to age 25, after taking account of family socioeconomic status and participants’ behaviors up to age 5. We had too few participants who had sustained a TBI prior to school entry to examine associations with criminality. However, TBIs sustained prior to age 13 did not predict juvenile convictions. Rather, juvenile convictions for any type of crime were predicted by FSS, age 10 hurtful behavior and IH, and age 12 uncaring behavior; juvenile convictions for a violent crime were predicted by FSS, age 10 IH, and age 12 uncaring and hurtful behavior. One previous study³⁴ compared arrests to age 33 of males and females who had and who had not sustained a TBI by age 7. The two groups were matched on poverty of the family of origin, maternal age, parental education level, maternal marital status, and race or ethnicity and gender. Consistent with our results, TBIs were not associated with an increased risk of arrest.

Sustaining a TBI prior to age 18 did not increase the risk of a conviction for any crime or for a violent crime from age 18 to age 24, regardless of whether participants had previous convictions. Rather, low FSS (defined as parents’ age at the birth of their first child, income and education, and whether or not they raised the participant) and childhood disruptive behaviors were consistently associated with increased risks of criminal convictions at all ages. A recent review⁵ asked whether TBIs were a cause of violent crime and reported that TBIs were a risk factor for “earlier, more violent, offending.” However, the review failed to distinguish studies that estimated the link between TBIs and offending after taking account of childhood behaviors that have been robustly associated with offending. Importantly, in our study, scores for hurtful behavior at age 6 were associated with an increased the risk of conviction for any type of crime, for a violent crime, and for a first conviction of any crime from age 18 to age 24. Age 6 uncaring behavior predicted a first conviction for any type of crime from age 18 to age 24. Similarly, the age 10 and age 12 scores for disruptive behaviors were associated with increases in the risk of convictions. These results are consistent with those from many studies showing that low FSS and childhood disruptive behaviors predict subsequent criminal offending.^17–25,29

Our measure of violent criminal convictions would not have captured all incidents of aggressive behavior. Two prior studies of children and adolescents observed high rates of aggressive behavior within 6 months³⁵ and 2 years³⁶ following a TBI. By contrast, a recent study of 103 individuals who had sustained a first TBI that included loss of consciousness, a Glasgow Coma Score of 15 or less, evidence of contusion or hemorrhage, and who required inpatient treatment reported that none of the participants engaged in physically aggressive behavior in the first 3 months after the TBI, two did in the next 3 months, and one did in the following 6 months.³⁷ Several studies have reported that aggressive behavior following a TBI was not predicted by the TBI but rather by factors present prior to the TBI, including aggressive behavior, arrest, substance misuse, lower level of education, mood disturbance, suicidality, and PTSD.^16,38–40

Strengths and Limitations

Physicians were required to record a diagnosis each time they saw a patient. Our coding of TBIs diagnosed within a period of 30 days as one TBI may have led to an underestimation of the number of TBIs. This may be the reason why we found no association between the number of TBIs and convictions. However, presence or absence of TBI at a specific age period was entered into regression models as a predictor of convictions. Further, it may be that diagnosing TBIs has changed over the decades during which the study took place. No information was available about the site and severity of injury or about neuropsychological symptoms and behaviors subsequent to the injury. It is possible that TBIs of a specific severity, in brain areas linked to antisocial and aggressive behavior,⁴¹ may lead to changes in cognition, emotion processing, and impulsivity that do increase the risk of criminality. Lastly, our analyses only considered criminal convictions up to age 24. Given that it is possible for individuals to commit a first offense after age 24, additional studies with longer follow-up periods are necessary. Strengths of the study include the use of health records to identify TBIs and official criminal records to document adolescent and adult convictions for a sample of 724 men. Additionally, information was available on social status of participants’ families in childhood and ratings of participants’ behavior at ages 6, 10, and 12 by different classroom teachers. Studying a sample enriched with offenders ensured sufficient statistical power to robustly identify links, or the absence of links, between TBIs and criminal convictions.

Conclusions

Using data collected prospectively to age 24 for 724 males, health file diagnoses of TBIs, and official records of criminal convictions from age 12 to age 24, there was no evidence that sustaining a TBI increased the subsequent risk of a conviction when regression models included family social status and scores for childhood disruptive behaviors. However, these childhood characteristics may explain, at least in part, the finding that offenders were more likely than nonoffenders to sustain TBIs from age 18 to age 24.