Autonomic activity in children and adolescents with obsessive-compulsive disorder

PSYCHIATRY RESEARCH ELSEVIER

Psychiatry Research 60 (1996) 67-76

Autonomic activity in children and adolescents with obsessivecompulsive disorder Theodore P. Zahn *a, Henrietta L. Leonard b, Susan E. Swedo b, Judith L. Rapoport b aLaboratory of Psychology and Psychopathology. Building !0. Room 4C! !0, National Institute of Mental Health, 10 Center Drive, M$C 1366, Bethesda. MD 20892-1366, USA bChild Psychiatry Branch. Building 10, Room 6N240, National Institute of Mental Health. !0 Center Drive, Bethesda, MD 20892, USA

Received 30 May 1995; revision received 22 September 1995; accepted 26 November 1995 Abstract

Electrodermal activity and heart rate were recorded from 55 children and adolescents with obsessive-compulsive disorder (OCD) and 58 normal subjects in a protocol that included rest and mild stress periods, and nonsignal and signal stimuli, to determine if autonomic activity might be involved in the pathogenesis of OCD or might be related to important clinical differences. Few differences were observed between OCD and normal subjects despite adequate power to detect small differences due to the large number of subjects. Thus, autonomic activity appears not to be an important etiological factor in childhood OCD. However, electrodermal activity showed consistent positive correlations with ratings of the severity of OCD symptoms (but not with anxiety or depression ratings), suggesting that severely afflicted cases are autonomically sensitive to OCD-related stimuli or, conversely, that low electrodermal activity may be prot~tire of syn,ptom severity. Patients with a coexisling tic disorder (not Tourette's syndrome) had larger electrodermal responses to a novel stimulus and higher heart rate variability than those without tics but did not differ from normal subjects. These few differences seem insufficient to support the hypothesis of a separate etiology of OCD cases with a coexisting tic disorder. Keywords: Electrodermal activity; Heart rate; Tic disorder

1. Intredaetion Obsessive-compulsive disorder (OCD) is classified as an anxiety disorder in D S M - I V (American Psychiatric Association, 1994)consistent with the commonplace clinical observations that exposure to sensitive stimuli, prevention of ritualistic behav* Corresponding author, Tel: +1 301 496-7672; Fax: +1 301 402.0921.

ior, and experience of obsessive thoughts produce increases in subjective anxiety. Also, the most effective type of behavior therapy for OCD, exposure with respon~,;e prevention, is similar to effective treatments for severe anxiety disorders such as agoraphobia (Foa and Kozak, 1985). As observed in other anxiety disorders, elevated autonomic nervous system (ANS) activity accompanies symptom expression and subjective anxiety (Boulougouris et al., 1977; Rabavilas et al., 1977),

0165-1781/96/$15.00 © 1996 Elsevier Science Ireland Ltd. All rights reserved SSD! 0 !65- ! 781(95)02846-0

68

T.P. Zahn et al. /Psychiatry Research 60 (1996) 67-76

and reductions in ANS activity after behavior therapy have been reported (Boulougouris et al., 1977). However, OCD differs from other anxiety disorders in at least two respects: (1) the specificity of serotonin uptake inhibitors as effective therapeutic agents rather than drugs that affect noradrenergic neurotransmission (Leonard et al., 1989) or anxiolytic drugs in general; and (2) evidence that OCD, at least in children, may be closely related to tic disorders and other neurological disorders of subcortical origin (Leonard et ai., 1992). This suggests that OCD might not be a typical anxiety disorder and/or that there may be more than one type of OCD. The present study seeks more evidence on these questions by looking at ANS activity in relation to the diagnosis and symptoms of OCD and coexisting tic disorders in children and adolescents. ANS markers in psychiatric disorders can be examined in relation to symptom expression. For example, low ANS base levels and responsivity are generalized accompaniments of melancholic or retarded depression, whereas elevated base levels may be seen in agitated or anxious depression (Lader and Wing, 1969; Kelly, 1980; reviewed in Zahn, 1986). In the anxiety spectrum, elevated ANS activity is found, of course, but specific disorders can be further categorized by the number or breadth of situations in which it occurs. Patients with agoraphobia and generalized anxiety disorders, in which severe anxiety occurs in a wide variety of situations, exhibit high sympathetic activity in the laboratory, manifested in such variables as increased spontaneous fluctuations of skin conductance (SC), elevated SC level (SCL) and heart rate (HR), and retarded habituation of orienting responses (ORs; Kelly and Walter, 1968; Zahn, 1986; Boucsein, 1992). In contrast, subjects with specific phobias exhibit high ANS activity in the presence of their feared objects but not necessarily in laboratory situations that do not involve those objects (Kelly and Walter, 1968; Zahn, 1986; Boucsein, 1992). It has not always been clear where to put OCD in this schema. An influential earlier theory of Beech and Perrigault (1977) hypothesized that persons with OCD have large fluctuations of ANS activity that become associated with symptoms when spontaneous reductions in arousal coincide for-

tuitously with certain behaviors; because such individuals also are characterized by fast conditioning and slow habituation and adaptation, these behaviors become rituals or obsessive thoughts. Thus, greater responsivity and slower adaptation and habituation in laboratory situations would be predicted. Several studies of adults have reported elevated ANS base levels in OCD patients compared with normal subjects. Although two studies of noradrenergic function in OCD yielded negative results (Benkelfat et al., 1991; Hollander et al., 1991), higher HR (Kelly, 1980; Insel et al., 1985; Benkeifat et al., 1991), higher diastolic blood pressure (Kelly, 1980), higher rates of spontaneous SC fluctuations and higher SCL (Insel et al., 1985) in OCD patients than in comparison subjects under baseline conditions have been reported. Lelliott et al. (1987) found slow habituation in OCD patients to 100-dB tones, which are probably aversive. However, Beech and Perrigault (1977) presented only minimal evidence for slow habituation in OCD, and Insel et al. (1985) failed to confirm this finding with tones of moderate intensity. Although challenges with imagery of sensitive scenes increase ANS activity and subjective anxiety in patients with OCD (Boulougouris et al., 1977; Rabavilas et al., 1977), these demonstrations do not address the question of whether OCD patients are generally hyperresponsive to stress. Laboratory challenges with mental arithmetic have actuaY~lyproduced smaller cardiovascular reactions (Kelly, 1980; Insel et al., 1985) and no differences in SC responses (Insel et ai., 1985) in OCD patients compared with normal subjects. This suggests, at a minimum, that OCD patients are not generally hypersensitive to stress. None of the previous studies investigated the relationship of the ANS markers to the severity of the disorder. Since childhood OCD is assumed to be an early stage of the same disorder that occurs in adults (e.g., Swedo et al., 1989a), one should expect the same ANS accompaniments of the disorder in children as have been found in adults. However, in a study of 21 boys and 9 girls with childhood OCD (Berg et al., 1986b), no differences were reported on baseline electrodermai activity (EDA) and HR compared with normal subjects in the group as a whole, but significant group × sex interactions

T.P. Zahn et al. / Psychiatry Research 60 (1996) 67-76

were frequently observed. Compared with controls, OCD boys were significantly higher on nonspecific fluctuations Of SC and mean HR, whereas OCD girls gave significantly fewer SC orienting responses (SCORs) and were marginally lower on nonspecific fluctuations and SCL. Thus, where OCD boys differed from controls, it was in the direction of higher sympathetic activation, while for girls the differences tended to be in the opposite direction. There are many possible interpretations of this unexpected result, perhaps the least arguable being that elevated sympathetic activity is not a necessary condition for the maintenance of OCD. Before the implications of this finding are pursued, however, it would be wise to attempt to replicate it. The present study assesses the role of ANS activity, under resting and activated conditions, in the pathology of childhood-onset OCD in a larger number of patients and control subjects than were included in the previous study. Autonomic activity, as indexed by SC and HR, was recorded during a simple protocol that includes three parts: (1) a re'st period to assess baseline autonomic activity (arousal) and its rate of adaptation to the testing situation; (2) a series of innocuous tones from which ORs and their habituation are evaluated; and (3) a simple warned reaction time (RT) task to measure the increase in autonomic activity during the challenge of task performance and responses to signal or meaningful stimuli. The effects of diagnosis and gender on ANS activity are tested by comparing the OCD subjects with a normal control group. The question of the role of ANS activity in symptom expression is examined by relating the ANS variables to the rated severity of OCD and other symptoms. In addition, a substantial subgroup of patients have a lifetime diagnosis of motor tics that may constitute an etiologically distinct variant of OCD (Leonard et al., 1992). This hypothesis should be testable since tic disorders would be expected to show differences in ANS activity from anxiety disorders. 2. Methods

2.1. Subjects Fifty-five unmedicated patients with DSM-III-R diagnoses (American Psychiatric Association,

69

1987) of OCD (34 males; mean age = 13.6 years, SD - 2.9) were compared with 58 control subjects (38 males; mean age = 13.3 years, SD = 3.0). Exclusion criteria were any significant neurological disorders or medical illnesses that required ongoing medication, a history of seizure disorders, IQ <80 on the Wechsler Intelligence Scale for Children-Revised, and psychosis. Diagnoses were made on the basis of clinical and structured interviews - - the Diagnostic Interview for Children and Adolescents (DICA; Herjanic and Campbell, 1977) and the Diagiostic Interview for the Parents of Children and Adolescents (DICA-P; Herjanic and Reich, 1982) - - of the child and of at least one parent about the child. All interviews were administered by a child psychiatrist. Available records from previous therapists, hospitals, residential treatment facilities, and schools were also consulted. Subjects with a co-diagnosis of Tourette's syndrome were excluded, but many subjects had DSM-III-R lifetime diagnoses of tics (n=30; 22 males; mean age= 13.5 years, SD = 3.3) as determined from the two DICA interviews, available medical records, and a neurological examination. Control subjects were recruited from the community, screened by the DICA and DICA-P interviews, and paid for their participation. They were excluded if any psychiatric diagnosis or medical illness was present. The patients were admitted as inpatients to a children's ward in the Clinical Center of the National Institutes of Health and were tested on the second or third day on the ward. They had been free of medication for at least 3 weeks before admission. The control subjects were admitted for 2 days and were tested on the second day. Two subjects in each group who met inclusion criteria refused participation in the study. All who were admitted to the ward participated in this protocol.

2.2. Clinical scales and ratings Clinical information available on the OCD group included the !.eyton Obsession InventoryChild Version (Berg et al., 1986a), a self-report scale that has been validated in clinical settings in children and has a reported test-retest reliability of 0.94-0.96 for the three subscales (Berg et al., 1986a), and several symptom rating scales completed by a child psychiatrist and a pediatrician

70

T.P. Zahn et al./ Psychiatry Research 60 (1996) 67-76

(H.L.L. and S.E.S.). Interrater reliability for the two raters was reported to be 0.91-0.98 (Swedo et al., 1989b). The measures included the ObsessiveCompulsive Rating-Ward Scale (Rapoport et al., 1980); the OCD subscale of the Comprehensive Psychiatric Rating Scale (CPRS; ~sberg et al., 1978); the National Institute of Mental Health Global Scales for OCD, Depression, Anxiety, and Global Functioning (Murphy et al., 1982); and the Insel OCD Scale (lnsel et al., 1983). 2.3. Procedures Physiological recording was done on a Grass polygraph (Model 7B; Quincy, MA, USA), the output of which was digitized by a PDP-I 1 computer for off-line editing and analysis. SC was recorded by a constant voltage (0.5 V) method from the distal phalanges of the middle and ring fingers of the left hand using Ag/AgCI recessed electrodes, electrode collars 0.8 cm in diameter, and 0.5% KC! electrode paste. HR was recorded by a Grass tachograph from an electrocardiographic signal (lead II). The procedure began at least 10 rain after the SC electrodes were applied to allow for hydration. Subjects were told that they should try to relax but stay awake, that after a few minutes there would be a series of tones or 'beeps', and that they did not have to do anything except to continue to relax. There ensued a 3-rain rest period followed by presentation through a speaker of ten 1000-Hz 80 dB (re 0.0002 dynes/cm 2) pure tones of 1.5-s duration at 20- to 30-s in~erstimulus intervals. This was followed by a simple warned RT task in which the subject was instructed as follows: On each trial a ready light would come on; when the subject was ready, he/she should depress a telegraph key and keep it down until a tone ('beep') sounded; at this time, the subject should release the key as quickly as possible; we were timing hew fast the subject responded to the beep. The instructions were accompanied by a demonstration and followed by at least four practice trials to ensure that the procedure was understood. The task consisted of nine trials with 4-s foreperiods followed by nine trials with 8-s foreperiods. The foreperiod was measured from the time the key was depressed to the onset of the RT stimulus. A premature key release

aborted the trial and turned on the ready light. The intertrial interval was irregularly distributed between 8 and 14 s (mean = 11 s). 2. 4. Data analysis For the rest and task instruction periods and the intervals between tones, the rate of 'spontaneous' SCRs per minute (SCR/min) for SCRs of at least 0.02/iS in amplitude; the mean amplitude of the SCRs; the mean SCL in ~S, measured at l-min intervals; and the change in SCL/min (slope) were evaluated. These periods were divided into 10-s 'epochs' and the mean, maximum, and minimum HRs were computed for each epoch. From these values, the mean HR and HR variability (defined as the mean difference between HR maxima and minima) were computed. The criteria for elicited SCRs were also a minimum amplitude of 0.02 ~S and an onset latency between 0.8 and 4 s, and their frequency, mean magnitude (total amplitude divided by number of stimuli), and latency were measured. The number of trials to habituation to a criterion of two consecutive no-response trials was evaluated for SCRs ~'o nonsignal tones. SCRs to the RT ready signal and stimulus were evaluated separately. HR decelerative and accelerative responses to the tones were measured separately by the difference between 10-s prestimulus minima or maxima and the corresponding poststimulus values. HR deceleration during the RT foreperiod was defined as the level at the time of the warning stimulus minus that at the time of the RT stimulus. Behavior measured during the RT task included median RT and the SD of the RTs in ms, and two measures of extraneous key presses: the number of intertrial presses of the RT key, and the number of trials that were aborted by premature release of the key. Because of the large numbers of variables generated by using multiple dependent variables in the several segments of the protocol, an attempt to reduce type I error was made, when possible, by analyzing data for several dependent variables and for repeated measures initially by group × sex multivariate analysis of covariance with age as a covariate (MANCOVA) with BMDP Program 4V (Dixon, 1990). These were followed up by

T.P. Zahn et aL / Psychiatry Research 60 (1996) 67-76

univariate analyses of covariance (ANCOVAs) with the Huynh-Feldt e correction of the degrees of freedom when appropriate. Relationships between ANS and clinical variables were tested by partial correlations that controlled for age and sex. 3. Results

3.1. OCD vs. control subjects A groups x sex x condition MANCOVA on SCR/min, SCL, and HR for rest, tones, and RT instruction periods produced no significant multivariate or univariate main effects or interactions involving groups. Thus, there was neither a difference in overall baselines nor a difference in the tonic response to the mild stress of task performance. However, the expected effects of elevated HR in female subjects (P < 0.0003) and greater mean levels of all variables in the R T instruction period (multivariate P < 0.0001) were obtained. A MANCOVA on the frequency and magnitude

of SCRs elicited by the nonsignal tones, the RT ready signal, and the RT stimulus produced a marginally significant group × sex × condition interaction (multivariate F = 2.12, dr= 4,434, P < 0.08) due to a significant univariate effect of magnitude ( F = 4.21, e = 0.86, dr= 2,218, P < 0.02). Tests of simple effects for males and females separately showed no effects for males but a group x condition effect for females (multivariate F = 3.03, dr= 4,434, P < 0.02) and a univariate effect for SCR magnitude F = 6.05, e = 0.86, dr= 2,218, P = 0.004). Fig. I shows that this result was due to an exceptionally high value of SCR magnitude in response to the nonsignal tones for control females compared with all other groups. A test on the data for the nonsignal tones only showed significant group effects for both frequency and magnitude of SCRs (F _ 4.89, df= 1,109, P < 0.03), and a group × sex interaction for magnitude (F = 4.47, P < 0.04). Electrodermal ORs to the simple tones were ex-

Control M OCD M I ~ Control F ['70CD F

0.7. AC.6-

co ~L "O

0.5-

mm

I

'1~ 0.4m

I¢ 0

0.a. I'

O~ 0.20.1

!

I

Tones

71

RT Rdy

RT Stim

Fig. I. Skin conductance response (SCR) magnitudes elicited by simple tones, the reaction time (RT) ready (Rdy) signal, and the RT stimulus (Sfim) for male (M) and female (F) control subjects and those with obsessive-compulsive disorder (OCD). Bar extensions represent standard errors of the mean.

72

T.P. Zahn et al. / Psychiatry Research 60 (1996) 67-76

plored further by a MANCOVA on the trials to habituation for the whole series and for the amplitude of the response to the first trial. Although the overall multivariate test for groups was marginally significant (F = 2.76, d f = 2,107, e < 0.07), both of the univariate tests were significant. These showed that the control subjects had more trials to habituation (F = 4.04, P < 0.05) and gave larger responses on trial 1 (F = 4.03, P < 0.05). Although in each case the results were stronger for women than men, the group × sex interactions were not significant. An ANCOVA showed a significant group × sex interaction for HR deceleration during the long foreperiod ( F = 7.02, d f = 1,95, e < 0.01). Tests of the simple effects of sex showed nonsignificantly greater deceleration in control males but a larger response in female OCD patients than in control females (F = 6.26, d f = 1,95, P < 0.02). There were, however, no differences in RT. 3.2. Correlations with clinical ratings

Another test of ANS involvement in OCD is whether the ANS variables relate to the severity of

the disorder within the OCD group. Since several of the 10 clinical rating scales assessed O C D symptoms, it was felt desirable to reduce the number of clinical indices by combining highly correlated measures. Although the ratings were not always substantially correlated with each other, there were four groups of scales with reasonably high within-group intercorrelations. Four composite scores were formed by averaging the standard scores for the individual measures. The three Leyton OCD scales form one group; the Ward OCD rating, the OCD rating from the CPRS, and the Insel scale form another, labeled ' O C D Severity'; the NIMH Global and OCD Scales form a third grouping, called 'Global Severity'; and a final grouping - - 'Anxiety-Depression' - - consisted of the NIMH Anxiety and Depression subscales. Table 1 shows those variables that had at least one significant (P < 0.05) partial correlation (with age and sex partialled out) with a composite score. It can be seen, in general, that EDA was positively correlated with OCD symptomatology but only marginally and inconsistently correlated with

Table 1 Partial correlations, controllingfor age and sex, between composite clinical ratings and psychophysiologicalvariables Period and variable

Rating Leyton OCD

OCD severity

Global severity

Anxiety-depression

0.35** 0.37***

0.36"* 0.43***

0.16 0.13

0.25* 0.27*

0.45*** 0.35** 0. !2

0.25* 0.22 0.35**

0.27* 0. ! ! 0.15

Orienting responses

Trial 1 SCR amplitude Total SCR magnitude Rest and tones

Nonspecific SCR amplitude SC level Heart rate

0.08 0.20 -0.06

Reaction time instructions

SCR frequency SCR amplitude SC level Increase in SCR frequency Increase in SC level

0.44*** 0.16 0.24

0.37** 0.17

0.25* 0.42*** 0.40*** 0.10 0.40***

-0.13 0.20 0.21 -0.22 0.08

0.37*** 0.40*** 0.17

-0.01 0.17 -0.35***

0.02 0. !6 0.15 0.00 0.17

$CR magnitude in reaction time task

Ready signal Reaction time stimulus Median reaction time

0.26* 0.30** 0.04

0. !5 0.28* -0.32**

Note. OCD, obsessive-compulsivedisorder; SC, skin conductance; SCR, skin conductance response; Asterisksdenote significance levels of correlations: ***P < 0.01; **P < 0.05; *P < 0.I0.

73

T.P. Zahn et al. /Psychiatry Research 60 (1996) 67-76

Table 2 Variables showing significant and marginal differences between OCD patients with and without tics and normal control subjects Variable

Group OCD with tics

OCD without tics

Control

Mean

SD

Mean

SD

Mean

SD

Amplitude of trial 1 SCOR (~S) T,~tal magnitude of

1.390

1.247

0.649

0.771

1.411"**

!.104

0.438

0.495

0.315

0.376

0.527*

0.555

SCL: rest and tones

7.569

4.649

5.707

2.485

8.080*

4.666

14.539

6.238

! !.529

4.206

14.541"*

6.593

SCOR ~s)

O,s) HR variability

(beats/min)

Note. OCD, obsessive-compulsive disorder; SCOR, skin conductance orienting response; SCL, skin conductance level; HR, heart rate. Asterisks indicate significance levels for group effects: *P < 0.10; **P < 0.05; ***P < 0.01.

ratings of ancillary symptoms. Both of the more specific OCD scales were correlated with the magnitudes of SCRs elicited by both the simple tones and the RT stimuli. However, while OCD severity consistently correlated with nonspecific SCR amplitudes and SC levels in both rest and RT instruction periods, the Leyton scales were correlated with the frequency of SCRs in the instruction period and with its increase from the preceding period in which no demands were made on the subjects. In contrast to the abundant significant correlations for the EDA variables, resting HR was correlated only with the Global Severity index. HR variability and responsivity were not correlated with any of the scales. Finally, rather surprisingly, the patients with higher ratings on the ancillary symptoms had faster RTs.

3.3. Effects of tics Group x sex analyses revealed only a few differences (see Table 2). There was a significant effect for group on SCOR amplitude to the first tone (F = 5.16, df= 2,102, P < 0.008). This was low in the subgroup without tics compared with both control subjects (P < 0.05 by Newman-Keuls post hoc test) and OCD patients with tics (P < 0.01), who did not differ. HR variability in the rest and tones periods was also low in the no-tic subgroup compared with the two other groups

(F = 3.49, d f - 2,102, P < 0.04), but the individual means did not differ according to the NewmanKeuls test. Similar trends were also present for mean SCL and the magnitudes of the SCORs across all trials (P < 0.10). 4. Discussion

There were neither overall differences in ANS baselines nor in the response to the challenge of task performance between OCD and control subjects. The likelihood that this is a Type II (falsenegative) error is diminished by the relatively large sample size. To the authors' knowledge, it is the largest such sample studied by these methods to date. The findings are similar to those of the previous study from this laboratory on childhood OCD (Berg et al., 1986b) and are inconsistent with the hypothesis that OCD is characterized by generally elevated ANS activity. Indeed, the smaller electrodermal ORs to nonsignal tones and faster habituation in the OCD subjects are in the opposite direction from that hypothesis. These findings held only in girls, providing some confirmation of the low EDA in OCD girls found previously (Berg et al., 1986b). However, the robust group x sex interactions and the elevated ANS activity in OCD boys found in the earlier study were not generally replicated, and the most

74

T.P. Zahn et aL / Psychiatry Research 60 (1996) 67-76

reliable result here, that for SCR magnitude, appeared to be due mainly to a high mean value for the control girls. The only other difference was greater HR deceleration in the RT foreperiods for OCD girls compared with control girls. It is thought that this reflects attention or preparation and that it facilitates sensory reception and motor responding (Jennings, 1986). While there is a certain amount of face validity to the notion that OCD subjects would develop a high degree of preparation during the RT foreperiod, there were no differences in RT itself, so it is difficult to know how to interpret the finding. Since significant findings for the overall group × sex analyses were so sparse, it would seem premature to attach too great a significaace to them before replication. Tic disorders are much more common in subjects with OCD than in the general population, and both disorders tend to coexist in the same families (Leonard et al., 1992), suggesting a genetic co-determination. It is possible, however, that there are subtypes of OCD with and without a concomitant risk for tics that may constitute etiologically distinct conditions. Psychophysiological differences between patients with and without tics would support the two-syndrome model. The tic group gave significantly larger Trial 1 SCORs and had marginally higher SCL and total SCOR amplitudes than OCD patients without tics. They also had significantly higher resting HR variability, possibly due to greater motor activity. These few differences out of the total number of possibilities provide only limited support for the separate etiology hypothesis, although it should be kept in mind that cases with Tourette's syndrome were omitted from the study group. Since in all these cases the no-tic OCD patients were deviant from, while the tic group was similar to, the control group, the data would seem more compatible with a hypothesis of selectively low EDA and HR variability in OCD which is attenuated by a concomitant tic disorder. Further evidence on this might come from studying patients with tic disorders but without OCD. Although the comparisons with control subjects do not support a role for elevated sympathetic activity in the etiology or maintenance of OCD, the

correlational analyses clearly do suggest that ANS activity is related to individual differences in symptom severity. Most of the significant relationships are between the scales measuring specific OCD symptoms and electrodermal responsivity to the stimuli in the protocol or to the task onset. It seems likely that in more severely ill patients, symptoms would also be accompanied by exaggerated ANS reactions. Subjects may differ in the sensitivity of the ANS to the internal or external stimuli that prompt symptom expression, and the symptoms may be or appear to be more severe when they are accompanied by strong ANS reactions. That is, low ANS reactivity may be protective against symptom severity. This is supported by evidence that successful behavior therapy (Boulougouris et al., 1977) and pharmacotherapy (Zahn et al., 1984) are accompanied by attenuation of ANS reactivity to OCD-relevant or OCDneutral stimuli (although this may not be a sufficient condition for clinical improvement [Kozak et al., 1989]). In the present unmedicated patients, it may be that the less symptomatic subjects have learned to attenuate their ANS reactions to stimuli. It is possible that EDA is a marker for or mediates some other important etiological factor. For example, in an OCD group that included most of the present cases, several indices of EDA were high in subjects whose parents were rated high on expressed emotion, a measure of criticism and overinvolvement (Hibbs et ai., 1992). This measure, in turn, was related to OCD severity (r = 0.44, P < 0.01). In addition, there was a very specific set of positive relationships between SCR amplitudes throughout the protocol and the presence of a psychiatric diagnosis in the mothers of the patients (Hibbs et al., 1992). The~e findings suggest that the correlations of EDA and symptom severity may reflect in part a familial and/or genetic influence. Similarly, if more speculatively, EDA may reflect differences in brain activity. Frontal lobe glucose metabolism correlated positively with symptom severity in adults with childhood-onset OCD (Swedo et al., 1989b), and there is evidence that similar areas of the frontal lobe are involved in EDA (Hazlett et al., 1993; Tranel and Damasio, 1994).

T.P. Zahn et al. / Psychiatry Research 60 (1996) 67-76

The finding that many electrodermal, but not HR, variables correlated significantly with OCD symptom severity constrains interpretations in terms of arousal or autonomic activity. However, in Fowles' (1980) translation of Gray's (1987) theory, EDA, but not HR, indexes the behaviorinhibition system. Thus, our data suggest that differences in symptom severity are accompanied by differences in behavioral inhibition. Although compulsions and rituals might be thought of as expressions of active avoidance, which is indexed by HR (Fowles, 1980), Gray (1987, p. 368) hypothesizes that in OCD the environment is frequently scanned and checked for danger, which is a function of the behavior-inhibition system, and if danger is found, ongoing behavior is indeed inhibited, much as in other anxiety disorders. The correlation data thus provide some limited support for Gray's conception of OCD. However, the finding that neither EDA nor HR was elevated in the OCD group as a whole would seem inconsistent with that hypothesis. Patients with OCD may be more like those with specific phobias who have elevated ANS activity only in the presence of feared stimuli. More severely symptomatic patients might be more likely to be sensitive to some aspect of the testing situation, which then triggers their behavior-inhibition system and elevates their EDA. Perhaps, challenge studies in OCD would help resolve this issue, but there may be formidable difficulties in equating challenges among different patients. The present negative results for the overall group comparisons, as well as those from Berg et ai. (1986b), are in some conflict with the studies reviewed in the introduction on adult OCD, especially one from this laboratory (Insel et al., 1985) which found elevated baselines for both EDA and HR. This might suggest that the illness is different in children and adolescents than in adults, but this seems unlikely in view of the many similarities of child and adult OCD (e.g., Swedo et al., 1989a). The correlations with severity suggest that the a ~ult samples with elevated ANS activity may be more severely afflicted with the illness. Support for this hypothesis comes from data showing that elevated EDA is related to poor 2- to 5-year outcome in these patients (Zahn et al.,

?5

1996). Elevated ANS activity in adult OCD may also be secondary to the life stress engendered by a chronic, life-disrupting illness that has resisted many attempts at treatment. Thus, studies of OCD in younger samples may be less 'contaminated' by these factors and give a truer picture of the role of ANS activity in the illness. Acknowledgments The authors are indebted to Thalene T. Mallus and David C. Rettew for expert technical assistalice.

References American Psychiatric Association. (1987) DSM-IlI-R: Diagnostic and Statistical Manual of Mental Disorders. 3rd rev. edn. American Psychiatric Press, Washington, DC. American Psychiatric Association. (1994) DSM-IV: Diagnostic and Statistical Manual of Mental Disorders. 4th edn. American Psychiatric Press, Washington, DC. Asberg, M., Montgomery, S.A., Perris, C., Schalling, D. and Sedvall, G. (1978) A comprehensive psychopathological rating scale. Acta Psychiatr Scand 271, 5-27. Beech, H.R.a.~d Perrigault, J. (1977) Toward a theory of obsessional disorder. In: Beech, H.R. (Ed.), Obsessional States. Methuen and Co., Ltd., London, pp. ! 13-141. Benkelfat, C., Mefford, I.N., Masters, C.F., Nordahl, T.E., King, A.C., Cohen, R.M. and Murphy, D.L. (1991) Plasma catecholamines and their metabolites in obsessivecompulsive disorder. Psychiatry Res 37, 321-331. Berg, C.J., Rapoport, J.L. and Flament, M.F. (1986a) The Leyton Obsessional Inventory - - child version. J 2m Acad Child Adolesc Psychiatry 25, 84-91. Berg, C.J., Zahn, T.P., Behar, D. and Rapoport, J.L. (1986b) Childhood obsessive-compulsive disorder: an anxiety disorder?. In: Gittelman, R. (Ed.), Anxiety Disorders of Childhood. Guilford Press, New York, pp. 126-135. Boucsein, W. (1992) Electrodermal Activity. Plenum Press, New York. Boulougouris, J.C., Rabavilas, A.D. and Stefanis, C. (1977) Psychophysiological responses in obsessive-compulsive patients. Behav Res Therapy 15, 221-230. Dixon, W.J. (1990) BMDP Statistical Software Manual. University of California Press, Berkeley, CA. Foa, E.B. and Kozak, M.J. (1985) Treatment of anxiety disorders: implications for psychopathology. In: Tuma, A.H. and Maser, J. (Eds.), Anxiety and the Anxiety Disorders. Lawrence Eribaum, Hillsdale, NJ, pp. 421-452. Fowles, D.C. (1980)The three arousal model: implications of Gray's two-factor learning theory for heart rate, electrodermal activity, and psychopathy. Psychophysiology 17, 87-104. Gray, J.A. (1987) The Psychology of Fear and Stress. 2nd edn. Cambridge University Press, Cambridge, UK.

76

T.P. Zahn et ai./ Psychiatry Research 60 (1996) 67-76

Hazlett, E.A., Dawson, M.E., Buchsbaum, M.S. and Nuechterlein, K.H. (1993) Reduced regional brain glucose metabolism assessed by positron emission tomography in electroderreal nonresponder schizophrenics: a pilot study. J Abnorm Psychol 102, 39--46. Herjanic, B. and Campbell, W. (1977) Differentiating psychiatrically disturbed children on the basis of a structured interview. J Abnorm Child Psychol 5, 127-134. Herjanic, B. and Reich, W. (1982) Development of a structured psychiatric interview for children: agreement between child and parent on individual symptoms. J Abnorm Child Psychol 10, 3O7-324. Hibbs, E.D., Zahn, T.P., Hamburger, S.D., Kruesi, MJ.P. and Rapoport, J.L. (1992) Parental expressed emotion and psychophysiological reactivity in disturbed and normal children. Br J Psychiatry 160, 504-510. Hollander, E., DeCaria, C., Nitescu, A., Cooper, T., Stover, B., Gully, R., Klein, D.F. and Liebowitz, M.R. (1991) Noradrenergic function in obsessive-compulsive disorder: behavioral and neurocndocrine responses to clonidine and comparison to healthy controls. Psychiatry Res 37, 161-177. lnsel, T.R., Murphy, D.L., Cohen, R.M., Alterman, I., Kilts, C. and Linnoila, M. (1983)Obsessive-compulsive disorder: a double-blind trial of domipramine and clorgyline. Arch Gen Psychiatry 40, 605-612. Insel, T.R., Zahn, T.P. and Murphy, D.L. (1985) Obsessivecompulsive disorder: an anxiety disorder? In: Tuma, A.H. and Maser, J. (Eds.), Anxiety and the Anxiety Disorders. Lawrence Erlbanm Associates, Hillside, NJ, pp. 577-590. Jennings, J.R. (1986) Bodily changes during attending. In: Coles, M.G.H., Donchin, E. and purges, S.W. (Eds.), Psychophysiology: Systems, Processes, and Applications. Guilford Press, New York, pp. 268-289. Kelly, D. (1980) Anxiety and Emotions. Charles C Thomas, Publisher, Springfield, IL. Kelly, D. and Walter, C.J.S. (1968) The relationship between clinical diagnosis and anxiety as assessed by forearm blood flow and other measurements. Br J Psychiatry i 14, 611-666. Kozak, M.J., Rossi, M., McCarthy, P.R. and Foa, E.B. (1989) Effects of imipramine on the autonomic responses of obsessive-compulsives to auditory tones. Biol Psychiatry 26, 707-716. Lader, M.H. and Wing, L. (1969) Physiological measures in agitated and retarded depressed patients. J Psychiatr Res 7, 89-95. Lelliott, P.T., Noshirvani, H.F., Marks, I.M., Monteiro, W.O., llasogiu, M. and Cohen, S.D. (1987) Relationship of skin conductance activity to clinical features in obsessivecompulsive ritualizers. Psychol Med 17, 905-914. Leonard, H.L., Lenane, M.C., Swedo, S.E., Rettew, D.C., Ger-

shon, E.S. and Rapoport, J.L. (1992) Tics and Tourette's disorder: a 2- to 7-year follow-up of 54 obsessive-compulsive children. Am J Psychiatry 149, 1244-1251. Leonard, H.L., Swedo, S.E., Rapoport, J.L., Koby, E.V., Lenane, M.C., Cheslow, D.L. and Hamburger, S.D. (1989) Treatment of obsessive-compulsive disorder with clomipramine and desipramine in children and adolescents: a double-blind crossover comparison. Arch Gen Psychiatry 46, 1088-1092. Murphy, D.L., Pickar, D. and Alterman, I.S. (1982) Methods for the quantitative assessment of depressive and manic behavior. In: Burdock, E.L., Sudilovsky, A. and Gershon, S. (Eds.), The Behavior of Psychiatric Patients. Marcel Dekker, Inc., New York, pp. 355-392. Rabavilas, A.D., Boulougouris, J.C. and Stefanis, C. (1977) Synchrony and concordance on subjective and psychophysiological measures after beta-blockade and flooding in obsessive-compulsive patients. In: Boulougouris, J.C. and Rabaviias, A.D. (Eds.), The Treatment of Phobic and Obsessive-Compulsive Disorders. Pergamon Press, Oxford, UK, pp. 115-125. Rapoport, J.L., Elkins, R. and Mikkelsen, E. (1980) Clinical controlled trial of clomipramine in adolescents with obsessive-compulsive disorder. Psychopharmacoi Bull 16, 61-63. Swedo, S.E., Rapoport, J.L., Leonard, H., Lenane, M. and Cheslow, D. (1989a) Obsessive-compulsive disorder in children and adolescents: clinical phenomenoiogy of 70 consecutive cases. Arch Gen Psychiatry 46, 335-341. Swedo, S.E., Schapiro, M.B., Grady, C.L., Cheslow, D.L., Leonard, H.L., Kumar, A., Friedland, R., Rapoport, S.I. and Rapoport, J.L. (1989b)Cerebral glucose metabolism in childhood-onset ob~ssive-compulsive disorder. Arch Gen Psychiatry 46, 518-523. Tranel, D. and Damasio, H. (1994) Neuroanatomical correlates of electrodermal skin conductance responses. Psychophysiology 31,427-438. Zahn, T.P. (1986) Psychophysiological approaches to psychopathology. In: Coles, M.G.H., Donchin, E. and purges, S.W. (Eds.), Psychophysiology: Systems, Processes, and Applications. Guilford Press, New York, pp. 508-610. Zahn, T.P., Flament, M., Leonard, H.L., Swedo, S.E. and Rapoport, J.L. (1995) Autonomic predictors of clinical course in children and adolescents with obsessive-compulsive disorder. Manuscript in preparation. Zahn, T.P., Insel, T.R. and Murphy, D.L. (1984) Psychophysiological changes during pharmacological treatment of patients with obsessive-compulsive disorder. Br J Psychiatry 145, 39-44.