Professional Documents
Culture Documents
The Impact of Experiential Avoidance and
The Impact of Experiential Avoidance and
The Impact of Experiential Avoidance and
a r t i c l e i n f o a b s t r a c t
Article history: The present study sought to replicate and extend a study by Abramowitz et al. (2009), who examined how
Received 27 October 2009 well experiential avoidance (EA) and obsessive beliefs predicted obsessive-compulsive (OC) symptoms
Received in revised form 5 May 2010 in a non-clinical sample. The current study utilized a severe, clinical, treatment-seeking sample (N = 108),
Accepted 5 May 2010
and examined how well EA and obsessive beliefs predicted changes in OC symptoms from pre- to post-
treatment. Findings were generally consistent with Abramowitz et al. EA was generally not related to
Keywords:
OC severity and did not add significantly to the prediction of OC symptom domains above and beyond
Obsessive-compulsive disorder
depression or general anxiety, whereas obsessive beliefs did. Pre- to post-treatment change in one type of
Experiential avoidance
Obsessive beliefs
obsessive belief (perfectionism/certainty), but not change in EA, predicted global change in OC severity.
Results suggest that EA as it is measured currently may not play a significant role in OC severity or changes
in OC severity across treatment.
© 2010 Elsevier Ltd. All rights reserved.
Experiential avoidance (EA) is defined as an unwillingness to lead to symptom improvements within an anxiety-disordered sam-
experience or remain in contact with unpleasant private experi- ple (Chawla & Ostafin, 2007).
ences (e.g., emotions, thoughts, images, etc.) through attempts to It has recently been suggested that ACT interventions target-
escape from or avoid these experiences (Hayes, Wilson, Gifford, ing EA may be a preferable alternative to exposure and response
Follette, & Strosahl, 1996). EA is a primary target of acceptance and prevention (ERP), the traditional cognitive-behavioral treatment
commitment therapy (ACT; Hayes, Strosahl, & Wilson, 1999) and method utilized in the treatment of obsessive-compulsive dis-
has received considerable research attention due to the prolifera- order (OCD; Eifert & Forsyth, 2005; Twohig, Hayes, & Masuda,
tion of ACT research in the past decade. EA is associated with higher 2006). OCD is characterized by obsessions (i.e., intrusive, unrea-
levels of general psychopathology, depression, anxiety, trauma- sonable thoughts, images, fear, or worry) and compulsions or
related symptoms, and lower quality of life (Hayes et al., 2004). rituals used to reduce the anxiety associated with the obsessions
There is evidence that decreases in EA are associated with improve- (American Psychiatric Association, 2000). Eifert and Forsyth (2005)
ments in psychopathology and that focusing treatment on EA may have suggested that compulsions function as attempts at EA, in that
individuals with OCD engage in compulsions in order to “. . .control
or reduce their unwanted thoughts because they want to reduce
the negative affect associated with them (p. 58).” In support of this
∗ Corresponding author at: Department of Psychology, University of Wisconsin- view, an eight-session ACT intervention devoid of within-session
Milwaukee, PO Box 413, Milwaukee, WI 53201, USA. Tel.: +1 414 229 5078; exposure in a sample of four participants led to reductions in EA
fax: +1 414 229 5219. and reductions in OCD symptoms from pre- to post-treatment, but
E-mail addresses: rcmanos@uwm.edu (R.C. Manos), cahill@uwm.edu
this relationship could not be explored in depth due to the nature
(S.P. Cahill), wetterneck@uhcl.edu (C.T. Wetterneck), cconelea@uwm.edu
(C.A. Conelea), ARoss@rogershospital.org (A.R. Ross), BRiemann@rogershospital.org of the study design (Twohig et al., 2006). Although treatment did
(B.C. Riemann). not explicitly involve therapist aided exposure techniques, a com-
1
Tel.: +1 414 229 5099. ponent of treatment involved having patients make “behavioral
2
Tel.: +1 281 283 3364.
3
commitments” to engage in values-guided behavior that seemed
Tel.: +1 414 229 6078.
4
Tel.: +1 262 646 1159.
to have led to ERP-like activities in some instances. For example,
5
Tel.: +1 262 646 1388. Twohig et al. (2006) list examples of behavioral commitments,
0887-6185/$ – see front matter © 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.janxdis.2010.05.001
R.C. Manos et al. / Journal of Anxiety Disorders 24 (2010) 700–708 701
Y-BOCS-SR has acceptable internal consistency (˛ = .78) for OCD The Beck Anxiety Inventory (BAI; Beck, Epstein, Brown, & Steer,
samples and adequate test-retest reliability over a one-week inter- 1988) is a self-report measure developed to distinguish cogni-
val (r = .79) in a non-clinical sample (Steketee, Frost, & Bogart, 1996). tive and somatic symptoms of anxiety from cognitive and somatic
Internal consistency for the current sample was unavailable, as only symptoms of depression. Participants are instructed to rate how
the subscale and total scores for the Y-BOCS-SR were included in much they have been bothered by a variety of anxiety symptoms
the database. during the previous month. The BAI has 21 items, and each item is
The Obsessive-Compulsive Inventory – Revised (OCI-R; Foa et rated from 0 (“not at all”) to 3 (“severely – it bothered me a lot”;
al., 2002) is an 18-item measure of distress associated with obses- range of scores = 0–63), with higher scores indicating greater anx-
sions and compulsions. Total scores range from 0 to 72 with high iety. The BAI has demonstrated high internal consistency (˛s = .92
scores indicating greater distress. The OCI-R has six subscales: and .94) and test-retest reliability over a 1-week interval (rs = .67
washing (WA; concern with contamination), checking (CH; check- and .75; Beck et al., 1988; Fydrich, Dowdall, & Chambless, 1992). The
ing locks, etc., checking if one has made a mistake), ordering (OR; BAI has been shown to be correlated with the modified Hamilton
ordering and arranging objects), obsessing (OB; intrusive violent, Rating Scale of Anxiety (r = .51, Beck et al., 1988) and the Trait and
sexual, blasphemous thoughts), hoarding (HO; collecting and sav- State scales from the State-Trait Anxiety Inventory-Form Y (rs = .58
ing objects), and neutralizing (NE; counting, mental counting). and .47, respectively; Fydrich et al., 1992). Internal consistency for
Subscale scores range from 0 to 12. For the current study, items the current sample was unavailable, as only the total score for the
from the original Obsessive-Compulsive Inventory (OCI; Foa, Kozak, BAI was included in the database.
Salkovskis, Coles, & Amir, 1998) distress scale were used to create
subscale scores of the OCI-R by selecting relevant items. All sub- 1.3. Analytic plan
scales demonstrated acceptable internal consistency in the current
sample (WA ˛ = .89, CH ˛ = .82, OR ˛ = .93, OB ˛ = .82, HO ˛ = .82, NE Analyses were conducted first with admission data, and second
˛ = .81). by examining change from admission to discharge in variables of
interest and how these variables related. In both cases, interactions
1.2.2. Predictor variables between variables of interest and treatment program (residen-
Two predictor variables were evaluated: OCD specific beliefs tial or intensive outpatient) were examined in order to determine
and EA. The Obsessive Beliefs Questionnaire – 44 (OBQ44; OCCWG, whether data from these two programs could be collapsed or
2005) is a 44-item self-report measure of obsessive belief domains needed to be examined separately.
associated with OCD: responsibility/threat estimation (referred For the admission data, we started by examining Pearson’s cor-
to as OBQ44-RT; “I often think things around me are unsafe”), relations. For significant correlations, multiple regression analyses
perfectionism/certainty (OBQ44-PC; “If I’m not absolutely sure of were conducted. Depression scores (BDI-II) were entered in Step
something, I’m bound to make a mistake”), and importance/control 1, EA scores (AAQ) were entered in Step 2, and obsessive belief
of thoughts (OBQ44-ICT; “For me, having bad urges is as bad scores (OBQ44) were entered in Step 3, with the Y-BOCS-SR and
as actually carrying them out”). Tolin, Worhunsky, and Maltby the six OCI-R subscale scores as separate dependent variables. If
(2006) reported good internal consistencies for the three subscales the AAQ significantly added to the prediction of the dependent
(OBQ44-PC ˛ = .93, OBQ44-ICT ˛ = .90, OBQ44-RT ˛ = .89). All three variable, regression analyses were repeated with Steps 2 and 3
subscales demonstrated good internal consistency in the current reversed in order to determine if results changed. Pending no sig-
sample (OBQ44-PC ˛ = .86, OBQ44-ICT ˛ = .82, OBQ44-RT ˛ = .87). nificant changes, only the first set of regression analyses (with the
The Acceptance and Action Questionnaire (AAQ; Hayes et al., AAQ in Step 2 and the OBQ44 in Step 3) was included. In addi-
2004) is a 9-item scale designed to measure EA. Items are rated tion to these analyses, we extended Abramowitz et al.’s (2009)
on a 7-point scale with total scores ranging from 9 to 63. Higher analyses by completing an additional set of regressions in which
scores on the AAQ indicate greater levels of EA. The AAQ has been general anxiety (BAI) was entered in Step 1 rather than depres-
found to possess adequate internal consistency (˛ = .70; Hayes et sion scores. In order to examine relationships among changes in
al., 2004); however, in the current sample, internal consistency variables of interest from admission to discharge, residual gain
at admission was lower (˛ = .58). Previous research on the AAQ in scores were calculated, which corrects for problems in using raw
OCD samples has also found lower than adequate internal consis- scores (see Steketee & Chambless, 1992). Next, Pearson’s corre-
tency (Sylvester et al., 2007). Hayes et al. (2004) found that higher lations in residual gain scores for all variables of interest were
scores on the AAQ significantly correlated with higher levels of gen- examined. For variables found to be significantly correlated, mul-
eral psychopathology, depression, anxiety, trauma, various specific tiple regression analyses were completed in the same manner
fears (e.g., agoraphobia, blood/injury phobia) and lower quality of as described above, with the difference being that in all cases
life, and found a test-retest reliability of r = .64 over a four-month residual gain scores were used rather than scores from admis-
period. sion.
Table 1
Descriptive statistics at admissions and comparison by treatment type.
Demographic variables
Gender (female) 59 (54.6%) 47 (53.4%) 12 (60.0%) Fisher’s Exact, ns
Age 32.10 (12.44) 32.31 (12.55) 31.20 (12.20) t(106) = −.358, ns
Measures of OCD severity
Admit Y-BOCS-SR 26.67 (6.66) 27.46 (6.37) 23.25 (6.96) t(105) = −2.620, .010
Admit OCI-R Total 26.23 (13.41) 27.02 (13.73) 22.75 (11.60) t(106) = −1.290, ns
Admit OCI-R WA 4.58 (4.34) 4.70 (4.40) 4.05 (4.11) t(106) = −.602, ns
Admit OCI-R CH 3.98 (3.32) 4.08 (3.37) 3.55 (3.12) t(106) = −.643, ns
Admit OCI-R OR 4.58 (3.86) 4.67 (3.96) 4.20 (3.47) t(106) = −.490, ns
Admit OCI-R OB 6.91 (3.71) 7.17 (3.67) 5.75 (3.73) t(106) = −1.557, ns
Admit OCI-R HO 3.02 (3.04) 3.15 (3.09) 2.45 (2.84) t(106) = −.993, ns
Admit OCI-R NE 3.16 (3.58) 3.25 (3.62) 2.75 (3.49) t(106) = −.561, ns
Predictor variables
Admit OBQ44-RT 62.94 (18.63) 63.42 (19.43) 60.90 (14.91) t(104) = −.543, ns
Admit OBQ44-PC 64.63 (18.11) 65.13 (18.62) 62.42 (15.90) t(102) = −.587, ns
Admit OBQ44-ICT 47.44 (13.60) 47.53 (14.00) 47.05 (12.07) t(102) = −.141, ns
Admit AAQ 40.55 (5.78) 40.86 (6.09) 39.20 (4.01) t(106) = −1.159, .ns
Control variables
Admit BDI-II 27.75 (11.22) 28.76 (10.98) 23.38 (11.52) t(105) = −1.960, ns
Admit BAI 22.94 (11.94) 23.49 (11.86) 20.50 (12.27) t(106) = −1.011, ns
Note: Y-BOCS-SR, Yale-Brown Obsessive-Compulsive Scale Self Report; OCI-R, Obsessive-Compulsive Inventory – Revised; WA, washing; CH, checking; OR, ordering; OB,
obsessing; NE, neutralizing; OBQ44, Obsessive Beliefs Questionnaire 44; RT, responsibility/threat estimation; PC, perfectionism/certainty; ICT, importance/control of thoughts;
AAQ, Acceptance and Action Questionnaire; BDI-II, Beck Depression Inventory – II; BAI, Beck Anxiety Inventory.
significant difference was found in Y-BOCS-SR, with residential par- AAQ (r values = .343 and .236, for the BDI-II and BAI respectively)
ticipants scoring significantly higher (more severe; see Table 1). and all subscales of the OBQ (r values ranging between .320 and
In order to determine whether it was appropriate to collapse res- .460).
idential and intensive outpatient data, treatment program was
dummy coded and separate multiple regression analyses were 2.1.3. Multiple regression analyses
performed for all predictor and control variables. With Y-BOCS- Results for the regression analyses when the BDI-II was entered
SR severity at admission as the dependent variable, regressions first, the AAQ second, and the OBQ44 subscales third are presented
were completed with the treatment program entered in Step 1, in the center columns of Table 3. These analyses revealed that
the predictor or control variable entered in Step 2, and the treat- neither the AAQ nor any of the OBQ44 subscales remained sig-
ment program × predictor variable interaction entered in Step 3. nificant predictors of the Y-BOCS-SR beyond BDI-II scores. With
The interaction term in all regressions was not significant, indicat- respect to OCI-R subscales, the analyses revealed that the AAQ
ing that level of treatment did not moderate correlations between did not predict any subscales beyond the BDI-II. The OBQ44-RT
predictor or control variables and Y-BOCS-SR severity. Therefore, subscale remained significantly associated with the WA subscale,
data across treatment programs were collapsed for all subsequent the OBQ44-PC subscale remained significantly associated with the
analyses examining admissions data. OB subscale, and the OBQ44-ICT subscale remained significantly
associated with the OR and HO subscales. In addition, the OBQ44-
2.1.2. Pearson’s correlations RT subscale was found to be a significant negative predictor of
Intercorrelations among all of the questionnaires are presented the HO subscale after first controlling for the BDI-II and the AAQ,
in Table 2. Of note, there were significant moderate correlations and simultaneously controlling for the OBQ44-ICT and PC sub-
between the two measures of OCD severity, with correlations scales.
between the Y-BOCS-SR and OCI-R subscales ranging between Tolerance for all regressions ranged from .152 to .849, with num-
r = .207 and .352. In addition, there was a significant moderate cor- bers approaching zero indicating a problem with multicollinearity.
relation between the AAQ and OBQ44-ICT subscale (r = .257), with As the reversal of the direction of the relationship between the
correlations between the AAQ and OBQ44-RT and PC subscales not OBQ44-RT and the OCI-R HO subscale was unexpected, additional
attaining significance (rs = .188 and .189, respectively). All subscales multiple regression analyses were conducted to better understand
of the OBQ44 had significant correlations with the Y-BOCS-SR (r when the reversal occurred. When the BDI-II was entered at Step 1,
values ranging between .202 and .278), and all but five correla- the AAQ entered at Step 2, and the OBQ44-RT subscale alone at Step
tions between the OBQ44 subscales and the OCI-R subscales were 3, the result was a significant positive relationship for the OBQ44-RT
significant (significant r values ranging between .212 and .416). (ˇ = .230, t = 2.119, p = .037). However, a significant negative rela-
Specifically, the OCI-R OR subscale was not correlated with the tionship was obtained when the OBQ44-RT subscale was entered
RT and PC subscales of the OBQ44 and the OCI-R NE subscale was simultaneously with the OBQ44-ICT subscale (ˇ = −.427, t = −2.121,
not significantly correlated with any of the OBQ44 subscales. The p = .036) and a non-significant negative relationship was obtained
AAQ did not significantly correlate with the Y-BOCS-SR (r = .159) when the OBQ44-RT subscale was entered simultaneously with the
or any of the OCI-R subscales (r values ranging between −.051 and OBQ44-PC subscale (ˇ = −.148, t = −.719, p = .474).
.113). Both control variables were significantly correlated with the Results for the regression analyses when the BAI was entered
Y-BOCS-SR (r values of .401 and .387 for the BDI-II and BAI). The first instead of the BDI-II (see right-hand columns of Table 3)
BDI-II was also significantly correlated with three of the six OCI- followed by the AAQ revealed the AAQ was not a significant
R subscales (CH, OR, and OB; r values ranging between .172 and predictor of the YBOCS-SR or any OCI-R subscale. By contrast,
.373); the BAI was significantly correlated five of the OCI-R sub- the OBQ44-ICT remained a significant predictor of the Y-BOCS-
scales (all but HO; significant r values ranging between .250 and SR and the CH, OR, and HO subscales of the OCI-R. In addition,
.412). Both control variables were significantly correlated with the the OBQ44-RT subscale was found to be a significant nega-
704 R.C. Manos et al. / Journal of Anxiety Disorders 24 (2010) 700–708
Table 2
Correlations at admission.
OCI-R WA OCI-R CH OCI-R OR OCI-R OB OCI-R HO OCI-R NE OBQ 44-RT OBQ 44-PC OBQ 44-ICT AAQ BDI-II BAI
** ** ** ** * * * * ** **
Y-BOCS-SR .352 .329 .258 .268 .207 .243 .220 .202 .278 .159 .401 .387**
OCI-R WA 1.0 .319** .060 .229* −.055 .146 .321** .212* .265** −.051 .153 .250**
OCI-R CH 1.0 .487** .151 .363** .308** .391** .400** .413** .113 .312** .389**
OCI-R OR 1.0 .105 .552** .540** .154 .121 .296** .039 .172* .270**
OCI-R OB 1.0 .011 .266** .375** .416** .304** .034 .373** .412**
OCI-R HO 1.0 .394** .236* .319** .358** .055 .130 .151
OCI-R NE 1.0 .082 .184 .190 −.036 .084 .259**
OBQ44-RT 1.0 .875** .885** .188 .460** .397**
OBQ44-PC 1.0 .840** .189 .431** 356**
OBQ44-ICT 1.0 .257** .423** 320**
AAQ 1.0 .343** .236*
BDI-II 1.0 .588**
BAI 1.0
Y-BOCS-SR, Yale-Brown Obsessive-Compulsive Scale Self Report; OCI-R, Obsessive-Compulsive Inventory – Revised; WA, washing; CH, checking; OR, ordering; OB, obsessing;
NE, neutralizing; OBQ44, Obsessive Beliefs Questionnaire 44; RT, responsibility/threat estimation; PC, perfectionism/certainty; ICT, importance/control of thoughts; AAQ,
Acceptance and Action Questionnaire; BDI-II, Beck Depression Inventory – II; BAI, Beck Anxiety Inventory.
*
p < .05.
**
p < .01.
tive predictor of the HO subscale. Tolerance for all regressions for the OBQ44-RT (ˇ = .206, t = 1.960, p = .053). However, a sig-
ranged from .150 to .888. Again, as reversal of the direction nificant negative relationship was obtained when the OBQ44-RT
of the relationship between the OBQ44-RT and the OCI-R HO subscale was entered simultaneously with the OBQ44-ICT subscale
subscale was unexpected, additional multiple regression analy- (ˇ = −.480, t = −2.357, p = .020) and a non-significant negative rela-
ses were conducted. When the BAI was entered at Step 1, the tionship was obtained when the OBQ44-RT subscale was entered
AAQ entered at Step 2, and the OBQ44-RT subscale entered alone simultaneously with the OBQ44-PC subscale (ˇ = −.202, t = −1.012,
at Step 3, the result was a non-significant positive relationship p = .315).
Table 3
Summary of the final step of significant regression equations predicting the Y-BOCS-SR and OCI-R subscale scores (BDI-II or BAI entered in Step 1, AAQ in Step 2, and OBQ44
subscales in Step 3).
R 2
ˇ t p R2 ˇ t p
Note: ˇ, standardized regression coefficient; Y-BOCS-SR, Yale-Brown Obsessive-Compulsive Scale Self Report; OCI-R, Obsessive-Compulsive Inventory – Revised; WA, washing;
CH, checking; OR, ordering; OB, obsessing; NE, neutralizing; OBQ44, Obsessive Beliefs Questionnaire 44; RT, responsibility/threat estimation; PC, perfectionism/certainty;
ICT, importance/control of thoughts; AAQ, Acceptance and Action Questionnaire; BDI-II, Beck Depression Inventory – II; BAI, Beck Anxiety Inventory.
R.C. Manos et al. / Journal of Anxiety Disorders 24 (2010) 700–708 705
Table 4
Variables of interest from admissions to discharge.
Measure Admissions Discharge Change (full Test of significance Change Change Test of significance
sample) (admissions to discharge) (residential) (partial) (residential compared to
partial)
M (SD) M (SD) M (SD) t(df), p M (SD) M (SD) t(df), p
Y-BOCS-SR 26.67 (6.66) 15.24 (7.03) 11.43 (7.30) 16.195 (106), <.001 11.66 (6.92) 10.45 (8.90) −.664 (105), ns
OCI-R WA 4.58 (4.34) 2.04 (2.82) 2.54 (3.38) 7.814 (107), <.001 2.47 (3.29) 2.85 (3.84) .450 (106), ns
OCI-R CH 3.98 (3.32) 2.17 (2.22) 1.81 (2.64) 7.151 (107), <.001 1.72 (2.63) 2.25 (2.69) .816 (106), ns
OCI-R OR 4.58 (3.86) 2.79 (3.15) 1.80 (2.74) 6.804 (107), <.001 1.78 (2.85) 1.85 (2.25) .097 (106), ns
OCI-R OB 6.91 (3.71) 4.38 (3.15) 2.53 (2.84) 9.250 (107), <.001 2.39 (2.85) 3.15 (2.80) 1.086 (106), ns
OCI-R HO 3.02 (3.04) 1.85 (2.46) 1.17 (2.31) 5.278 (107), <.001 1.20 (2.34) 1.05 (2.19) −.259 (106), ns
OCI-R NE 3.16 (3.58) 1.45 (2.40) 1.70 (2.82) 6.287 (107), <.001 1.63 (2.77) 2.05 (3.07) .607 (106), ns
OBQ44-RT 62.94 (18.63) 48.11 (18.24) 14.81 (16.13) 9.457 (105), <.001 14.15 (15.57) 17.65 (18.51) .873 (104), ns
OBQ44-PC 64.63 (18.11) 50.50 (18.36) 14.06 (16.39) 8.749 (103), <.001 13.55 (15.94) 16.32 (18.55) .663(102), ns
OBQ44-ICT 47.44 (13.60) 36.69 (13.77) 10.61 (10.82) 10.001 (103), <.001 10.26 (10.52) 12.10 (12.17) .683 (102), ns
AAQ 40.55 (5.78) 37.60 (5.36) 2.95 (6.03) 5.080 (107), <.001 3.09 (6.04) 2.35 (6.11) −.490 (106), ns
BDI 27.75 (11.22) 11.33 (10.38) 16.45 (12.01) 14.166 (106), <.001 17.16 (11.57) 13.38 (13.67) −1.274 (105), ns
BAI 22.94 (11.94) 13.53 (9.99) 9.40 (11.28) 8.060 (107), <.001 9.38 (10.89) 9.50 (13.21) .042 (106), ns
Note: Y-BOCS-SR, Yale-Brown Obsessive-Compulsive Scale Self Report; OCI-R, Obsessive-Compulsive Inventory – Revised; WA: washing; CH: checking; OR: ordering;
OB: obsessing; NE: neutralizing; OBQ44: Obsessive Beliefs Questionnaire 44; RT: responsibility/threat estimation; PC: perfectionism/certainty; ICT: importance/control
of thoughts; AAQ: Acceptance and Action Questionnaire; BDI-II: Beck Depression Inventory – II; BAI: Beck Anxiety Inventory.
2.2. Change over the course of treatment data were collapsed across treatment programs for all subsequent
analyses.
2.2.1. Descriptives
See Table 4 for descriptive data on variables of interest as well as
results of t-tests for correlated samples for the change from admis- 2.2.3. The relationship of change in predictor variables and
sion to discharge. All variables examined were significantly lower outcome measures
(i.e., less severe) at discharge than at admission. Furthermore, t- Pearson correlations were calculated between the residualized
tests for independent samples comparing admissions to discharge gain scores for all variables of interest (see Table 5). Changes in
change scores across the two treatment programs found no signif- obsessive beliefs, as measured by the OBQ44 subscales, were sig-
icant differences in the amount of improvement for participants in nificantly correlated with changes in OC symptoms, as measured by
the residential program compared to intensive outpatient. the Y-BOCS-SR and OCI-R subscales (r values ranging between .380
and .609, median r value = .480, all p values < .001). Change in EA
was significantly correlated with change in one facet of OC sever-
2.2.2. Differences by treatment program ity, as indicated by OCI-R OR subscale (r = .227, p = .018). Change in
To examine the hypothesized relationship between changes in EA was not significantly correlated with change in OC severity on
predictor variables (i.e., EA, obsessive beliefs) and changes in OC the Y-BOCS-SR or any other OCI-R subscale.
severity, we calculated residualized gain scores for the Y-BOCS- Admissions to discharge residualized gain scores were used in
SR and AAQ total scores and the OCI-R and OBQ44 subscales in multiple regression analyses in which changes in Y-BOCS-SR and
order to account for shortcomings of using raw change scores (see OCI-R subscale scores were dependent variables, changes in BDI
Steketee & Chambless, 1992). In order to determine whether it was or BAI were entered in Step 1, changes in EA were entered in
again appropriate to collapse data from the residential and inten- Step 2, and changes in obsessive beliefs (OBQ44-RT, OBQ44-ICT,
sive outpatient programs, treatment program was dummy coded and OBQ44-PC) were entered in Step 3 (see Table 6). As the AAQ
and multiple regressions were performed for all predictor change was never significant when entered in Step 2, additional analyses
variables (BDI-II, BAI, AAQ, and OBQ44 subscales), wherein the with the OBQ subscales entered at Step 2 and the AAQ entered at
treatment program was entered in Step 1, the residual gain score Step 3 were not conducted. All seven regressions were significant.
of the predictor variable was entered in Step 2, and the treatment Change on the OBQ44-PC was a significant predictor of change on
program by predictor change variable (residual gain score) inter- the Y-BOCS-SR and OCI-R WA subscale when the BDI-II was entered
action was entered in Step 3. For all cases, the interaction term was at Step 1, but not when the BAI was entered at Step 1, whereas
not significant, indicating that treatment level did not moderate change on the OBQ44-PC was a significant predictor of OCI-R NE
correlations between predictor and outcome variables. Therefore, when the BAI was entered at Step 1 but not when the BDI-II was
Table 5
Pearson correlations between admissions to discharge standardized residual predictor and outcome variables.
Note: Y-BOCS-SR: Yale-Brown Obsessive-Compulsive Scale Self Report; OCI-R: Obsessive-Compulsive Inventory – Revised; WA: washing; CH: checking; OR: ordering;
OB: obsessing; NE: neutralizing; OBQ44: Obsessive Beliefs Questionnaire 44; RT: responsibility/threat estimation; PC: perfectionism/certainty; ICT: importance/control of
thoughts; AAQ: Acceptance and Action Questionnaire.
*
p < .05.
**
p < .001.
706 R.C. Manos et al. / Journal of Anxiety Disorders 24 (2010) 700–708
Table 6
Summary of the final step of significant regression equations predicting change in Y-BOCS-SR and OCI-R subscale scores from admission to discharge standardized residual
gain scores (BDI-II or BAI entered in Step 1, AAQ in Step 2, and OBQ44 in Step 3).
R2 ˇ t p R2 ˇ t p
Note: ˇ, standardized regression coefficient; Y-BOCS-SR, Yale-Brown Obsessive-Compulsive Scale Self Report; OCI-R, Obsessive-Compulsive Inventory – Revised; WA, washing;
CH, checking; OR, ordering; OB, obsessing; NE, neutralizing; OBQ44, Obsessive Beliefs Questionnaire 44; RT, responsibility/threat estimation; PC, perfectionism/certainty;
ICT, importance/control of thoughts; AAQ, Acceptance and Action Questionnaire; BDI-II, Beck Depression Inventory – II; BAI, Beck Anxiety Inventory.
entered at Step 1. The OBQ44-RT and OBQ44-ICT subscales did not and perfectionism/certainty predicted CH. By contrast, compara-
significantly predict any of the OCI-R subscales or the Y-BOCS-SR ble analyses in our study found that when first controlling for
regardless of whether the BDI-II or BAI was entered at Step 1. Tol- depression and EA, perfectionism/certainty predicted OB and nei-
erance for all regressions ranged from .198 to .917. ther importance/control of thoughts nor perfectionism/certainty
predicted CH. We also found several associations between obses-
3. Discussion sive beliefs and OCD domains after first controlling for depression
and EA that were not found in the Abramowitz study. Specifically,
3.1. Admission data importance/control of thoughts predicted OR and HO and respon-
sibility/threat estimation predicted WA, OR, and HO. Notably in
Our results indicated that EA had limited association with mea- the case of HO, although the zero-order relationship was positive,
sures of OCD severity and did not add significantly to prediction the relationship became negative in multiple regression analyses in
of OC symptom dimensions above and beyond obsessive beliefs which depression or anxiety, EA, and at least one additional OBQ44
and depression. By contrast, we found that moderate associations subscale were included. Our results also differ from findings by
between a range of obsessive beliefs and OCD severity, and cer- Myers, Fisher, and Wells (2008), who reported that after controlling
tain obsessive beliefs continued to predict specific OCD domains for worry, responsibility/threat estimation significantly predicted
above and beyond EA and depression. These findings broadly repli- all OCD domains on the original OCI (Foa et al., 1998), perfection-
cate and extend findings of Abramowitz et al. (2009). Interestingly, ism/certainty significantly predicted WA, Doubting, and OR, and
the specific types of obsessive beliefs that predicted specific symp- importance/control of thoughts significantly predicted CH, OB, and
tom dimensions differed from Abramowitz et al.’s (2009) findings. NE in a student sample.
For example, Abramowitz et al. found that, after controlling for It is not clear why these differences exist, although it may be
depression and EA, importance/control of thoughts predicted OB that relationships among these variables differ for severe, clini-
R.C. Manos et al. / Journal of Anxiety Disorders 24 (2010) 700–708 707
cal samples, as in the current study, compared to student samples paring mechanisms of change in CBT and ACT is needed to examine
as in Abramowitz et al. (2009) and Myers et al. (2008), although whether change in obsessive beliefs will continue to have a sig-
this would not account for differences between the Abramowitz nificant impact on change in OC symptoms in other treatment
and Myers studies. Further research is needed to examine why the modalities such as ACT, or if other treatment-specific mechanisms
pattern, and not just strength, of these relationships differs among (e.g., change in EA) will play a significant role.
these investigations. It is also important to note that measures of depression (BDI-
We also investigated these relationships when controlling for II) and general anxiety (BAI) performed as well, and in some cases
general anxiety. For the most part, the pattern remained the same better, than the AAQ and OBQ44 in several analyses. In the exam-
as when we controlled for depression with respect to which obses- ination of admission scores, the BDI-II was significantly correlated
sive belief domains predicted OC severity. The three exceptions with half of the OCI-R subscales, and the BAI was significantly cor-
were that importance/control of thoughts significantly predicted related with almost all OCI-R subscales. Changes in BDI-II and BAI
OC severity on the Y-BOCS-SR and OCI-R CH, and that perfection- scores were significantly correlated with changes in the Y-BOCS-SR
ism/certainty did not significantly predict OCI-R NE. and changes in all OCI-R subscales. One reason that these mea-
sures may have “outperformed” the AAQ and OBQ44 scores may
3.2. Change data be that depression and general anxiety are more relevant to OCD
symptoms and severity than the theoretical constructs of obses-
We also investigated whether changes in obsessive beliefs and sive beliefs and EA. On the other hand, it might be that obsessive
EA were associated with changes in OC severity from admission to beliefs and/or EA are quite relevant but that their correspond-
discharge. Change in all three categories of obsessive beliefs was ing measures are inadequate assessment tools. Indeed, the AAQ
associated with change in OCD severity on the Y-BOCS-SR and all has been criticized for measurement problems, such as low inter-
OCI-R subscales. Change in EA was associated with change on only nal consistency, which was a concern in the current study. It has
the OCI-R OR subscale. However, our multiple regression analy- also been suggested that the AAQ may be too general to function
ses revealed that change in EA did not continue to predict OCI-R OR as an appropriate measure of disorder-specific EA (Hayes et al.,
scores above changes in anxiety or depression, whereas perfection- 2004). To date, numerous versions of the AAQ have been devel-
ism/certainty remained a significant predictor of total Y-BOCS-SR oped targeting specific conditions, such as smoking (Gifford et al.,
scores and OCI-R WA when the BDI-II was the control variable 2002) and chronic pain (Westin, Hayes, & Andersson, 2008). To
and perfectionism/uncertainty predicted OCI-R NE scores when help clarify the relationship between OCD and EA, further research
the BAI was the control variable. None of the other OBQ44 sub- is needed using newer versions of the AAQ with better psycho-
scales (i.e., importance/control of thoughts or responsibility/threat metric properties, such as the AAQ-II (Bond et al., submitted for
estimation) significantly predicted changes in OCD severity. publication), or to develop a more specific measure of EA in OCD.
Overall, our pattern of results supports the notion that EA, at For the latter suggestion, the MetaCognitions Questionnaire, which
least as measured by the AAQ, may not play a significant role in assesses processes from a theory similar to ACT and also useful for
OC severity in a severe, clinical sample. Alternatively, it may be the treating OCD (i.e., Metacognitive Therapy; cf. Fisher, 2009), may
case that changes in OC severity over treatment were not reflected serve as an example of how to design a measurement tool that
by changes in EA because treatment did not directly target EA. matches both the underlying theory and the specific aspects of the
The treatment used, ERP plus thought challenging and other cogni- disorder in question (cf. Solem, Haland, Vogel, Hansen, & Wells,
tive techniques, is thought to produce change through mechanisms 2009).
that may be distinct from EA (e.g., habituation to anxiety, cognitive An important limitation of our results is that, although they are
change). From the ACT perspective, cognitive change strategies may broadly consistent with the cognitive theory of OCD, all of our mod-
be viewed as further attempts to control and modify OC symptoms, els accounted for less than 50% of the variance in outcome, and most
which conflicts with the notion of accepting thoughts as thoughts accounted for substantially less. It is not clear to what extent this
(i.e., as part of “cognitive defusion”) while behaving in accor- reflects limitations in cognitive theory or limitations in the OBQ to
dance with one’s values. Had treatment been focused on processes measure the hypothesized beliefs underlying OCD. Thus research
emphasized in ACT, such as defusion, acceptance, mindfulness, self to further refine the OBQ or to test alternative models may help to
as context, contact with the present moment, and values-directed resolve this question.
behavioral commitments, then perhaps changes in EA may have
had a stronger relationship with changes in OC symptoms.
References
In contrast to acceptance theory, cognitive theory would predict
strong associations between specific obsessive beliefs and specific Abramowitz, J. S. (1997). Effectiveness of psychological and pharmacological treat-
OC symptom domains and contend that treatment brings about ments for obsessive – compulsive disorder: a quantitative review. Journal of
symptom change through changes in beliefs. Accordingly, changes Consulting and Clinical Psychology, 65, 44–52.
Abramowitz, J. S., Lackey, G. R., & Wheaton, M. G. (2009). Obsessive-compulsive
in obsessive beliefs should be strongly correlated with changes symptoms: the contribution of obsessional beliefs and experiential avoidance.
in OC symptoms. Broadly speaking, our results provide modest Journal of Anxiety Disorders, 23, 160–166.
support for the tenants of cognitive theory, in that: (1) obses- American Psychiatric Association. (2000). Diagnostic and statistical manual of men-
tal disorders IV-TR (4th ed.). Washington, DC: Author.
sive beliefs were correlated with OC symptoms at admission, (2) Baer, L., Brown-Beasley, M., Sorce, J., & Henriques, A. (1993). Computer-assisted
there was some evidence for specificity between specific belief telephone administration of a structured interview for obsessive-compulsive
domains and specific OC symptoms, and (3) change in obsessive disorder. American Journal of Psychiatry, 150, 1737–1738.
Beck, A. T., Epstein, N., Brown, G., & Steer, R. A. (1988). An inventory for measuring
beliefs was in some cases predictive of changes in OC symptoms.
anxiety: psychometric properties. Journal of Consulting and Clinical Psychology,
We noted in our introduction that, at least in some cases, treat- 56, 893–897.
ment with ACT may result in interventions that appear to overlap Beck, A. T., Steer, R. A., & Brown, G. K. (1996). Manual for the BDI-II. San Antonio, TX:
Psychological Corporation.
with ERP procedures. To the extent that ACT is efficacious in the
Björgvinsson, T., & Hart, J. (2007). Obsessive-compulsive disorder. In: M. Hersen, &
treatment of OCD, perhaps similar processes are occurring as those J. Rosqvist (Eds.), Handbook of assessment, conceptualization, and treatment. New
hypothesized by cognitive theory. In other words, rather than pro- York: John Wiley & Sons.
moting acceptance and values-based action, interventions based Bond, F. W., Hayes, S. C., Baer, R. A., Carpenter, K. M., Orcutt, H. K., Waltz, T., et al. (sub-
mitted for publication). Preliminary psychometric properties of the Acceptance
on ACT principles may actually promote symptom change through and Action Questionnaire – II: a revised measure of psychological flexibility and
habituation or cognitive changes. Further research directly com- acceptance.
708 R.C. Manos et al. / Journal of Anxiety Disorders 24 (2010) 700–708
Chawla, N., & Ostafin, B. (2007). Experiential avoidance as a functional dimensional Myers, S. G., Fisher, P. L., & Wells, A. (2008). Belief domains of the Obsessive Beliefs
approach to psychopathology: an empirical review. Journal of Clinical Psychology, Questionnaire-44 (OBQ-44) and their specific relationship with obsessive-
63, 871–890. compulsive symptoms. Journal of Anxiety Disorders, 22, 475–484.
Dozois, D. J., Dobson, K. S., & Ahnbert, J. L. (1998). A psychometric evaluation of the Obsessive Compulsive Cognitions Working Group. (2005). Psychometric validation
Beck Depression Inventory-II. Psychological Assessment, 10(10), 83–89. of the Obsessive Belief Questionnaire and Interpretation of Intrusions Inventory.
Eifert, G. H., & Forsyth, J. P. (2005). Acceptance and commitment therapy for anxiety Part 2. Factor analyses and testing of a brief version. Behaviour Research and
disorders: a practitioner’s guide to using mindfulness, acceptance, and values-based Therapy, 43, 1527–1543.
behavior change strategies. Oakland: New Harbinger. Rachman, S. (1997). A cognitive therapy of obsessions. Behaviour Research and Ther-
Fisher, P. L. (2009). Obsessive compulsive disorder: a comparison of the CBT and apy, 35, 793–802.
metacognitive approach. International Journal of Cognitive Therapy, 2, 107–122. Solem, S., Haland, A. T., Vogel, P. A., Hansen, B., & Wells, A. (2009). Change in metacog-
Foa, E. B., Huppert, J. D., Leiberg, S., Langner, R., Kichic, R., Hajcak, G., et al. (2002). nitions predicts outcome in obsessive-compulsive disorder patients undergoing
The Obsessive-Compulsive Inventory: development and validation of a short treatment with exposure and response prevention. Behavior Research and Ther-
version. Psychological Assessment, 14, 485–496. apy, 47, 301–307.
Foa, E. B., Kozak, M. J., Salkovskis, P. M., Coles, M. E., & Amir, N. (1998). The valida- Steer, R. A., Kumar, G., Ranieri, W. F., & Beck, A. T. (1998). Use of the Beck Depression
tion of a new obsessive-compulsive disorder scale: the obsessive-compulsive Inventory-II with adolescent psychiaric outpatients. Journal of Psychopathology
inventory. Psychological Assessment, 10, 206–214. and Behavioral Assessment, 20, 127–137.
Fydrich, T., Dowdall, D., & Chambless, D. L. (1992). Reliability and validity of the Beck Steketee, G., & Chambless, D. L. (1992). Methodological issues in prediction of treat-
Anxiety Inventory. Journal of Anxiety Disorders, 6, 55–61. ment outcome. Clinical Psychology Review, 12, 387–400.
Gifford, E. V., Antonuccio, D. O, Kohlenberg, B. S., Hayes, S. C., & Piasecki, M. M. (2002). Steketee, G., Frost, R., & Bogart, K. (1996). The Yale-Brown obsessive-compulsive
Combining Bupropion SR with acceptance and commitment-based behavioral ther- scale: interview versus self-report. Behavior Research and Therapy, 34, 675–684.
apy for smoking cessation: preliminary results from a randomized controlled trial. Sylvester, B. D., Entricht, T. L., Wetterneck, C. T., Hart, J. M., & Björgvinsson, T. (2007).
Paper presented at the annual meeting of the Association for Advancement of The relationships among acceptance, anxiety sensitivity, and depression in severe
Behavioral Therapy, Reno, NV. OCD? Poster presented at the 41st Annual Convention of the Association for
Goodman, W. K., Price, L. H., Rasmussen, S. A., Mazure, C., Fleischmann, R. L., Hill, C. Behavioral and Cognitive Therapies, Philadelphia, PA.
L., et al. (1989). The Yale-Brown Obsessive Compulsive Scale. Archives of General Tolin, D. F., Worhunsky, P., & Maltby, N. (2006). Are ‘obsessive’ beliefs specific to
Psychiatry, 46, 1006–1011. OCD?: a comparison across anxiety disorders. Behaviour Research and Therapy,
Hamilton, M. (1960). A rating scale for depression. Journal of Neurology, Neurosurgery 44, 469–480.
and Psychiatry, 23, 56–62. Twohig, M. P., Hayes, S. C., & Masuda, A. (2006). Increase willingness to experience
Hamilton, M. (1967). Development of a rating scale for primary depressive illness. obsessions: acceptance and commitment therapy as a treatment for obsessive
British Journal of Social and Clinical Psychology, 6, 278–296. – compulsive disorder. Behavior Therapy, 37, 3–13.
Hayes, S. C., Strosahl, K. D., & Wilson, K. G. (1999). Acceptance and commitment Westin, V., Hayes, S. C., & Andersson, G. (2008). Is it the sound or your relationship
therapy: an experiential approach to behavior change. New York: Guilford Press. to it? The role of acceptance in predicting tinnitus impact. Behaviour Research
Hayes, S. C., Strosahl, K. D., Wilson, K. G., Bissett, R. T., Pistorello, J., Toarmino, D., and Therapy, 46, 1259–1265.
et al. (2004). Measuring experiential avoidance: a preliminary test of a working Whittal, M. L., Robichaud, M., Thordarson, D. S., & McLean, P. D. (2008). Group and
model. The Psychological Record, 54, 553–578. individual treatment of obsessive-compulsive disorder using cognitive therapy
Hayes, S. C., Wilson, K. W., Gifford, E. V., Follette, V. M., & Strosahl, K. (1996). Expe- and exposure plus response prevention: a 2-year follow-up of two randomized
riential avoidance and behavioral disorders: a functional dimensional approach trials. Journal of Consulting and Clinical Psychology, 76, 1003–1014.
to diagnosis and treatment. Journal of Consulting and Clinical Psychology, 64,
1152–1168.