Startle Reactivity and PTSD Symptoms in a Community Sample of Women

Psychiatry Research 101Ž. 2001 157᎐169

Startle reactivity and PTSD symptoms in a community sample of women

Anna Marie Medinaa,U, Veronica Y. Mejiaa, Anne M. Schellb, Michael E. Dawsona, Gayla Margolina

aDepartment of Psychology, Uni¨ersity of Southern California, Los Angeles, CA 90089-1061, USA bDepartment of Psychology, Occidental College, Los Angeles, CA 90041, USA

Received 22 December 1999; received in revised form 14 December 2000; accepted 14 January 2001

Abstract

Exaggerated startle and PTSD symptoms have been investigated primarily in relation to acute or Type I stressors. The present study examined PTSD symptoms and startle eyeblink response in relation to chronic or Type II stressors. Type II stressors were operationally deﬁned as high levels of childhood corporal punishment and high levels of current partner aggression. This study recruited a sample of 52 women from a metropolitan community and administered several questionnaires assessing experience of corporal punishment in childhood, current intimate partner aggression and level of PTSD symptoms. Following questionnaires, women were presented with eight auditory startle probesŽ. white noise . Results showed that both childhood corporal punishment and intimate partner aggression were associated with women’s PTSD symptom scores. However, only PTSD symptom scores were associated with reduced startle. Results are discussed in light of Type I and Type II stressors, and recent suggestions in the PTSD literature that a subgroup of individuals may experience physiological suppression rather than heightened physiological reactivity. ᮊ 2001 Elsevier Science Ireland Ltd. All rights reserved.

Keywords: PTSD symptoms; Startle; Chronic stressor

1. Introduction Ž.PTSD has been the study of increased arousal and physiological reactivity in persons suffering An important focus of psychophysiological in- from this conditionŽ see ver Ellen and van Kam- vestigations into post-traumatic stress disorder men, 1990; Shalev and Rogel, 1993, for reviews. . Increased reactivity to both trauma-related cues and unconditioned stimuliŽ e.g. sudden or loud U Corresponding author. Tel.: q1-213-740-2309. noises. has been the subject of numerous investi- E-mail address: [email protected]Ž. A.M. Medina . gations over the last decade and a half. Elevated

0165-1781r01r$ - see front matter ᮊ 2001 Elsevier Science Ireland Ltd. All rights reserved. PII: S 0 1 6 5 - 1 7 8 1Ž. 0 1 00221-9 158 A.M. Medina et al. rPsychiatry Research 101() 2001 157᎐169 physiological arousal elicited by audiovisual and shock in the paradigmŽ. Grillon et al., 1996 failed imaginal reminders of the original trauma have to obtain results supporting the hypothesis of been found in studies of combat veterans exaggerated baseline startle in PTSD patients. In Ž.Blanchard et al., 1986; Pitman et al., 1987, 1990 addition, other investigations of the startle re- and traumatized civilians, both male and female sponse in men and women afflicted with PTSD Ž.Shalev et al., 1992, 1993 . In addition, studies that have presented startle stimuli in neutral set- with PTSD-afflicted individuals have demon- tings without the threat of shock have failed to strated that they also exhibit increased sympa- find differences between PTSD and normal groups thetic responses, such as increased heart rate and in startle habituationŽ. Ross et al., 1989 and star- skin conductance responses, to simple, strong sti- tle amplitudeŽ Shalev et al., 1992; Morgan et al., muli such as loud tonesŽ Paige et al., 1990; Shalev 1997. . Indeed, one study actually reported smaller et al., 1992, 1997; Orr et al., 1995, 1997. . baseline startle responsesŽ although less pre-pulse Recently, psychophysiological studies of PTSD inhibition of startle. in children with PTSD have begun to focus more closely on the physio- Ž.Ornitz and Pynoos, 1989 . logical hyperreactivity of patients to uncondi- In the discussion of trauma, it may be impor- tioned stimuli, such as the exaggerated startle tant to distinguish between aversive situations response to sudden or loud noises. Exaggerated which occur only once or are clearly delimited in startle responsivity is a criterion for diagnosis of time and where the individual perceives or experi- PTSD as listed in the Diagnostic and Statistical ences some level of control and can adopt active Manual of Mental Disorders, 4th ed.Ž American coping responses, and aversive situations of long Psychiatric Association, 1994. . The startle eye- or apparently unlimited duration in which the blink response, a robust and reliable component individual feels helpless and lacking adequate of the startle reflexŽ see reviews by Filion et al., coping responses. In writing of childhood trau- 1998; Dawson et al., 1999. , is a particularly useful mas, TerrŽ. 1991 distinguished between sudden, instrument in the investigation of increased physi- acute stressorsŽ. Type I trauma and chronic stres- ological reactivity in PTSD. Some studies employ- sorsŽ. Type II trauma . According to Terr Ž. 1991 , ing electromyographicŽ. EMG measurement of an acute stressor is a single external blow charac- the startle eyeblink response have shown that terized by intense surprise, whereas chronic stres- persons suffering from PTSD symptoms exhibit sors are marked by ‘prolonged and sickening an- physiological hyperreactivity to such uncondi- ticipation’Ž. p. 11 . Unlike an acute stressor such tioned stimuli as novel loud tones or loud bursts as combat, Type II or chronic stressors are char- of white noiseŽ. Butler et al., 1990 . acterized by repeated victimization over a long However, other investigations of startle respon- period of timeŽ. months or even years , where the sivity in PTSD-afflicted samples have produced situation is frequently perceived to be utterly equivocal results, with some studies finding evi- aversive and inescapable, and where the individ- dence of hyperreactivity and others failing to de- ual can do little or nothing to effect an outcome tect differences in startle between PTSD and other than victimization. Given the distinction control samples. Morgan and colleaguesŽ 1995, between types of stressors, we suggest that trau- 1996. found greater startle responsivity in both matic stress during combat is usually an acute Gulf War and Vietnam veterans with PTSD when experience, rendering this a Type I stressor. compared to, in the first study, civilians and veter- Moreover, the scripted coping responses and ans without PTSD, and in the second study, agency possessed by soldiersŽ in the form of or- ‘healthy controls’. The Gulf War veteran study ders and weapons. are less characteristic of a presented the startle stimuli in a neutral setting, Type II stressor. whereas the study with Vietnam veterans in- In the discussion of trauma, psychophysiologi- cluded the threat of aversive shock stimuli. cal studies of startle in individuals suffering from Notably, though, a subsequent study using the PTSD have, heretofore, focused largely on indi- same Vietnam veteran cohort without threat of viduals exposed to an acute or Type I stressor A.M. Medina et al. rPsychiatry Research 101() 2001 157᎐169 159

Ž.e.g. combat, rape . Investigations into PTSD and has been observed in abused children in the form startle in individuals exposed to chronic or Type of lower electrodermal response to emotional and II stressors have yet to be undertaken. TerrŽ. 1991 visual stimuliŽ. Carrey et al., 1995 . Although nei- notes that the two types of trauma are character- ther PTSD symptomatology nor startle were as- ized by distinct psychological sequelae. The ques- sessed in this latter study, the Type II stressor of tion of whether PTSD associated with Type II growing up in a hostile home environment ap- stressors is also characterized by distinct physio- pears to have resulted in physiological inhibition logical sequelae, including distinct startle re- rather than sensitization in these children. Thus, sponse patterns, remains to be addressed. it may be that an individual’s age of exposure to a The animal literature provides indirect evi- stressor, as well as type of stressor, affects physio- dence that Type I and II stressors may be charac- logical responding. terized by distinct physiological profiles. Research Given the findings of Ornitz and PynoosŽ. 1989 , with rats has shown that the type of stressor plays and of Carrey et al.Ž. 1995 of hyporesponsiveness a critical role in determining the type of physio- in traumatized children, we might expect adults logical andror behavioral responses. For exam- who experienced frequent corporal punishment ple, Mormede et al.Ž. 1990 observed that male as children to display similar hyporesponsive star- rats displayed different neuroendocrine changes tle reactions. However, other researchers have to different types of social stress. Furthermore, determined that adults exposed to Type I stres- Williams and ScottŽ. 1989 determined that, under sors, who were also exposed to high levels of certain circumstances, conspecific stressorsŽ i.e. childhood corporal punishment, appear more vul- the presence of alpha male odors for rats previ- nerable to developing PTSDŽ Donovan et al., ously defeated by alpha males. increase freezing 1996; King et al., 1996. . Keeping in mind startle behavior in rats, but predatory stressorsŽ i.e. cat studies by Morgan and colleagues, which gener- odors for rats. do not. In light of the fact that rats ally found heightened startle in adult PTSD are social animals, the presence of alpha males patients, we might conversely expect hyperrespon- may have been an inescapable and aversive expe- sive startle reactivity. This study was designed to rience for the rats in this study. Thus, the conspe- test these opposing predictions by examining the cific stressor can be understood as a chronic or influence of Type II stressors and PTSD related Type II stressor, whereas the predatory stressor to child and adulthood Type II stressors, on star- appears to represent an acute or Type I stressor. tle reactivity. Although animal models of Type I and Type II The present investigation assesses startle in a stressors have yet to be explicitly studied, evi- group of women with PTSD symptoms associated dence within the animal literature implicitly sup- with chronic, non-catastrophicŽ. Type II stressors, ports the notion that Type I and Type II stressors defined in this study as childhood corporal pun- may be characterized by distinct physiological and ishment and intimate partner aggression. A non- behavioral profiles. help-seeking sample of women were recruited In addition to type of stressor influencing phys- and evaluated for the level of corporal punish- iological and behavioral responses, age of expo- ment in childhood, level of current partner ag- sure to traumatic stressors may likewise distin- gression and PTSD symptoms. We hypothesized guish physiological reactions such as startle re- that exposure to Type II stressors would be asso- sponses. Findings of hyporesponsiveness to trau- ciated with PTSD symptoms. Moreover, consis- matic stressors have, to our knowledge, only been tent with past startle reactivity studies of PTSD detected in childrenŽ Ornitz and Pynoos, 1989; patientsŽ. Butler et al., 1990 , we hypothesized that Carrey et al., 1995.Ž. . Ornitz and Pynoos 1989 women suffering from more severe PTSD symp- found reduced startle in children exposed to toms would display larger startle responses than sniper fire. Moreover, whereas PTSD in adults women suffering from less severe or no PTSD exposed to acute stressors has been linked to symptoms. We also speculated that women who autonomic hyperarousal, autonomic hypoarousal had aggressive partners andror who were fre- 160 A.M. Medina et al. rPsychiatry Research 101() 2001 157᎐169 quently struck as children would display different discussed here. The final sample of women con- patterns of startle eyeblink reactivity than women sisted of 22 African-AmericansŽ. 47.8% , 11 Cau- without these experiences. However, given find- casiansŽ. 23.9% , 7 Hispanics Ž. 15.2% , and 6 ings in childrenŽ Ornitz and Pynoos, 1989; Carrey women who identified themselves as ‘Other’ et al., 1995. , and in light of the conflicting PTSD Ž.13.0% . The median family income was literature on adults, it was difficult to predict US$55 000, with the minimum at US$7 000 and a whether facilitation or inhibition of startle eye- maximum of US$800 000. The mean length of blink reactivity might occur as a function of child- current relationship was 10.93 years, with the hood experiences. minimum being 9 months and the maximum being 29.5 years. The mean length of current relationship for women in the ‘high partner aggres- 2. Methods sion’ group was 8.06 years, again with the minimum length being 9 months and the maximum being 29.5 years. Seventeen womenŽ. 37% from 2.1. Subjects the total sample reported having been separated from their partner at least once. Fifty-two women from the Los Angeles Metropolitan community were recruited as part 2.2. Procedures of a larger study investigating communication patterns between couples and their effects on children. Notices were placed in parental magazines Upon arrival, subjects were asked to sign a and church bulletins, and there was a wide dis- consent form giving an explanatory overview of tribution of flyers to YMCAs, public and private the testing session. The experimenter presented schools, recreation and community centers, police the subject with a brief introduction followed by activity leagues, parks and librariesŽ Medina et the attachment of electrodes for the recording of al., 2000. . The advertisement read: skin conductance and startle eyeblink, and the ‘Seeking mothers and children. The family attachment of a photoelectric plethysmograph on studies project at the University of Southern Cali- the right index finger for the recording of blood fornia seeks volunteers for a study on the effects pulse volume and heart rate. During the first of couple’s communication styles on mothers and phase of the experiment, the subjects’ skin con- children. Must be native English speakers. Chil- ductance reactivity, blood pulse volume and heart dren must be 8᎐13 years old. Motherrchild pairs rate were recorded while they listened to three to be paid $30 for approximately 2 h of their conversations between a man and a woman, vary- time.’ ing in degree of verbal and physical conflict Eligibility criteria for the study included:Ž. a between the couple, that was developed by Mar- Having at least one biological child between the golin and colleaguesŽ. 1992 . The effects of the ages of 8 and 13;Ž. b having lived with a spouse or conversations on autonomic responses are re- partner of the opposite sex within the last 6 ported by Medina et al.Ž manuscript in prepara- months;Ž. c being a native English speaker; and tion. . Ž.d having no hearing difficulties. After the subject had listened to all three con- Data related to cognitive performance and versations, the experimenter entered the room emotional responsivity of these mothers and chil- and verbally administered the Los Angeles Symp- dren will be reported elsewhereŽ Medina et al. tom ChecklistŽ. LASC; Foy, 1993 , a questionnaire 2000, manuscript in preparation. . This investiga- regarding the final vignette, and a brief question- tion examined the startle eyeblink reactivity of naire which contained questions regarding demo- the mothers. Due to equipment difficulties, blink graphic information and traumatic childhood ex- data could not be scored for six participants, periences. While the participant followed along leaving a sample of 46 women whose data are on her own copies, all directions and questions A.M. Medina et al. rPsychiatry Research 101() 2001 157᎐169 161 were read aloud by the experimenter to control reporting on her own. For each behavior, she for reading level differences that may have ex- rated:Ž. a whether the behavior had ever occurred isted between subjects. in the relationship with the partner; andŽ. b Upon completion of these questionnaires, sub- whether it had ever occurred in front of the child. jects were told that they would be presented with To report frequency in the previous year, the a series of short static bursts over their head- woman endorsed one of six categories, represent- phones. Subjects were told that they need not ing: ‘not at all’Ž. 0 ; ‘once’ Ž. 1 ; ‘2᎐5 times a year’ respond in any way, just sit quietly and listen. A Ž.2; ‘6᎐12 times per year’ Ž. 3 ; ‘2᎐4 times per total of eight startle stimuli were presented at month’Ž. 4 and ‘more than once per week’ Ž. 5 . A random intervals of 28᎐33 s apart. The startle previous study that involved 136 married women eyeblink magnitudes elicited by these stimuli found the test᎐retest reliability for husband’s to- served as the primary dependent variable in this tal aggression over a 2-week interval was 90 based study. After this startle phase of the experiment, on wives’ reportsŽ. Margolin, et al., 1990 . electrodes were removed and the participant was A partner aggression score was created for this led into another room where other questionnaires study by summing the frequency in the past year were administered, including one which inquired of each woman’s experience of physical, emotio- about current partner aggression. nal andror verbal aggression by her partner. Scores ranged from 0 to 77Ž means14.96; S.D.s 2.3. Instruments 18.70. . Of the total participants, 8% did not experience any aggression by their partners in the last 2.3.1. Childhood corporal punishment year. Women were asked ‘How often did your par- ents hit or slap you?’ Women responded by indi- 2.3.3. Post-traumatic stress disorder symptoms cating a number between 0 and 6. This number PTSD symptoms were assessed with the Los corresponded to the following frequency: 0s Angeles symptom checklistŽ. LASC; Foy, 1993 . never; 1sonce; 2s2᎐5 times; 3s5᎐10 times; The LASC is a 43-item scale that has been used 4s10᎐20 times; 5s20᎐50 times; 6smore than to measure PTSD and associated features in a 50 times. Scores ranged from 0Ž.Ž never to 6 more number of different trauma groups including than 50 times.Ž means2.93, S.D.s1.84, median Vietnam veteransŽ. Carroll et al., 1985 , survivors s3.Ž. , with 39 of the total 46 women 84.78% of childhood sexual abuseŽ. Rowan et al., 1994 indicating they had been hit at least once as a and battered womenŽ. Astin et al., 1993 . A subset child. of 17 items target the PTSD symptom categories of ‘re-experiencing’, ‘avoidance and numbing’ and 2.3.2. Partner aggression ‘hyperarousal.’ Each item is a word or phrase Partner aggression conflict was assessed using delineating a PTSD symptom or other attribute of the Domestic Conflict IndexŽ.Ž DCI Margolin et distress, which is accompanied by a 5-point Lik- al., 1990. . This 51-item inventory of intimate part- ert-type rating scale. The respondent is asked to ner conflict behaviors assesses the incidence and indicate the extent to which the item is a problem frequency of physical, emotional and verbal ag- for her by providing a rating from 0Ž. no problem gression, and anger between partners. The DCI to 4Ž. extreme problem . For this study, the 17 contains all eight physical aggression items and PTSD subscale items were summed for each sub- all six verbal symbolic aggression items from the ject, resulting in a continuous score for PTSD Conflict Tactics ScaleŽ. Straus, 1979 , but has been symptom severity. PTSD symptom scores ranged supplemented with additional items concerning from 0 to 40Ž means12.72, S.D.s9.43, median physical abuse, emotional abuse and general s10.00.Ž. , with 44 of the total 46 women 95.65% anger. reporting at least one PTSD symptom as being a Each woman was presented with the list twice, slight problem. first reporting on her partner’s behavior and then For a categorical diagnosis of PTSD, the re- 162 A.M. Medina et al. rPsychiatry Research 101() 2001 157᎐169 spondent must endorse at least one item on the a custom algorithm. In this algorithm, the ampli- PTSD subscale assessing re-experiencing of the tude of each EMG response is scored in micro- trauma, three items indexing avoidance and volts Ž.␮V as the difference between the mean numbing, and two items reflecting increased rectified EMG activity in the 200 ms preceding arousalŽ. King et al., 1995 . Of the 46 participants, the onset of the startle probe, and the mean seven qualified as PTSD-positive and two quali- EMG activity in the 10 ms following the peak of fied as PTSD-partialŽ meeting any two of the the response to the startle probe. The peak of the above three criteria. . None of the women en- response is defined as the highest, average micro- dorsed having sustained any severe or major ill- volt reading calculated across three EMG sam- nesses or injuries in the past year or having plesŽ because raw EMG was collected at 2000 Hz, sought therapy in the past year. Although this the average was calculated across 1.5 ms. . does not preclude an isolated Type I stressor that may have contributed to high levels of PTSD 2.6. Statistical approach symptoms, it does suggest that the likelihood of such a stressor, in the absence of injury, illness or Firstly, Pearson correlations were generated to therapy seeking, is low. confirm that subjects’ PTSD symptom scores were related to the Type II stressors of interest in this 2.4. Experimental stimuli study, namely childhood corporal punishment and intimate partner aggression. Pearson correlations The startle stimulus consisted of a 104-dBŽ. A were also used to explore associations between white noise that was 40 ms in duration. The white the blink average and each of the risk variables. noise was generated by a Grason-Stadler 901B Next, a simultaneous regression was conducted to noise generator and was gated at a near instanta- determine the relative contribution of childhood neous riserfall time. The eight stimuli were pre- and adult Type II stressors to subjects’ PTSD sented binaurally through headphonesŽ TDH- symptom scores. In addition, although extent of 49P. . Auditory stimulus intensities were cali- childhood corporal punishment, partner aggres- brated on a realistic sound level meterŽ cat. no. sion and PTSD symptomatology are all continu- 33-2055. using a Quest Electronics Earphone ous constructs, previous investigators have em- CouplerŽ. model EC-9A appropriate for the ployed the upper quartile to represent risk status supra-aural cushions. A 486 computer with a when dichotomizing distributions of risk variables METROBYTE DAS-16 ArD board and a cus- Ž.Klein and Forehand, 2000 . Thus, this investiga- tom program written in C controlled the onsets of tion adopted a categorical approach in order to and intervals between stimuli. compare extreme levels of the independent variables. Childhood corporal punishment, partner 2.5. Recording and scoring of dependent äriables aggression and PTSD symptomatology were grouped into the lowest, two middle and the Startle eyeblink amplitude was measured as highest quartiles of each distribution, and the electromyographic activityŽ. EMG from two lowest and highest quartiles were then compared miniature electrodesŽ. 4 mm in diameter placed with respect to the dependent variables. In other over the orbicularis oculi muscle of the left eye, words, to examine the influence of the indepen- one electrode centered below the pupil and the dent variables on startle response, groups repre- other approximately 10 mm lateral to the first. senting the lowest and highest quartiles of each The EMG signal was fed to a Grass 7P3 wide independent variable distribution were compared. band pre-amplifierrintegrator and a 7DA-driver Descriptions of group composition are included amplifier. The EMG signal was digitized at a rate in the results. of 2000 Hz for 200 ms before to 300 ms following Startle eyeblink magnitude scores were aver- the presentation of each startle-eliciting stimulus. aged across blocks of two trials, producing four The startle eyeblink was then scored off-line with trial blocks. Also, in order to evaluate the possi- A.M. Medina et al. rPsychiatry Research 101() 2001 157᎐169 163 bility that any group differences in startle eye- startle eyeblink magnitude. Moreover, this associ- blink magnitude might be due to group differ- ation between startle eyeblink magnitude and ences in baseline muscle tension of the orbicu- PTSD symptoms is a negative one, such that as laris oculi, mean baseline EMG activity Ž.␮V was PTSD symptom scores increase, startle eyeblink calculated for the 200 ms preceding each startle magnitude scores decrease. To investigate the probe onset for each trial block. Finally, because relative contributions of childhood corporal pun- the response to the startle stimulus on the first ishment and current partner aggression to PTSD, trial block may be dominated by the novel charac- a simultaneous multiple regression analysis was teristics of the stimulus rather than by group conducted to determine whether these variables differences in general startle reactivity, an aver- contribute uniquely to PTSD symptoms. Child- age of startle responses was calculated across the hood and adult stressors accounted for 37% of three subsequent trial blocks, excluding the first. the PTSD symptom varianceŽ adj R2 s0.37, d.f.s Thus, in addition to employing repeated measure 2,43, P-0.001. , with both betas significant at the - ␤ s analyses of variance to examine blink response P 0.05 level Ž Childhood corporal punishment 0.30; ␤ s across groups and all trial blocks, extreme groups Partner aggression 0.45. . Thus, for this community also were compared for differences on this modi- sample, both childhood and adult Type II stres- fied average of overall blink response, which ex- sors contribute uniquely to PTSD symptoms. cluded the first trial block.

3.2. Baseline EMG and startle eyeblink response

3. Results Comparisons of groups representing the highest and lowest quartiles of each risk variable were 3.1. Type II stressors and PTSD performed with respect to both baseline EMG levels and startle eyeblink magnitude scores. The Table 1 reveals associations between Type II lowest quartile of the childhood corporal punish- stressors and PTSD. Specifically, links can be ment distribution comprised women who never seen between childhood corporal punishment and experienced corporal punishment during child- PTSD symptoms, as well as between partner’s hood Ž.ns11 . The highest quartile of the dis- aggression and PTSD symptoms. Table 1 also tribution included women who reported being hit presents correlations among the modified blink at least 20 times during childhood Ž.ns10 . A average and risk variables, revealing that only groupŽ lowest vs. highest quartiles for childhood PTSD symptoms are significantly associated with corporal punishment.Ž.=trial block 1᎐4 repeated

Table 1 Relations among risk variables and modiﬁed blink averagea

Childhood Partner PTSD symptom Modiﬁed blink corporal aggression scores average punishment

Childhood corporal punishment ᎐ Partner aggression 0.40b ᎐ PTSD symptom scores 0.48bb 0.57 ᎐ Modiﬁed blink average y0.10 y0.16 y0.32c ᎐

a Two-tailed correlations. bCorrelation is significant at the 0.01 level. c Correlation is significant at the 0.05 level. 164 A.M. Medina et al. rPsychiatry Research 101() 2001 157᎐169 measures analysis of varianceŽ. ANOVA1 was In order to determine whether differential sen- conducted for both baseline EMG and eyeblink sitization or suppression from the first to the magnitude scores. No significant effects, neither second trial blocks occurred between the groups, main nor interaction effects, were detected. a groupŽ high vs. low quartiles for PTSD symptom For the partner aggression distribution, the scores.Ž.=trial blocks 1᎐2 repeated measures lowest quartile included women whose partners ANOVA was performed on eyeblink magnitude were reported to have an aggression score of 2 or scores. A main effect for group was observed lower Ž.ns10 , whereas the highest quartile com- indicating that the startle eyeblink magnitude prised women whose partners were reported to scores of the high and low quartile PTSD symp- have an aggression score of 19 or higher Ž.ns12 . tom groups significantly differed across both trial = s - Again, a group trial block repeated measures blocks Ž.F1,21 5.81, P 0.05 . In addition, a s ANOVA for baseline EMG and eyeblink magni- marginal interaction effect was detected ŽF1,21 tude revealed no significant findings for either 2.67, P-0.12. , despite very low observed power main or interaction effects. Ž.observed powers0.35 , showing that the two The lowest quartile of the PTSD symptom score groups differed almost significantly in the direc- distribution included women with symptom scores tion of their eyeblink magnitude response change of 6 or lower Ž.ns12 , and the highest quartile from trial block 1 to trial block 2. Whereas eye- contained women with symptom scores of 20 or blink magnitude scores of the low quartile PTSD higher Ž.ns11 . The seven women who met crite- symptom group increased slightly from trial block ria for a categorical diagnosis of PTSD and the 1Ž. means29.23, S.D.s23.97 to trial block 2 two women whom the LASC identified as Ž.means32.00, S.D.s28.65 , the eyeblink magni- ‘PTSD-partial’ were included in the highest quar- tude scores of the high quartile PTSD symptom tile group. That is, in addition to their symptom group decreased substantially from trial block 1 profile matching a ‘PTSD-positive’ or a ‘PTSD- Ž.mean s 13.02, S.D.s 16.18 to trial block 2 partial’ diagnosis, these women also scored in the Ž.means7.55, S.D.s8.39 . highest quartile of the PTSD symptom score distribution. A group=trial block ANOVA for base- 3.3. Modified blink aërage and risk äriables line EMG revealed only a significant main effect s - ␧s for trial block Ž.F3,63 6.41, P 0.05, 0.45 , As indicated earlier, a modified blink average showing that baseline EMG declined over trial was computed which omitted the first trial block. blocks. The low and high quartile PTSD groups This was done to circumvent problems raised by did not significantly differ with respect to baseline the possibility that response to the first trial block EMG, nor was an interaction effect detected. may have been dominated by novel characteristics A similar ANOVA conducted to examine star- of the stimulus, rather than group differences in tle eyeblink magnitude revealed main effects for startle reactivity. A one-way ANOVAŽ high vs. both trial block and group. That is, results indi- low quartiles of PTSD symptom scores. indicated cated that startle blink magnitude diminished over that the modified blink average of the high quar- s - ␧s s s blocks Ž.F3,63 4.55, P 0.05, 0.84 , and that tile groupŽ. mean 6.13, S.D. 6.16 was signifi- the high PTSD symptom group displayed significantly less than that of the low quartile group cantly smaller startle eyeblinks than the low PTSD wmeans26.53, S.D.s23.78, FŽ.1,21 s7.60 and s - - x symptom group Ž.F1,21 6.77, P 0.05 . Fig. 1 de- P 0.05 . picts the eyeblink magnitude scores for both high and low quartile PTSD symptom groups over the four trial blocks. 4. Discussion

The aim of this investigation was to highlight 1 Huynh-Feldt probabilities were used for the repeated potential links between Type II stressors, PTSD measures effects. symptoms and startle eyeblink reactivity. Results revealed that Type II stressors experienced both A.M. Medina et al. rPsychiatry Research 101() 2001 157᎐169 165

Fig. 1. Average startle blink magnitude across trial blocks for PTSD groups. in child and adulthood contribute uniquely to have developed PTSDŽ Grillon et al., 1996; Orr et women’s PTSD symptom scores. Contrary to ex- al., 1997. . pectation, results also suggested that PTSD symp- PTSD is often understood in terms of a diathe- tom scores associated with Type II stressors were sis᎐stress model, wherein genetic or acquired in- negatively related to startle eyeblink magnitude. dividual differences lead to differences in emotio- That is, higher PTSD symptom scores were re- nal reactivity and information-processing biases lated to smaller eyeblink responses. with respect to stressorsŽ Rasmusson and Char- No direct links were detected between Type II ney, 1997; Thrasher and Dalgleish, 1999. . This stressors and startle eyeblink reactivity. That Type model has been based primarily on research in- II stressors did not change startle magnitude is volving individuals exposed to Type I or acute perhaps not surprising in light of earlier ﬁndings stressors. The present investigation extends our where traumatized individuals suffering from understanding of PTSD and the generalizability PTSD differ from traumatized individuals not of the diathesis᎐stress model by showing that afﬂicted with the disorder. For example, previous high levels of PTSD symptoms are found in some, studies of individuals exposed to Type IŽ. acute but not all, individuals exposed to chronic or Type traumas have shown that physiological changes II stressors. It is only when these chronic stres- are most frequently detected in individuals who sors, like acute stressors, are associated with high 166 A.M. Medina et al. rPsychiatry Research 101() 2001 157᎐169

PTSD symptom scores that abnormalities in star- reducer, who tends to diminish what is perceived, tle reactivity manifest themselves. In this study, and the moderate, who does neither. Whereas the women with the highest, compared with the augmenter amplifies incoming stimulation, the re- lowest, PTSD scores demonstrated lower startle ducer immediately begins to limit hisrher intake eyeblink magnitude. of it. Petrie further stated that although these Researchers have noted that stimuli with mod- conditions may be genetically determined to some erate to strong intensities generally elicit initial extent, it was also likely they could be greatly sensitization of the startle response in animals altered by environmental conditions. In fact, she and in humans, followed by habituationŽ Groves suggested that one biological function that these and Thompson, 1970; Davis et al., 1982. . Results traits may serve is as an adaptation for survival. of this study indicate, however, that whereas This theory provides one possible explanation women with the lowest PTSD symptom scores for the results of this study, wherein women with seem to display sensitization from trial block 1 to high PTSD scores not only show the absence of trial block 2, women with the highest PTSD symp- exaggerated startle, but exhibit diminished startle. tom scores display an abrupt diminution in startle An investigation relevant to both Petrie’s theory response from trial block 1 to 2. This pattern of of reducers and findings from the present study, is response is consistent with the findings of Carrey Griffin and colleaguesŽ. 1997 study of peritrau- et al.Ž. 1995 regarding downward regulation in matic dissociation in rape survivors. These re- children exposed to Type II stressors. searchers observed suppression of autonomic The lower magnitude startle responses in the physiological responses for rape survivors classi- high PTSD group are especially striking given the fied into a high dissociation group. A person who dissociates in response to a stressor appears to fit moderately stressful context in which the eyeblink Petrie’s description of ‘reducers’. Perhaps Type II data were collected. Prior investigations have stressors, which are chronic and perceived to be found that the recall of trauma-related thoughts inescapable, more frequently elicit coping respon- or imagery is linked to heightened physiological ses of dissociation, and thus PTSD arising from reactivityŽ. Shalev et al., 1993 , and Grillon et al. Type II stressors is more likely to be related to Ž.1996 have speculated that ‘...the degree to suppression of physiological reactivity. The pre- which the experimental test conditions are per- sent findings, in conjunction with conflicting received as stressful or reminiscent of prior trauma sults from previous studies investigating startle may lead to a state of heightened startle during reactivity, highlight the possibility that psy- Ž. testing’ p. 170 . In this study, smaller startle chophysiological differences may exist among eyeblink responses were elicited in women from different subgroups of traumatized individuals. the high PTSD group following both an audio- An important feature of the present study was taped presentation of aggressive marital conflict the use of a non-treatment-seeking sample. Previ- and a questionnaire concerning their partner’s ous investigations into startle and PTSD have aggressive conflict behaviors. Findings of this in- relied primarily on treatment-seeking samples ex- vestigation thus suggest that PTSD arising from posed to Type I stressors. With the exception of Type II stressors experienced in either childhood Orr and colleaguesŽ. 1997 and Ornitz and Pynoos or adulthood may be related to suppression rather Ž.1989 , studies of startle and PTSD generally have than sensitization of physiological reactivity. employed treatment-seeking civilian hospital One theory consistent with this possibility is patientsŽ. Shalev et al., 1992, 1993 , Vietnam and that of augmentation vs. reduction of environ- Gulf War combat veterans at a Veterans Admin- ment perception. PetrieŽ. 1967 argued for a neu- istration hospitalŽ Grillon et al., 1996; Morgan et rological basis for individual differences in meth- al., 1995, 1996. , or victims of sexual assault partic- ods of perceiving the environment. Three types of ipating in an outpatient women’s trauma program persons were defined: the augmenter, who tends Ž.Morgan et al., 1997 . to subjectively increase what is perceived, the The prevailing use of treatment-seeking sam- A.M. Medina et al. rPsychiatry Research 101() 2001 157᎐169 167 ples in past studies of PTSD has been criticized as power, replication of these results is clearly war- being ‘seldom representative of the target popula- ranted. Nonetheless, these findings encourage tion’Ž. Solomon, 1993, p. 53 , with the use of such further investigation of types of stressors associ- samples placing ‘severe limitations on the generated with PTSD symptoms and additional inquiry alizability of findings to larger populationsŽ e.g. into the startle eyeblink response of persons af- the population of all patients with PTSD.Ž ’ Carl- flicted with PTSD stemming from Type II stres- son, 1996, p. 109.Ž. . Interestingly, Orr et al. 1997 sors. did not find PTSD vs. non-PTSD group differ- Previous studies examining the presence of ences in startle responding to loud tones in a PTSD symptoms and physiological reactivity have non-treatment seeking sample. Orr and his col- focused on PTSD associated with acute stressors leaguesŽ. Orr et al., 1995, 1997 have suggested and have relied on institutionalized or that exaggerated startle eyeblink responses treatment-seeking samples from hospitals or other observed in other PTSD subjects may be due to social service agencies. These samples may repre- an accompanying state of anxiety. As Orr et al. sent only one type of PTSD and also may reflect Ž.1997 pointed out, treatment-seeking samples of only the most extreme presentation of that type. convenience are more likely to include more se- By assessing the linkages between Type II or vere cases of PTSD than are the non-treatment- chronic stressors, PTSD symptoms, and the star- seeking samples that theyŽ. and also we studied. tle eyeblink response in a community sample, our Alternatively, a treatment-seeking, PTSD-af- study broadens understanding of the physiological flicted sample may be simply less effective in effects of both less severe levels of PTSD and the suppressing physiological responses than a similar influence of type of stressor associated with PTSD non-treatment-seeking sample. Whatever the on physiological reactivity. source of difference between treatment-seeking and non-treatment-seeking PTSD samples, the observations of Orr and colleagues and the find- Acknowledgements ings of this study suggest that the physiological profile of PTSD is more variable and complex This study has been supported in part by NIMH than previously thought. This possibility bears Grants 1 R01 36595 awarded to Dr Gayla Mar- further investigation. golin, and R01 MH46433rK02 MH01086 awarded Another important feature to consider in un- to Dr Michael E. Dawson. The authors thank derstanding the link between reduced startle re- Charles A. Morgan for his valuable suggestions activity and PTSD symptom scores is the expo- regarding this manuscript and the women who sure to the stressor. That is, whereas previous participated in this study for sharing their experi- studies of PTSD and startle eyeblink have investi- ences. gated individuals in the weeks, months and years following a particular traumatic stressor, this study References investigates PTSD symptoms and startle eyeblink within a population still exposed to one of the American Psychiatric Association, 1994. Diagnostic and Statis- Type II stressors. Moreover, in those cases where tical Manual of Mental Disorders, 4th Ed. Author, Wash- high PTSD symptom scores may be more related ington DC. Astin, M.C., Lawrence, K.J., Foy, D.W., 1993. Post-traumatic to childhood exposure to corporal punishment, stress disorder among battered women: risk and resiliency exposure to this Type II stressor is presumed to factors. Violence and Victims 8, 17᎐28. have encompassed years instead of the shorter Blanchard, E.B., Kolb, L.C., Gerardi, R.J., Ryan, P., Pall- durations of exposure to stressors in many previ- meyer, T.P., 1986. Cardiac response to relevant stimuli as ous studies. an adjunctive tool for diagnosing post-traumatic stress disorder in Vietnam veterans. Behavior Therapy 17, 592᎐606. A primary limitation of the present study is Butler, R.W., Braff, D.L., Rausch, J.L., Jenkins, M.A., Sprock, related to the small sample size. Although effects J., Geyer, M.A., 1990. Physiological evidence of exagger- were detected despite relatively low statistical ated startle response in a subgroup of Vietnam veterans 168 A.M. Medina et al. rPsychiatry Research 101() 2001 157᎐169

with combat-related PTSD. American Journal of Psychiatry of Psychology, University of Southern California, Los An- 147, 1308᎐1312. geles, CA 90089᎐1061. . Carlson, E.B., 1996. Trauma Research Methodology. Sidran, Medina, A.M., Margolin, G., Wilcox, R.R., 2000. Family hostil- Lutherville, MD. ity and children’s cognitive processes. Behavior Therapy 31, Carrey, N.J., Butter, H.J., Persinger, M.A., Bialik, R.J., 1995. 667᎐689. Physiological and cognitive correlates of child abuse. Jour- Medina, A.M., Torrey, M.L., Margolin, G.Ž. in preparation . nal of the American Academy of Child and Adolescent Psychological and Physiological Correlates of Domestic Psychiatry 34, 1067᎐1075. Hostility Partner aggresion. Carroll, E.M., Rueger, D.B., Foy, D.W., Donahoe, C.P., 1985. Morgan, C.A., Grillon, C., Lubin, H., Southwick, S.M., 1997. Vietnam combat veterans with post-traumatic stress dis- Startle reflex abnormalities in women with sexual assault- order: an analysis of marital and cohabitating adjustment. related post-traumatic stress disorder. American Journal of Journal of Abnormal Psychology 94, 329᎐337. Psychiatry 154, 1076᎐1080. Davis, M., Parisi, T., Gendelman, D.S., Tischler, M., Kehne, Morgan, C.A., Grillon, C., Southwick, S.M., Davis, M., Char- J.H., 1982. Habituation and sensitization of startle reflexes ney, D.S., 1995. Fear-potentiated startle in post-traumatic elicited electrically from the brainstem. Science 218, stress disorder. Biological Psychiatry 38, 378᎐385. 688᎐689. Morgan, C.A., Grillon, C., Southwick, S.M., Davis, M., Char- Dawson, M.E., Schell, A.M., Bohmelt,¨ A.H.Ž. Eds. , 1999. Star- ney, D.S., 1996. Exaggerated acoustic startle reflex in Gulf tle Modification: Implications for Neuroscience, Cognitive War veterans with post-traumatic stress disorder. American Science, and Clinical Science. Cambridge University Press, Journal of Psychiatry 153, 64᎐68. Cambridge. Mormede, P., Lemaire, V., Castanon, N., Dulluc, J., Laval, M., Donovan, B.S., Padin-Rivera, E., Dowd, D., Blake, D., 1996. Le Moal, M., 1990. Multiple neuroendocrine responses to Childhood factors and war zone stress in chronic PTSD. chronic social stress: interaction between individual charac- Journal of Traumatic Stress 9, 361᎐368. teristics and situational factors. Physiology and Behavior Filion, D.L., Dawson, M.E., Schell, A.M., 1998. The psycholog- 47, 1099᎐1105. ical significance of human startle eyeblink modification: a Orr, S.P., Lasko, N., Shalev, A.Y., Pitman, R.K., 1995. Physio- review. Biological Psychology 47, 1᎐43. logic responses to loud tones in Vietnam veterans with Foy, D., 1993. Los Angeles Symptom ChecklistŽ Unpublished PTSD. Journal of Abnormal Psychology 104, 75᎐82. Instrument. . Pepperdine University, Culver City, CA 90230. Orr, S.P., Solomon, Z., Peri, T., Pitman, R.K., Shalev, A.Y., Griffin, M.G., Resick, P.A., Mechanic, M.B., 1997. Objective 1997. Physiologic responses to loud tones in Israeli veterans assessment of peritraumatic dissociation: psychophysiologi- of the 1973 Yom Kippur war. Biological Psychiatry 41, cal indicators. American Journal of Psychiatry 154, 319᎐326. 1081᎐1088. Ornitz, E.M., Pynoos, R.S., 1989. Startle modulation in chil- Grillon, C., Morgan, C.A., Southwick, S.M., Davis, M., Char- dren with post-traumatic stress disorder. American Journal ney, D.S., 1996. Baseline startle amplitude and prepulse of Psychiatry 146, 866᎐870. inhibition in Vietnam veterans with post-traumatic stress Paige, S.R., Reid, G.M., Allen, M.G., Newton, J.E.O., 1990. disorder. Psychiatry Research 64, 169᎐178. Psychophysiological correlates of post-traumatic stress dis- Groves, P.M., Thompson, R.F., 1970. Habituation: a dual order in Vietnam veterans. Biological Psychiatry 27, process theory. Psychological Review 77, 419᎐450. 419᎐430. King, D.W., King, L.A., Foy, D.W., Gudanowski, D.M., 1996. Petrie, A., 1967. Individuality in Pain and Suffering. University Prewar factors in combat-related post-traumatic stress dis- of Chicago Press, Chicago. order: structural equation modeling with a national sample Pitman, R.K., Orr, S.P., Forgue, D.F., Altman, B., de Jong, of female and male Vietnam veterans. Journal of Consult- J.B., Herz, L.R., 1990. Psychophysiologic responses to com- ing and Clinical Psychology 64, 520᎐531. bat imagery of Vietnam combat veterans with post-trau- King, L.A., King, D.W., Leskin, G., Foy, D.W., 1995. The Los matic stress disorder vs. other anxiety disorders. Journal of Angeles symptom checklist: a self-report measure of post- Abnormal Psychology 99, 49᎐54. traumatic stress disorder. Assessment 2, 1᎐17. Pitman, R.K., Orr, S.P., Forgue, D.F., de Jong, J.B., Claiborn, Klein, K., Forehand, R., 2000. Family processes as resources J.M., 1987. Psychophysiologic assessment of post-traumatic for African-American children exposed to a constellation stress disorder imagery in Vietnam combat veterans. of sociodemographic risk factors. Journal of Clinical Child Archives of General Psychiatry 44, 970᎐975. Psychology 29, 53᎐65. Ross, R.J., Ball, W.A., Cohen, M.E., Silver, S.M., Morrison, Margolin, G., Burman, B., John, R.S., O’Brien, M., 1990. A.R., Dinges, D.F., 1989. Habituation of the startle reflex Domestic Conflict InventoryŽ. Unpublished Instrument . in post-traumatic stress disorder. Journal of Neuropsychia- University of Southern California, Los Angeles. try and Clinical Neurosciences 1, 305᎐307. Margolin, G.Ž. Executive ProducerrResearch Director , Rott- Rowan, A.B., Foy, D.W., Rodriguez, N., Ryan, S., 1994. Post- man, B.Ž.Ž. WriterrDirector , Jones, D. Sound Engineer , traumatic stress disorder in adults sexually abused as chil- 1992. Quarreling Couples: Five Examples of Conflict Style dren. International Journal of Child Abuse and Neglect 18, Ž.ŽAudiotapes . Available from Gayla Margolin, Department 51᎐61. A.M. Medina et al. rPsychiatry Research 101() 2001 157᎐169 169

Shalev, A.Y., Orr, S.P., Peri, T., Schreiber, S., Pitman, R.K., Terr, L.C., 1991. Childhood traumas: an outline and overview. 1992. Physiologic responses to loud tones in Israeli patients American Journal of Psychiatry 148, 10᎐20. with post-traumatic stress disorder. Archives of General Thrasher, S., Dalgleish, T., 1999. The use of information- Psychiatry 49, 870᎐875. processing paradigms to investigate post-traumatic stress Shalev, A.Y., Orr, S.P., Pitman, R.K., 1993. Physiologic assess- disorder: a review of the evidence. In: Yule, W.Ž. Ed. , ment of traumatic imagery in Israeli civilian post-traumatic Post-Traumatic Stress Disorders: Concepts and Therapy. stress disorder patients. American Journal of Psychiatry John Wiley and Sons Ltd, Chichester, UK. 150, 620᎐624. Rasmusson, A.M., Charney, D.S., 1997. Animal models of Shalev, A.Y., Peri, T., Orr, S.P., Bonne, O., Pitman, R.K., relevance to PTSD. In: McFarlane, A.C., Yehuda, R.Ž. Eds. , 1997. Auditory startle responses in help-seeking trauma Annals of the New York Academy of Sciences: Psychobi- survivors. Psychiatry Research 69, 1᎐7. ology of posttraumatic stress disorder, Vol. 821. New York Shalev, A.Y., Rogel, F.Y., 1993. Psychophysiology of the post- Academy of Science, New York, pp. 332᎐358. traumatic stress disorder: from sulfur fumes to behavioral ver Ellen, P., van Kammen, D.P., 1990. The biological findings genetics. Psychosomatic Medicine 55, 413᎐423. in post-traumatic stress disorder: a review. Journal of Ap- Solomon, Z., 1993. Combat Stress Reaction: The Enduring plied Social Psychology 20, 1789᎐1821. Toll of War. Plenum Press, New York. Williams, J.E., Scott, D.K., 1989. Influence of conspecific and Straus, M.A., 1979. Measuring intrafamilial conflict and vio- predatory stressors and their associated odors on defensive lence: the Conflict TacticsŽ. CT Scales. Journal of Marriage burying and freezing responses. Animal Learning and Be- and the Family 41, 75᎐88. havior 17, 383᎐393.