In 1980, posttraumatic stress disorder (PTSD) was introduced into the official classification of psychiatric disorders, the DSM-III (1). This marked the beginning of contemporary research on the psychiatric response of traumatic event victims. The DSM-III and subsequent DSM definitions of PTSD (2,3) are based on a conceptual model that brackets traumatic events from other stressful experiences and brackets PTSD from other responses to stress. In contrast to “ordinary” stressful experiences, traumatic or catastrophic events are linked etiologically in the DSM to a specific syndrome—PTSD. The disorder’s criterion symptoms are defined in terms of their connection, in time and in content, with a distinct traumatic event. They include reexperiencing the event through intrusive thoughts and dreams, avoidance of stimuli that symbolize the event, numbing of general responsiveness, and increased arousal not present before the event. Many of these symptoms are among the characteristic features of other psychiatric disorders. However, it is the connection with a distinct event that renders the list of PTSD symptoms a specific syndrome.
Since 1980, research on PTSD has focused primarily on Vietnam War veterans and, to a lesser extent, on victims of specific traumatic events, such as disaster or rape. In recent years, several epidemiologic studies have been conducted in the general population (4–11). These studies describe the prevalence of traumatic events and PTSD and their distribution across population subgroups. They identify risk factors for trauma exposure and PTSD and describe aspects of the course of PTSD, including duration of symptoms and comorbidity with other psychiatric disorders.
This paper reviews results from these studies. It focuses also on core methodologic issues for research in this growing field. It addresses changes in the definition of the stressor criterion in the DSM-IV and presents data on the implications of the DSM-IV shift toward an emphasis on the subjective experience of trauma victims and a more inclusive variety of stressors. Approaches to measuring exposure to traumatic events and PTSD are discussed. This review also provides a summary of findings on the prevalence of trauma exposure and PTSD and a brief summary of findings on risk factors. Research on other psychiatric disorders that might be attributed to traumatic events, the logic of the approach to this inquiry, and the implications of recent findings are discussed in more detail.
The Definition of Trauma and PTSD
According to the current nosology expressed in the DSM-IV, the core features of PTSD comprise a stressor criterion that defines the etiologic event and a configuration of symptoms, drawn from 3 groups, that define the characteristic PTSD syndrome (3). The 3 symptom groups that constitute the PTSD syndrome are as follows: reexperiencing the trauma in nightmares, intrusive memories, or “flashbacks” (Criterion B); numbing of affect and avoidance of thoughts, acts, and situations that symbolize the trauma (Criterion C); and symptoms of excessive arousal (Criterion D). The diagnosis requires the persistence of symptoms for at least 1 month (Criterion E) and clinically significant distress or impairment (Criterion F).
The DSM-IV definition of the PTSD syndrome is little changed from earlier DSM editions. However, the stressor criterion in DSM-IV clearly departs from earlier DSM editions. Traumatic experience was originally defined as an overwhelming experience outside the usual range. The DSM-IV redefined traumatic experiences in subjective terms. The current definition is 2-part: the first part (Criterion A1) defines the range of qualifying stressors, and the second part (Criterion A2) requires that “the person’s response involved intense fear, helplessness, or horror.” Criterion A1 broadens the variety of qualifying traumatic events. In addition to the original core category of traumas used to define PTSD (military combat, disaster, and criminal violence), the expanded DSM-IV definition attempts to cover all possible events that clinicians might regard as potentially culminating in PTSD symptoms. Although stressors classified as less extreme are explicitly excluded (for example, spouse leaving or being fired [3, p 427]), the DSM-IV list of examples is clearly more inclusive than earlier DSM editions. For the first time, death of a loved one from any cause (including natural causes) qualifies as a stressor, as long as it was “sudden and unexpected.” Being diagnosed with a life-threatening illness is another example of the wider range of traumatic events included in the new definition. The DSM-IV revision—the broader range of qualifying traumatic events and the added criterion of a specific emotional response—deemphasizes the objective features of the stressors and highlights the clinical principle that people may perceive and respond differently to outwardly similar events.
A recent report from a general population study that used the DSM-IV suggests that the net effect of the wider variety of events and the added subjective component has been to increase by more than 20% the total number of qualifying events, compared with previous DSM definitions (12). The subjective component of the DSM-IV stressor criterion did little to offset the effect of the broadened range of qualifying stressors: approximately 90% of those who have ever been exposed to 1 or more stressors have responded with “intense fear, helplessness or horror” to 1 (the worst) of their stressors. There has also been a considerable increase in PTSD cases, with most of the added cases attributable to a single type of trauma; namely, “learning about unexpected death of a close relative/friend” (12). The DSM-IV definition requires, for the first time, that the syndrome cause significant distress or impairment—an addition that renders the diagnostic definition of PTSD more stringent. This is particularly true in epidemiologic surveys of the general population, as opposed to clinical practice, where impairment or distress are generally the reasons for seeking treatment. In the study described above (10,12), the requirement that the syndrome cause significant distress or impairment led to a reduction by approximately 25% in the number of PTSD cases identified (unpublished data).
Assessing Exposure to Traumatic Events and PTSD in Community Studies
The standard measurement procedure in contemporary epidemiologic studies of psychiatric disorders (including PTSD) has been the National Institute of Mental Health-Diagnostic Interview Schedule (NIMH-DIS) and the World Health Organization-Composite International Diagnostic Interview (WHO-CIDI), which is based on the DIS (13–15). These structured interviews are designed to be administered by experienced interviewers without clinical training. The PTSD section inquires about lifetime history of traumatic events and asks respondents to nominate the worst event ever experienced. It then elicits information about PTSD symptoms connected with that event. Some epidemiologic studies have introduced modifications to the NIMH-DIS—chiefly, in the approach to eliciting information about exposure to traumatic events (5,6).
Estimates of the prevalence of exposure to traumatic events vary according to the inclusiveness of the stressor criterion and the methods used to measure exposure to qualifying stressors. In the DSM-III and DSM-III-R, qualifying stressors were defined as events that would be “distressing to almost everyone” and as “generally outside the range of usual human experience.” Typical PTSD traumas were military combat, rape, physical assault, natural disaster, witnessing violence, and learning about violent injury or a violent death of a loved one. DSM-IV has broadened the stressor criterion beyond the earlier definition, as described above. According to one study, the impact of the revision on the lifetime prevalence has been an increase from 68.1% to 89.6% (12).
Differences in estimates of the prevalence of exposure to traumatic events across epidemiologic studies that predate the 1994 publication of the DSM-IV reflect differences in measurement approaches. The key difference is between studies that used the revised NIMH-DIS, which elicits history of exposure to traumatic events with a single question incorporating examples of typical PTSD events (13), and studies that used a list of events and inquired about each event separately (5,6). The use of a list of events and the number of events included in the list have important implications for estimating the prevalence of traumatic events. Using a list, compared with using a single question, and using a long vs a short list, yields higher prevalence estimates of trauma exposure and higher estimates of the average number of traumas per exposed person. A list of events has become the standard measurement procedure, incorporated into current versions of the major structured interviews—the WHO-CIDI and the DIS for DSM-IV (14,15).
Estimates of the conditional probability of PTSD are derived from information on the prevalence of exposure to 1 or more qualifying events (the denominator) and the proportion of those exposed who meet criteria for PTSD in connection with 1 of their events (the numerator). Estimates of the overall conditional risk for PTSD depend in part on the stressor definition. A definition that includes only rare and highly traumatic events will yield a lower prevalence of exposure and a higher conditional risk for PTSD, compared with an inclusive definition that encompasses a wide range of events. In addition, a measurement procedure such as a single question, which yields a lower prevalence of exposure, yields a higher conditional risk for PTSD, compared with procedures that yield a higher prevalence of exposure, such as a list of events. It appears that using a list vs a single item (or a long vs a short list) enhances recall of events that are less memorable but also less likely to have led to PTSD.
Kessler and others identified a methodological issue in community studies of PTSD (7,16). When 2 or more qualifying traumas were reported, previous studies focused on traumatic events selected by the respondent as the worst or the most upsetting ever experienced. This approach is an efficient way to identify persons with PTSD and estimate the prevalence of the disorder. Only a few respondents who fail to meet PTSD diagnostic criteria for their worst trauma are likely to meet criteria for other traumas they experienced (17). The worst-event method has also been used to estimate the conditional risk for PTSD. The worst events yield estimates of the probability of PTSD in persons who have experienced 1 or more traumas in their lifetime and allow the investigation of risk factors and consequences of PTSD among this group. However, the approach has been suspected of overstating the conditional risk for PTSD associated with the class of DSM-qualifying events as a whole (that is, typical trauma) (7,16).
An alternative to inquiring about PTSD symptoms in relation to all reported events, which would impose too great a respondent’s bias, involves selecting a random event from the complete list of traumatic events reported by each respondent. Such an approach, together with a weighting procedure to adjust for differences in the selection probabilities of events across respondents, provides a representative sample of qualifying, or typical, traumatic events. This method was used in a recent survey of the general population, and several reports have been published based on it (for example, 10,12,18,19). A comparison of the estimates based on the 2 methods—the randomly selected events and the worst events—suggests that the worst events moderately overstate the conditional risk for representative (typical) traumas. Specifically, the conditional probability of PTSD based on the sample of the worst events was 13.6%; based on the representative sample of events, it was 9.2% (10). This new method provides an estimate of the conditional probability of PTSD following typical traumas. However, it is incapable of identifying all (or nearly all) those who met criteria for PTSD following exposure to traumatic events, either in a lifetime or during a specified time period. In other words, the randomly selected events cannot replace the worst events as a shortcut when an inquiry about all traumatic events is not feasible.
Archives in English | Archives
RCP en français