Panic Attacks and Suicide Attempts in Mid-Adolescence

This investigation represents a cross-sectional study of the association between a lifetime history of panic attacks and suicidal ideation and suicide attempts in a community sample of adolescents, most of whom were 13–14 years old at the time of assessment.

Estimated lifetime prevalence proportions for major depression, panic attacks, suicidal ideation, and suicide attempts are shown in Table 1. Being female and race/ethnicity were not found to be significantly associated with suicidal ideation and attempts in the bi­variate analysis (Table 2). Higher rates of suicidal ideation and suicide attempts were associated with a lifetime history of panic attacks and major depression (Table 2).

The highest prevalence of suicidal ideation was evident among the 12 adolescents with both major depression and panic attacks (Table 3). The highest prevalence of suicide attempts was evident among the 13 adolescents with major depression and no panic attacks. These figures should be viewed cautiously because the number of adolescents in each category was small (Table 3).

As shown in Table 4, the logistic regression models revealed that adolescents with history of panic attacks were approximately three times more likely to have expressed suicidal ideation and about two times more likely to have made suicide attempts than adolescents without panic attacks. Adolescents with a history of major depression were approximately six times more likely to have expressed suicidal ideation and almost five times more likely to have made suicide attempts than adolescents with no history of major depression. These odds were adjusted for demographic variables (sex and race-ethnicity), lifetime history of major depression, history of alcohol use, and the use of any of four illicit drugs (Table 4).

As a check on the validity of our logistic regression adjustment procedures, we conducted an exploratory search for possible interactions between variables. We did not find that interaction terms improved the fit of the model (alpha=0.10).

This study adds new population-based estimates of the association between a lifetime history of panic attacks and suicidal ideation and suicide attempts in the middle years of adolescence. The logistic regression models indicated that adolescents with a lifetime history of panic attacks were more likely to have expressed suicidal ideation and to have made suicide attempts than those without a lifetime history of panic attacks. These findings were obtained with statistical control for demographic variables and other characteristics suspected to affect the occurrence of suicide attempts (e.g., major depression, alcohol use, and illicit drug use).

Although this study has a number of important strengths, including the use of a structured interview and standard diagnostic criteria to ascertain psychiatric disturbances, several limitations are especially noteworthy and deserve comment. A principal limitation of this study is the reliance on cross-sectional lifetime prevalence data, which cannot disentangle the temporal sequences that underlie observed associations between a lifetime history of panic attacks and a history of suicide attempts. Of course, this limitation also applies to the work of Weissman et al. (1), which remains the most cited published article on this association. Another noteworthy limitation is the lack of control or adjustment for other specific psychiatric disorders such as anxiety disorders and schizophrenia, which were not assessed in this investigation. However, it should be noted that the prevalence of schizophrenia in early- to mid-adolescence is very low (36). Last, the Composite International Diagnostic Interview has been validated only in adult populations. Since, according to DSM-IV, adolescent and adult depression present with similar affective, cognitive, motivational, and vegetative symptoms, the Composite International Diagnostic Interview might well be valid among adolescents. Nonetheless, we acknowledge that our adapted question on suicide attempts, designed to be responsive to parental concerns about a sensitive topic, might be broad enough to encompass self-mutilation and self-harming behaviors in general, rather than suicide attempts specifically. This is an issue we are probing in a new follow-up assessment of this sample of adolescents.

Notwithstanding limitations such as these, the association between panic attacks and suicide attempts found in this study parallels the ECA findings reported by Weissman et al. (1). Anthony and Petronis (37) expressed concern that reliance on cross-sectional lifetime prevalence ECA data might have created a spurious association between a history of panic attacks and a history of suicide attempts (1, 38). For this reason, they completed a prospective nested case-control study of 40 ECA respondents who had recently attempted suicide and 160 matched control subjects. Controlling for demographic factors, psychiatric disorders, and drug use, they also found that adults with a lifetime history of panic attacks were at greater risk of suicide attempts (37). A recent study based on a community sample of children and adolescents aged 9 to 17 years (31) revealed that panic attacks were significant predictors of suicide risk, after adjustments for relevant psychiatric disorders.

The present study adds to the accumulating evidence of an association between panic attacks and suicide attempts and observes this association in a community sample of adolescents. Two types of studies might confirm that these findings are not the result of a spurious association: 1) conventional epidemiologic case-control studies of adolescents with panic attacks and 2) experimental studies aimed at ascertaining whether preventive treatment of panic attacks might reduce the risk of ensuing suicide attempts.

Panic attacks were ascertained in this investigation in essentially the same manner as in the ECA survey. In all likelihood, some students who reported panic attacks in this study, as well as some adults who reported them in the ECA study, would have met only one DSM-IV criterion for panic attacks. DSM-IV criteria require at least four from a list of 11 symptoms of panic attacks. The list includes symptoms such as palpitations, sweating, and trembling, among others. Had we ascertained the presence of panic attacks using DSM-IV criteria, a group with more severe panic attacks would have been selected. In such a group, the odds of reporting suicide attempts might well have been higher than is conveyed by the estimates reported here.

Andrews and Lewinsohn (26), analyzing data from their community study of suicide attempts among high school students, found no association between panic disorder and suicide attempts. However, their sample included only 14 adolescents with a lifetime history of panic disorder, which constrained the statistical power of their analysis and the information value of their study. Given the low prevalence of panic disorder among adolescents (16), it is not surprising that only a relatively small number of students met the criteria for a diagnosis of panic disorder. Focusing on students who reported panic attacks, a more frequent occurrence, has enabled us to base our estimates on more than 300 students with a lifetime history of panic attacks.

This new epidemiologic research extends to adolescents an important finding from adult investigations, i.e., that individuals with panic attacks seem to be more likely to experience suicidal ideation and make suicide attempts than those without panic attacks. Given the current concern about adolescent suicide, research findings that suggest potential risk factors for suicide attempts in this population may contribute to better-informed screening and prevention efforts. Clinicians are well advised to inquire routinely about past suicidal ideation and suicide attempts during the initial interview with adolescents who present with a history of panic attacks and to monitor for emergence of these symptoms during treatment.

Readers also may be interested to know that recent epidemiologic data suggest an increased occurrence of suicidal deaths among African American youths in the United States (39). Our epidemiologic sample consisted primarily of African American adolescents. Hence, these empirical findings may have some value to clinicians and public health practitioners who seek to curb this trend.

Received June 30, 1998; revision received Jan. 14, 1999; accepted March 4, 1999. From the Kennedy Krieger Institute; the Johns Hopkins University School of Medicine; and the Johns Hopkins University School of Hygiene and Public Health, Baltimore, Md. Address reprint requests to Dr. Anthony, Johns Hopkins School of Hygiene and Public Health, Hampton House 893, 624 N. Broadway, Baltimore, MD 21205; [email protected] (e-mail). Supported primarily by NIMH grants MH-47447 and MH-01456 and training grant award MH-14592 and by National Institute of Drug Abuse grant DA-09592.