Psychoneuroendocrinology (2005) 30, 373–381

www.elsevier.com/locate/psyneuen

PTSD symptoms predict waking salivary cortisol levels in police officers Thomas C. Neylana,b,c,*, Alain Brunetd,e, Nnamdi Polef, Suzanne R. Bestc, Thomas J. Metzlerb,c, Rachel Yehudag, Charles R. Marmara,b,c a

Department of Psychiatry, University of California, San Francisco, CA, USA San Francisco Veterans Administration Medical Center, San Francisco, CA, USA c Northern California Institute for Research and Education, San Francisco, CA, USA d ´al, Que., Canada Department of Psychiatry, McGill University, Montre e Douglas Hospital Research Center, 6875 LaSalle Boulevard, Montreal, Que., Canada H4H 1R3 f Department of Psychology, University of Michigan, 525 East University, Ann Arbor, MI 48109-1109, USA g Department of Psychiatry, Mount Sinai School of Medicine, P.O. Box 1230, One Gustave L. Levy Place, New York, NY 10029, USA b

Received 2 December 2003; received in revised form 28 October 2004; accepted 28 October 2004

KEYWORDS Stress disorders; Post-traumatic; Hydrocortisone; Dexamethasone; Police; Saliva; Pituitary-adrenal system

Summary This study examines whether pre- or post-dexamethasone salivary cortisol is related to cumulative critical incident exposure, peritraumatic responses, or post-traumatic stress disorder (PTSD) symptom severity. Thirty active duty police officers completed the study protocol, which included measures of peritraumatic emotional distress, peritraumatic dissociation, duty-related trauma exposure, and PTSD symptoms. Salivary cortisol was consolidated into three outcome variables: (1) pre-dexamethasone free cortisol levels at 1, 30, 45, and 60 min after awakening, (2) post-dexamethasone cortisol levels at the identical wake times, and (3) percentage of cortisol suppression. Control variables included age, gender, average daily alcohol use, night shift work, routine work environment stressors, and salivary dexamethasone levels. Zero order correlations showed that greater levels of PTSD symptoms, peritraumatic distress, and peritraumatic dissociation were associated with lower levels of pre-dexamethasone cortisol levels on awakening, but were not associated with the other two cortisol variables. A trend was also noted for older subjects to have lower pre-dexamethasone cortisol on awakening. When these four predictors were entered simultaneously in a regression analysis, only age and PTSD symptom severity significantly predicted pre-dexamethasone awakening cortisol levels. These results replicate previous research indicating a relationship between greater PTSD symptoms and lower levels of basal cortisol on awakening, and extend this finding to a previously unstudied non-treatment seeking population, urban police. Q 2004 Elsevier Ltd. All rights reserved.

* Corresponding author. Address: PTSD Program, Psychiatry Service 116P, VA Medical Center, 4150 Clement Street, San Francisco, CA 94121, USA. Tel.: C1 415 750 6961; fax: C1 415 751 2297. E-mail address: [email protected] (T.C. Neylan). 0306-4530/$ - see front matter Q 2004 Elsevier Ltd. All rights reserved. doi:10.1016/j.psyneuen.2004.10.005

374

1. Introduction Active duty police officers, of which there are over one million in the United States (Reaves and Hickman, 2002), are a uniquely important population for the study of responses to traumatic stressors. Not only are they routinely at risk for exposure to critical incident stressors such as being injured, injuring others in the line of duty, or witnessing death or injuries to civilians and other officers, but they are also understudied relative to such groups as combat veterans and sexual assault victims. Critical incident stress exposure, long a concern of police managers, police unions, and the friends and families of police officers, is all the more salient in light of current ongoing threats of domestic terrorism. Further, studies of police officers may broaden our understanding of the biology of stress response syndromes, including post-traumatic stress disorder (PTSD), because police officers are for the most part medically healthy and psychologically resilient. Many studies have reported an association between PTSD and lower cortisol levels, though this has not been a consistent finding (for recent reviews, see Yehuda (2002) and Rasmusson et al. (2003)). Compared to those without PTSD, individuals with PTSD have in several studies shown greater suppression of cortisol release following the administration of low doses of dexamethasone (Stein et al., 1997; Yehuda et al., 2002, 2004; Grossman et al., 2003; Newport et al., 2004), supporting a hypothesis that PTSD is associated with enhanced negative feedback regulation of the hypothalamic-pituitary-adrenal (HPA) axis. To date, however, most studies have examined HPA axis alterations by comparing a sample of chronic, highly symptomatic PTSD patients with healthy controls. It is not known whether and to what extent a relationship between basal cortisol levels and PTSD symptoms could be discerned across a wide continuum of symptom severity in persons who are experiencing both acute and chronic stress yet still functioning reasonably in society. Police officers represent a population exposed to a high rate of traumatic stressors (as defined in the DSM-IV criterion A for PTSD), as well as a high degree of routine, non-traumatic work environment stressors (Liberman et al., 2002). Thus, they are an important group in which to examine the relationship between cortisol levels, exposure to both routine and traumatic stressors, and PTSD symptoms. Given the multiple traumatic experiences that are likely to occur over the course of police work, this group also affords the opportunity to

T.C. Neylan et al. examine whether cortisol is correlated with cumulative history of trauma exposure. Indeed, recent studies have suggested that lower cortisol levels in both adult female sexual assault victims (Resnick et al., 1995) and combat veterans (Yehuda et al., 1995) may be due to exposure to events predating these adulthood experiences and may thus constitute risk factors for PTSD. Accordingly, in the current study, we were interested in examining whether cumulative critical incident exposure, independent from PTSD symptoms, would predict lower cortisol levels. Active duty police officers are an ideal study population to test these questions because they express a broad range of PTSD symptomatology (Pole et al., 2001), have high rates of both critical incident (Weiss et al., 1999) and routine work environment stressors (Liberman et al., 2002), are ambulatory, and are physically active. In a previous study involving San Francisco Bay Area and New York City police officers, we showed that peritraumatic distress (i.e. negative emotions and arousal at the time of the trauma including panic reactions) and peritraumatic dissociation were related to PTSD symptom level but cumulative critical incident exposure was not (Brunet et al., 2001). High peritraumatic emotional reactivity, including panic reactions at the time of exposure, may be a strong predictor of PTSD (Galea et al., 2002). Lower cortisol levels may also be a risk factor that affects peritraumatic reactivity and increases the likelihood for developing greater PTSD symptoms (Yehuda et al., 1998; Delahanty et al., 2000). We therefore examined whether the cortisol response to awakening, which has a high degree of heritability and intraindividual stability (Wust et al., 2000; Bartels et al., 2003), would be related to retrospective reports of peritraumatic distress and peritraumatic dissociation experienced at the time of the worst duty-related critical incident. We were particularly interested in determining whether higher peritraumatic responses predicted lower cortisol levels independent of PTSD symptoms. Data were collected on a sample of active duty police officers exposed to a wide range of critical incidents in the line of duty. The outcome variables included; (1) area under the curve (AUC) of four salivary cortisol samples collected during the first hour of awakening prior to dexamethasone administration, (2) waking AUC cortisol levels following the ingestion of dexamethasone, and (3) the percentage of dexamethasone suppression of cortisol. The primary analyses examined the contribution of cumulative critical incident exposure,

PTSD symptoms predict waking salivary cortisol levels in police officers peritraumatic distress, peritraumatic dissociation and current PTSD symptoms in accounting for preand post-dexamethasone cortisol levels. Our secondary analyses examined group difference in pre- and post-dexamethasone cortisol levels in those officers with and without the diagnosis of PTSD. Our primary hypotheses were that higher peritraumatic distress, peritraumatic dissociation, and PTSD symptoms would be associated with lower cortisol levels, before and after dexamethasone suppression, and greater percent suppression of cortisol by dexamethasone.

2. Methods 2.1. Subjects Participants were a sub-sample of respondents to a larger survey on risk and resilience factors for PTSD in police officers (nZ747) recruited from New York, NY and Oakland and San Jose, CA. Potential participants were selected by each department’s personnel section from departmental personnel rosters. A proportionally higher pool of minority and women officers was identified. In all other respects, selection was random and no systematic strata were specified. The group of 747 participants is best described as a convenience sample. More details about the specific sample characteristics can be found elsewhere (Brunet et al., 2001; Pole et al., 2001; Liberman et al., 2002; Neylan et al., 2002; Mohr et al., 2003). The study protocol and consent form was approved by the Committee on Human Research at the University of California, San Francisco (UCSF). Participants in the sub-sample presented in this report were officers from the San Francisco Bay Area selected on the basis of reporting either high or low PTSD symptoms. All subjects with high PTSD symptoms were selected for study inclusion. Each subject with high PTSD symptoms was matched with a subject with low PTSD symptoms levels who had a comparable level of critical incident exposure. The sub-sample was not selected to be representative of the full study sample or active duty police officers in general. Selected police officers participated in a clinical research interview and were given take-home saliva collection kits (see Instruments for details) to obtain pre- and post-dexamethasone cortisol measures. Fifteen participants were excluded from the final data analyses for a variety of reasons including: no dexamethasone level/failure to take the dexamethasone (nZ4), failure to adhere to the protocol

375

(nZ4), missing cortisol data (nZ2), and failure to complete the full battery of self-report questionnaires (nZ5). Subjects were also excluded if they had cortisol values greater than 3000 ng/dl (beyond the reliable range of the assay) or greater than MG 2.3SD (i.e. a of 99%). This led to the rejection of 12 of the 304 cortisol samples (4%). Rejected samples (3098–7709 ng/dl) were spread across eight cases (7 males, 1 female) and across the two collection days (7 for day 1 and 5 for day 2). These exclusions brought the sample size to a final nZ30 (24 males, 6 females). There were no differences in the mean total clinician-administered PTSD symptom (CAPS) scores between the 30 included cases (MZ12.2, SDZ16.9) and the eight excluded cases (MZ11.3, SDZ15.5). The CAPS scores included in the analyses ranged from 0 to 54 (MZ12.2, SDZ16.9). Five (16.7%) officers met full CAPS criteria for current PTSD. One of these officers also met criteria for current major depression as indexed by the Structured Clinical Interview for DSM-IV (SCID; First et al., 1996). None of the officers without PTSD met current criteria for major depression. Eighteen of our participants were Caucasian, two were African–American, six were Hispanic, and five were Other. None of the subjects in the full sample met current criteria for alcohol abuse/dependence or substance abuse/ dependence. Average daily alcohol use ranged from 0 to 2.5 drinks per day (MZ0.7, SDZ0.6). None of the subjects smoked cigarettes. Three subjects reported smoking one to two cigars per week.

2.2. Instruments and procedure 2.2.1. Clinical interview and self-report measures To assess the extent of cumulative critical incident exposure in the line of duty, we used the critical incident history questionnaire (Weiss et al., submitted for publication). This 34-item self-report measure yields a quantitative measure of cumulative exposure to potentially traumatic duty-related events experienced over the course of a career in policing. The total score incorporates both frequency of exposure to each event and a rating of how difficult each event would be for the average officer to cope with, summed across items. After completing this questionnaire, participants were asked to select the most troublesome, disturbing, or distressing event that they had ever experienced in their career as a police officer as a reference point for completing incident-specific questionnaires. These incidents included events such as

376 being seriously injured intentionally or making a mistake that lead to the serious injury or death of a fellow officer (Weiss et al., submitted for publication). The peritraumatic distress inventory (Brunet et al., 2001) and peritraumatic dissociative experiences questionnaire (Marmar et al., 1997) were used to retrospectively measure self-reported distress (panic and related negative emotions) and dissociation at the time of their most disturbing duty-related critical incident. A PhD level psychologist administered the clinician-administered PTSD scale for DSM-IV (Blake et al., 1995) to assess current symptoms of PTSD to the same incident as the above peritraumatic measures. The structured clinical interview for DSM-IV (SCID; First et al., 1996) was used to determine presence of Axis I disorders including alcohol and substance abuse. The work environment inventory (Liberman et al., 2002) was used to assess routine work stress separate from critical incident stressor exposure in police officers. The scale includes items related to police-specific work stressors such as court decisions and safety, as well as generic work stressors, racial discrimination, and shift work. 2.2.2. Measurement of salivary cortisol Saliva collection is a well-validated and unobtrusive method of measuring cortisol that strongly reflects level of serum cortisol (Kirschbaum and Hellhammer, 1989). The saliva samples were collected according to a method outlined in detail elsewhere (Pruessner et al., 1997) which involves collecting samples with reference to the individual’s usual time of awakening rather than at a fixed clock time. This procedure accounts for the fact that cortisol levels at awakening typically are at the circadian peak of activity (Gallagher et al., 1973). This methodology was utilized because the cortisol response to awakening has recently been found to have more test–retest stability (Pruessner et al., 1997; Wust et al., 2000; Edwards et al., 2001) than single point measures yoked to clock time (Coste et al., 1994), particularly in a group, like police, that may show high variability in sleep and wake time. Subjects were instructed to maintain a stable sleep-wake schedule during the testing, to perform this procedure in the middle of their working week, to awaken at the same clock time on the pre- and post-dexamethasone saliva collection days, and to abstain from activities such as brushing or flossing their teeth, smoking, exercising, and ingesting food or drink prior to or during the collection. Four saliva samples were collected at 1, 30, 45 and 60 min after awakening on both days. Samples were collected at home using Salivettes (Sarstedt, Inc., Newton, NC), returned by mail, and deep-frozen (K78 8C) until

T.C. Neylan et al. assay. We relied on the subjects to inform us about the time of awakening and sampling. 2.2.3. Low dose dexamethasone suppression test (DST) The DST was combined with the procedure outlined above for collecting basal cortisol samples. Patients were given instructions by the research team on the exact time to take 0.5 mg of dexamethasone 15 h after awakening on day 1. The saliva collection procedure was then repeated on the following morning. Because dexamethasone bioavailability may affect cortisol levels, we measured dexamethasone levels for covariance analysis (O’Sullivan et al., 1997) and to check compliance with the protocol. Salivary cortisol and dexamethasone were assayed by Dr Yehuda’s laboratory using radioimmunoassay as previously described (Goenjian et al., 1996) by technicians blind to clinical data.

2.3. Statistical analyses The salivary cortisol data were consolidated into three variables: (1) pre-dexamethasone or basal cortisol level (the AUC created by the four samples collected after awakening on day 1), (2) postdexamethasone cortisol level (the AUC created by the four samples collected after awakening on day 2), and (3) the percentage of dexamethasone suppression of cortisol (pre-dexamethasone minus post-dexamethasone AUC/pre-dexamethasone AUC)!100 (Pruessner et al., 1997). Log transformed AUC cortisol data were used in all correlation and regression analyses, which were two-tailed with type-I error (a) set to 0.05. The first step was to examine the zero order correlations between pre- and post-AUC cortisol values with our main predictor variables (PTSD symptoms, peritraumatic distress, peritraumatic dissociation, and cumulative critical incident exposure) as well as potentially confounding factors such as age, gender, average daily alcohol use, night shift work, routine work environment stressors, and dexamethasone bioavailability. Second, we constructed hierarchical regression models with waking cortisol levels as the outcomes and the following set of predictors: cumulative critical incident exposure, peritraumatic dissociation, peritraumatic distress, and PTSD symptoms. Finally, a repeated measures ANOVA with log transformed cortisol AUC preand post-cortisol, covarying for dexamethasone levels, was used to examine group differences between participants with and without PTSD in pre- and post-dexamethasone cortisol levels.

PTSD symptoms predict waking salivary cortisol levels in police officers

3. Results We first examined the correlations between our predictors, potential confounds, and outcome measures. As can be seen in Table 1, gender, average daily alcohol use, nightshift work, and dexamethasone levels did not correlate significantly with our cortisol measures. Further, we did not find a relationship between cumulative critical incident exposure or routine work environment stressors and either pre- or post-dexamethasone cortisol levels. In addition, the partial correlation coefficients for cumulative critical incident exposure and routine work environment stressors with both pre- and post-dexamethasone cortisol levels controlling for total CAPS score all had absolute values of less than rZK0.20 and were all non-significant (data not shown). A trend was noted for an inverse association of age with predexamethasone cortisol levels. As predicted, higher peritraumatic distress, peritraumatic dissociation and PTSD symptoms were associated with lower pre-dexamethasone cortisol levels, but counter to our predictions were not associated with postdexamethasone cortisol levels or with amount of post-dexamethasone cortisol suppression. The strongest correlation was between pre-dexamethasone awakening cortisol levels and PTSD symptoms (rZK0.57, p!0.005). See Fig. 1. Repeating the correlation with the 8 subjects who were dropped from the primary analyses because of outlier cortisol levels, modestly reduced the strength of the correlation between pre-dexamethasone awakening cortisol levels and PTSD symptoms (rZK0.47, p!0.005). With the outliers included,

Table 1

there were no significant correlations with predexamethasone awakening cortisol levels and any of the other predictor variables. The second step of the analyses involved examining the relative contribution of the predictors to explaining variance in pre-dexamethasone cortisol levels by means of linear multiple regression. Because of the sample size, we opted to include only four predictors in our regression analysis (Tinsley and Tinsley, 1987): age, peritraumatic distress, peritraumatic dissociation, and PTSD symptom level. Prior to constructing the model, we first examined the intercorrelations among the four predictors. Current PTSD symptoms were moderately correlated with both peritraumatic distress (rZ0.43, pZ0.02) and peritraumatic dissociation (rZ41, pZ0.02). Age was not significantly correlated with peritraumatic distress (rZ0.18, pZ0.35), peritraumatic dissociation (rZK0.02, pZ0.93), or PTSD symptom level (rZK0.14, pZ0.46). Peritraumatic distress and peritraumatic dissociation were highly correlated with each other (rZ0.63, p!0.001) and therefore were entered simultaneously into our model. As can be seen in Table 2, the only variables significantly accounting for unique variance in pre-dexamethasone waking cortisol levels in the complete model were current PTSD symptom level and age. The overall model accounted for 45% of the variance in pre-dexamethasone waking cortisol levels. We originally chose age as a control variable because of the prior work showing a relationship between age and cortisol levels. Although most studies show that aging is associated with a progressive increase in basal cortisol levels

Pearson correlations between our predictors and outcome variables (nZ30).

Age Gender Average daily alcohol use Night shift Routine work stressorsc Dexamethasoned Critical incident exposure Peritraumatic distress Peritraumatic dissociation PTSD symptoms †

377

Pre-dexamethasone salivary cortisola

Post-dexamethasone salivary cortisola

Amount of cortisol suppressionb

K0.32† 0.13 0.21 0.04 K0.26 K K0.20 K0.49* K0.41* K0.57**

K0.10 0.07 0.11 K0.16 K0.10 0.11 0.04 K0.11 K0.31 K0.27

K0.06 0.09 0.03 0.17 K0.13 K K0.26 K0.05 0.14 0.03

pZ0.08, *p!0.05, and **p!0.005. a Area under the curve (AUC): time 1, 30, 45, 60 min after waking (log transformed). b (Pre- minus post-dexamethasone AUC/pre-dexamethasone AUC)!100. c Measured with the work environment inventory. d Levels in ng/dl, 9.5 h after ingestion.

378

T.C. Neylan et al. interaction (F(1,27)Z0.1, pZ0.72). Post hoc examination of the means revealed that those officers with PTSD (nZ5), compared to the participants without the diagnosis (nZ25), had lower predexamethasone waking cortisol values, M (SD)Z 697.7 (232.9) vs. 1066.1 (246.4), t(28)Z3.08, p!0.005. However, compared to those without the diagnosis, there was a non-significant trend for participants with PTSD to have lower post-dexamethasone cortisol values, M (SD)Z233.6 (159.2) vs. 348.7 (210.8), t(28)Z1.15, pZ0.26. The extent of cortisol suppression was 66% in both groups, reflecting normal suppression.

Figure 1 Waking cortisol (AUC, area under the curve for salivary cortisol levels obtained at 1, 30, 45, and 60 min after awakening) and PTSD symptoms.

(reviewed in Seeman and Robbins (1994)), there are studies that have shown the opposite (Drafta et al., 1982). Because our data were not consistent with the majority of studies; that is, we found that greater age was associated with lower pre-dexamethasone waking cortisol levels, we considered the possibility that age was acting as a proxy for another variable. One possible candidate was years of service in police work, which was highly correlated with age (rZ0.92, p!0.001). We repeated the regression model substituting years of service for age and found nearly identical results (beta weight for years of serviceZK0.33, p!0.05). A repeated measures ANOVA examining log transformed AUC cortisol levels pre- and postdexamethasone covarying for dexamethasone bioavailability found a significant group effect with PTSD subjects exhibiting lower cortisol levels (F(1,27)Z5.8, pZ0.02), as well as an effect of treatment with dexamethasone (F(1,27)Z61.9, p!0.001). However, there was no group by time Table 2

4. Discussion This study demonstrates that in a group of occupationally functional active duty police officers exposed to both chronic and intermittent traumatic and non-traumatic stressors, baseline awakening cortisol levels are inversely related to PTSD symptom levels. Cortisol levels were not associated with either cumulative critical incident exposure or routine work environment stressors. Though peritraumatic distress and dissociation shared significant associations with pre-dexamethasone waking cortisol, these relationships were fully accounted for by current PTSD symptom severity, which is not surprising because both peritraumatic variables are putative risk factors for PTSD (Ozer et al., 2003). What police officers have in common with combat veterans, Holocaust survivors, mothers whose children have cancer, and other populations in which low cortisol levels have been observed, is that these individuals have experienced repeated traumatic events against a background of chronic non-traumatic environmental stressors. This may be important in understanding the link between post-traumatic stress and lowered cortisol.

Variables predicting pre-dexamethasone cortisol levels (nZ30).

Step 1 Age Step 2 Peritraumatic distress Peritraumatic dissociation Step 3 PTSD symptoms

Adj. R2

DR2

F

df

Step 1 b

Step 2 b

Step 3 b

0.07

–

3.2

1.29

K0.32

K0.27

K0.37*

0.25

0.18

4.3*

2.27

K0.31

K0.13

K0.22

K0.12

0.45

*p!0.05 and **p!0.005.

0.20

6.8**

3.26

K0.51**

PTSD symptoms predict waking salivary cortisol levels in police officers Younger age by itself was only weakly associated with higher cortisol levels. There was also a weak trend suggesting that younger police officers reported higher PTSD symptom levels. Interestingly, age and PTSD symptom levels exhibit what Cohen and Cohen (1983) refer to as cooperative suppression. The predictive power of each variable is enhanced in the presence of the other variable. The beta weight for age increases from K0.27 to K0.37 when PTSD symptoms are added to the model, and the beta weight for PTSD symptoms increases from K0.42 to K0.51 when the steps are reversed and age is added to the model. Age alone accounts for 7% of variance in cortisol, and PTSD symptoms alone account for 29% of variance, but age and PTSD together account for 44% of variance. Peritraumatic dissociation and distress account for only 1% of additional variance, presumably due to their strong correlation with PTSD symptom levels. In a careful examination of the data we found that much, but not all, of the cooperative suppression effect was due to the scores of a subject who was both at the top of the age range and near the bottom of both the cortisol and PTSD symptom ranges. We did not eliminate him from analysis because we had no reason to suspect his scores to be erroneous, and to do so would only serve to increase both the total R-squared of the regression model and the beta weight of the PTSD symptom variable, biasing the results in favor of our hypothesis. As noted above, most of the published literature suggests that advanced age is associated with higher basal cortisol levels (Seeman and Robbins, 1994). Since our data were not consistent with most of the literature, we considered in post hoc analyses that age was serving as a proxy for years of police service. If years of service represents the true predictor of waking cortisol than this suggests two possibilities: the relationship is driven by a relative increase in morning cortisol levels in junior officers; or the finding is caused by a relative decrease in more senior officers. Regarding the first possibility, junior officers may have increased cortisol levels because they have more variable work shifts. However, we did not find that night shift work was significantly related to awakening cortisol levels. An alternative explanation is that junior officers perceive more job strain, including anticipatory anxiety to starting their next shift, and that cortisol levels are increased as part of the normative response to chronic stress. A previous study reported that waking cortisol levels were elevated in a sample of junior high and high school teachers with increased job strain (Steptoe et al., 2000). Another study examining university students found that

379

increased cortisol secretion after awakening was associated with perceived work overload (Schulz et al., 1998). However, we are doubtful that job strain explained higher cortisol levels in less experienced officers because there was no association between years of service and perceived routine work environment stressors (rZK0.05, pZ0.82) as indexed by the Work Environment Inventory developed for police officers (Liberman et al., 2002). A final consideration is that officers are thought to pass through a series of stages during their careers that might help to explain the finding of higher waking cortisol in more junior officers. Violanti and Aron (1993), for example, suggested that junior officers begin their careers in an ‘alarm stage’ after which they commonly experience a period of disenchantment. They proposed that eventually, the majority of officers enter into a ‘personalization stage’ in which personal rather than occupational concerns come to the forefront. A second possibility was that the negative relationship between years of service and cortisol levels might be driven by a relative decrease in cortisol levels in the more senior officers. One speculation to consider is that senior officers are more likely to experience higher levels of professional burnout. Pruessner and co-workers found that burnout in teachers was associated with a lower cortisol response to awakening (Pruessner et al., 1999). Unfortunately, we did not include a measure of burnout in our study and this question will need to be examined in future studies. The waking cortisol response was utilized in this study because it has a greater degree of intraindividual stability (Pruessner et al., 1997; Wust et al., 2000; Edwards et al., 2001) than single point measures and may be particularly useful in a group with a high degree of interindividual variability in sleep-wake schedules. This may have increased our power to detect associations between baseline cortisol and our predictor variables. However, there are potential limitations to this method in the context of the dexamethasone suppression test. There were minor variations in wake times despite instructions given to subjects to maintain identical wake times. In the 23 of the 30 subjects who returned the time sheets, the differences in day 1/day 2 wake times were 0 min for 17 subjects, and ranged from K60 to 60 min in the remaining 6 subjects. We found no differences in percent suppression of cortisol or level of PTSD symptoms in those that did and did not complete the time sheets. However, the reliance on subjects to report the saliva collection times is a limitation of this study. Dexamethasone levels were obtained in all of our subjects and this measure partially

380 accounts for time since administration. Our repeated measures analysis covarying for dexamethasone levels did not show a significant effect of PTSD on cortisol suppression. However, future studies will be needed to covary for both time differences between the two collection points and dexamethasone levels in order to fully evaluate the utility of this new method for testing cortisol suppression to dexamethasone. This study does have limitations that affect the generalizability of the findings. The correlations among post-dexamethasone cortisol and peritraumatic dissociation and PTSD symptoms, which were in the same direction as the pre-dexamethasone correlations, may have reached statistical significance in a larger sample. The small number of subjects meeting case criteria for PTSD limit the ability to conduct case-control comparisons such as cortisol suppression from dexamethasone. The focus on critical incident exposure and cortisol levels is limited to police duty-related incidents and did not include trauma exposure outside of police work. This cross-sectional study could not address the temporal (i.e. causal) relationship between low cortisol and PTSD symptom development. It is possible that individuals with lower cortisol levels prior to police service were more prone to experiencing distress following critical incident exposure and hence were at greater risk for developing PTSD symptoms. It is also possible that subjects with more severe current PTSD symptoms have a recall bias for remembering higher levels of peritraumatic distress and dissociation. Prospective and longitudinal studies in which cortisol and PTSD symptom measures are obtained prior to trauma exposure, peritraumatically, and following trauma exposure are needed to resolve these issues. In summary, our results were consistent with previous research indicating a relationship between PTSD and lower levels of basal cortisol. This relationship was found in a functional and ambulatory group of active duty police officers who were sampled to reflect a broad range of severity in PTSD symptom levels. Our results also illustrate that under certain circumstances other variables such as age may suppress the relationship between PTSD symptoms and pre-dexamethasone waking cortisol levels.

Acknowledgements This research was supported by NIMH grant MH56350 (Marmar), the Fonds de Recherche en

T.C. Neylan et al. Sante ´ du Que ´bec (Brunet), a Veterans Affairs Merit Review grant (Yehuda), and the Sierra Pacific (VISN 21) Mental Illness Research, Education, and Clinical Center (Marmar and Neylan). The authors gratefully acknowledge Cynthia Rogers and Maryanne Lenoci for helping to coordinate this study.

References Bartels, M., Van den Berg, M., Sluyter, F., Boomsma, D., de Geus, E., 2003. Heritability of cortisol levels: review and simultaneous analysis of twin studies. Psychoneuroendocrinology 28, 121–137. Blake, D.D., Weathers, F.W., Nagy, L.M., Kaloupek, D.G., Gusman, F.D., Charney, D.S., Keane, T.M., 1995. The development of a clinician-administered PTSD scale. J. Trauma. Stress 8, 75–90. Brunet, A., Weiss, D.S., Metzler, T.J., Best, S.R., Neylan, T.C., Rogers, C., Fagan, J., Marmar, C.R., 2001. The peritraumatic distress inventory: a proposed measure of PTSD criterion A2. Am. J. Psychiatry 158, 1480–1485. Cohen, J., Cohen, P., 1983. Applied multiple regression/correlation analysis for the behavioral sciences, second ed. Erlbaum, Hillsdale, NJ. Coste, J., Strauch, G., Letrait, M., Bertagna, X., 1994. Reliability of hormonal levels for assessing the hypothalamic-pituitaryadrenocortical system in clinical pharmacology. Br. J. Clin. Pharmacol. 38, 474–479. Delahanty, D.L., Raimonde, A.J., Spoonster, E., 2000. Initial posttraumatic urinary cortisol levels predict subsequent PTSD symptoms in motor vehicle accident victims. Biol. Psychol. 48, 940–947. Drafta, D., Schindler, A.E., Stroe, E., Neacsu, E., 1982. Agerelated changes of plasma steroids in normal adult males. J. Steroid Biochem. 17, 683–687. Edwards, S., Clow, A., Evans, P., Hucklebridge, F., 2001. Exploration of the awakening cortisol response in relation to diurnal cortisol secretory activity. Life Sci. 68, 2093–2103. First, M.B., Spitzer, R.L., Williams, J.B.W., Gibbon, M., 1996. Structured Clinical Interview for DSM-IV (SCID-I), Patient Version ed. New York State Psychiatric Institute, Biometrics Research, New York. Galea, S., Ahern, J., Resnick, H., Kilpatrick, D., Bucuvalas, M., Gold, J., Vlahov, D., 2002. Psychological sequelae of the September 11 terrorist attacks in New York City. N. Engl. J. Med. 346, 982–987. Gallagher, T.F., Yoshida, K., Roffwarg, H.D., Fukushima, D.K., Weitzman, E.D., Hellman, L., 1973. ACTH and cortisol secretory patterns in man. J. Clin. Endocrinol. Metab. 36, 1058–1068. Goenjian, A.K., Yehuda, R., Pynoos, R.S., Steinberg, A.M., Tashjian, M., Yang, R.K., Najarian, L.M., Fairbanks, L.A., 1996. Basal cortisol, dexamethasone suppression of cortisol, and MHPG in adolescents after the 1988 earthquake in Armenia. Am. J. Psychiatry 153, 929–934. Grossman, R., Yehuda, R., New, A., Schmeidler, J., Silverman, J., Mitropoulou, V., Sta Maria, N., Golier, J., Siever, L., 2003. Dexamethasone suppression test findings in subjects with personality disorders: associations with posttraumatic stress disorder and major depression. Am. J. Psychiatry 160, 1291–1298.

PTSD symptoms predict waking salivary cortisol levels in police officers Kirschbaum, C., Hellhammer, D.H., 1989. Salivary cortisol in psychobiological research: an overview. Neuropsychobiology 22, 150–169. Liberman, A., Best, S., Metzler, T., Fagan, J., Weiss, D., Marmar, C., 2002. Routine occupational stress and psychological distress in police. Policing: Int. J. Police Strategies Manage. 25, 421–439. Marmar, C.R., Weiss, D.S., Metzler, T.J., 1997. The peritraumatic dissociative experiences questionnaire. In: Wilson, J.P., Keane, T.M. (Eds.), Assessing Psychological Trauma and PTSD: A Handbook for Practitioners. Guilford Press, New York, pp. 412–428. Mohr, D.C., Vedantham, K., Neylan, T.C., Metzler, T.J., Best, S.R., Marmar, C.R., 2003. The mediating effects of sleep in the relationship between traumatic stress and health symptoms in urban police officers. Psychosom. Med. 65, 485– 489. Newport, D.J., Heim, C., Bonsall, R., Miller, A.H., Nemeroff, C.B., 2004. Pituitary-adrenal responses to standard and low-dose dexamethasone suppression tests in adult survivors of child abuse. Biol. Psychol. 55, 10–20. Neylan, T.C., Metzler, T.J., Best, S.R., Weiss, D.S., Fagan, J.A., Liberman, A., Rogers, C., Vedantham, K., Brunet, A., Lipsey, T.L., Marmar, C.R., 2002. Critical incident exposure and sleep quality in police officers. Psychosom. Med. 64, 345– 352. O’Sullivan, B.T., Cutler, D.J., Hunt, G.E., Walters, C., Johnson, G.F., Caterson, I.D., 1997. Pharmacokinetics of dexamethasone and its relationship to dexamethasone suppression test outcome in depressed patients and healthy control subjects. Biol. Psychol. 41, 574–584. Ozer, E.J., Best, S.R., Lipsey, T.L., Weiss, D.L., 2003. Predictors of posttraumatic stress disorder and symptoms in adults: a meta-analysis. Psychol. Bull. 129, 52–73. Pole, N., Best, S., Weiss, D., Metzler, T., Liberman, A., Fagan, J., Marmar, C., 2001. Effects of gender and ethnicity on duty-related posttraumatic stress symptoms among urban police officers. J. Nerv. Ment. Dis. 189, 442–448. Pruessner, J.C., Wolf, O.T., Hellhammer, D.H., BuskeKirschbaum, A., von Auer, K., Jobst, S., Kaspers, F., Kirschbaum, C., 1997. Free cortisol levels after awakening: a reliable biological marker for the assessment of adrenocortical activity. Life Sci. 61, 2539–2549. Pruessner, J.C., Hellhammer, D.H., Kirschbaum, C., 1999. Burnout, perceived stress, and cortisol responses to awakening. Psychosom. Med. 61, 197–204. Rasmusson, A.M., Vythilingam, M., Morgan 3rd., C.A., 2003. The neuroendocrinology of posttraumatic stress disorder: new directions. CNS Spectr. 8, 651–656; see also pp. 657–665. Reaves, B., Hickman, M., 2002. Census of state and local law enforcement agencies, 2000. Bur. Justice Stat. Bull. 2002;. October 2002. Resnick, H.S., Yehuda, R., Pitman, R.K., Foy, D.W., 1995. Effect of previous trauma on acute plasma cortisol level following rape. Am. J. Psychiatry 152, 1675–1677. Schulz, P., Kirschbaum, C., Pruesner, J., Hellhammer, D., 1998. Increased free cortisol secretion after awakening in

381

chronically stressed individuals due to work overload. Stress Med. 14, 91–97. Seeman, T.E., Robbins, R.J., 1994. Aging and hypothalamicpituitary-adrenal response to challenge in humans. Endocr. Rev. 15, 233–260. Stein, M.B., Yehuda, R., Koverola, C., Hanna, C., 1997. Enhanced dexamethasone suppression of plasma cortisol in adult women traumatized by childhood sexual abuse. Biol. Psychol. 42, 680–686. Steptoe, A., Cropley, M., Griffith, J., Kirschbaum, C., 2000. Job strain and anger expression predict early morning elevations in salivary cortisol. Psychosom. Med. 62, 286–292. Tinsley, H.E., Tinsley, D.J., 1987. Uses of factor analysis in counseling psychology research. J. Counseling Psychol. 34, 414–424. Violanti, J.M., Aron, F., 1993. Sources of police stressors, job attitudes and psychological distress. J. Psychol. Rep. 72, 899– 904. Weiss, D.S., Brunet, A., Metzler, T.J., Best, S.R., Fagan, J.A., Marmar, C.R., 1999. Critical incident exposure in police officers: frequency, impact, and correlates, Proceedings of the International Society of Traumatic Stress Studies, Miami, FL. p. 52. Weiss, D.S., Brunet, A., Best, S.R., Metzler, T.J., Liberman, A., Rogers, C., Fagan, J.A., Marmar, C.R., submitted for publication. The Critical Incident History Questionnaire: A Method for Measuring Total Cumulative Exposure to Critical Incidents. Wust, S., Federenko, I., Hellhammer, D.H., Kirschbaum, C., 2000. Genetic factors, perceived chronic stress, and the free cortisol response to awakening. Psychoneuroendocrinology 25, 707–720. Yehuda, R., 2002. Current status of cortisol findings in posttraumatic stress disorder. Psychiatr. Clin. North Am. 25, 341–368. Yehuda, R., Boisoneau, D., Lowy, M.T., Giller Jr.., E.L., 1995. Dose-response changes in plasma cortisol and lymphocyte glucocorticoid receptors following dexamethasone administration in combat veterans with and without posttraumatic stress disorder. Arch. Gen. Psychiatry 52, 583–593. Yehuda, R., McFarlane, A.C., Shalev, A.Y., 1998. Predicting the development of posttraumatic stress disorder from the acute response to a traumatic event. Biol. Psychol. 44, 1305–1313. Yehuda, R., Halligan, S.L., Grossman, R., Golier, J.A., Wong, C., 2002. The cortisol and glucocorticoid receptor response to low dose dexamethasone administration in aging combat veterans and holocaust survivors with and without posttraumatic stress disorder. Biol. Psychol. 52, 393–403. Yehuda, R., Halligan, S.L., Golier, J.A., Grossman, R., Bierer, L.M., 2004. Effects of trauma exposure on the cortisol response to dexamethasone administration in PTSD and major depressive disorder. Psychoneuroendocrinology 29, 389–404.