May 2013  << Back  

  Can't open the PDF? Click here for help.

 

 RESEARCH ARTICLE

A prospective study of the onset of PTSD symptoms in the first month after trauma exposure

Jeannie B. Whitman, PhD

Department of Psychiatry/Division of Crisis and Disaster Psychiatry, University of Texas Southwestern Medical Center, Dallas, TX, USA

Carol S. North, MD, MPE

VA North Texas Health Care System, Departments of Psychiatry and Surgery/Division of Emergency Medicine, University of Texas Southwestern Medical Center, Dallas, TX, USA

Dana L. Downs, MA, MSW

Department of Psychiatry/Division of Crisis and Disaster Psychiatry, University of Texas Southwestern Medical Center, Dallas, TX, USA

Edward L. Spitznagel, PhD

Department of Mathematics, Division of Biostatistics, Washington University in St. Louis, St. Louis, MO, USA

BACKGROUND: The course of posttraumatic stress disorder (PTSD) symptoms in the month after trauma exposure has not been determined adequately. Symptom group C (avoidance/numbing) has been identified retrospectively as a marker for PTSD, but prospective studies are needed to determine whether these symptoms can provide substantially earlier identification of those who will have PTSD 1 month after trauma exposure.

METHODS: We evaluated 42 patients hospitalized for traumatic injuries over the first post-injury month to track development of posttraumatic symptoms.

RESULTS: Symptoms emerged rapidly, with group B (intrusion) and group D (hyperarousal) symptoms occurring earlier than group C symptoms. At 1 week, group C criteria accurately predicted who would develop PTSD by 1 month, and by 2 weeks, group C criteria also predicted who would not develop PTSD by 1 month.

CONCLUSIONS: The findings, if replicated, may permit earlier identification of PTSD and more timely, appropriate treatment.

KEYWORDS: trauma, posttraumatic stress disorder, PTSD, emergency medicine

ANNALS OF CLINICAL PSYCHIATRY 2013;25(2):E8-E17

  INTRODUCTION

DSM-III introduced the diagnosis of posttraumatic stress disorder (PTSD) more than 30 years ago,1 and voluminous literature about the structure and characteristics of this disorder has accumulated since then. Breslau et al2 recently noted that existing knowledge about the structure and function of symptom groups in relation to diagnosis of PTSD is based largely on retrospective lifetime studies. They identified a need for “prospective research…with fine-grained assessment of clinical features as they emerge” in the early development of PTSD following trauma.2 No studies have charted the course of PTSD symptoms prospectively throughout the first post-trauma month.

The most active period for development of new PTSD symptoms is the first post-trauma month. A study of directly-exposed Oklahoma City (OK, USA) bombing survivors interviewed approximately 6 months after the 1995 bombing found that 94% of those with PTSD developed symptoms in the first post-disaster month.3 Interestingly, the first month after trauma also is the time period during which PTSD cannot be diagnosed because diagnostic criteria require a 1-month waiting period.4

Conducting prospective studies of the early course of PTSD requires systematic access to traumatized populations in the first few days after their trauma exposure and repeated assessment of PTSD symptoms across the next few weeks. These methodological challenges likely have contributed to the relative lack of existing prospective data on this issue. Three prospective studies of the early course of PTSD in hospitalized trauma patients5-7 collected data on PTSD symptoms at approximately 1 week post-trauma and a full diagnostic interview after 1 month. All 3 studies5-7 found that early symptom scores were associated with the likelihood of later development of PTSD, but the numerical symptom variable presented in the results used in these studies did not dichotomize subgroups needed to differentiate those destined to develop PTSD from those who would not develop the disorder. To address the need to establish such a dichotomy, Shalev et al6 compared the ability of a range of symptom score thresholds to identify PTSD cases earlier than 1 month. The authors concluded that symptom score thresholds accurately identified those who would not develop PTSD (specificity) but failed to distinguish those who would develop PTSD (sensitivity). In explaining these conclusions, they emphasized the relatively low positive predictive values (PPV) achieved by these methods (generally in the 20% to 50% range).

Rothbaum et al8 attempted to identify individuals who later would develop PTSD (diagnosis approximated using a symptom scale) in a 2-step classification procedure exploring combinations of thresholds for 2 posttraumatic symptom scales administered an average of 13 days post-rape to determine the best PTSD-predictive fit. This 2-step procedure yielded an optimal pair of predictive thresholds achieving 90% sensitivity, 61% specificity, 58% PPV, and 80% negative predictive value (NPV). Nishi et al9 established a threshold score on a peritraumatic distress scale administered approximately 2 days post-motor vehicle accident in an attempt to identify those who would later develop PTSD (assessed by a full diagnostic interview), with 77% sensitivity, 82% specificity, 53% PPV, and 93% NPV.

There is substantial clinical value in differentiating patients who will develop PTSD early in the first month following trauma exposure from those who will not. That is the case because clinicians may be confronted with pressures to intervene well before 1 month has passed—before clinicians formally can diagnose PTSD. This period of diagnostic uncertainty presents dilemmas for assessment and hence treating psychiatric symptoms because effective treatment decisions are based on accurate diagnosis.10 Withholding treatment while awaiting diagnostic determination may delay the relief of a patient’s suffering and prolong the hazards associated with an untreated illness.

Conversely, treatment applied before a diagnosis can be established may be inappropriate, unnecessary, or even harmful. Because most trauma-exposed patients will not develop PTSD, it’s inadvisable to treat everyone for the disorder.11 Clearly, the profession needs a reliable procedure for confident identification of PTSD cases in the first few days or weeks after trauma exposure.

The diagnosis of acute stress disorder originally was created to fill a clinical gap in the first post-trauma month before clinicians could diagnose PTSD. It also predicts PTSD in the first days or weeks after trauma exposure. However, Bryant12 found the sensitivity of acute stress disorder for early identification of PTSD is unacceptably low.

To identify PTSD cases reliably in the first few days and weeks after exposure to trauma, clinicians must differentiate incipient PTSD from symptoms and distress that are highly prevalent—and normative—in that period.11,13-17 Research has established that group B (intrusion) and D (hyperarousal) symptoms are normative and, in the absence of group C (avoidance/numbing) criteria, do not indicate psychopathology by themselves.3,18 Several retrospective studies have identified the fulfillment of Group C criteria as a marker for PTSD.2,3,14,19-23 A consistently large majority (75% to 100%) of individuals meeting symptom group C meet full PTSD criteria.3,14,20-24 Prospective studies have not been conducted to determine whether fulfillment of group C criteria in the early post-trauma days and weeks can differentiate people who will or will not develop PTSD by 1 month. Although group C is an established indicator of PTSD in studies measuring these constructs after 1 month, it’s not known if group C occurring before 1 month can predict PTSD accurately.

We designed this study to prospectively track the temporal onset of posttraumatic symptoms emerging in the first post-injury days and weeks through repeated measurements over time among hospitalized trauma patients. An additional aim was to examine prospectively the potential for PTSD symptom group C to identify PTSD early in the first post-trauma month.

  METHODS

Sample

Adult trauma patients consecutively admitted to the Parkland Memorial Hospital Trauma Services (Dallas, TX, USA) following a medically serious injury—excluding intentionally self-induced injuries—were recruited during 4 active enrollment periods of approximately 3 weeks each, corresponding to study personnel availability. To be eligible, patients had to speak English, recall the traumatic incident, and be sufficiently alert, coherent, and medically stable to participate. All 47 study-eligible patients admitted to the trauma service during the study enrollment periods provided informed consent for participation; 5 (11%) were lost to follow-up during the course of the study. These 5 individuals did not differ significantly (by sex, age, race, education) level of injury (trauma activation level, injury severity score, type of incident), baseline symptom level (number of group B, group C, group D, or PTSD symptoms), or psychiatric history. The institutional review boards of the participating institutions—University of Texas Southwestern Medical Center (Dallas, TX, USA), Parkland Memorial Hospital (Dallas, TX, USA), and Capella University (Minneapolis, MN, USA)—approved the study.

Procedures

Study procedures could not always be completed during the initial hospital days because of patients’ unavailability during surgical or medical procedures, ventilator dependence, unconsciousness, or medical instability precluding participation. We enrolled patients and received a baseline interview on the first possible post-accident day.

During the baseline interview, participants briefly described what they recalled of the accident and confirmed or denied each of 17 current-day DSM-IV PTSD symptoms using a structured diagnostic instrument. Personal interviews in the hospital or by telephone after discharge queried all 17 PTSD symptoms daily for the first 7 days and weekly thereafter, with the final interview occurring at 1 month. During week 2, week 3, and week 4 interviews, we also asked patients to report symptoms they had experienced on the days of the week since the last interview. A full diagnostic interview for PTSD was completed 1 month after the incident.

Patients enrolling in the study after their first hospital day provided interview data for the previous days. We asked patients not able to participate in interviews on consecutive days in the first post-accident week to provide information about current day symptoms and then to report symptoms they had experienced for each of the previous day(s) that had not been assessed. In summary, prospective symptom data were collected for the first 7 days and for the current day at the week 2, week 3, and 1-month assessments; retrospective symptom data were collected for missed interview days and non-interview days.

Medical record review data

With patient consent, we obtained information on the types of incidents that caused the injuries and findings from the Injury Severity Score (ISS) and Trauma Activation Level (TAL) in the medical records. The ISS sums up multiple injury scores in 6 anatomic regions, with a total score of 10 to 14 considered moderate and ≥15 considered serious or critical.25 The TAL reflects global injury severity: Level I indicates severe injuries with abnormal vital signs, Level II indicates severe injuries with normal vital signs, and Level III indicates serious but less severe injuries.26

Psychiatric assessment instruments

All participants’ medical records documented that the traumatic incident prompting the current hospitalization fulfilled PTSD criteria A, signifying sufficient exposure to a qualifying traumatic event. Therefore, we did not need to question patients about their exposure to trauma, and all symptom queries were referenced to the incident leading to the current injuries. These procedures were followed both in repeated measurements of symptoms throughout the month and in the diagnostic assessment at the end of the month.

We used the PTSD section of the Diagnostic Interview Schedule for DSM-IV (DIS-IV)27 for both repeated assessment of current PTSD symptoms over the course of the month and for diagnostic assessment of PTSD conducted at the end of the month. The DIS-IV is a fully structured diagnostic interview with well-established reliability and validity.28 It has been used extensively to assess trauma in general populations,28,29 disaster survivors,21,30 combat veterans,31 motor vehicle accident victims,32 and other trauma-exposed groups.33,34

Symptom groups B (requiring ≥1 intrusion symptoms), C (requiring ≥3 avoidance/numbing symptoms), and D (requiring ≥2 hyperarousal symptoms) criteria for the diagnosis of PTSD at 1 month were determined from the prospective symptom data obtained in the repeated assessments throughout the month, including the final interview. The 1-month symptom duration (criterion E) was prospectively determined by examining the complete dataset for the occurrence of posttraumatic symptoms on all days of the entire 1-month assessment period (ie, absence of symptom-free days). The final interview queried symptom-related clinical significance or functional impairment during the month (criterion F).

Data analysis

To do our data analysis, we used SAS 9.2 software (Cary, NC, USA). Demographics were summarized using raw numbers, percentages, means, standard deviations (SDs), medians, and ranges. Daily individual symptom prevalence and symptom-count variables were computed for all days of the first post-injury month. The number of days to onset of each symptom was determined based on the date the symptom was first reported. The average number of days to the onset of symptoms of a particular symptom group (mean, SD) was established from the mean onsets of all individual symptoms in the symptom group. The average number of days elapsed before criteria were cumulatively fulfilled was calculated for symptom groups B (≥1 of 5 possible symptoms), C (≥3 of 7 possible symptoms), and D (≥2 of 5 possible symptoms), and for PTSD.

Cumulative frequency distribution curves were constructed depicting development of symptoms over the course of 1 month. A cumulative curve for meeting PTSD symptom criteria represents the cumulative proportions meeting PTSD symptom criteria over time. Curves depicting the mean numbers of groups B, C, and D days to first symptom onset were compared with one another in pairs using nonparametric survival analysis (PROC LIFETEST in SAS). Cumulative frequency distribution curves documenting the length of time to fulfill criteria for groups B, C, and D were similarly compared with one another in pairs using nonparametric survival analysis. Both comparisons of cumulative frequency distribution curves accounted for right-censored observations, using stratification on individual participant to address co-occurrence of multiple symptom groups within individuals.

The proportions of participants meeting groups B, C, and D criteria are provided at weeks 1, 2, and 3, and at 1 month; the proportion meeting PTSD criteria is provided at 1 month. Criteria met for groups B, C, and D at the 3 weekly time points were compared with 1-month PTSD status using McNemar tests that addressed repeated testing of the same individuals. Sensitivity and specificity analyses and PPV and NPV were provided for each comparison. Finally, comparisons with 1-month PTSD were made with substitution of 1, 2, 3, or 4 numbing symptoms for symptom group criteria at 1 month. The numbing variable with the optimally predictive number of symptoms was compared at weekly time points with 1-month PTSD, with provision of sensitivity, specificity, and PPV and NPV.

This study examines established DSM-IV diagnostic criteria for PTSD symptom groups B, C, and D for their ability to predict PTSD at 1 month, rather than attempting to establish a global symptom threshold for prediction of PTSD. A recognized hazard of establishing data-driven thresholds to identify cases from continuous measures is the potential to arrive at overly optimistic estimates of sensitivity and specificity, especially in studies with sample sizes of <100 to 200.35 Leeflang et al35 suggested that the problem of overestimation might be overcome by starting with a prespecified cutoff value—such as that provided by the accepted diagnostic criteria—rather than deriving it by identifying an optimal cutoff from an existing database, as previous research has done.

  RESULTS

TABLE 1 summarizes demographic characteristics of the sample, which was approximately one-half male, median age 35, predominantly white, high-school educated, and two-thirds employed. Nearly one-third (31%, 13/42) had received treatment for a preexisting psychiatric condition. TABLE 1 also summarizes the type of trauma sustained by patients and indicators of their injury severity. Nearly two-thirds of the injuries were from motor vehicle accidents. Most injuries were rated as TAL II, reflecting moderate severity. The average ISS score was moderately high. There were no skull fractures or severe head injuries in this sample.


TABLE 1

Characteristics of the sample group

Variable N % Mean SD Median Range
Demographics
Sex
Male 20 47.6%        
Female 22 52.4%        
Age (years)     35.9 12.8 35 18 to 65
Ethnicity
White 28 66.7%        
African American 8 19.0%        
Hispanic 5 11.9%        
Native American 1 2.4%        
Education
Years of education     13.0 2.1 13 7 to 17
Education level            
  <High school 6 14.3%        
  High school graduate 32 76.2%        
  College graduate 4 9.5%        
Employment status
Employed 28 66.7%        
  Full time 22 52.4%        
  Part time 6 14.3%        
Unemployed 6 14.3%        
Housewife 4 9.5%        
Student 2 4.8%        
Retired 1 2.4%        
Other 1 2.4%        
Disabled 0 0.0        
Injury characteristics
Incident type
Motor vehicle accident 27 64.3%        
Motor vehicle or pedestrian accident 4 9.5%        
Gunshot wound 3 7.1%        
Fall 3 7.1%        
Burn 3 7.1%        
Stabbing 1 2.4%        
Assault 1 2.4%        
Injury severity level in emergency department
Trauma team activation
Level I 11 26.2%        
Level II 28 66.7%        
Level III 3 7.1%        
Injury Severity Score     11.7 1.5 12 3 to 12
SD: standard deviation.

Study enrollment occurred on the first hospital day for 5 patients; other patients entered the study on hospital day 2 (n = 11), 3 (n = 7), 4 (n = 6), 6 (n = 3), or 7 (n = 10). All 42 patients were enrolled within the first hospital week. In the first 6 hospital days, 53% of symptom data were collected prospectively. Virtually all of the weekly assessment data were collected prospectively: all 42 participants were interviewed on study day 7, 98% (n = 41) on day 14, and 95% (n = 40) on days 21 and 30. However all missing prospective data were obtained retrospectively on the next available day; the resulting dataset had no missing values.

FIGURE 1 depicts the prevalence of individual posttraumatic symptoms occurring during the first post-injury month. The most prevalent symptoms were irritability or anger outbursts (95%), insomnia (86%), and psychological distress after reminders of the event (86%). Full criteria for PTSD were met by 59% of the patients. Symptom group criteria were met for groups B, C, and D by 95%, 62%, and 95% respectively. Significantly fewer patients met criteria for group C than for groups B and D (for both comparisons, McNemar χ2 = 12.25, df = 1, P < .001).

FIGURE 1: Proportions reporting individual PTSD symptoms during the first 30 post-injury days
PTSD: posttraumatic stress disorder.

FIGURE 2 displays the mean number of days to onset of each individual symptom (represented by bars), the mean of the mean number of days to onset of individual symptoms by group (dotted horizontal lines), and mean number of days to first meeting symptom group criteria (solid horizontal lines), separately for each symptom group. Group C symptoms had a slower average onset (mean = 6.8, SD = 4.3 days) relative to symptoms in groups B (mean = 5.5, SD = 5.4 days; Wilcoxon χ2 = 7.81, P = .005) and D (mean = 4.7, SD = 3.7 days; Wilcoxon χ2 = 10.00, P = .002). Times to onset of symptoms were equivalent for groups B and D (Wilcoxon χ2 = .90, P = .343).

FIGURE 2: Mean number of days to onset of each individual PTSD symptom, to mean onset of individual symptoms by group, and to first meeting symptom group criteria
PTSD: posttraumatic stress disorder.

FIGURE 3 displays cumulative frequency distribution curves reflecting the cumulative proportion of cases meeting symptom group and diagnosis criteria over the course of the first post-injury month. The frequency distribution curves for first meeting symptom criteria show a slower accumulation rate for group C relative to the curves for groups B and D, reflecting a longer time to develop full criteria for symptom group C (mean = 6.4, SD = 5.4 days) than for groups B (mean = 3.8, SD = 5.1 days; Wilcoxon χ2 = 11.56, P < .001) and D (mean = 2.8, SD = 4.4 days; Wilcoxon χ2 = 6.76, P = .009). The curves for first meeting criteria were equivalent for groups B and D (Wilcoxon χ2 = .12, P = .749). Group B and D criteria were met by approximately two-thirds of patients on the first day, leaving little room for ascent of the cumulative onset curve to its ultimate 95% saturation, which occurred in the third week. Symptom group criteria for Group C had a considerably slower onset with only 62% maximal saturation, also occurring in the third week. The cumulative PTSD curve closely followed the group C curve.

FIGURE 3: Cumulative proportion meeting PTSD symptom criteria and symptom group criteria over 1 month
PTSD: posttraumatic stress disorder.

PTSD was diagnosed at 1 month in 63%, 96%, and 63%, respectively, among those who satisfied B, C, and D group criteria during the month. FIGURE 4 shows relationships between cumulative group C criteria since the trauma, assessed at the end of each week, and full PTSD criteria at 1 month. At each weekly assessment, virtually all those who had met group C criteria proceeded to fulfill 1-month PTSD criteria. Nearly one-third (32%) of those without group C criteria by the end of the first week would develop PTSD by 1 month, but by 2 weeks, only a small fraction (6%) of those without group C criteria would develop PTSD. The optimal prediction point was at 2 weeks, when the presence of group C criteria correctly predicted 1-month PTSD in 41 of 42 patients. PTSD did not occur by 1 month in individuals who had not met group C criteria by 2 weeks.

FIGURE 4: Proportions of patients cumulatively meeting vs not meeting group C criteria by week who met PTSD criteria at 1 month
PTSD: posttraumatic stress disorder.

TABLE 2 provides summary statistics for prediction of 1-month PTSD from cumulative fulfillment of group C criteria at weekly intervals. In the first week, group C lacked specificity, but by the second week, group C performed optimally in predicting PTSD, with 100% specificity and 94% sensitivity. The performance of group C in predicting 1-month PTSD continued to be strong in subsequent weeks. Groups B and D had low specificity in predicting 1-month PTSD at all weekly time points.

Because the group C (numbing) symptoms are considered to be a robust marker of pervasive posttraumatic disturbance in retrospective studies,2 additional analyses explored the potential for numbing symptoms alone to predict 1-month PTSD prospectively (TABLE 2). The cutoff for the number of numbing symptoms assessed at 1 month that was most closely associated with 1-month PTSD was 2 symptoms (sensitivity = 92%; specificity = 83%). Thus, group C criteria had better specificity than the 2-symptom numbing measure in specificity at all 4 weeks and equal or better in sensitivity in all except the first week.


TABLE 2

Groups B, C, and D criteria for likelihood of meeting full PTSD criteria at 1 month: weekly sensitivity/specificity and PPV/NPV

  Week 1
(%)
Week 2
(%)
Week 3
(%)
1 month
(%)
Group B
Sensitivity 96 100 100 100
Specificity 53 29 12 12
PPV 75 68 63 63
NPV 90 100 100 100
Group C
Sensitivity 68 96 100 100
Specificity 100 100 94 94
PPV 100 100 96 96
NPV 68 94 100 100
Group D
Sensitivity 96 96 100 100
Specificity 24 24 12 12
PPV 65 65 63 63
NPV 80 80 100 100
2 numbing symptoms
Sensitivity 100 95 90 92
Specificity 65 70 71 83
PPV 64 72 76 88
NPV 100 94 88 88
NPV: negative predictive value; PPV: positive predictive value; PTSD: posttraumatic stress disorder.

  DISCUSSION

This study successfully charted the prospective development of posttraumatic symptoms throughout the first month following exposure to trauma. Meeting criteria for symptom groups B and D occurred quickly, with approximately two-thirds meeting group B and D criteria on the first post-trauma day, to plateau at 95% saturation in the third week. In contrast, group C symptoms began more slowly, with only a fraction meeting group C criteria on the first post-trauma day and reaching 62% of saturation, also in the third week. The group C and PTSD cumulative onset curves tracked closely together throughout the month, demonstrating the critical role of group C in the development of PTSD in the first post-trauma month.

Although symptom group C has been well established as a robust marker for PTSD in retrospective studies, the current study is the first to our knowledge to demonstrate this association prospectively across the 1-month post-trauma time frame: fulfillment of group C criteria in the first days and weeks after trauma exposure successfully identified PTSD well before 1 month in this sample. At 1 week, group C criteria accurately predicted who would develop PTSD by 1 month, and by 2 weeks group C criteria also predicted who would not develop PTSD by 1 month. Group C demonstrated high specificity (94% to 100%) for 1-month PTSD at all weekly assessments and high sensitivity (96% to 100%) from the second week onward. These levels of sensitivity and specificity are comparable with acceptable values for predictors of disease, such as those reported in biologic tests for diagnosis of medical disorders—89% to 99% in dipstick tests for malaria36 and 90% to 96% in serologic tests for influenza.37

The overall design of this study had 3 core strengths. First, we did multiple prospective assessments over the course of the first post-trauma month, creating detailed cumulative symptom development curves. Second, this study not only predicted the likelihood of PTSD, it provided a means to identify specific individuals destined to develop PTSD and to differentiate them from those who would not develop the disorder. Third, to avoid overestimation problems that can result from identifying thresholds as predictors from the study’s own database, this study instead used a priori predictors based on established DSM-IV symptom group criteria. Additional methodological strengths of this study include the high participation rate and low attrition rate, nearly complete prospective data at weekly assessments, and the use of structured interviews to assess full diagnostic criteria for both early symptoms and for the 1-month PTSD diagnosis. This was a highly trauma-exposed sample: all study participants met the PTSD trauma exposure criterion and experienced substantial physical trauma as indicated by the severity of their injuries.

The study had an 11% attrition rate; we could identify no attrition bias in baseline variables, but possibly undetectable attrition bias was present. A potential study limitation was discontinuity in study enrollment: we consecutively recruited the patients from admissions to a trauma service during 3 specified enrollment periods with intervening breaks. These breaks were for study personnel-related reasons rather than patient-related reasons, thus not representing an imaginable source of sampling bias, although this possibility cannot be completely ruled out. Additionally, gaps in the prospective symptom data were supplemented with retrospective symptom report data from subsequent days (approximately one-half of the data collected in the first 6 days and all the days between subsequent weekly interviews). However, almost all weekly assessment points—those used for predicting 1-month PTSD—had nearly complete (95% to 100%) prospective data. Recall bias for symptoms on the days of retrospective data collection could have caused loss of precision in the cumulative symptom group onset curves for data points representing the days outside of the weekly assessment points.

This study’s sample represented hospitalized trauma patients with injuries sustained in community incidents (especially motor vehicle accidents). Therefore, the findings may not apply to survivors of other types of trauma (eg, combat trauma, community-wide disasters, or sexual assault), less severely injured individuals (eg, patients discharged to home from the emergency department and those who did not seek treatment), or uninjured people who narrowly escaped harm in a traumatic incident. This study was limited to adult patients at 1 urban medical center; therefore, the results may not apply to other age groups or to other geographic and cultural settings. Because symptoms were not assessed beyond 1 month, additional research is needed to determine whether the relationships between early symptom groups and a later diagnosis of PTSD hold over a longer period of time. Repeated interviewing of study participants regarding their current symptoms over the course of a month may have influenced the findings, such as by facilitating reflection rather than avoidance, or by stimulating arousal through revisiting trauma-associated memories.

Finally, this study was limited by the relatively small size of its sample. It’s well established that statistically significant findings may emerge from small studies that may not withstand replication.35,38,39 Further research is needed to replicate and confirm this study’s prediction of PTSD from group C symptoms within the first month after trauma exposure. Replication of these findings will further establish the potential clinical use of group C for early identification of people who will develop PTSD by 1 month.

Having the confident ability to identify PTSD cases as early as the first post-trauma week may allow commencement of treatment well before 1 month, which has the potential to: 1) prevent extra weeks of suffering and 2) permit prophylaxis against further development of symptoms into full PTSD. Being able to rule out PTSD by 2 weeks among those without group C symptoms can further provide reassurance that PTSD is unlikely to develop in these individuals in upcoming weeks, avoiding the risk of inappropriately treating symptoms not representing PTSD.

  CONCLUSIONS

In the first post-trauma month, symptoms emerge rapidly, especially group B and D symptoms. At 1 week, group C criteria accurately predicted who would develop PTSD by 1 month, and by 2 weeks, group C criteria also predicted who would not develop PTSD by 1 month. The findings, if replicated, may permit earlier identification of PTSD and more timely application of specific treatment for the disorder.

DISCLOSURES: The authors report no financial relationship with any company whose products are mentioned in this article of with manufacturers of competing products.

ACKNOWLEDGEMENT: Points of view in this document are those of the authors and do not necessarily represent the official position of the Department of Veterans Affairs or the United States government.

    REFERENCES

  1.  Diagnostic and statistical manual of mental disorders, 3rd ed. Washington DC: American Psychiatric Association; 1980.
  2. Breslau N, Reboussin BA, Anthony JC, et al. The structure of posttraumatic stress disorder: latent class analysis in 2 community samples. Arch Gen Psychiatry. 2005;62:1343–1351.
  3. North CS, Nixon SJ, Shariat S, et al. Psychiatric disorders among survivors of the Oklahoma City bombing. JAMA. 1999;282:755–762.
  4.  Diagnostic and statistical manual of mental disorders, 4th ed, text revision. Washington DC: American Psychiatric Association; 2000.
  5. Shalev AY, Freedman S. PTSD following terrorist attacks: a prospective evaluation. Am J Psychiatry. 2005;162:1188–1191.
  6. Shalev AY, Freedman S, Peri T, et al. Predicting PTSD in trauma survivors: prospective evaluation of self-report and clinician-administered instruments. Br J Psychiatry. 1997;170:558–564.
  7. O’Donnell ML, Creamer M, Elliott P, et al. Tonic and phasic heart rate as predictors of posttraumatic stress disorder. Psychosom Med. 2007;69:256–261.
  8. Rothbaum BO, Foa EB, Riggs DS, et al. A prospective examination of post-traumatic stress disorder in rape victims. J Trauma Stress. 1992;5:455–475.
  9. Nishi D, Matsuoka Y, Yonemoto N, et al. Peritraumatic Distress Inventory as a predictor of post-traumatic stress disorder after a severe motor vehicle accident. Psychiatry Clin Neurosci. 2010;64:149–156.
  10. Guze SB. Why psychiatry is a branch of medicine. New York NY: Oxford University Press; 1992.
  11. Bryant RA. Early predictors of posttraumatic stress disorder. Biol Psychiatry. 2003;53:789–795.
  12. Bryant RA. Acute stress disorder as a predictor of posttraumatic stress disorder: a systematic review. J Clin Psychiatry. 2011;72:233–239.
  13. Breslau N. Epidemiologic studies of trauma posttraumatic stress disorder, and other psychiatric disorders. Can J Psychiatry. 2002;47:923–929.
  14. Ehlers A, Mayou RA, Bryant B. Psychological predictors of chronic posttraumatic stress disorder after motor vehicle accidents. J Abnorm Psychol. 1998;107:508–519.
  15. Kessler RC, Sonnega A, Bromet E, et al. Posttraumatic stress disorder in the National Comorbidity Survey. Arch Gen Psychiatry. 1995;52:1048–1060.
  16. North CS. Epidemiology of disaster mental health response. In: Ursano RJ Fullerton CS, Wiesaeth L, et al, eds. Textbook of disaster psychiatry. New York, NY: Cambridge University Press; 2007:29–47.
  17. Ozer EJ, Best SR, Lipsey TL, et al. Predictors of posttraumatic stress disorder and symptoms in adults: a meta-analysis. Psychol Bull. 2003;129:52–73.
  18. North CS, Suris AM, Davis M, et al. Toward validation of the diagnosis of posttraumatic stress disorder. Am J Psychiatry. 2009;166:34–41.
  19. Norris FH. Epidemiology of trauma: frequency and impact of different potentially traumatic events on different demographic groups. J Consult Clin Psychol. 1992;60:409–418.
  20. McMillen JC, North CS, Smith EM. What parts of PTSD are normal: intrusion avoidance, or arousal? Data from the Northridge, California, earthquake. J Trauma Stress. 2000;13:57–75.
  21. North CS, Pfefferbaum B, Narayanan P, et al. Comparison of post-disaster psychiatric disorders after terrorist bombings in Nairobi and Oklahoma City. Br J Psychiatry. 2005;186:487–493.
  22. Maes M, Delmeire L, Schotte C, et al. Epidemiologic and phenomenological aspects of post-traumatic stress disorder: DSM-III-R diagnosis and diagnostic criteria not validated. Psychiatry Res. 1998;81:179–193.
  23. North CS, Kawasaki A, Spitznagel EL, et al. The course of PTSD, major depression, substance abuse, and somatization after a natural disaster. J Nerv Ment Dis. 2004;192:823–829.
  24. Verger P, Dab W, Lamping DL, et al. The psychological impact of terrorism: an epidemiologic study of posttraumatic stress disorder and associated factors in victims of the 1995-1996 bombings in France. Am J Psychiatry. 2004;161:1384–1389.
  25. Baker SP, O’Neill B, Haddon W Jr, et al. The injury severity score: a method for describing patients with multiple injuries and evaluating emergency care. J Trauma. 1974;14:187–196.
  26.  American College of Surgeons Committee on Trauma. Resources for optimal care of the injured patient. Chicago IL: American College of Surgeons; 2006.
  27. Robins LN, Cottler LB, Bucholz KK, et al. The Diagnostic Interview Schedule for the DSM-IV (DIS-IV). St. Louis, MO: Washington University School of Medicine; 2000.
  28. Breslau N, Kessler RC, Chilcoat HD, et al. Trauma and posttraumatic stress disorder in the community: the 1996 Detroit Area Survey of Trauma. Arch Gen Psychiatry. 1998;55:626–632.
  29. Kessler RC, McGonagle KA, Zhao S, et al. Lifetime and 12-month prevalence of DSM-III-R psychiatric disorders in the United States. Results from the National Comorbidity Survey. Arch Gen Psychiatry. 1994;51:8–19.
  30. North CS. Approaching disaster mental health research after the 9/11 World Trade Center terrorist attacks. Psychiatr Clin North Am. 2004;27:589–602.
  31. Fu Q, Koenen KC, Miller MW, et al. Differential etiology of posttraumatic stress disorder with conduct disorder and major depression in male veterans. Biol Psychiatry. 2007;62:1088–1094.
  32. Green MM, McFarlane AC, Hunter CE, et al. Undiagnosed post-traumatic stress disorder following motor vehicle accidents. Med J Aust. 1993;159:529–534.
  33. Falck RS, Wang J, Siegal HA, et al. The prevalence of psychiatric disorder among a community sample of crack cocaine users: an exploratory study with practical implications. J Nerv Ment Dis. 2004;192:503–507.
  34. Teplin LA, Abram KM, McClelland GM. Prevalence of psychiatric disorders among incarcerated women. I. Pretrial jail detainees. Arch Gen Psychiatry. 1996;53:505–512.
  35. Leeflang MM, Bossuyt PM, Irwig L. Diagnostic test accuracy may vary with prevalence: implications for evidence-based diagnosis. J Clin Epidemiol. 2009;62:5–12.
  36. Jelinek T, Grobusch MP, Schwenke S, et al. Sensitivity and specificity of dipstick tests for rapid diagnosis of malaria in nonimmune travelers. J Clin Microbiol. 1999;37:721–723.
  37. Veguilla V, Hancock K, Schiffer J, et al. Sensitivity and specificity of serologic assays for detection of human infection with 2009 pandemic H1N1 virus in U.S. populations. J Clin Microbiol. 2011;49:2210–2215.
  38. Ioannidis JP. Why most published research findings are false. PLoS Med. 2005;2:e124.
  39. Ioannidis JP. Contradicted and initially stronger effects in highly cited clinical research. JAMA. 2005;294:218–228.

CORRESPONDENCE: Jeannie B. Whitman, PhD Department of Psychiatry University of Texas Southwestern Medical Center at Dallas 6363 Forest Park Road, Suite 651 Dallas, TX 75390 USA E-MAIL: Jeannie.Whitman@UTSouthwestern.edu