Elsevier

General Hospital Psychiatry

Volume 24, Issue 6, November–December 2002, Pages 375-380
General Hospital Psychiatry

Psychiatry and primary care
Validation of the PTSD checklist in an HMO sample of women

https://doi.org/10.1016/S0163-8343(02)00203-7Get rights and content

Abstract

Although Post-traumatic Stress Disorder (PTSD) is common among patients seeking care at medical clinics, little is known about the performance of screening instruments for this disorder in these settings. Previous studies of acute trauma populations using the PTSD Checklist (PCL) have suggested that scores of 45–50 provide the best discrimination between cases and noncases. We gave the PCL to 1,225 randomly selected women enrolled in an HMO. After interviewing a sample of 261 of these women using a structured, clinician-administered PTSD interview, we compared the results of the PCL to the clinician interviews over a range of possible cut scores using Receiver Operating Characteristic analysis. The optimum balance of sensitivity and specificity for this population was a score of 30, yielding a sensitivity of .82 and specificity of .76. The positive and negative likelihood ratios for this cut score were 3.40 and 0.24, respectively. By comparison, the use of 45 as a cut score would result in very low sensitivity (.36) in this setting. The lower cut score found in this study may indicate that the use of previously published cut scores of 45–50 may not optimize the function of the PCL as a screening tool outside of acute trauma settings due to an unacceptably high number of false negative cases.

Introduction

Exposure to traumatic experiences is surprisingly common in North America [1], [2], [3]. For example, 90% of a community sample of over 2,181 individuals had previous exposure to a range of potentially traumatic events, including such experiences as military combat, torture or kidnapping (2%), mugging or serious car crash (25%), or the sudden, unexpected death of a relative or friend (60%) [3]. One consequence of such exposure can be the development of Post-traumatic Stress Disorder (PTSD). As described in the DSM IV, this diagnosis requires both the experience of a significant trauma and the presence of a constellation of symptoms that fall into three general clusters: re-experiencing the trauma, avoiding reminders of the trauma, and autonomic hyperarousal [4]. PTSD is a common disorder in the US population, with estimates suggesting lifetime prevalence rates from 1–14% [1], [2], [3], [4], [5]. Among primary care outpatients, rates of PTSD have been estimated at 7–10% [6]. Both pharmacological and behavioral approaches to treatment have shown great promise [7].

Although the majority of trauma exposures do not result in the development of the full clinical diagnosis of PTSD, there is preliminary evidence to suggest that “partial” or subsyndromal PTSD is associated with significant functional disability in the general population [5]. Many individuals with cases of subsyndromal PTSD do not come to the attention of mental health providers and remain anonymous in the community or receive symptomatic treatment in primary care without complete recognition of the presence of the disorder. Although significant advances in the detection and treatment of major depression have occurred in medical settings, a similar awareness of the high prevalence of PTSD has been hampered by the absence of efficient diagnostic tools. Hence, the development of efficient instruments to screen for PTSD in a large sample assumes greater importance.

The Post-traumatic Stress Disorder Checklist (PCL) is a self-report instrument for measuring PTSD [8] that has been extensively used in research [9], [10] and is well regarded [11]. Using a cut-off score of 50, the instrument showed a sensitivity of .82 and specificity of .83 for male war veterans [12], and a sensitivity of .69 and a specificity of .99 among women with breast cancer [6]. For civilians exposed to motor vehicle accidents and adult sexual assault, a cut score of 44 demonstrated .94 sensitivity and .86 specificity [13]. Despite this extensive use, however, its psychometric properties have been examined only in samples of veterans or selected groups of civilians exposed to acute trauma. Hence, it is difficult to determine optimal cut points for civilian populations that might have a wider range of traumatic experiences over a longer time period. As Blanchard and colleagues recommend [13], PCL cut-off scores need to be understood in the context of gender differences and trauma type differences.

We are unaware of any studies that have validated this instrument against a gold standard in a population sample. There are several ways in which a population sample might yield lower optimal cut scores. First, although the full range of traumatic experiences may be found in such a sample, it is likely that lower severity traumas might be better represented and that more severe traumatic experiences that have been previously studied (e.g., war experiences, recent motor vehicle accidents) would have less influence on the sensitivity and specificity measures. Secondly, in addition to examining individuals with recent trauma, a population sample would likely also contain individuals who have had traumatic experiences that may be years or decades old. At least some of these individuals have a more complex form of PTSD that may affect the performance of the instrument. Finally, one of the most common traumatic experiences in the population is childhood maltreatment, and the psychometric properties of the PCL for adult survivors of childhood maltreatment have yet to be determined.

We included the PTSD Checklist in large population survey examining correlates of childhood maltreatment in a random sample of 1225 female HMO enrollees. In this study we examined the validity of the PCL against a gold-standard clinician interview for PTSD, and computed a sensitivity analysis showing the sensitivity, specificity, positive and negative predictive values and likelihood ratios for the instrument at various cut scores.

Section snippets

Setting

The study took place during 1996–97 among the enrolled members of Group Health Cooperative of Puget Sound (GHC), a large staff model HMO that provides primary and specialty medical and mental health services to approximately 400,000 individuals in the Puget Sound area of Washington State. The enrollees are broadly representative of the population of greater Seattle on all demographic variables. The project was a joint collaboration between the University of Washington Department of Psychiatry

Results

Of the 1,963 surveys initially mailed, 1,225 were completed and an additional 51 were ineligible (e.g., no longer insured by the HMO, deaths, etc.) or returned as undeliverable, for a final adjusted return rate of 62%. The 1,225 participants had a mean ± S.D. age of 41.8 ± 11.5 years. Fifty one percent of the women were married and 57% had completed college. The median income was approximately $40,000. The participants were 79% Caucasian, 6% African-American, 8% Asian, 2% Hispanic and 1% Native

Discussion

In this HMO sample of women we found that a cut score of approximately 30 provides the best combination of sensitivity and specificity for the PCL in screening for current PTSD as determined by the CAPS. This is lower than cut scores previously reported for other military and civilian populations [12], [13]. The finding of a lower PCL threshold may be due to the use of a population sample where there may be less severe traumatic experiences than those studied elsewhere as well as trauma

References (28)

  • E.B. Foa et al.

    Effective treatments for PTSD

    (2000)
  • F.W. Weathers et al.
  • G.L. Roberts et al.

    How does domestic violence affect women’s mental health?

    Women Health

    (1998)
  • S.L. Manne et al.

    Posttraumatic stress disorder among mothers of pediatric cancer survivorsdiagnosis, comorbidity, and utility of the PTSD Checklist as a screening instrument

    J Pediatr Psychology

    (1998)
  • Cited by (183)

    • Elucidating posttraumatic stress symptom dimensions and health correlates among postpartum women

      2021, Journal of Affective Disorders
      Citation Excerpt :

      The PCL had excellent internal consistency in our sample (Cronbach's α=0.91). For descriptive purposes, we examined PTSD as a unitary construct, using a cut-off score of ≥30 to indicate elevated PTSD symptom severity; this cut-off was developed and validated in an HMO sample of women (Walker et al., 2002). For the CFA, responses to the 17 PCL items were used as indicators of dimensional latent variables.

    View all citing articles on Scopus

    This study was funded by a grant from the National Institutes of Mental Health K-20 MH-01106.

    View full text