Physical activity, especially vigorous aerobic exercise, has been found to reduce mild to moderate depression (18,33). Few studies have examined the relationship between physical activity and suicidal behavior; this, despite the fact that depression can be linked to suicide (20,34), and that suicide is the third leading cause of death among youth 15 to 24 yrs of age, the second leading cause of death among adults 25 to 34 yrs of age, and the fourth leading cause of death among adults 35 to 44 yrs of age (1). The risk of suicide among depressed persons of all ages is 30 times higher than for those in the general population (13). Theoretically, if persons who are physically active are less likely to be depressed than persons who are not active, or if depression or depressive symptomatology can be reduced or alleviated by physical activity interventions, then physical activity may potentially reduce the risk of suicide. It has also been reported that “athletes and nonathletes differ significantly on selected psychological states and traits, and these differences have consistently been observed to favor athletes in terms of positive mental health (16, p. 170).” Therefore, persons participating in sports may be at lower risk for suicidal behavior than those who are not involved in sports.
A few studies have examined the association between physical activity or sports participation and suicidal behavior. Simon and Powell (29) found among people 13 to 34 yrs of age that nearly lethal suicide attempts were related to self-reports of no exercise one month before the attempts. During a 23 to 27 yr follow-up study, Paffenbarger, Lee, and Leung (24) found that suicides among male Harvard alumni were not associated with physical activity habits and recreational sport participation at a younger age. However, Paffenbarger and Asnes (23) indicate that participation in varsity athletics as a Harvard University student had a protective effect against suicide in later life (4% of eventual suicides and 10% of controls were athletes). Similar findings were not observed for students from the University of Pennsylvania (20% of eventual suicides and 17% of controls were collegiate varsity athletes) (23). Sports participation has been found to be protective against suicide ideation and attempts among adolescents (22,25,30,31), with the exception of one study conducted in France and Quebec, Canada that reported finding no relationship (6). Tomori and Zalar (30) indicate that low participation in sport/physical activity is linked to a high rate of suicide attempts among adolescent boys in Slovenia. Additionally, the attitude “that sport is not important for health” was significantly higher among suicide attempters than nonattempters for boys and girls, and girl nonattempters were more likely to view sport/physical activity as a strategy for coping with distress compared with suicide attempters (30). Unger (31) found that adolescent boys who did not exercise or participate in sports had higher rates of suicidal behavior than peers who were more physically active, especially those whose activity included sports participation. However, adolescent girls who exercised 6 to 7 times per wk and who were not involved in sports had the highest rates of suicidal behavior compared with girls reporting no activity and no sports participation (31).
The purpose of this research is to evaluate the independent effect and association in college students between physical activity and suicidal behavior, and between sports participation and suicidal behavior. Overall, the evidence about whether physical activity is protective against suicidal behavior is equivocal. We hypothesized that college men and women who did no weekly physical activity were at greater risk of suicidal behavior than students who participated in weekly activity. This hypothesis was made tentatively, however, since the findings from Unger (31) showed an increased risk of suicidal behavior among adolescent girls who exercised frequently. Findings have more consistently shown that sports participation is associated with reduced risk of suicidal behavior. We also hypothesized that students who participated in sports were at lower risk of suicidal behavior compared with students who did not participate in sports.
We analyzed data from the 1995 National College Health Risk Behavior Survey (NCHRBS), a self-administered mail-in survey. The NCHRBS questionnaire and survey methodology have been described previously (10). According to Douglas et al.(10), the survey employed a two-stage cluster design that resulted in “a nationally representative sample of undergraduate college students aged 18 yr or older. The first-stage sampling frame contained 2,919 two- and four-year colleges and universities as the primary sampling units.” Of these, 148 (74 two-year and 74 four-year) institutions were selected from 16 strata including a representative number of Black and Hispanic students in the schools. Of the 148 colleges and universities, 136 (92%) of the institutions participated in the survey. The second sampling stage involved a random selection of students from full-time and part-time undergraduate students aged 18 yr or older who were enrolled in the selected colleges and universities. Differential sampling rates ensured that adequate representation of minority students was included, resulting in 7,442 students being eligible for the study. Students were assured that their participation in the survey was voluntary, that their responses were confidential, and that no identifiers would be linked to their responses. Of 7,442 students, 4,838 (65%) completed the survey. Twenty-four questionnaires were deleted due to incomplete or too few responses. Data from an additional 86 respondents were deleted from our analyses because of missing data specific to physical activity or suicide. Thus, information from 4,728 students was used in our data analyses.
The NCHRBS questionnaire was developed through collaborative efforts among the Centers for Disease Control and Prevention (CDC), other federal agencies, pertinent national organizations, and universities. The NCHRBS questionnaire underwent CDC institutional review and the data are available for use in the public domain. The 96-item multiple-choice questionnaire was designed to assess health risk behaviors among college students, including alcohol and drug use, tobacco use, dietary behaviors, physical activity, sexual behavior, suicidal behavior, safety, violence, and AIDS/HIV awareness. Draft questionnaire items were pilot-tested using focus groups of Black, White non-Hispanic, and Hispanic students. The questionnaire was available in both English and Spanish versions.
Many of the items on the NCHRBS questionnaire are also included as part of the Youth Risk Behavior Surveillance System (YRBSS), a CDC epidemiologic surveillance system designed to monitor health risk behaviors of high school students on a nationwide and biennial basis. Reliability for many of the NCHRBS/YRBSS items have been reported previously (2), and physical education, sports participation, and suicide items have shown acceptable reliabilities; Kappa percentages ranging from 60 to 84% over a 14-d test-retest period (2). To date, no validity studies have been done using the NCHRBS/YRBSS items. Thus, our findings reflect students’ perceptions of their physical activity behavior and suicidal tendencies, and not physical activity patterns that were assessed by using objective measures (e.g., accelerometers, doubly labeled water, etc.), or verified suicide attempts.
Definition of Physical Activity
Students who completed the NCHRBS questionnaire were asked two physical activity questions: 1) On how many of the past 7 d did you exercise or participate in sports activities for at least 20 min that made you sweat and breathe hard, such as basketball, jogging, swimming laps, tennis, fast bicycling, or similar aerobic activities? and 2) On how many of the past 7 d did you walk or bicycle for at least 30 min at a time? The first question was designed to capture student participation in vigorous activity. The second question was designed to provide an estimate of participation in two popular types of moderate intensity activities. Based on their responses, students were categorized into one of five activity levels:
- Frequent Vigorous Activity, 6–7 d·wk−1. Students who reported doing exercise that made them sweat and breathe hard for 20 min or more on 6 or 7 d of the week preceding the survey were defined as doing frequent vigorous activity.
- Vigorous Activity, 3–5 d·wk−1. Students who reported doing exercise that made them sweat and breathe hard for 20 min or more on 3 to 5 d of the week preceding the survey were defined as doing vigorous activity.
- Moderate Activity. Students who did less than 3 d of vigorous physical activity during the week, but who did a combined total of 5 or more d a wk of walking or bicycling for 30 min duration each day and 1 or 2 d of vigorous activity of 20 min or more duration, were defined as moderately active. Included in this category were students who reported walking or bicycling for 30 min or greater on 5 or more d during the week preceding the survey. Students in this group could also have done 1 d of vigorous activity and 4 or more d of walking/bicycling, or 2 d of vigorous activity and 3 or more d of walking/bicycling.
- Low Activity. Students who reported doing more than “no activity” and less than the moderately and vigorously active groups were defined as low active. This group did some activity, but less than amounts typically recommended to achieve health and fitness gains.
- No Activity. Persons in this category reported doing no days of vigorous activity, or walking or bicycling.
Definition of Sports Participation
Students were asked “During this school year, on how many college sports teams (intramural or extramural) did you participate?” Sports participation was dichotomized into groups of students who indicated that they participated on one or more college sports teams and those students reporting no sports involvement. The sports participation item did not differentiate between students who participated in sports relying primarily on individual performance (e.g., running; swimming) and sports relying on team play or strategy (e.g., volleyball; soccer; basketball).
Definition of Suicidal Behavior
Students were characterized as suicidal if they indicated that during the past 12 months they seriously thought about attempting suicide, made plans about how they would attempt suicide, or actually attempted suicide. Three NCHRBS survey items assessed suicide thoughts, plans, and attempts separately. These data were collapsed into a single variable, however, because the number of students who reported attempting suicide was small (N = 68).
Sociodemographic, Health Risk Behaviors, and Suicide
Several potential correlates of suicidal behavior could affect the physical activity, sports participation, and suicidal behavior associations. Age and race are associated with suicide. Suicide is the fourth, third, second, fourth, sixth, and eighth leading cause of death among the age ranges 10–14, 15–24, 25–34, 35–44, 45–54, and 55–64, respectively (1). These rates may differ depending on race/ethnicity. Higher rates of suicide have been reported for White men with increasing age, but prevalence of suicide among White women and non-Whites regardless of sex seems to peak in young adulthood and then decline with age (15). It has been estimated that alcohol is linked to approximately 30% of all suicides and is a critical factor in adolescent suicides (20). Substance abuse (6,12) and cigarette smoking (6,7,12,14) have also been associated with suicide behavior in high school students (6,12) and adults (7,14). Unger (31) found that very frequent exercise among adolescent girls was associated with suicide thoughts, plans, and attempts. If young women are overly or inappropriately concerned about their weight, this could confound any physical activity association with suicidal behavior. Thus, we created a four-category variable based on students’ Body Mass Index (BMI) and perception of their own weight (i.e., students with a BMI < 25 and ≥ 25 who do or do not perceive themselves as overweight). BMI is based on self reports of height and weight calculated as weight in kilograms divided by height in meters (2). A BMI of ≥ 25 was defined as overweight using clinical guidelines established by the National Institutes of Health (32). Perception of weight was defined by the single item, “How do you describe your weight?” A dichotomous variable was used to define perceived overweight. “Yes,” included students who defined themselves as slightly or very overweight, and “no” included those students who did not perceive themselves as overweight.
We previously published an abstract evaluating physical activity (but not sports participation) and suicide association using one level of vigorous activity rather than two (3). All chi square P-values were nonsignificant, indicating no association between levels of physical activity and suicide. Despite these findings, we subsequently decided to do regression modeling analyses that serve as the basis for this research for two reasons. First, we have data about our subjects that could mediate the physical activity and suicidal behavior relationship (such as drug and alcohol use). Regression modeling procedures enabled us to obtain a more precise picture of any association between physical activity and suicidal behavior by controlling for possible confounding variables. Second, as pointed out by Fleiss (11), “even though chi square is excellent as a measure of the significance of the association, it is not at all useful as a measure of the degree of association.” Because we are most interested in determining the degree of the association between different levels of activity and suicidal behavior, we conducted logistic regression modeling analyses that are reported here. Furthermore, for this research we decided to include two levels of vigorous activity (3 to 5 d and 6 to 7 d), because results from one study (31) indicate that “very frequent exercise (6 or 7 times per wk) was associated with higher rates of suicidal thoughts, plans, and attempts” among adolescent girls. This research also expands upon our earlier work in that we now include sports participation in our analyses.
We stratified our data by sex because a related study reported sex differences among the behaviors and associations that we were investigating (31). Multivariable logistic regression modeling was used to calculate crude and adjusted odds ratios for suicidal behavior as associated with physical activity. Similar analyses were performed to determine the association between suicidal behavior and sports participation. We modeled suicidal behavior as a dichotomous variable (yes, there were self-reports of suicidal thoughts, plans, or attempts; and, no, there were not), and five levels of physical activity as the independent variable (no activity, low activity, moderate activity, vigorous activity [3 to 5 d], and frequent vigorous activity [6 to 7 d]) We also modeled suicidal behavior and two levels of sports participation as the independent variable (participation on one or more team sports and no participation). Odds ratios were adjusted for age (18 to 24 yr; ≥ 25 yr), race/ethnicity (White non-Hispanic, Black non-Hispanic, Hispanic, Asian/Pacific Islander non-Hispanic, other), smoking (never, former, current), previous 30-d episodic heavy drinking/alcohol use (i.e., having ≥ 5 drinks in a row within a 2 h period = yes), lifetime ever drug use (0, 1–9, 10–39, ≥ 40 times), and BMI/weight perception (BMI < 25/not perceived overweight; BMI < 25/perceived overweight; BMI ≥ 25/not perceived overweight; BMI ≥ 25/perceived overweight). Sports participation was controlled for during the physical activity analyses, and physical activity was controlled for during the sports participation analyses. Chi square analyses were done to evaluate differences within the sociodemographic and health risk behavior categories, and t-tests determined differences among physical activity prevalence rates for men and women. The final analyses were performed using the SUDAAN software to take into account the complex sample design of the NCHRBS. All percentages were calculated using sampling weights.
For men, prevalence rates of no activity, low activity, moderate activity, vigorous activity (3 to 5 d·wk−1), and frequent vigorous activity (6 to 7 d·wk−1) were 19.2%, 27.3%, 10.1%, 31.6%, and 11.9%, respectively. For women, prevalence rates for these groups were 25.2%, 29.9%, 12.0%, 26.3%, and 6.6%, respectively. Approximately 10% of women and 26% of men participated in sports. Women were significantly more inactive (P < 0.0001), less vigorously active for 3 to 5 d (P < 0.0009) and 6 to 7 d·wk−1 (P < 0.0001), and less involved in sports (P < 0.0001) than men.
As indicated in Table 1, the prevalence of suicidal behavior among men and women students was similar. This was true for the suicide data collapsed as one variable (P < 0.62), as well as for suicide thoughts (M = 9.4%, W = 10.6%;P < 0.21), plans (M = 7.1%, W = 6.0%;P < 0.13), and attempts (M = 1.6%, W = 1.3%;P < 0.36) analyzed separately. Chi square analyses revealed several significant findings among sociodemographic characteristics and selected health risk behaviors. The prevalence of suicidal behavior was higher among younger than among older students (P < 0.0001). The prevalence of suicidal behavior was highest among students who reported that they were of Asian/Pacific Islander or “Other” descent. However, the survey did not attempt to oversample Asian/Pacific Islanders, so this group (and the group termed “Other”) has the smallest sample size and thus widest confidence intervals. Students reporting current smoking (P < 0.0001), one or more episodes of heavy drinking during the past 30 d (P < 0.0002), and lifetime ever drug use 10 or more times (P < 0.0001) are associated with higher prevalence rates of suicidal behavior compared with students reporting no involvement with these behaviors. The above findings were similar for men and women.
Men and women who were overweight and perceived themselves as overweight report a higher prevalence of suicidal behavior than students who were overweight and did not view themselves as overweight. Men who were not overweight but perceived themselves as overweight had a higher prevalence of suicidal behavior compared with men who were not overweight and did not view themselves as overweight. However, a small sample size and large confidence interval associated with findings for overweight women who did not perceive themselves as overweight and men who were not overweight but perceived themselves as being overweight restricts the significance of these findings.
The crude prevalence ratios for suicidal behavior by activity level are listed for men in Table 2. Results indicate that men who reported low activity had lower odds of suicidal behavior compared with those who reported no activity. Adjusting for age, race/ethnicity, smoking, previous 30 d episodic heavy drinking, lifetime ever drug use, BMI/weight perception, and sports participation did not substantially change our findings. Adjusted ORs show that men who did low activity were at almost half the odds (adjusted OR = 0.54;P < 0.015) of reporting suicidal behavior than men not active. Men who were moderately active, or vigorously active 3 to 5 or 6 to 7 d a wk, were not at either significant or reduced risk of suicidal behavior compared with the “no-activity” referent group.
Women exhibited a different profile than men, as indicated by the findings in Table 3. Adjusted ORs indicate that women who were vigorously active 6 to 7 d a wk had two times the odds of suicidal behavior compared with women reporting no activity (P < 0.034). Findings also indicate that moderately active women were 1.76 times more likely to report suicidal behavior than women not active (P < 0.035). Women who were low active and who did vigorous activity 3 to 5 d a wk had higher odds of suicidal behavior compared with the referent group, but these findings were not significant.
The crude prevalence and adjusted ORs for suicidal behavior by sports participation are listed in Tables 4 and 5. The crude prevalence ratios indicate that men who did not participate in sports were more likely to report suicidal behavior compared with men involved in sports. The adjusted ORs show that men not participating in sports had approximately 2.5 times the odds of reporting suicidal behavior (P < 0.0003) over men participating in sports. As shown in Table 5, women who did not participate in sports were 1.67 times more likely to report suicidal behavior than women sports participants, and this finding was borderline significant (P < 0.05).
We found no interaction between physical activity and sports participation in any of our logistic regression models for either men or women. We also observed no evidence of multicollinearity.
The occurrence of suicide among adolescents and young adults is high and the risk of suicide is greatest during the Spring season or semester of school (5). Efforts to understand the determinants of suicidal behavior or interventions that can reduce the risk of suicidal behavior are needed. It is possible that a health risk behavior such as physical inactivity can be a marker for suicidal behavior, and, if modified, be an intervention strategy to reduce risk of suicide. In this regard, our findings show that participation in intramural or extramural sports is protective against suicidal behavior among college men and women. While participation in low activity is more protective against suicidal behavior than no activity for men, participation in regularly taken moderate or vigorous activity is not. Women who reported doing frequent vigorous activity or moderate activity were found to be at increased odds of suicidal behavior compared with inactive women. A question remains as to whether the associations we found are due to physical activity (energy expenditure) per se, or other possible factors. Much of the remaining discussion explores these issues for both physical activity and sports participation.
In the absence of an association between moderate or vigorous activity and suicidal behavior among men, it is unclear why men who reported doing some low activity were at reduced risk of suicidal behavior compared with inactive men. It is possible that low physical activity is a covariant of another variable we did not control for that reduces suicidal tendencies. Two hypothetical examples, will illustrate this point. First, it is possible that college students experiencing moderate to high distress (e.g., anxiety or depressive symptomatology) may be those most likely to be either inactive or regularly active. In other words, high levels of distress may be linked to inactivity among some students, and frequent activity among others as a means of “feeling better” or coping with distress. Students who are low in activity may also tend to possess low distress and be at reduced risk of suicide. They are neither inactive nor are they regularly or frequently active. Second, it is possible that men in the low activity group are doing a higher amount of purposeful activity (which did not meet our definition of moderate or vigorous activity of longer duration) than men who are inactive. Their physical activity may be associated with school performance such as going to and from class, library, etc. and, therefore, low activity may merely be a covariate of academic achievement and suicidal behavior. Purposeful activity has been identified with very high adherence rates (19), yet health outcomes linked to purposeful activity need to be further studied. Certainly, these examples can be criticized for their plausibility; however, the examples make the point that a variable that we did not measure or control for may be mediating the relationship between low activity and suicidal behavior among men. This possibility, however, does not explain why men but not women of low activity were at reduced risk of suicidal behavior. Alternative explanations include the possibility that our findings may be due to chance, or that they reflect an artifact of our measurement items. Students were asked to report on suicidal tendencies during the year previous to taking the survey. Physical activity behavior was assessed by self-report items that asked students about the amount of physical activity they engaged in during the week before survey administration. A 7-d recall measure of physical activity could have led to the classification of students as active or not active at a current point in time, and their responses may not reflect the true association with suicidal behaviors at earlier points in time. Physical activity self-report measures may also be influenced by respondents’ motivation and ability to accurately recall their behaviors, and such problems can result in the over- or underreporting of physical activity habits (9). In addition, the tendency by some people to answer in a socially desirable manner may lead to inaccurate appraisals of both physical activity and suicidal tendencies that could have influenced our findings. We cannot completely rule out that problems inherent with self-report data are responsible for the study outcomes.
We hypothesized that college students who reported no physical activity would be at greater odds for suicidal behavior than students who did regular vigorous activity, regular moderate physical activity (primarily due to walking or bicycling) or low activity. Our findings pertaining to young women were opposite the hypothesized direction and are worrisome, especially in view of the fact that they also lend some support to findings of Unger (31). Contrary to prediction, Unger found that girls in 9th through 12th grade who participated in frequent physical activity reported higher rates of suicidal behavior than those who were inactive.
Our measures of physical activity do not allow for the determination of whether intensity or frequency of physical activity is associated with suicidal behavior. Theoretically, frequent physical activity could be a marker for an increased risk of suicidal behavior among some students. Excessive or compulsive exercise (as well as, overtraining encountered by endurance athletes such as swimmers) has resulted in physiological and psychological problems among some people (28). The outcomes associated with compulsive exercise have not included an increased risk of suicide ideation or attempt. However, our research (to a degree) and Unger’s suggest that suicidal behavior should be studied as a possible stimulus or consequence of excessive exercise. Such efforts will require distinguishing among people who are exercising frequently, but at levels healthy for them, from people who are exercising at levels placing them at serious physical, social, and psychological risk, such as described by Morgan (17). Although compulsive exercise may ultimately be linked to suicidal behavior, it is extremely unlikely that a large number of women in this study are compulsive exercisers, especially among women in the moderate activity group. Furthermore, frequent vigorous activity did place women at significantly higher odds for suicidal behavior. However, women who performed moderate activity, which included a low frequency of vigorous activity, were also at greater odds of suicidal behavior than inactive women. Thus, it is unlikely that vigorous activity is responsible for our findings.
Unger (31) proposed that many young women may “perceive themselves to be overweight and may exercise frequently in order to lose weight.” Thus, factors related to poor body image may contribute to low self esteem, depression, and suicidal feelings (31) that could influence the physical activity-suicidal behavior relationship. To get at issues related to BMI, weight perception, and suicidal behavior we developed a BMI/perceived weight variable, for two reasons. First, the correlation coefficient between BMI and weight perception was 0.71 (R2 = 0.50). Using the BMI/weight perception index helped to avoid potential problems related to multicollinearity, which may have resulted from controlling for BMI and weight perception separately in our regression analyses. Second, since “fear of becoming fat” is a common problem among women with eating disturbances, including an eating disorder such as anorexia nervosa (21), we speculated that the greatest risk for suicidal behavior may be among women who are not overweight, but who perceive themselves as being overweight. The index enabled us to evaluate this and other different BMI/weight perception profiles. Although not the primary focus of our study, we conducted regression analyses separately for men and women, using suicidal behavior as the dependent variable and the BMI/weight perception index as the independent variable. The only BMI/weight perception category that was significant was for women who were overweight and perceived themselves as overweight. Women in this group had 1.49 (P < 0.002) times the odds of suicidal behavior compared with women not overweight who accurately perceived themselves as not being overweight. Despite this finding, the association between physical activity and suicidal behavior seems independent of BMI/weight perception. Odds ratios remained significant between moderate physical activity and suicidal behavior and between frequent vigorous physical activity and suicidal behavior after adjusting for BMI/weight perception in our multivariate regression analyses. Our findings suggest that collegiate women who are at risk for suicidal behavior may be using exercise to cope with distress or stressors, independent of concerns about being overweight. Future research should attempt to identify other factors that may be mediating an association between physical activity and suicidal tendencies among college-age women. However, such efforts should also include better measures of body image or weight concerns, as it is premature to completely rule out these factors as trigger mechanisms for frequent physical activity or suicide thoughts, plans, or attempts. Similar to findings for men, we cannot rule out chance occurrence or our inability to more precisely measure physical activity behavior as factors influencing our findings.
Similar to other research, our findings support the hypothesis that students participating in sports were at lower risk of suicidal behavior than students not involved in sports. The reasons for lower rates of suicidal behavior among sports participants is unclear, but positive mental health and sports participation have been linked previously in a review paper by Morgan (16). As early as the 1950’s and 60’s, student athlete populations were noted to present fewer psychological problems than the nonathletic population. Davie (8), Carmen et al. (4), and Pierce (26) reported that fewer athletes than nonathletes sought psychotherapeutic help from student health services at Yale University (8), Harvard University (4), and the University of Rochester (26). Davie (8) questioned whether athletes who have less psychological problems than nonathletes tend to sublimate their problems through involvement in sports or receive some form of “therapy” from the coach. The “therapeutic” support base for contemporary athletes extends beyond the coach and may include athletic trainers, strength coaches, academic counselors, and other team members. The students in our study who participated in intramural or extramural sports had, at a minimum, access to a network of peers who could provide informal social or emotional support that reduced the risk for suicidal behavior.
Our data do not allow for conclusions to be made about “team” sports participation primarily relying on individual performance rather than overall team play. It is possible, however, that the image of the solitary or lonely long distance runner may be offset by a running subculture that is reinforcing and supportive in numerous ways, equivalent to the support received by members of a volleyball, soccer, or basketball team, for example. This is a testable research question that remains to be answered.
Our data are cross-sectional and preclude our ability to identify cause and effect relationships. For example, sport participation may mediate the lower risk of suicidal behavior, or people who do not have suicidal tendencies may be more likely to be involved in sports than those who possess suicidal tendencies.
By necessity, we also relied on self-reported data rather than actual measures of behaviors. It is unknown how the associations we found would differ if we conducted a longitudinal or prospective study using exposure and outcome measures such as physical activity assessments based on movement devices (e.g., accelerometers, pedometers) or doubly-labeled water, and information from medical records documenting attempted suicides or suicides verified by death certificates. Such measures would result in greater measurement precision (9).
Because only a small number of students reported that they had attempted suicide, our definition of suicidal behavior included students who both thought about or planned suicide attempts and actual attempters. People who report suicide ideation only may be unique from attempters. Ideally, it would have been beneficial if we were able to stratify these groups.
Our measure of moderate physical activity is also limited in that it primarily takes into account only two behaviors. However, walking and bicycling are likely to capture a significant portion of college students’ daily moderate activity.
Prevalence rates among the inactive college men (19.2%) and women (25.2%) in this study are consistent with rates reported for 18 to 29 yr-old men (19.7%) and women (26.3%) from the general population (27). Our physical activity prevalence data indicate that college students participate in higher rates of vigorous activity than young adults of the same age in the population-at-large, and this finding has been reported previously (27). While our data for moderate activity appears low compared with prevalence data from other national surveys, this is most likely because respondents were not placed into more than one physical activity category. Students defined as vigorously active may also meet the definition of moderate activity. However, if students were performing vigorous activity 3 or more times a wk along with moderate activity, they were classified as being either vigorously active or as doing frequent vigorous activity. The rate of moderate activity for the NCHRBS population is 19.5% overall (10), compared with 10% and 12% for men and women, respectively, in our analyses. These rates compare to 27.8% of young adults (18–29 yr of age) from the population-at-large who report doing regular moderate activity (27). The discrepancy (19.5% vs 27.8%) may be due to differences in the populations surveyed or the survey methods and items. For example, the moderate activity items from the NCHRBS are based on the prevalence of walking and bicycling behaviors only, while the prevalence rates generated by other national surveys assess participation in a greater variety of activities.
Although the physical activity groups in this study are generally distinct with regard to the frequency of student participation in vigorous activity. As noted above, the vigorous activity groups did include some students who also performed moderate activity regularly. This represents a degree of imprecision in our definitions of physical activity that could be affecting our findings. On the other hand, physical activity is a complex behavior, and most people do a mix of daily low, moderate, and vigorous activity that make it virtually impossible to completely classify people on a single criterion. Because of this problem we conducted a post hoc analysis using three physical activity groups: no activity, low activity, and recommended activity. The recommended activity group represented the combination of moderate and vigorous activity groups. These groups obtained physical activity that closely approximates national recommendations for moderate and vigorous activity. Of course, the findings for the low activity group remained as reported in Tables 2 and 3. Physical activity at the “recommended” level was not related to suicidal behavior among men (adjusted OR = 0.97;P = 0.91). This finding is consistent with those obtained using our five definitions of physical activity. Women doing the “recommended” level of physical activity were 1.47 times the odds of suicidal behavior, and these findings are nonsignificant at the P = 0.068 level. This differs from our research showing that women who did moderate and frequent vigorous activity were at greater odds of conducting suicidal behavior, and this is because the data for women performing vigorous activity 3 to 5 d per wk attenuated the findings for the moderate and two vigorous groups combined.
CONCLUSIONS AND FUTURE RESEARCH DIRECTIONS
We found that selected physical activity patterns were associated in a nonsystematic manner with decreased or increased odds of suicidal behavior (thoughts, plans, and attempts) among men and women college students, respectively. Sport participation among men and women was protective against suicidal behavior. Based on the available evidence, including this research, we are unable to conclude that physical activity (energy expenditure) per se is causally related to suicidal behavior. Future research is needed to determine what factors are causing the observed associations between physical activity or sport participation and a decreased or increased risk of suicidal behavior. Such factors may include fitness or biological factors related to leading an active versus an inactive lifestyle (e.g., depression, psycho-physiological reductions in tension or stress associated with a single session of exercise, changes in body composition or percentage of body fat, etc.). Alternatively, social or behavioral factors related to physical activity or the sport experience may mediate the risk of suicidal behavior (e.g., social support, body image, perceived physical competence, etc.).
Research is also needed to determine if the association between physical activity or sports performance and suicidal behavior differs among distinct groups such as men and women, different age groups, individual and team sports participants, purposeful (e.g., transportation to and from class, library, and restaurants) and leisure-time activity, etc. If so, why? Dose-response issues also need to be evaluated to identify what types, frequency, intensity, and duration of activity are linked to suicidal behavior.
The population-at-large, including college students, should participate in regular physical activity and benefit from the numerous physical and mental health benefits that result. However, exercise that is too frequent or too intense may result for some people in detrimental health outcomes such as injuries or increases in mood disturbance (28). We recommend that health care professionals serving young adults assess their physical activity behaviors and what motivates them to exercise. Health care professionals are in a unique position to recommend that students increase their level of physical activity, when necessary, to improve health status and to determine whether physical activity is being used excessively or inappropriately in order to mask problems that may require intervention.
Curtis Blanton is a biostatistician with the Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health, Mailstop K-50, 4770 Buford Highway, N.E., Atlanta, GA 30341–3724.
Address for correspondence: David Brown, PhD, Centers for Disease Control and Prevention, Mailstop K-46, 4770 Buford Highway, N.E., Atlanta, GA 30341-3724; E-mail: DBrown@cdc.gov
1. Anderson, R. N. Deaths: leading causes for 1999. Natl Vital Stat Rep. 49: 1–87, 2001.
2. Brener, N. D., J. L. Collins, L. Kann, C. W. Warren, and B. I. Williams. Reliability of the Youth Risk Behavior Survey Questionnaire. Am. J. Epidemiol. 141: 575–580, 1995.
3. Brown, D. R, and C. J. Blanton, Physical activity and suicide ideation/attempts among undergraduate college students [abstract]. Med. Sci. Sports Exerc.
30(suppl):S221, May 1998. Abstract 1260.
4. Carmen, L. R., J. L. Zerman, and G. B. Blaine, Jr. Use of the Harvard psychiatric service by athletes and nonathletes. Ment. Hyg. 52: 134–137, 1968.
5. Centers for Disease Control and Prevention. Temporal variations in school-associated student homicide and suicide events—United States, 1992–1999. MMWR Morb Mortal Wkly Rep. 50: 657–60, 2001.
6. Choquet, M., V. Kovess, and N. Poutignat. Suicidal thoughts among adolescents: an intercultural approach. Adolescence 28: 649–659, 1993.
7. Clayton, P. Smoking and suicide. J. Affect. Disord. 50: 1–2, 1998.
8. Davie, J. S. Who uses a college mental hygiene clinic? In: Psychosocial Problems of College Men. B. M. Wedge (Ed.). New Haven, CT: Yale University Press, 1958, pp. 140–149.
9. Dishman, R. K., R. A. Washburn, and D. A. Schoeller. Measurement of physical activity. Quest 53: 295–309, 2001.
10. Douglas, K. A., J. L. Collins, C. Warren, et al. Results from the 1995 National College Health Risk Behavior Survey. Coll. Health. 46: 55–66, 1997.
11. Fleiss, J. L. Statistical Methods for Rates and Proportions, 2nd Ed. New York: John Wiley & Sons, 1981, p.58.
12. Garrison, C. Z., R. E. McKeown, R. F. Valois, and M. L. Vincent. Aggression, substance abuse, and suicidal behaviors in high school students. Am. J. Public Health 83: 179–184, 1993.
13. Guze, S., and E. Robins. Suicide and primary affective disorders. Br. J. Psychiatry. 117: 437–438, 1970.
14. Hemenway, D., S. J. Solnick, and G. A. Colditz. Smoking and suicide among nurses. Am. J. Public Health. 83: 249–251, 1993.
15. Kastenbaum, R. Suicide. In: The Encyclopedia of Aging, G. L. Maddox, R. C. Atchley, L. W. Poon, G. S. Roth, I. C. Siegler, and R. M. Steinberg (Eds.). New York: Springer Publishing Company, 1987, pp. 654–655.
16. Morgan, W. P. Athletes and non-athletes in the middle years of life. In: Sport and Aging, B. D. McPherson (Ed.). Champaign, IL: Human Kinetics Publishers, 1985, pp.167–186.
17. Morgan, W. P. Negative addiction in runners. Phys. Sportsmed. 7: 57–70, 1979.
18. Morgan, W. P. Physical activity, fitness and depression. In: Physical Activity, Fitness, and Health: International Proceedings and Consensus Statement. C. Bouchard, R. J. Shephard, and T. Stephens (Eds.). Champaign, IL: Human Kinetics Publishers, 1994, pp. 851–867.
19. Morgan, W. P. Prescription of physical activity: a paradigm shift. Quest. 53: 366–382, 2001.
20. Mrazek, P. J., and R. J. Haggerty. Reducing Risks for Mental Disorders: Frontiers for Prevention Intervention Research. Washington, D.C: National Academy Press, 1994, pp. 91, 95.
21. O’Connor, P. J., and J. C. Smith. Physical activity and eating disorders. In: Lifestyle Medicine. J. M. Rippe (Ed.). Malden, MA: Blackwell Science, Inc., 1999, pp. 1005–1015.
22. Oler, M. J., A. G. Mainous 3rd, C. A. Martin, et al. Depression, suicidal ideation, and substance use among adolescents: are athletes at less risk? Arch. Fam. Med. 3: 781–785, 1994.
23. Paffenbarger, R. S., Jr., and D. P. Asnes. Chronic disease in former college students: III. Precursors of suicide in early and middle life. Am. J. Public Health Nations Health. 56: 1026–1036, 1966.
24. Paffenbarger, R. S., Jr., I. M. Lee, and R. Leung. Physical activity and personal characteristics associated with depression and suicide in American college men. Acta. Psychiatr. Scand. Suppl. 377: 16–22, 1994.
25. Page, R. M., J. Hammermeister, A. Scanlan, and L. Gilbert. Is sports participation a protective factor against adolescent health risk? J. Health Educ. 29: 186–192, 1998.
26. Pierce, R. A. Athletes in psychotherapy: how many, how come? J. Am. Coll Health Assoc. 17: 244–249, 1969.
27. Pratt, M., C. A. Macera, and C. Blanton. Levels of physical activity and inactivity in children and adults in the United States: current evidence and research issues. Med. Sci. Sports Exerc.
31(Suppl.), S526–S533, November 1999.
28. Raglin, J. S., and L. Moger. Adverse consequences of physical activity: when more is too much. In: Lifestyle Medicine. J. M. Rippe (Ed.). Malden, MA: Blackwell Science, Inc., 1999, pp. 998–1004.
29. Simon, T. R., and K. E. Powell. Involvement in physical activity and risk for nearly lethal suicide attempts [abstract]. Med. Sci. Sports and Exerc.
31(Suppl.):S85, May 1999. Abstract 271.
30. Tomori, M., and B. Zalar. Sport and physical activity as possible protective factors in relation to adolescent suicide attempts. Int. J. Sport Psychol. 31: 405–413, 2000.
31. Unger, J. B. Physical activity, participation in team sports, and risk of suicidal behavior in adolescents. Am. J. Health Promot. 12: 90–93, 1997.
32. U.S. Department of Health and Human Services. Clinical Guidelines on the Identification, Evaluation and Treatment of Overweight and Obesity in Adults: The Evidence Report. Bethesda, MD: National Institutes of Health, National Heart, Lung, and Blood Institute, 1998, pp. xiv.
33. U.S. Department of Health and Human Services. Physical Activity and Health: A Report of the Surgeon General. Atlanta, GA: Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, 1996.
34. Winokur, G., and D. W. Black. Suicide—what can be done? N. Engl. J. Med. 327: 490–491, 1992.
Keywords:©2002The American College of Sports Medicine
EXERCISE; ATHELETICISM; MENTAL HEALTH; WELL-BEING; UNIVERSITY STUDENTS