Test of a single-item posttraumatic stress disorder screener in a military primary care setting

Available online at www.sciencedirect.com

General Hospital Psychiatry 30 (2008) 391 – 397

Psychiatry and Primary Care Recent epidemiologic studies have found that most patients with mental illness are seen exclusively in primary care medicine. These patients often present with medically unexplained somatic symptoms and utilize at least twice as many health care visits as controls. There has been an exponential growth in studies in this interface between primary care and psychiatry in the last 10 years. This special section, edited by Jürgen Unutzer, M.D., will publish informative research articles that address primary care-psychiatric issues.

Test of a single-item posttraumatic stress disorder screener in a military primary care setting☆ Kristie L. Gore, Ph.D. a,b,⁎, Charles C. Engel, M.D., M.P.H. a,b , Michael C. Freed, Ph.D. a,b , Xian Liu, Ph.D. a,b , David W. Armstrong III, Ph.D., F.A.C.S.M. a,b a

Deployment Health Clinical Center, Walter Reed Army Medical Center, Washington, DC 20307, USA b Uniformed Services University of the Health Sciences, Bethesda, MD 20814, USA Received 28 January 2008; accepted 11 May 2008

Abstract Background: Posttraumatic stress disorder (PTSD) is prevalent in primary care, frequently goes undetected and can be highly debilitating when untreated. Objective: We assessed the operating characteristics of a single-item PTSD screener (SIPS) for primary care and compared it to a commonly used four-item primary care PTSD screener (PC-PTSD). The SIPS asks: “Were you recently bothered by a past experience that caused you to believe you would be injured or killed … not bothered, bothered a little, or bothered a lot?” Methods: A total of 3234 patients from three Washington, DC, area military primary care clinics completed the SIPS. Independent, blinded assessments using a structured diagnostic PTSD interview were completed in 213 of these patients. Results: The SIPS yielded a reasonable range of likelihood ratios, suggesting capacity to discriminate between low- and high-probability PTSD patients. However, the SIPS sensitivity was only 76% for those reporting “bothered a little” and the four-item PC-PTSD yielded significantly better test characteristics on Receiver–Operator Curve analysis. Conclusion: A single, user-friendly primary care PTSD screening question with three response options, while sensible and worth further investigation, failed to offer sound test characteristics for PTSD screening. Ways of improving SIPS performance are discussed. © 2008 Elsevier Inc. All rights reserved. Keywords: PTSD; Screening; Primary care

1. Background Posttraumatic stress disorder (PTSD) is prevalent in primary care [1–3], frequently goes undetected [2] and can ☆

Disclaimer: The views expressed herein are those of the authors and do not necessarily represent the official policy or position of the Deployment Health Clinical Center, Walter Reed Army Medical Center, Uniformed Services University of the Health Sciences, Department of Defense or the United States Government. ⁎ Corresponding author. Tel.: +1 202 356 1074; fax: +1 202 356 1090. E-mail address: [email protected] (K.L. Gore). 0163-8343/$ – see front matter © 2008 Elsevier Inc. All rights reserved. doi:10.1016/j.genhosppsych.2008.05.002

be highly debilitating when untreated [1,4–6]. Approximately 7–8% of adults in the general population will suffer from the disorder during their lifetime [5,7]. PTSD prevalence in a recently deployed military population is perhaps even higher [8–10]. People with PTSD experience significant social, occupational and physical impairment [4]; have high rates of co-occurring mental disorders [5,7]; and are disproportionately high users of health care [1,6]. Effective treatments exist [11–15] and some evidence suggests that early identification and intervention improves outcomes [16].

392

K.L. Gore et al. / General Hospital Psychiatry 30 (2008) 391–397

Primary care represents an opportunity to identify untreated individuals with PTSD. PTSD is relatively common in primary care settings, with rates in various studies ranging from 7.5% to 36% [2,3,17,18]. Seal et al. [10] recently found that 13% of Iraq or Afghanistan conflict veterans receiving VA care had a chart diagnosis of PTSD, and 42% of all mental health diagnoses were made in primary care settings. In addition, recently traumatized individuals may be more likely to present to primary care than to specialty mental health settings. For example, 72.6% of sexual assault victims seeking crisis assistance had sought medical treatment within the previous year while only 19.1% had sought specialty mental health care [19]. Unfortunately, studies find that the likelihood of PTSD detection in routine primary care practice is low. In one study of 26 Israeli primary care clinics involving 2975 patients, only 2% of those with current PTSD were given a PTSD diagnosis by their primary care physician [17]. Nearly half of patients with PTSD in a large systematic, multisite VA primary care sample were found to be undiagnosed [2]. Low rates of detection suggest the need to pursue new and efficient ways to recognize PTSD in primary care settings. Primary care screening measures should be brief, user friendly, require minimal training to administer and score, be acceptable to both patient and provider and contain the minimum number of items necessary for sensitive but accurate case identification [20–22]. If busy primary care clinics and providers could effectively screen patients for PTSD with one easily memorized question, it may increase the likelihood of successful screening program implementation. Therefore, this report describes our effort to develop a brief, easy-to-remember, simple-to-administer, valid and reliable PTSD screen that can be routinely implemented in primary care settings to better detect patients with PTSD. We hypothesized that a single-item PTSD screener (SIPS) designed for primary care use would show sound criterion validity, likelihood ratios (LRs) and test–retest reliability and that it would perform comparably to the briefest, validated PTSD screener in common use, the primary care PTSD screener (PCPTSD) [18], a four-item measure widely used in the VA and military health care systems. An effective single-item PTSD screen, accompanied by an adequately resourced, deliverable plan to provide appropriate care for patients who screen positive, could ultimately reduce the burden of PTSD — on patients and on the health care system.

2. Methods 2.1. Screening measures 2.1.1. Single-item PTSD screener The SIPS we tested was selected from among several candidate questions. Candidate SIPS were written, discussed among investigators and then reviewed with patients and providers to insure that they were easily understood, intuitive and acceptable. Candidate questions were refined based on

feedback and reviewed again by the investigators, with the selection of one SIPS for further testing. This question read as follows: “Were you recently bothered by a past experience that caused you to believe you would be injured or killed? (For example: witnessed or experienced a serious accident or illness, threatened with a weapon, physically or sexually assaulted, experienced a natural disaster, participated in wartime combat)?” Response options were “not bothered at all,” “bothered a little,” and “bothered a lot.” The SIPS did not systematically assess traumatic events. Traumatic events are reported by the majority of adults in the general population and only a minority develop PTSD, suggesting traumatic events lack PTSD diagnostic specificity [5,23]. 2.2. Comparison screening measure We selected the shortest available validated PTSD screener to compare to the SIPS. The VA mandated the use of the four-item PC-PTSD [18] as a primary care screening tool in 2004 [24]. The Department of Defense is using it as a key part of its postdeployment health screening programs [9,25]. The PC-PTSD reads, “In your life, have you ever had any experience that was so frightening, horrible, or upsetting that, in the past month you ... 1) Had nightmares about it when you did not want to? 2) Tried hard not to think about it or went out of your way to avoid situations that reminded you of it? 3) Were constantly on guard, watchful, or easily startled? 4) Felt numb or detached from others, activities, or your surroundings?” Each question uses yes/no response options yielding a screener score of 0 (all questions marked “no”) to 4 (all questions marked “yes”). Like the SIPS, the measure is brief and easy to score; however, it is challenging for providers to remember and implement without visual aid. Prins et al. [18] evaluated the PC-PTSD in 188 VA general medical and women's health clinic patients. The measure evidenced strong sensitivity (0.78) and specificity (0.87) at the optimal cutoff score of 3 versus the ClinicianAdministered PTSD Scale (CAPS) [26] diagnostic interview and performed nearly as well as the psychometrically sound 17-item PTSD Checklist (PCL) [27]. We wanted to create a SIPS that was user friendly and easy to remember and that offers comparable test characteristics to the PC-PTSD. 2.3. Participants and procedures An estimated 3700 consecutive adult (age range, 18–65) patients in three military health system primary care clinics in the Washington, DC, area were asked to voluntarily complete the SIPS in the clinic waiting room. Patients included a mix of uniformed service members, retirees and family members. Patients also completed five demographic questions (age, sex, race, military status and rank) plus a question assessing potential interest in further participation. Only a few patients refused these brief questions. A subgroup of individuals consented to a longer assessment involving completion of the PC-PTSD, the PCL and an independent, blinded, structured diagnostic interview to assess our criterion variable, PTSD

K.L. Gore et al. / General Hospital Psychiatry 30 (2008) 391–397

diagnosis. Of those who endorsed a willingness for further participation, approximately 1 of every 10 patients endorsing “not bothered,” 1 out of every 4 endorsing “bothered a little” and 1 out of every 2 endorsing “bothered a lot” were invited to participate in a diagnostic interview and questionnaire packet (see Section 2.4). This oversampling of patients endorsing “bothered a little” and “bothered a lot” responses was done to insure an adequate sample of individuals with PTSD. Interviewers were trained mental health professionals, either research psychologists or advanced psychology doctoral candidates, and were blinded to SIPS responses. All interviews were audiotaped, and a random 10% (n=20) were independently reviewed for quality control purposes and to insure interview reliability (100% concordance in 19 interviews; the quality of one tape was too poor to evaluate). To assess the test–retest reliability of the SIPS, we mailed participants the SIPS and the PCL to complete and return approximately 1 week after the diagnostic interview in a postage-paid envelope. This research was fully approved by the Walter Reed Army Medical Center Department of Clinical Investigations and by the Institutional Review Board at the Uniformed Services University of the Health Sciences. 2.4. Other study measures 2.4.1. PTSD Checklist The PCL [28] is a widely used 17-item self-report questionnaire that asks respondents to rate how bothersome PTSD symptoms were in the past month (1=not at all; 5=extremely). The internal consistency of the PCL is 0.94, and it possesses strong convergent validity with other self-report measures of PTSD [27]. At the optimal cutoff score of 44 in a convenience sample of largely female (92%) motor vehicle accident and sexual assault trauma victims, the PCL diagnostic efficiency (i.e., the number of false-positive results plus the number of false-negative results divided by the number of all screening results) was 0.90, sensitivity was 0.94 and the specificity was 0.86 using the CAPS as the criterion comparison measure [27]. There are three PCL versions [28], a military version that asks veterans about symptoms “in response to a stressful military experiences,” a civilian version that asks people about symptoms “in response to stressful life experiences” and a single-event version that asks respondents to identify a single event and relates symptoms to the identified event. We opted to use the civilian version because most military health system patients are not currently serving in the military and because we wanted to assess symptoms due to traumatic events other than military service per se. 2.4.2. PTSD Symptom Scale — Interview Version (PSS-I) The criterion diagnosis was assessed using the PSS-I, a 17item semistructured interview that can be administered in approximately 20 min [29]. It evaluates both the presence and severity of the 17 symptoms of PTSD using a four-point Likert scale (0=not at all; 3=five or more times a week/very much) to

393

derive a severity score ranging from 0 to 51. The PSS-I has strong psychometric properties. Against the PTSD module of the Structured Diagnostic Interview for DSM-IV [30], the sensitivity and specificity of the PSS-I were 0.88 and 0.96, respectively, and the PSS-I correctly identified the PTSD status of 94% of the 64 participants [29]. Foa et al. [29] also found that the test–retest reliability of the PSS-I over a 1-month period was 0.80 in 93 participants and the interrater reliability (k) was 0.91 for 43 women who were interviewed twice. Compared to the CAPS [26], which requires approximately 1 h for administration, one study found the PSS-I and the CAPS derive equivalent diagnoses [31]. Interviewers used a modified version of the Life Events Checklist from the CAPS [26,32] to identify the respondents' index trauma and queried symptom presence over the last 1 month. 2.5. Analysis plan We examined demographics and SIPS responses from the consecutive sample to characterize the primary care population. Intercorrelations among the SIPS, the PC-PTSD and the PCL were evaluated to assess convergent validity. We developed weighted Receiver Operating Characteristic (ROC) curves [33] and calculated weighted [34] and unweighted sensitivity (the probability of a positive test among patients with the disorder), specificity (the probability of a negative test among patients without the disorder), positive predictive values (PPVs; the probability of disorder among patients with a positive test), negative predictive values (NPVs; the probability of no disorder among patients with a negative test) and multilevel LRs (the ratio of the probability of true screening positives over the probability of false screening positives [35,36]) based on the criterion PTSD diagnosis from the interviewed sample. LRs combine sensitivity and specificity to describe the impact of test results on the odds that the disease is present in the respondent. They allow for the assessment of multiple levels of test scores and are not impacted by pretest probability of disease [36]. An LR equal to 1 indicates no increased or decreased odds of disease; the odds of a disease increase when LRs are greater than 1 and decrease when LRs are less than 1 [36]. Unweighted, unadjusted operating characteristics were calculated using only the SIPS and PSS-I data from the interviewed participants. To reduce bias associated with the oversampling of “bothered a little” and “bothered a lot” respondents, we calculated test characteristics with weighted data. Specifically, we calculated weights for each participant in the interviewed sample to adjust for the biased distribution of the three SIPS response options and correct for the oversampling of “bothered a little” and “bothered a lot” respondents [34]. The weights for the “not bothered,” “bothered a little” and “bothered a lot” respondents were 2.25, 0.44 and, 0.29 respectively. With such weighted data, the calculation of sensitivity, specificity, PPV, NPV and multilevel LRs derived population estimates. In order to evaluate the impact of demographic variables on SIPS performance with a small sample size, we calculated a

394

K.L. Gore et al. / General Hospital Psychiatry 30 (2008) 391–397

propensity score [37,38] to balance the differences in individual demographic characteristics while evaluating the diagnostic probability of a given SIPS response. The propensity score in this study was defined as the conditional probability of being diagnosed with PTSD given values of age, sex, race, military status and rank. We compared the adjusted and unadjusted (including the propensity score or not) LRs to determine whether demographic variables affected operating characteristics of the SIPS. All analyses were completed using SAS version 9.1.3. 3. Results 3.1. Sample characteristics Fig. 1 visually displays dropout, response rates and distribution of responses at various study stages. It shows the number of participants interviewed and the presence or absence of PTSD by SIPS response category. In the consecutive primary care patient sample, 74% said that they were “not bothered” by a past traumatic experience, 19% said that they were “bothered a little” and 7% said that they were “bothered a lot.” Of 229 people consenting to interview, 213 (93%) were interviewed. Sixteen individuals dropped out prior to interview because of time constraints but completed other study measures. Among those interviewed, 33% reported that they were “not bothered,” 44% “bothered a little” and 23% “bothered a lot.” Characteristics of the consecutive and interviewed samples are presented in

Table 1. The 213 interviewed individuals were comparable to the consecutive primary care sample in sex, age, race and military status. There was a significantly greater proportion of enlisted individuals in the interviewed sample than in the consecutive sample. However, this was in part due to the fact that we oversampled individuals who were bothered by a recent experience and in part because one of the primary care clinics we recruited from mainly serves officers, and therefore, the percentage of officers in the consecutive sample (40%) is greater than that of the military at large (15%) [39]. Within SIPS response categories, characteristics of the consecutive and interviewed samples were similar, the only exception being in the “not bothered” category, where we interviewed a greater percentage of active duty military and fewer retirees than were represented in the consecutive sample. 3.2. SIPS criterion validation ROC curves comparing the SIPS to the four-item screener versus the criterion PTSD diagnosis are presented in Fig. 2 using weighted sensitivity and specificity values. The PC-PTSD captured significantly more area under the curve [0.89, 95% confidence interval (CI)=0.84–0.94] than did the SIPS (0.77, 0.70–0.84), suggesting that the fouritem screener was a significantly more efficient measure [χ2(1)=10.38, Pb.001]. Table 2 presents the unweighted and weighted sensitivity, specificity, PPV, NPV and LRs and their 95% CIs for the SIPS and the PC-PTSD. The optimal SIPS cutoff score was

Fig. 1. Diagram of study participation.

K.L. Gore et al. / General Hospital Psychiatry 30 (2008) 391–397

395

Table 1 Sample demographics by SIPS response: percentage and number of participants in consecutive and interviewed samples Demographic variables

Sex Male Age (years) ≤30 31–40 41–50 51–60 61+ Race White Black Hispanic Other Military status Active Retired Family Rank Officer Enlisted

“Not bothered”

Total samples Consecutive (n=3234)

Interviewed (n=213)

Consecutive (n=2392)

Interviewed (n=70)

“Bothered a little”

“Bothered a lot”

Consecutive (n=629)

Interviewed (n=94)

Consecutive (n=213)

Interviewed (n=49

60 (1972)

61 (130)

61 (1459)

61 (43)

63 (395)

65 (61)

48 (102)

53 (26)

21 (686) 24 (788) 31 (1023) 16 (521) 8 (257)

24 (52) 23 (49) 31 (65) 18 (38) 4 (9)

20 (470) 24 (585) 32 (762) 16 (388) 8 (187)

19 (13) 29 (20) 33 (23) 19 (13) 1 (1)

25 (155) 24 (153) 29 (181) 15 (95) 7 (45)

26 (24) 22 (21) 28 (26) 19 (18) 5 (5)

27 (58) 19 (41) 30 (64) 15 (31) 9 (19)

31 (15) 16 (8) 33 (16) 14 (7) 6 (3)

60 (1951) 28 (909) 3 (102) 9 (287)

60 (128) 25 (53) 3 (6) 12 (26)

63 (1498) 27 (634) 3 (61) 7 (175)

67 (47) 23 (16) 3 (2) 7 (5)

53 (334) 30 (185) 5 (30) 12 (78)

62 (58) 22 (21) 4 (4) 12 (11)

42 (89) 39 (83) 4 (9) 15 (32)

47 (23) 33 (16) 0 (0) 20 (10)

62 (1978) 18 (577) 19 (611)

68 (145) 14 (30) 18 (38)

62 (1422) 19 (444) 19 (445)

71 ⁎ (50) 6 (4) 23 (16)

66 (403) 16 (100) 17 (104)

63 (59) 21 (20) 16 (15)

64 (132) 14 (28) 23 (47)

73 (36) 12 (6) 14 (7)

40 (1275) 39 (1241)

32 ⁎ (67) 49 (102)

38 (26) 38 (26)

36 (220) 46 (283)

33 (31) 49 (46)

22 (44) 52 (106)

21 (10) 63 (30)

42 (990) 36 (847)

Each cell displays a percentage with the number of participants in parentheses. Comparisons of consecutive versus interviewed participants revealed that only rank for the total samples and military status for the “not bothered at all” groups are significantly different. ⁎ Pb.05.

“bothered a little,” yielding a 76% sensitivity. In contrast, a score of 2 or more was optimal on the PC-PTSD and it exhibited 91% sensitivity. The PPV of the SIPS “bothered a little” response was considerably lower than that of the PCPTSD (0.26 vs. 0.37 at a PC-PTSD cutoff score of 2). Thus, the SIPS “bothered a little” cutoff score was associated with a larger number of false positives. To put this in perspective, the SIPS PPV was lower than the pooled PPV reported for

single-item depression screens (0.56) [40]. The spread in LRs across SIPS response categories indicated that the measure effectively identified patient groups with relatively low and high probability of PTSD; however, the precision associated with the “bothered a lot” LR was low. For interviewed participants, the PCL means and standard deviations were 21.37 (8.14) in “not bothered at all” respondents, 28.66 (12.16) in “bothered a little” respondents and 47.04 (17.34) in “bothered a lot” respondents. 3.3. Convergent validity and test–retest reliability Spearman correlations between the SIPS, the four-item screener and PCL scores (SIPS and PC-PTSD=.59; SIPS and PCL=.63) using weighted data from the interviewed sample were statistically significant but reflect relatively low convergent validity. The test–retest reliability of the SIPS was 0.63 (n=104, Pb.001; median, 13 days to return), indicating fair SIPS reliability. The limited range of SIPS scores can be problematic when calculating correlation coefficients with other measures that have a wider range of scores and use interval scales. We anticipate that increasing the number of response options will in turn increase reliability and validity.

4. Discussion

Fig. 2. ROC curves for the SIPS and PC-PTSD scores against criterion PTSD diagnosis.

A sufficiently reliable and valid SIPS could significantly improve the implementation of PTSD screening in a busy primary care setting. Properly devised, it would be versatile and user friendly, requiring no scoring and little training to

396

K.L. Gore et al. / General Hospital Psychiatry 30 (2008) 391–397

Table 2 Sensitivity, specificity, PPV, NPV and multilevel LRs with 95% CIs Screener and score Unweighted values SIPS Not bothered Bothered a little Bothered a lot PC-PTSD 0–1 2 3 4 Weighted values SIPS Not bothered Bothered a little Bothered a lot PC-PTSD 0–1 2 3 4

Sensitivity (true positives)

Specificity (true negatives)

PPV

NPV

LRs and 95% CIs

1.0 0.95 (0.89, 1.00) 0.56 (0.41, 0.71)

– 0.40 (0.33, 0.47) 0.85 (0.80, 0.91)

– 0.29 (0.21, 0.36) 0.49 (0.35, 0.63)

– 0.97 (0.93, 1.00) 0.88 (0.84, 0.93)

0.12 (0.03, 0.46) 0.87 (0.58, 1.31) 3.80 (2.42, 5.95)

1.00 0.91 (0.82, 0.99) 0.79 (0.67, 0.91) 0.51 (0.36, 0.66)

– 0.71 (0.64, 0.78) 0.85 (0.79, 0.90) 0.94 (0.91, 0.98)

– 0.44 (0.34, 0.55) 0.57 (0.44, 0.69) 0.69 (0.53, 0.85)

– 0.97 (0.94, 1.00) 0.94 (0.90, 0.98) 0.88 (0.84, 0.93)

0.13 (0.05, 0.33) 0.85 (0.34, 2.12) 2.95 (1.51, 5.76) 8.65 (4.43, 16.87)

1.00 0.76 (0.57, 0.95) 0.36 (0.15, 0.58)

– 0.79 (0.73, 0.85) 0.96 (0.94, 0.99)

– 0.26 (0.14, 0.37) 0.49 (0.23, 0.75)

– 0.97 (0.95, 1.00) 0.94 (0.91, 0.97)

0.30 (0.14, 0.68) 2.28 (1.21, 4.29) 9.90 (3.88, 25.22)

1.00 0.91 (0.79, 1.00) 0.70 (0.50, 0.91) 0.47 (0.26, 0.70)

– 0.84 (0.79, 0.90) 0.92 (0.88, 0.96) 0.98 (0.96, 1.00)

– 0.37 (0.23, 0.50) 0.46 (0.27, 0.64) 0.71 (0.36, 0.96)

1.00 0.99 (0.98, 1.00) 0.97 (0.94, 0.99) 0.95 (0.92, 0.98)

0.10 (0.02, 0.44) 2.89 (1.06, 7.86) 3.64 (1.36, 9.78) 24.90 (8.07, 76.79)

Note: SIPS=Single Item PTSD Screener; PC-PTSD=the Primary Care PTSD Screen [18].

implement. Unfortunately, the question we tested, accompanied by three ordinal response options, failed to perform, as well as the most commonly used PC-PTSD, a fourquestion screen that is arguably harder to remember, implement and score. Nevertheless, the SIPS identified primary care patients at increased risk of meeting a PTSD diagnosis and performed similarly to existing single-item screens for anxiety disorders [41] and depression. Mitchell and Coyne [40] reported pooled operating characteristics for 22 studies of single-item depression screens and found sensitivity to be 32% and specificity to be 97% compared to two- and three-item screens with sensitivity at 74% and specificity at 75%. They concluded that single-item screens were useful for ruling out those without depression but not for ruling in people likely to be depressed. The SIPS' high NPV (0.97) is consistent with this notion. Our SIPS showed an acceptable range of LRs (from 0.30 for those “not bothered” by a past experience to 9.90 for those “bothered a lot”), suggesting that it effectively delineated primary care patient groups with relatively low and high probabilities of PTSD. For example, at the estimated PTSD prevalence of 9.0% noted in our primary care clinic population, the 7% of patients reporting they are “bothered a lot” have a 49% prevalence of PTSD, and we estimate that patient report of “bothered a lot” is 96% specific. In contrast, 3% met criteria for PTSD among the 74% of primary care patients reporting they were “not bothered” by a past traumatic experience. Therefore, routine administration of our SIPS in a primary care setting can at least enable providers to rapidly identify patients at high, intermediate and low probability of PTSD, an important feature when weighing the importance of extended PTSD assessment in the face of competing medical problems.

The adjusted and unadjusted LRs were nearly identical, suggesting that the SIPS performed similarly across demographic groups. Unfortunately, our interviewed sample was not large enough to produce statistically stable estimates of SIPS performance in different demographic subgroups. However, propensity score analyses [37,38] suggested that the overall impact of age, sex, race, military status and rank on SIPS results was small and statistically insignificant. Future studies with larger interviewed samples are needed to assess SIPS performance in specific demographic subgroups. Although the SIPS did not perform as well as the PCPTSD did for identifying primary care patients with PTSD, our experience using and evaluating the SIPS suggests that strategic refinements may result in a significant performance improvement. First, the time frame the question addresses could be improved so the focus is clearly on current illness. The version we tested asked if respondents were “recently bothered” and perhaps “currently bothered” would reduce the number of individuals with recent but not current PTSD that are identified as high-probability PTSD patients. Second, increasing the number of response categories using, for example, an ordinal 0 to 10 scale similar to the one routinely used in health care settings to rapidly assess pain severity may improve SIPS sensitivity. Third, simplified wording may reduce the frequency that the question is misread or misunderstood and thereby decrease rates of response misclassification. Therefore, further testing to develop an optimal SIPS remains warranted. Acknowledgments The authors would like to thank Cheryl Blount, Elizabeth Harper Cordova and Phoebe Kuesters for assistance with

K.L. Gore et al. / General Hospital Psychiatry 30 (2008) 391–397

data collection and Leigh Johnson, Rob Lippy and Kris Morris for help with diagnostic interviews. References [1] Stein MB, McQuaid JR, Pedrelli P, Lenox R, McCahill ME. Posttraumatic stress disorder in the primary care medical setting. Gen Hosp Psychiatry 2000;22(4):261–9. [2] Magruder KM, Frueh BC, Knapp RG, Davis L, Hamner MB, Martin RH, et al. Prevalence of posttraumatic stress disorder in Veterans Affairs primary care clinics. Gen Hosp Psychiatry 2005;27(3):169–79. [3] Dobie DJ, Kivlahan DR, Maynard C, Bush KR, McFall M, Epler AJ, et al. Screening for post-traumatic stress disorder in female Veteran's Affairs patients: validation of the PTSD checklist. Gen Hosp Psychiatry 2002;24(6):367–74. [4] Davidson JR, Stein DJ, Shalev AY, Yehuda R. Posttraumatic stress disorder: acquisition, recognition, course, and treatment. J Neuropsychiatry Clin Neurosci 2004;16(2):135–47. [5] Kessler RC, Sonnega A, Bromet E, Hughes M, Nelson CB. Posttraumatic stress disorder in the National Comorbidity Survey. Arch Gen Psychiatry 1995;52(12):1048–60. [6] Schnurr PP, Friedman MJ, Sengupta A, Jankowski MK, Holmes T. PTSD and utilization of medical treatment services among male Vietnam veterans. J Nerv Ment Dis 2000;188(8):496–504. [7] Kessler RC, Chiu WT, Demler O, Walters EE. Prevalence, severity, and comorbidity of 12-month DSM-IV disorders in the National Comorbidity Survey Replication. Arch Gen Psychiatry 2005;62(6): 617–27. [8] Hoge CW, Castro CA, Messer SC, McGurk D, Cotting DI, Koffman RL. Combat duty in Iraq and Afghanistan, mental health problems, and barriers to care. N Engl J Med 2004;351(1):13–22. [9] Hoge CW, Auchterlonie JL, Milliken CS. Mental health problems, use of mental health services, and attrition from military service after returning from deployment to Iraq or Afghanistan. JAMA 2006;295 (9):1023–32. [10] Seal K, Bertenthal D, Miner C, Sen S, Marmar C. Bringing the war back home. Arch Intern Med 2007;167:476–82. [11] Davidson JR, Weisler RH, Malik ML, Connor KM. Treatment of posttraumatic stress disorder with nefazodone. Int Clin Psychopharmacol 1998;13(3):111–3. [12] Foa EB, Rothbaum BO, Riggs DS, Murdock TB. Treatment of posttraumatic stress disorder in rape victims: a comparison between cognitive–behavioral procedures and counseling. J Consult Clin Psychol 1991;59(5):715–23. [13] Foa EB, Hembree EA, Cahill SP, Rauch SA, Riggs DS, Feeny NC, et al. Randomized trial of prolonged exposure for posttraumatic stress disorder with and without cognitive restructuring: outcome at academic and community clinics. J Consult Clin Psychol 2005;73(5):953–64. [14] Harvey AG, Bryant RA, Tarrier N. Cognitive behaviour therapy for posttraumatic stress disorder. Clin Psychol Rev 2003;23(3):501–22. [15] Schnurr PP, Friedman Matthew J, Engel Charles C, Foa Edna B, Shea MT, Resick PM, et al. Issues in the design of multisite clinical trials of psychotherapy: VA Cooperative Study No. 494 as an example. Contemp Clin Trials 2005;26:626–36. [16] Bryant RA. Early interventions following psychological trauma. CNS Spectr 2002;7(9):650–4. [17] Taubman-Ben-Ari O, Rabinowitz J, Feldman D, Vaturi R. Posttraumatic stress disorder in primary-care settings: prevalence and physicians' detection. Psychol Med 2001;31(3):555–60. [18] Prins A, Ouimette P, Kimerling R, Cameron RP, Hugeishofer DS, Shaw-Hegwer J, et al. The primary care PTSD screen (PC-PTSD): development and operating characteristics. Prim Care Psychiatry 2004; 9(1):9–14.

397

[19] Kimerling R, Calhoun KS. Somatic symptoms, social support, and treatment seeking among sexual assault victims. J Consult Clin Psychol 1994;62(2):333–40. [20] Brewin CR. Systematic review of screening instruments for adults at risk of PTSD. J Trauma Stress 2005;18(1):53–62. [21] Nelson EC, Berwick DM. The measurement of health status in clinical practice. Med Care 1989;27(3 Suppl):S77–S90. [22] Roy-Byrne PP, Dagadakis C, Ries R, Decker K, Jones R, Bolte MA, et al. A psychiatrist-rated battery of measures for assessing the clinical status of psychiatric inpatients. Psychiatric Services 1995;46(4):347–52. [23] Breslau N, Kessler RC, Chilcoat HD, Schultz LR, Davis GC, Andreski P. Trauma and posttraumatic stress disorder in the community: the 1996 Detroit Area Survey of Trauma. Arch Gen Psychiatry 1998;55:626–32. [24] Department of Veterans Affairs, V.H.A. Clinical reminder regarding veterans of the recent conflicts in Afghanistan and Iraq. 2004 Feb 6. [25] Milliken CS, Auchterlonie JL, Hoge CW. Longitudinal assessment of mental health problems among active and reserve component soldiers returning from the Iraq war. JAMA 2007;298(18):2141–8. [26] Blake DD, Weathers FW, Nagy LM, Kaloupek DG, Gusman FD, Charney DS, et al. The development of a Clinician-Administered PTSD Scale. J Trauma Stress 1995;8(1):75–90. [27] Blanchard EB, Jones-Alexander J, Buckley TC, Forneris CA. Psychometric properties of the PTSD Checklist (PCL). Behav Res Ther 1996;34(8):669–73. [28] Weathers FW, Litz BT, Herman JA, Huska JA, Keane TM. The PTSD Checklist (PCL): reliability, validity, and diagnostic utility. Paper presented at the Annual Convention of the International Society for Traumatic Stress Studies, San Antonio, TX; 1993. [29] Foa EB, Riggs DS, Dancu CV, Rothbaum BO. Reliability and validity of a brief instrument for assessing post-traumatic stress disorder. J Trauma Stress 1993;6(4):459–73. [30] First MB, Spitzer RL, Gibbon M, Williams JB. Structured clinical interview for DSM-IV Axis I disorders-clinician version (SCID-IV). 1996. [31] Foa EB, Tolin DF. Comparison of the PTSD Symptom Scale — Interview Version and the Clinician-Administered PTSD Scale. J Trauma Stress 2000;13(2):181–91. [32] Gray MJ, Litz BT, Hsu JL, Lombardo TW. Psychometric properties of the life events checklist. Assessment 2004;11(4):330–41. [33] Metz CE. Basic principles of ROC analysis. Semin Nucl Med 1978;8 (4):283–98. [34] Kish L. Survey sampling. New York: John Wiley & Sons; 1965. [35] Deeks JJ, Altman DG. Diagnostic tests 4: likelihood ratios. BMJ 2004; 329(7458):168–9. [36] Simel DL, Samsa GP, Matchar DB. Likelihood ratios with confidence: sample size estimation for diagnostic test studies. J Clin Epidemiol 1991;44(8):763–70. [37] Robins JM, Mark SD, Newey WK. Estimating exposure effects by modelling the expectation of exposure conditional on confounders. Biometrics 1992;48(2):479–95. [38] D'Agostino Jr RB. Propensity score methods for bias reduction in the comparison of a treatment to a non-randomized control group. Stat Med 1998;17(19):2265–81. [39] U.S. Governmental Accountability Office (GAO). Military Personnel: reporting additional servicemember demographics could enhance congressional oversight. Government Accountability Office Report GAO-05-952. Washington (DC): GAO; 2005. [40] Mitchell AJ, Coyne JC. Do ultra-short screening instruments accurately detect depression in primary care? A pooled analysis and meta-analysis of 22 studies. Br J Gen Pract 2007;57(535):144–51. [41] Means-Christensen AJ, Sherbourne CD, Roy-Byrne PP, Craske MG, Stein MB. Using five questions to screen for five common mental disorders in primary care: diagnostic accuracy of the Anxiety and Depression Detector. Gen Hosp Psychiatry 2006;28(2):108–18.