Sensation-Seeking, Criminality, and Spinal Cord Injury: A Case

American Journal erf EpidemiologyCopyright O 1996 by The Johns Hopkins University School of Hygiene and Public HealthAll rights reserved

Vol. 144, No. 5Printed in U.S.A

Sensation-Seeking, Criminality, and Spinal Cord Injury: A Case-ControlStudy

Anthony R. Mawson,1-2 Joseph J. Biundo, Jr.,3 Dorothy I. Clemmer,4 Keith W. Jacobs,56

Virginia K. Ktsanes,4 and Janet C. Rice4

A retrospective case-control study was performed in New Orleans, Louisiana, in 1985-1986 to test thehypotheses that 1) criminality is a risk factor for severe injury, and 2) the association between criminality andinjury can be explained in terms of a common underlying factor—increased sensation-seeking tendencies. Atotal of 140 males with spinal cord injury were individually matched with 140 driver's license holders on age,race, sex, educational attainment, and zip code of residence and were interviewed by telephone. Criminalityprior to spinal cord injury was measured by seff-report and police records, and sensation seeking wasmeasured by the Disinhibition and Boredom Susceptibility subscales of Zuckerman's Sensation-SeekingScale (Form V). Those with spinal cord injuries were significantly more likely than controls to report a historyof juvenile delinquency, adult criminality, and incarceration prior to the time of spinal cord injury. Statisticallysignificant but modest differences were also found between cases and controls with respect to Disinhibition,Boredom Susceptibility, and the combined Sensation-Seeking Scale score. Matched-pairs logistic regressionanalysis indicated that the association between sensation seeking and spinal cord injury remained significantafter controlling for criminality, with an estimated relative risk of 2.05 (95% confidence interval 1.67-2.53).However, the association between criminality and spinal cord injury also remained significant after controllingfor sensation seeking (estimated relative risk = 2.04, 95% confidence interval 1.09-3.82). On the basis of theseresults, criminality and sensation seeking may be statistically significant but independent predictors of spinalcord injury. Am J Epidemiol 1996;144:463-72.

behavior; crime; spinal cord injuries

This investigation was prompted by the informalobservation that a high proportion of patients withspinal cord injuries admitted to the Louisiana Reha-bilitation Institute, New Orleans, Louisiana, had his-tories of criminality and/or narcotics addiction. Re-ports from rehabilitation centers serving other inner-city populations (1, 2) suggested a similar patientprofile. However, since patients with spinal cord inju-ries at the Institute are mostly young, indigent, andmale—the population groups at greatest risk of crim-

Receivecl for publication December 18, 1995, and in final formMay 31, 1996.

Abbreviations: Cl, confidence interval; OR, odds ratio; SD, stan-dard deviation.

1 Post Office Box 1329, Abita Springs, LA2 Former address: Department of Medicine, Louisiana State Uni-

versity Medical Center, New Orleans, LA3 Section of Physical Medicine and Rehabilitation, Department of

Medicine, Louisiana State University Medical Center, New Orleans,LA

4 Department of Biostatistlcs and Epidemiology, School of PublicHealth and Tropical Medicine, Tulane University, New Orleans, LA

5 Department of Psychology, Loyola University, New Orleans, LA"Deceased.Reprint requests to Dr. Anthony R. Mawson, P. O. Box 1329,

Abita Springs, LA 70420.

inality—many would be expected to have histories ofcriminality. Common sense might suggest that theapparent link between criminality and spinal cord in-jury in this population is due to the fact that patientsare injured while engaging in crime (3). However, areview of the literature indicated that adult criminalsand juvenile delinquents are at increased risk of injuryin general, not only crime-related injury. Moreover,juvenile delinquents are at increased risk of injuryprior to becoming delinquent (4) and not only ofdeliberately inflicted injuries (5). Conversely, peoplewith a history of "accidents" are at increased risk ofsubsequent criminality (5-7). The literature thus sug-gests a complex association between criminality andinjury.

We hypothesize that increased vulnerability to in-jury and to criminality are related to a common factor,increased sensation-seeking behavior, which manifestsitself very early in life. It is known that criminality isrelated to sensation seeking (8). We suggest that in-creased sensation-seeking behavior, manifested as atendency to place oneself in potentially hazardoussituations, heightens the risk of occurrence of either

463

Dow

nloaded from https://academ

ic.oup.com/aje/article/144/5/463/66509 by guest on 02 M

arch 2022

464 Mawson et al.

criminality or injury, with the actual outcome depend-ing on situational exigencies and/or opportunities forobtaining sensory stimulation.

The purpose of this investigation was to test thehypotheses that spinal cord injury is associated withcriminality and sensation seeking and that a tendencytoward increased sensation seeking accounts for theassociation between criminality and spinal cord injury.

MATERIALS AND METHODS

Study population

Cases eligible for the study were individuals withacute, traumatic spinal cord injuries admitted to theLouisiana Rehabilitation Institute between January 1,1965, and December 31, 1984, who were male, En-glish speaking, and currently alive. Only those patientswho could be reached by telephone and who consentedto be interviewed were included. The study wasrestricted to English-speaking males because therewere too few females to allow analysis and becausethe study involved English-speaking telephoneinterviewers.

Controls were selected from holders of Louisianadriver's licenses, matched with individual cases onage, sex, race, zip code of residence, and educationalattainment. Computer tapes, obtained free of chargefrom the Louisiana Department of Public Safety, con-tained the names, current addresses, dates of birth,sexes, and races of all Louisiana driver's license hold-ers. A COBOL program was written (by K. W. J.) toselect every individual who matched one or more ofthe cases on age, race, sex, and zip code of residence.Potential controls were further screened by telephoneto ensure that cases and controls were matched oneducational level (high school graduation or its equiv-alent and beyond high school graduation).

Definitions and measures

Spinal cord injury was defined as "an acute, trau-matic lesion of the spinal cord, including trauma tonerve roots that results in sensory or motor deficit, orboth" (9, p. 126). Injuries due to chronic degenerative,neoplastic, bone, and central nervous system diseaseswere excluded. The diagnosis of spinal cord injury andthe level of lesion were accepted as recorded by anattending physician in the hospital admission or dis-charge summary. The term "paraplegia" refers to pa-ralysis of the lower limbs, "quadriplegia" refers toparalysis of all four limbs, with a "complete" lesionbeing one in which there is a total loss of sensation andmotor ability below the level of the lesion. The level ofthe spinal cord injury is classified in terms of the

associated spinal vertebrae, i.e., C (cervical), T (tho-racic), L (lumbar), and S (sacral).

Sensation-seeking preferences were measured bytwo subscales of the Sensation-Seeking Scale (FormV) of Zuckerman (8): Disinhibition and Boredom Sus-ceptibility, with each subscale consisting of 10 binary-choice items. To control for the possible effect ofspinal cord injury on sensation-seeking preferences,we asked cases and their matched controls to respondto the questions as they felt they would have before thedate of the case's spinal cord injury. Estimates ofinternal consistency (split-half reliability) of the totalscore of the Form V Sensation-Seeking Scale rangedfrom 0.83 to 0.86 (8). The 10-item Disinhibition andBoredom Susceptibility subscales had internal consis-tencies for US males of 0.74 and 0.57, respectively.The test-retest reliability of Form V was 0.94 for thetotal score at 3 weeks and 0.91 and 0.71 for Disinhi-bition and Boredom Susceptibility, respectively, at thesame 3-week intervals. The decision to use Disinhibi-tion was based on the fact that it is the subscale mostclosely associated with psychopathy and hence withcriminality (8). The Boredom Susceptibility subscalereflects "an aversion for repetitive experiences of anykind, routine work, or dull and boring people andextreme restlessness under conditions when escapefrom constancy is impossible" (8, p. 103).

Criminality was measured by self-report and officialrecords. Subjects were asked if they had ever beenarrested, convicted of a crime, or placed in a correc-tional institution at age 16 years or younger and at age17 or older, but before the date of their spinal cordinjury (or, for controls, before the date of the spinalcord injury of the patient with whom they had beenmatched). To supplement the self-report data, criminalarrest records were supplied by the New Orleans Po-lice Department for consenting cases and controls whohad always lived in Orleans Parish.

Data collection

Demographic, social, and diagnostic information onpatients with spinal cord injuries was abstracted fromthe medical records of the Louisiana RehabilitationInstitute and its affiliated Charity Hospital. Attemptsto contact the cases began with telephone calls andwritten inquiries to the telephone numbers and ad-dresses listed in the medical records and continuedwith a variety of methods for those who were notreached. Written requests for information were madeto the Department of Social Security, the LouisianaPower and Light Company, and New Orleans PublicService, Incorporated (all of which were denied) andto the Louisiana Department of Corrections, whichresulted in locating one patient at a state correctional

Am J Epidemiol Vol. 144, No. 5, 1996

Dow



arch 2022

Sensation-Seeking, Criminality, and Spinal Cord Injury 465

facility who subsequently completed the interview.Patients admitted while the project was under waywere contacted upon admission.

From the computer list of 2.5 million driver's li-cense holders, 31,655 potential controls were identi-fied who matched the cases on year of birth, race, sex,and zip code. The number of potential controls foreach case ranged from zero to 513. There were 31instances in which two cases matched on the abovecriteria, four instances in which three cases matched,and 392 unique cases. For cases that were identical onthe matching variables, the available controls weredivided equally among the cases.

A standardized procedure (available from the firstauthor upon request) was used to select 25 or fewerpotential controls. For cases who lacked potential con-trols, alternate controls were selected by identifyingother patients from the same zip code region whosebirth was within 5 years of that of the case in question.Potential controls who had originally been selected forthese patients were then assigned to the patients wholacked them. For comparability with regard to expo-sure to criminality and sensation seeking before thedate of the spinal cord injury, controls who were not ofthe same age as the case were given a reference date(i.e., an upper month/year cutoff point) that wouldhave placed them at the same age as the patient whenthe latter was injured; the controls were then askedabout their sensation-seeking preferences and crimi-nality before the date in question.

The interviews were conducted by the first authorand 12 part-time student assistants, all but one ofwhom were female. Patients and controls were tele-phoned and invited to participate in the study. Theywere told that the interview would take about 20minutes and assured of the confidentiality of theirreplies. A consent form, approved by the University'sInstitutional Review Board, was read to them.

Since telephone numbers are not provided on thedriver's license record, the telephone numbers of thecontrols were obtained from directories in the NewOrleans Public Library. Interviewers began callingpotential controls when 6-20 telephone numbers hadbeen found. The first person who completed the inter-view served as the matching control. When the tele-phone call to a potential control was answered, theinterviewer began by verifying that the speaker wasthe person whose name was listed as the control andthen outlined the nature of the study and screened thespeaker on educational attainment prior to the date ofthe case's spinal cord injury.

To check the validity of self-reports, a letter wassent to cases and controls who said that they hadalways lived in Orleans Parish requesting written per-

mission to examine their criminal record (if any) withthe New Orleans Police Department. Enclosed withthe letter was a stamped addressed envelope and twocopies of a consent form, one of which the subjectswere asked to sign and return.

Statistical analyses were performed using standardsoftware packages (10, 11) on a DEC-2060 computer(Digital Equipment Corporation, Framingham, Massa-chusetts). Odds ratios and confidence intervals werecomputed to compare the extent of criminality amongpersons with spinal cord injuries and controls prior tothe date of spinal cord injury. After we determinedwhether patients differed from controls on the sub-scales of the Sensation-Seeking Scale by usingStudent's t test, we tested the hypothesis that theexpected association between sensation seeking andspinal cord injury would remain significant after con-trolling for the effect of criminality using matched-pairs logistic regression (12).

The number of pairs needed to provide an adequatesample size for the survey was calculated using ap-propriate formulae for a matched-pairs case-controlstudy. A total of 136 pairs were required for the surveyto detect a threefold increased risk of criminality witha statistical power of 90 percent in the patients withspinal cord injuries compared with the controls.

RESULTS

A total of 526 persons with spinal cord injuries wereadmitted to the hospital during the 20-year study pe-riod, of whom 453 (86 percent) were males. Of thesemales, 169 (37 percent) were lost to follow-up, and 14(3 percent) were living outside Louisiana; hence, bothgroups were ineligible for inclusion. Of the remaining270 males, 91 (34 percent) could not be contacted; 55of the 91 were known to be alive but could not bereached despite numerous attempts to contact them, 24were known to be deceased, and 12 were too ill at thetime of the study to participate. Of the 179 patientsactually contacted, 28 (16 percent) refused, and 151patients gave their informed consent. Eleven of the151 were interviewed but did not have a matchingcontrol. The remaining 140 male patients were inter-viewed and formed the population for the survey. Atotal of 140 matching Louisiana driver's license hold-ers were contacted and interviewed. On average, sixpersons were contacted before a successful match wasfound.

Of the 140 male patients interviewed, over half (56percent) were admitted to the hospital after January 1,1980. The mean age at injury was 28.6 years (standarddeviation (SD) = 11.0 years), and the majority (55percent) were black. Diagnostically, the greatest pro-portion (40 percent) consisted of complete paraplegics


Dow



arch 2022

466 Mawson et al.

injured between spinal levels Tl and T12. Fifty-fivepercent of the cases had not graduated from highschool, and 75 percent had received no additionaleducation since their spinal cord injury (14).

To ensure that the survey participants were repre-sentative of the male patients with spinal cord injuriesadmitted to the Louisiana Rehabilitation Institute (i.e.,to compare respondent and nonrespondent cases), the140 respondents were compared with 1) all 453 malepatients minus the 140 interviewed, and 2) nonpartic-ipant but study-eligible males who were living inLouisiana who were not lost to follow-up and whowere known to be alive, less the 140 survey partici-pants (n = 106). No significant differences (p < 0.05)were found for either comparison with respect to race,education, diagnosis, or age of onset of the spinal cordinjury.

Criminality

Self-reported criminality was significantly morecommon among those with spinal cord injuries thanamong matched controls prior to spinal cord injury(table 1). The cases were more Likely than the controlsto report being arrested by the police before age 17years (29 vs. 17 percent; odds ratio (OR) = 2.06, 95percent confidence interval (CI) 1.05-4.12), beingconvicted (14 vs. 6 percent; OR = 2.50, 95 percent CI0.97-6.89), and being placed in a correctional institu-tion before age 17 years (14 vs. 4 percent; OR = 3.33,95 percent CI 1.17-10.75). Similarly, at age 17 yearsor older, prior to injury, the cases were significantlymore likely than the controls to have been arrested (48

vs. 19 percent; OR = 4.55, 95 percent CI 2.19-9.92),convicted (24 vs. 5 percent; OR = 5.33, 95 percent CI1.99-16.38), and placed in a correctional institution(25 vs. 3 percent; OR = 31.0, 95 percent CI 30.86-38.33).

The odds ratios are significant in all but one com-parison, convictions before age 17 years; they aregreater in the comparisons for those over age 17 thanin the comparisons for those under 17, and they in-crease progressively within each age/group as compar-isons are made in terms of arrests, convictions, andimprisonment.

Former patients were also interviewed about thecircumstances of their spinal cord injury. One questionwas whether the subject was committing a crime at thetime of his spinal cord injury. Contrary to what hadbeen expected, only seven (5 percent) responded in theaffirmative. Of these, four said they were dealing orbuying drugs, one was "drinking and driving," onewas gambling, and one was committing a robbery.Among the 65 cases with self-reported histories ofcriminal arrest, only six (9.2 percent) said they werecommitting a crime at the time of their spinal cordinjury. Hence, the association between criminality andspinal cord injury cannot be explained simply as adirect outcome of criminal behavior.

Official arrest data

We had originally hoped to compare the self-reportdata on criminality with official arrest data supplied bythe New Orleans Police Department. Unfortunately,this was impossible, as only 12 respondents (nine of

TABLE 1. Paired comparisons of spinal cord-injured males and controls in terms of serf-reportedcriminality before and after age 17 years, prior to spinal cord Injury*, New Orleans, Louisiana, 1985-1986

vartaHe CasesControls

YBS NoOdds ratio

95%confidence

Interval

Before age 17 yearsArrests

Convictions

Incarcerations

Age 17 years or olderArrests

Convictions

Incarcerations

"resNo

YeaNo

No

YesNo

No

YesNo

716

08

06

1511

16

31

3383

20112

20114

5059

3296

31100

2.06

2.50

3.33

4.55

5.33

31.00

1.05-4.12

0.97-6.89

1.17-10.75

2.19-9.92

1.99-16.38

30.86-38.33

* Odds ratios for matched pairs are calculated by using the tvro discordant pairs of cells only in each of the2 x 2 tables.


Dow



arch 2022


34 cases and three of 40 controls) returned their con-sent forms. No police record could be found for five ofthe six who reported no arrests, while reports wereconfirmed for four of the six patients who said theyhad been arrested at least once.

Sensation seeking

The literature indicates that criminality is associatedwith an increased risk of injury as a whole, not just ofinjuries linked to violent crime, and that the risk existsfrom an early age (4-6). This suggests that some asyet unidentified factor accounts for both the increasedrisk of criminality and the increased risk of injury. Oursuggestion is that this factor may be a proclivity to-ward high levels of sensory stimulation. If this hypoth-esis is correct, 1) persons with spinal cord injurieswould be expected to have higher Sensation-SeekingScale scores than would matched controls; 2) personswith spinal cord injuries who had criminal back-grounds would be expected to have higher Sensation-Seeking Scale scores than those lacking such back-grounds; and 3) if increased sensation-seekingtendencies explain the association between criminalityand spinal cord injury, the association between sensa-tion seeking and spinal cord injury would be expectedto remain significant after controlling for the effect ofcriminality.

We turn now to the evidence for each of these threepredictions.

1. Paired t test analysis indicates that those withspinal cord injuries and controls differ signifi-cantly with respect to Disinhibition (mean differ-ence = 1.057, p = 0.0003), Boredom Suscepti-bility (mean difference = 0.529, p = 0.007), andthe two subscales combined (mean difference =1.6, p = 0.0001).

2. Two sample t tests (table 2) indicate that personswith spinal cord injuries who had self-reportedhistories of arrest at age 17 years or older, beforetheir spinal cord injury, had significantly higherDisinhibition scores than the did those who were

not arrested {p = 0.04), but did not differ sig-nificantly on Boredom Susceptibility from thosewho were not arrested {p = 0.45). Arrestees alsohad higher scores on the combined Sensation-Seeking Scale, but the difference (1.01 units) wasonly marginally significant {p = 0.07).

3. Matched-pairs logistic regression analysis wasused to test the hypothesis that sensation seekingaccounts for the association between criminalityand spinal cord injury by determining whetherthe association between sensation seeking andspinal cord injury remained significant after con-trolling for the effect of criminality. The predic-tor variables were 1) other injury before thespinal cord injury of sufficient severity to resultin an emergency room visit or hospitalization; 2)self-reported arrests at age 17 years or older,before the spinal cord injury; and 3) the totalSensation-Seeking Scale score (Disinhibition andBoredom Susceptibility combined).

Table 3 presents the strength of the associationbetween each predictor (other severe injury, criminalarrest, sensation seeking) and spinal cord injury, withand without adjustment for the effect of the otherpredictors. The unadjusted analysis shows that, withrespect to other severe injury prior to spinal cordinjury, the difference between cases and controls didnot attain statistical significance (OR = 1.52, 95 per-cent CI 0.92-2.52). On the other hand, cases andcontrols differed significantly in self-reported arrests(OR = 2.13, 95 percent CI 1.17-3.85) and sensationseeking (OR = 2.15, 95 percent CI 1.75-2.64). Theodds ratios for sensation seeking are based on a 1standard deviation (SD) difference of approximatelyfive units on the 20-unit combined scale.

Regarding the association between each predictorvariable and spinal cord injury, after adjustment forthe effect of the other variables, the main hypothesis ofthe study was supported by the finding that the asso-ciation between sensation seeking and spinal cordinjury remained significant after controlling for the

TABLE 2. Sensation-Seeking Scale scores of arrested and never-arrested spinal cord4n|ured males,New Orleans, Louisiana, 1985-1986

Variable

Disinhibition

Boredom Susceptibility

Combined Sensation-Seeking Scalef 8.29

* SD, standard deviation,t Zuckerman (8).

Mean

4.86

3.43

8.29

Arrested(/I = 65)

(SD)«

(2.09)

(1.71)

(3.27)

Never arrested

Mean

4.07

3.21

7.28

(SD)

(2.36)

(1.74)

(3.29)

Pvalue

0.04

0.45

0.07


Dow



arch 2022

468 Mawson et al.

TABLE 3. Strength of association between each predictorvariable (prior injury, criminal arrest*, and sensation seeking)and spinal cord injury, by multiple logistic regressionanalysis, with and without adjustment for the effect of otherpredictors, New Orleans, Louisiana, 1985-1986

Variable

Injury (yea/no)

Criminal arrest (yes/no)

Sensation seeking(per 5-unit increasein a 20-unit scale)*

• Zuckerman (8).

Unadjustedoddsratio

1.52

2.13

2.15

Adjusted

ratio

1.66

2.04

^05

95%confidence

Interval

0.95-2.89

1.09-3.82

1.67-2.53

effect of criminality. Thus, given a mean difference of1 SD (4.56 units on the Sensation-Seeking Scale), theestimated relative risk of experiencing a spinal cordinjury was calculated as the exponential function ofthe regression coefficient, which was 2.05 (95 percentCI 1.67-2.53). However, the association betweencriminality and spinal cord injury also remained sig-nificant after controlling for sensation seeking; that is,the estimated relative risk of experiencing a spinalcord injury given one or more arrests at age 17 yearsor older was 2.04 (95 percent CI 1.09-3.82). In sum,a mean difference of 1 SD on the combined Sensation-Seeking Scale or being arrested one or more times atage 17 years or older were both associated with aslightly more than twofold increased risk of having aspinal cord injury.

DISCUSSION

Prompted by several lines of evidence suggesting anassociation between criminality and an increased riskof injury (3-7), a retrospective case-control study wasundertaken to find out whether this association existedfor persons with spinal cord injury. We also sought totest the hypothesis that the association between crim-inality and spinal cord injury was due to their commonassociation with increased sensation-seeking tenden-cies. Consistent with previous reports of an associationbetween criminality and injury (3-7), in our studypersons with spinal cord injuries were significantlymore likely than the controls to report being arrested,convicted of a crime, and incarcerated before the timeof the spinal cord injury, both before and after age 17years. Indeed, the cases were 31 times more likely tohave been incarcerated than were the controls after age17 and prior to spinal cord injury. The number ofconsenting subjects was too low to allow for case-control comparisons of official arrest records.

In contrast to a report that found no associationbetween sensation seeking and spinal cord injury butthat was based on a relatively small sample and did notuse a control group (15), our study found that thosewith spinal cord injuries scored significantly higherthan the controls on both the Disinhibition and Bore-dom Susceptibility subscales of the Sensation-SeekingScale. Persons with spinal cord injuries arrested at age17 years or older, before their injury, also had signif-icantly higher Disinhibition scores than did the non-arrested, thus confirming previous findings of an as-sociation between self-reported as well as officialcriminality and the Sensation-Seeking Scale, particu-larly Disinhibition (16-21).

Logistic regression analysis indicated that criminal-ity and sensation seeking were both significant butindependent predictors of spinal cord injury (i.e., bothfactors remained significant after controlling for theeffects of the other).

Potential limitations

Selection of appropriate controls. Several optionsfor control groups were considered during the plan-ning stages of the survey, including neighborhoodcontrols, school controls, hospital controls, and ran-dom digit dialing, but all were rejected for variousreasons. Louisiana Department of Public Safety driv-er's license records seemed most feasible for our pur-poses, particularly as we were able to obtain all 2.5million records without charge. An attempt was madeto control for socioeconomic status by matching casesand controls on educational attainment; income andoccupation were inappropriate indicators since the vastmajority of patients had only their disability incomeand were unemployed.

It could be argued that possession of a driver'slicense is itself a basic indicator of socioeconomicstatus. To check the comparability of cases and con-trols in this regard, a follow-up telephone call wasmade to all patients after the interviews had beencompleted. The results showed that only 13 percent ofthose recontacted (n = 124) did not have a driver'slicense at the time of their injury. Thus, the cases andcontrols seemed roughly comparable in terms of edu-cational attainment and the possession of a driver'slicense.

A potentially more serious problem with regard tocomparability was the fact that to be selected as acontrol the individual had to have a personal telephonenumber, which implies some degree of social stability.Although no attempt was made to determine whatproportion of cases had a telephone listed in theirname before the spinal cord injury, at the time ofsurvey it appeared that most patients did not have their


Dow



arch 2022


own telephone number. Thus, having a personal tele-phone could mean that the controls were more stableand therefore, presumably, at a lower risk of criminal-ity than were the cases. To eliminate this possibility, anew study would have to be carried out, restricted topatients with spinal cord injuries who had a telephonelisted in their name at the time of the injury.

Because of the known association between crimi-nality and injury (3), it was decided not to obtain acontrol group of strictly uninjured persons since thiswould have biased the study toward finding an asso-ciation between criminality and spinal cord injury. Forthis reason, a general population sample was selectedthat consisted of driver's license holders. The onlycriterion for excluding a potential control was that hehad been a patient with spinal cord injuries at theLouisiana Rehabilitation Institute.

Definition and ascertainment of exposure. Thequestion might be raised about whether asking peopleover the telephone about such sensitive matters ascriminal behavior would be expected to produce reli-able and valid results. There has been much discussionabout the relative merits of official versus self-reported data on criminality (22) and traffic violations(23, 24). Public records tend to underestimate theactual occurrence of crimes and traffic violations whencompared with self-reports, since only the more severeor serious incidents are reported and filed (24). On theother hand, public records are more reliable than self-reports for events occurring more than 5 years beforethe time of interview (24).

Self-reports have been found to be reliable and validfor measuring criminality among adolescents as agroup (22), but they are far less reliable as a measureof the volume of crime, as well as for black males andfor more serious delinquents, who are less likely toreport offenses known to the police. Although thevalidity coefficients for black males are smaller thanthose for white males, black males can be rankedamong themselves with marginal validity. Thus, "self-reports may have some utility in studying etiologicalquestions within black male populations ... [but inves-tigators] must stratify by race in sampling and inanalyzing self-report data" (22, p. 214). Self-reportsare also less useful when accurate counts of the vol-ume of crime are required; the strongest correlationswith official data are found with self-report measuresthat involve asking subjects whether or not they have"ever committed" a given crime. These points (22)were taken into account in our study insofar as casesand controls were matched on race and subjects wereasked whether they had ever been arrested, convictedof a criminal offense, or served time in a correctional

institution, rather than how many crimes of a particu-lar sort they had ever committed.

An attempt was made to check the validity of self-reported arrests by comparing these data with NewOrleans Police Department records for cases and con-trols who had always lived in Orleans Parish. Thenumber of consenting subjects (n = 12) was too smallfor valid comparisons, but the results of sensitivity andspecificity tests for validity were in the expected di-rection. The positive predictive value of the self-reports was 80 percent.

With regard to the reliability and validity of tele-phone interviews compared with personal surveys, thetelephone method is capable of producing generallycomparable data and at considerably less cost than thepersonal interview (25). One study in which the twomethods were compared indicated that the telephonemethod yielded a lower response rate but at less thanhalf the cost of personal interviews. Although therewas some evidence of nonresponse bias in the tele-phone survey and some relatively minor differences inresponses between the two methods, there was noconclusive evidence that these differences were due tothe mode of data collection (26).

The finding of significant differences in Sensation-Seeking Scale scores between persons with spinal cordinjuries and controls (27) is contrary to the uncon-trolled study by Ditunno et al. (15), but is consistentwith other literature suggesting a tendency towardextraversion and impulsivity in persons with spinalcord injuries, particularly those considered to haveplayed an active role in their injury and those whosebehavior was judged imprudent (28-30). Only two ofthe four subscales were used in this study, comparedwith all four in the study by Ditunno et al., but Dis-inhibition and Boredom Susceptibility both yieldedsignificant associations with spinal cord injury. Thevalidity of Disinhibition in particular is supported bythe fact that it distinguishes between arrested andnonarrested cases, as is found in previous research onthe association between the Sensation-Seeking Scaleand criminality (8). This scale is also said to be lessvalid for blacks than for whites (31, 32), and blackscomprised 55 percent of the present sample of personswith spinal cord injuries. Nevertheless, cases and con-trols were matched on race, thereby eliminating thepossibility of bias due to racial differences.

Although significant differences emerged on theSensation-Seeking Scale between cases and controls,the magnitude of the differences was generally mod-est, yielding an overall estimated relative risk of ex-periencing a spinal cord injury of 2.05 (95 percent CI1.67-2.53), given a mean difference of 1 SD (4.56units) on the sensation-seeking scale. Thus, if sensa-


Dow



arch 2022

470 Mawson et al.

tion seeking or some allied concept relating to risktaking, impulsivity, or extraversion accounts for theassociation between criminality and spinal cord injury,a more robust measure than the Sensation-SeekingScale is needed.

Suggestions for further research

The major hypothesis of this study was that certainindividuals may be at risk both of criminality andinjury from an early age because of a tendency, pos-sibly constitutional in origin, to be more active in theenvironment and intolerant of sensory restriction. Thisfactor, as measured by the Sensation-Seeking Scale ofZuckerman (8), proved to be associated with spinalcord injury after controlling for criminality, but thelatter was also associated with spinal cord injury aftercontrolling for sensation seeking.

Several options might be considered regarding thedevelopment of a possibly more powerful measure ofsensation-seeking tendencies. First, a new scale mightbe developed that avoids the use of value-laden termsand phrases, includes reference to the frequencies ofactual behaviors, and assesses aversions to sensoryrestriction as well as sensory preferences. Second, anexisting measure could be considered as a substitutefor the Sensation-Seeking Scale, e.g., extraversion, asmeasured by the Eysenck Personality Inventory (33),which is closely related to the concept of sensationseeking (34). A study using the Eysenck PersonalityInventory to assess 111 patients undergoing rehabili-tation (30) showed that traumatically disabled personswith spinal cord injury were significantly more extro-verted than persons disabled due to nontraumaticcauses. Third, certain presumed surrogates of sensa-tion seeking could be studied in relation to criminalityand spinal cord injury. For instance, the medical his-tories of cases and controls could be studied to deter-mine the prevalence of childhood hyperactivity in eachgroup; early school records could also be used todetermine the prevalence of behavioral problems andrelated indicators of excessive motor activity or im-pulsivity. Such data would suggest the presence inearly childhood of tendencies that potentially reflectsensation seeking and are known to be associated withjuvenile delinquency (35) as well as an increased riskof injury (3, 36, 37). The higher level of motor activitynoted in male children compared with female children(38) also suggests increased sensation-seeking tenden-cies in males; moreover, activity levels are associatedwith accidents (39), with males having higher rates ofvarious types of accidents than do females (36). Con-sistent differences between males and females in sen-sation seeking (8) may help to account for sex differ-ences in the rates of certain types of accidents. The

fact that rates of violent criminality and injury bothpeak in the late teens and early twenties (7) suggeststhat the presumed underlying causal factor also peaksat this time (40).

A more fundamental issue is the meaning and def-inition of the term sensation seeking itself. Organismshave a physiologic need for sensory stimulation (41,42), but it remains uncertain how sensation seekingrelates to other need-related behaviors. Zuckerman (8)considers sensation seeking to be a distinct form ofbehavior manifested in thrill-seeking activities, ten-dencies toward substance abuse, and preferences forspicy foods. Traditionally, eating, drinking, sex, andaggression have been considered separate systems,each with a relatively independent set of activatingconditions and physiologic mechanisms. An alterna-tive model proposed by the first author (A. R. M.) isthat stimulation seeking (a preferred term) is inherentin all forms of motor-motivational activity (43).Stimulation-seeking behavior is defined as any activitythat enhances or facilitates contact between an organ-ism's sensory receptors and external objects or sur-faces (44). Just as the color green represents an arbi-trarily defined wavelength on the spectrum of visiblelight, so the different forms of motivational behaviorrepresent arbitrarily defined and overlapping "bands"on a continuous spectrum of stimulation-seeking be-havior. Thus, activities such as eating and drinking arestimulation-seeking behaviors of relatively mild inten-sity, while sex, aggression, flight, and self-mutilationrepresent bands of stimulation seeking at progressivelyhigher levels. According to this view (43), involve-ment in criminality, substance abuse, and an increasedrisk of injury are varying manifestations of a chroni-cally elevated level of stimulation-seeking behavior.Consistent with this broader concept of sensation-seeking behavior, there is evidence that, in addition tocriminality, persons with spinal cord injury are also atincreased risk of alcoholism and other forms of sub-stance abuse prior to injury (45, 46).

With regard to other avenues for research on cau-sation, it would be of interest to explore the role ofacute environmental conditions such as situationalstress and the availability of supporting social relation-ships. A large volume of literature attests to the im-portance of such factors to mental and physical health(47-49).

The results of our study indicate that criminality andsensation seeking are independently associated withspinal cord injury. About 50 percent of those withspinal cord injuries reported being arrested by thepolice at age 17 years or older, compared with only 19percent of the controls, and 23 percent reported beingimprisoned at age 17 or older, prior to injury, com-


Dow



arch 2022


pared with 3 percent of the controls. In terms of riskestimates, those with spinal cord injuries were nearlyfive times as likely to have been arrested as an adultand 31 times as likely to have been imprisoned as werethe controls. With regard to sensation seeking, patientsand controls differed on average by 1.6 units on thecombined Sensation-Seeking Scale. Given a 4.5-unitdifference in score (corresponding to 1 SD), the esti-mated relative risk of spinal cord injury, after adjust-ment for criminality, was 2.05 (95 percent CI 1.67-2.53).

From diese observations, one might infer either thatthe Sensation-Seeking Scale is insufficient as a mea-sure of stimulation-seeking preferences or that crimi-nality may in fact be an independent predictor ofspinal cord injury. Reservations have been noted aboutthe Sensation-Seeking Scale, and suggestions havebeen made for improving it. Even if the association ofcriminality with spinal cord injury is independent ofstimulation seeking, the question still remains unre-solved about how criminality relates to spinal cordinjury. There may well be a common antecedent factorthat accounts for the association (40), but that is notbeing adequately "tapped" by the Sensation-SeekingScale.

Some have argued that the search for behavioralantecedents of injury should be accorded low prioritysince it had little potential for being translated intopractical countermeasures (37). The focus should in-stead be on environmental factors, since they are moreoften amenable to change (50) and influence injuryrates irrespective of individual differences in suscep-tibility. Others (36) have suggested that behavioralstudies of injury will receive greater attention andrecognition after more obvious changes have beenmade to the physical environment to reduce the risk ofinjury (e.g., firearms control, air bags in motor vehi-cles). Interestingly, while the emphasis in injury epi-demiology has switched from psychosocial to environ-mental factors (50, 51), the reverse is true for chronicdiseases, where increasing emphasis is now beingplaced on behavioral contributions to disease preven-tion (36). While environmental changes have thegreatest potential for reducing injuries in the generalpopulation, the need remains to develop explanatorymodels of injury; to identify risk factors for severeinjury; and to develop psychosocial, educational, orother types of measures aimed at reducing such risksin individuals (52).

ACKNOWLEDGMENTS

Supported by a grant from the US Department of Educa-tion to Joseph J. Biundo, Jr., Principal Investigator.

The authors are indebted to Deborah Bates, Liz Berzas,Art Borja, Lisa Gansar, Doris Gautier, Jeanette Hill, JulieLevine, Delia Penalba, Diana Webster, and BronwynWilliams for assistance with data collection and to YvetteWinchester and Jeffrey Williams for assistance with dataanalysis. They also thank the Louisiana Department ofPublic Safety for providing data tapes on Louisiana driver'slicence holders.

An earlier version of this paper was presented at theAnnual Meeting of the American Public Health Associa-tion, Las Vegas, Nevada, October 1986.

REFERENCES

1. Khella L, Stoner EK. 101 cases of spinal cord injury. 15-yearfollow-up study in a large city hospital. Am J Phys Med1977;6:21-32.

2. Macauley CA, Weiss L. Spinal cord injury in an inner-cityhospital. Arch Phys Med Rehabil 1978,59:76-80.

3. Waller JA. Accidents and violent behaviors: are they related?In: Mulvhill DJ, Tumin MM, eds. Crimes of violence. Vol. 13.Appendix 33. A staff report submitted to the National Com-mission on the Causes and Prevention of Violence. Washing-ton, DC: US GPO, 1969:1525-58.

4. Palfrey JS, Kamiski W, Clarke S, et al. Health profiles of earlyadolescent delinquents. Public Health Rep 1983;98:449-57.

5. Glueck S, Glueck E. Delinquents and nondelinquents in per-spective. Cambridge, MA: Harvard University Press, 1968.

6. Lewis DO, Shanok SS. Medical histories of delinquent andnondelinquent children: an epidemiological study. Am J Psy-chiatry 1977; 134:1020-5.

7. Wilson M, Daly M. A lifespan perspective on homicidalviolence: the young male syndrome. In: Block CR, Block R,eds. Questions and answers on lethal and nonlethal violence.Washington, DC: National Institute of Justice, 1993:29-38.(NIJ publication no. 147480).

8. Zuckerman M. Sensation-seeking: beyond the optimal level ofarousal. Hillsdale, NJ: L. Erlbaum Associates, 1979.

9. Kewalramani LS, Kraus JF. Acute spinal-cord lesions fromdiving—epidemiological and clinical features. West J Med1977;126:353-61.

10. Dixon WJ, Brown MB, Engelman L, et al., eds. BMDPstatistical software. Berkeley, CA: University of CaliforniaPress, 1983.

11. Nie N, Hull C, et al., eds. Statistical package for the socialsciences (SPSS). New York, NY: McGraw-Hill Book Co.,1975.

12. KJeinbaum DG, Kupper LL, Morgenstem H. Epidemiologicresearch: principles and quantitative methods. Belmont, CA:Lifetime Learning Publications, 1982.

13. Schlesselmann JJ. Case-control studies: design, conduct, anal-ysis. New York, NY: Oxford University Press, 1982.

14. Mawson AR, Snelling L, Winchester Y, et al. 526 spinal cordinjuries: experience of the Louisiana Rehabilitation Institute,1965-1984. J La State Med Soc 1991 ;143:31-7.

15. Ditunno P, McCauley C, Marquette C. Sensation-seeking be-havior and the incidence of spinal cord injury. Arch Phys MedRehabil 1985;66:152-5.

16. Farley FH, Farley SV. Stimulus-seeking motivation and de-linquent behavior among institutionalized delinquent girls. JConsult Clin Psychol 1972;39:140-7.

17. Farley FH. A theory of delinquency. Paper presented at theAnnual Meeting of the American Psychological Association.Montreal, Quebec, Canada, 1973.

18. Leblanc RF, Tolor A. Alienation, distancing, externalizing andsensation-seeking in prison inmates. J Consult Clin Psychol1972;39:514.


Dow



arch 2022

472 Mawson et al.

19. Farley FH, Sewell T. Test of an arousal theory of delinquency.Criminal Justice Behav 1976;3:315-20.

20. Perez J, Torrubia R. Sensation-seeking and antisocial behaviorin a student sample. Personality Individual Differences 1985;6:401-3.

21. White HR, Labouvie EW, Bates ME. The relationship be-tween sensation-seeking and delinquency: a longitudinal anal-ysis. J Res Crime Delinquency 1985;23:197-211.

22. Hindelang MJ, Hirschi T, Weis JG. Measuring delinquency.Beverly Hills, CA: Sage, 1981.

23. Zylman R. Driver's records: are they a valid measure ofdriving behavior? Accid Anal Prev 1972;4:333-49.

24. Smith DI. Official driver's records and self reports as a sourceof accident and conviction data for research purposes. AccidAnal Prev 1976,8:207-11.

25. Aneshensel CS, Frerichs RR, Clark VA, et al. Measuringdepression in the community: a comparison of telephone andpersonal interviews. Public Opinion Q 1982;72:1017-21.

26. Weeks MF, Kulka RA, Lessler JT, et al. Personal versustelephone surveys for collecting household data at the locallevel. Am J Public Health 1983;73:1389-94.

27. Mawson AR, Jacobs KW, Winchester Y, et al. Sensation-seeking and traumatic spinal cord injury: case-control study.Arch Phys Med Rehabil 1988;69:1039-43.

28. Fordyce W. Personality characteristics of men with spinalcord injury as related to manner of onset of disability. ArchPhys Med Rehabil 1964;45:321-5.

29. Nehemkis A, Groot H. Indirect self-destructive behavior inspinal cord injury. In: Farberow N, ed. The many faces ofsuicide. New York, NY: McGraw-Hill Book Co., 1981:99-115.

30. Malec J. Personality factors associated with severe traumaticdisability. Rehab Psychol 1985 ;30:165-71.

31. Karoly P. Comparison of "psychological styles" in delinquentand nondelinquent females. Psychol Rep 1975;36:567-7O.

32. Kaestner E, Rosen L, Appel P. Patterns of drug abuse: rela-tionships with ethnicity, sensation-seeking and anxiety. J Con-sult Clin Psychol 1977;45:462-8.

33. Eysenck HJ, Eysenck SBG. Manual for the Eysenck Person-ality Inventory. San Diego, CA: Educational and IndustrialTesting Service, 1968.

34. Eysenck S, Zuckerman M. The relationship betweensensation-seeking and Eysenck's dimensions of personality.

Br J Psychol 1978;69:483-7.35. Rutter M, Giller H. Juvenile delinquency: trends and perspec-

tives. New York, NY: Penguin Books, 1983.36. Matheny AP Jr, Fisher JE. Behavioral perspectives on chil-

dren's accidents. Adv Dev Behav Pediatr 1984;5:221-64.37. Langley J. Injury control—psychosocial considerations. J

Child Psychiatr Psychol 1984;25:349-56.38. Maccoby EE, Jacklin CN. The psychology of sex differences.

Stanford, CA: Stanford University Press, 1974.39. Manheimer DI, Mellinger GD. Personality characteristics of

the child accident repeater. Child Dev 1967;38:491-514.40. Piazza PV, Deroche V, Deminiere JM, et al. Corticosterone in

the range of stress-induced levels possesses reinforcingproperties: implications for sensation-seeking behaviors. ProcNatl Acad Sci U S A 1993;9O:11738-42.

41. Zubek JP, ed. Sensory deprivation: fifteen years of research.New York, NY: Appleton-Century-Crofts, 1969.

42. Riesen AH, ed. The developmental neuropsychology of sen-sory deprivation. New York, NY: Academic Press, Inc., 1975.

43. Mawson AR. Transient criminality: a model of stress-inducedcrime. New York, NY: Praeger Publishers, 1987.

44. Mawson AR. Attenuating effect of fighting on shock-inducedgastric ulceration and hypertension; hypothesis of inhibitionby sensory feedback. Med Hypoth 1978;4:403-10.

45. Sweeney TT, Foote JE. Treatment of drug and alcohol abusein spinal cord injury veterans. Int J Addict 1982;17:897-9O4.

46. Heinemann AW, Keen M, Donohue R, et al. Alcohol use bypersons with recent spinal cord injury. Arch Phys Med Reha-bil 1988;69:619-24.

47. Cohen S, Wills TA. Stress, social support, and the bufferinghypothesis. Psychol Bull 1985;98:310-57.

48. House JS, Landis KR, Umberson D. Social relationships andhealth. Science 1988;241:540-5.

49. Coyne JC, Downey G. Social factors and psychopathology:stress, social support, and coping processes. Ann Rev Psychol1991;42:401-25.

50. Haddon W, Baker SP. Injury control. In: Clark JD, MacMahonB, eds. Preventive and community medicine. Boston, MA:Little, Brown, and Co., 1981:109-40.

51. Robertson LA. Injuries: causes, control strategies and publicpolicy. Lexington, MA: Lexington Books, 1983.

52. Runyan CW. Progress and potential in injury control. Am JPublic Health 1993;83:637-9.


Dow



arch 2022

Documents

Sensation-Seeking, Criminality, and Spinal Cord Injury: A Case