The Multidimensionality of Fear of Pain: Construct Independence for the Fear of Pain Questionnaire-Short Form and the Pain Anxiety Symptoms Scale-20

The Journal of Pain, Vol 10, No 1 (January), 2009: pp 29-37 Available online at www.sciencedirect.com

Original Reports The Multidimensionality of Fear of Pain: Construct Independence for the Fear of Pain Questionnaire-Short Form and the Pain Anxiety Symptoms Scale-20 R. Nicholas Carleton and Gordon J. G. Asmundson The Anxiety and Illness Behaviour Laboratory, University of Regina, Regina, Saskatchewan.

Abstract: Current fear-anxiety-avoidance models of chronic pain emphasize pain-related fear and anxiety as potential precursors for disabling chronic pain; however, anxiety and fear are often used interchangeably when discussing pain. Fear is a present-oriented emotive state associated with an imminent threat (eg, a patient about to receive an injection), whereas anxiety is a more general, future-oriented emotive state, that occurs in anticipation of threats without requiring an objective stimulus (eg, the possibility of receiving an injection). Theoretical and empirical evidence suggests pain-related fear and anxiety represent distinct cognitive constructs. Moreover, pain-related anxiety has been posited as a manifestation of anxiety sensitivity, which has implications for several theoretical models as well as treatment. The Fear of Pain Questionnaire and the Pain Anxiety Symptoms Scale-20 are popular measures, often used comparably, that were designed to measure pain-related fear and anxiety, respectively. These measures, along with the Anxiety Sensitivity Index, were administered to an undergraduate sample (N ⴝ 268; 66% women). Results of confirmatory factor analyses suggest each measure represents a related, but distinct, construct. Furthermore, correlations with anxiety sensitivity suggest that pain-related anxiety may be better conceptualized as a fundamental fear. Implications and directions for future research are discussed. Perspective: Fear-anxiety-avoidance models of chronic pain posit pain-related fear and anxiety as diatheses for disabling chronic pain. This research suggests theoretical and clinical distinctions between pain-related fear and anxiety. Moreover, pain-related anxiety appears more complex than a manifestation of anxiety sensitivity; pain-related anxiety may be better conceptualized as a fundamental fear. © 2009 by the American Pain Society Key words: Fear of pain, pain-related fear, pain-related anxiety, FPQ-SF, PASS-20.

C

urrent fear-anxiety-avoidance models of chronic pain7,47 emphasize pain-related fear as a key diathesis for disabling chronic pain. Pain-related anxiety facilitates a protective function; without some anxi-

Received December 4, 2007; Revised March 19, 2008; Accepted June 23, 2008. Supported by a Canadian Institute of Health Research Investigator’s (CIHR) Award (Dr. Asmundson) and by a CIHR Canada Graduate Scholarship Doctoral Research Award (R. N. Carleton). Address reprint requests to Dr. Gordon J. G. Asmundson, Anxiety and Illness Behaviours Laboratory, University of Regina, Regina, Saskatchewan, S4S 0A2. E-mail: [email protected] 1526-5900/$34.00 © 2009 by the American Pain Society doi:10.1016/j.jpain.2008.06.007

ety, nociceptive stimulation might fail to engage avoidance behaviors, thereby reducing the utility of pain. Typically, fear- and anxiety-related pain cognitions are collectively called fear of pain5; however, fear of pain is probably comprised of present- and future-oriented cognitions related to pain. The current amalgamation may be an important inaccuracy requiring delineation.31 The Fear of Pain Questionnaire-Short Form (FPQ-SF4) is a self-report inventory, derived from the Fear of Pain Questionnaire-III (FPQ-III34,50), that assesses situationally-specific fears of painful stimuli (eg, receiving an injection in your mouth).34,37,50 Conversely, the Pain Anxiety Symptoms Scale-Short Form (PASS-2032) is a short version of the Pain Anxiety Symptoms Scale (PASS33) that 29

30 measures latent, nonspecific pain-related anxiety (eg, I worry when I am in pain). Item endorsement on either scale is not necessarily indicative of pathology; instead, each construct is thought to occur along a continuum.3,7,34 Correlational analyses have found moderate relationships between the FPQ-SF and the PASS-20 and their respective longer versions.4,37 Together these measures should capture different dimensions of pain-related cognitions alluded to by previous researchers.19,20,29,31 Independently they should represent pain-related fear and anxiety, respectively, assessing related but distinct constructs (ie, present- vs future-oriented pain-related cognitions10,16). Precedent differences have been found between painrelated fear and anxiety using the FPQ and the PASS.31 Results of correlation and regression analyses using a sample of chronic pain patients (n ⫽ 45) indicated that the PASS correlated more with pain severity and disability, whereas the FPQ was correlated primarily with pain complaints. Regression analyses controlling for pain severity also demonstrated differing relationships between the FPQ, the PASS, and measures of pain-related disability. Moreover, relative to the FPQ, the PASS accounted for substantially more variance in pain severity, disability, and behavior. Pain-related anxiety has also been posited as a manifestation of anxiety sensitivity (ie, the tendency to fear anxiety sensations based on the belief they may have harmful consequences6,24). Treatments for reducing anxiety sensitivity (eg, interoceptive exposure) have been shown to reduce fear of pain48; however, preliminary evidence from clinical and nonclinical samples3,34 indicates pain-related anxiety may be continuous in nature (ie, occurring along a latent continuum ranging from low to high), whereas anxiety sensitivity appears taxonic (ie, having qualitatively distinct normative and pathological forms11,12). Such discrepancies suggest the constructs are related but distinct. The current study evaluated whether pain-related fear and anxiety, as measured by the FPQ-SF and PASS-20, represent independent constructs, related but distinct constructs, or the same construct. This study also assessed whether the FPQ-SF or the PASS-20 represent independent constructs relative to the Anxiety Sensitivity Index (ASI39), potentially contributing to current knowledge pertinent to the postulate that fear of pain is a manifestation of anxiety sensitivity.6,24 The measures are expected to be highly correlated, but the constructs best conceptualized as distinct.

Materials and Methods Participants and Procedure Participants for this investigation included a convenience sample of 268 undergraduate students [90 men, aged 18 –37 (M ⫽ 20.6; SD ⫽ 3.0) and 178 women, aged 18 – 45 (M ⫽ 20.5; SD ⫽ 3.7)] from the University of Regina who participated in a large questionnaire-based study approved by the University Research Ethics Board. Using this more general sample ensures an appropriate

Multidimensionality of Fear of Pain range of responses; conversely, a clinical sample would likely provide a restricted range of relatively higher responses.1 There was no statistically significant difference between men and women on age, t(264) ⫽ .38, P ⬎ .10. The majority of participants self-declared as Caucasian (88%), First Nations (3%), or Asian (3%), as well as being single (87%) or married (12%). Participants were recruited through brief in-class presentations and poster advertisements that directed them to a secure website for web-based completion of the questionnaire package. Web-based data collection has been demonstrated to be a valid approach for questionnaire-based research.23 The ASI was administered first, followed by the PASS-20 and the FPQ-SF. This study was approved by the University of Regina Research Ethics Board, and all participants provided informed consent. Participants received course credit for participation.

Measures The Fear of Pain Questionnaire-Short Form (FPQ-SF4) is a 20-item short form of the empirically derived Fear of Pain Questionnaire-III (FPQ-III34). Each item is responded to using a 5-point Likert scale ranging from 1 (not at all) to 5 (extreme). The FPQ-SF consists of 4 factorially distinct subscales, each related to a specific type of pain: Minor Pain (eg, biting your tongue while eating), Severe Pain (eg, having someone slam a heavy car door on your hand), Injection Pain (eg, having a blood sample drawn with a hypodermic needle), and Dental Pain (eg, having one of your teeth drilled). There are still questions regarding the additive utility of the Injection Pain and Dental Pain factors, such that they have been argued to be optional.4 The FPQ-SF has good factorial validity and internal constancy (␣ ⫽ .914). Although not established for the FPQ-SF, there is evidence of good test-retest reliability with the FPQ-III.34,37 For this sample, the internal consistency was good for the total score (␣ ⫽ .91) and subscale scores (Minor Pain, ␣ ⫽ .87; Severe Pain, ␣ ⫽ .86; Injection Pain ␣ ⫽ .91; Dental Pain ␣ ⫽ .83). The Pain Anxiety Symptoms Scale-20 (PASS-2030) is a 20-item measure developed as a short-form of the original 40-item measure.32,33 Each item is responded to using a 6-point Likert scale anchored from 0 (never) to 5 (always). The PASS and the PASS-20 comprise 4 factorially distinct components of pain-related anxiety, including (1) Cognitive Anxiety (eg, I can’t think straight when in pain), (2) Pain-related Fear, (eg, pain sensations are terrifying), (3) Escape and Avoidance (eg, I try to avoid activities that cause pain), and (4) Physiological Anxiety (eg, pain makes me nauseous). The PASS-20 measures more general pain-related statements (eg, pain seems to cause my heart to pound or race), rather than statements directly associated with situational experience (eg, biting your tongue while eating). This permits a subtle measure of latent pain-related anxiety that contrasts the direct, situation-specific FPQ-SF measure. The PASS-20 has an ␣ ⫽ .81,30 and correlates highly, r ⫽ .95, with the original. Factorial validity for both the total and subscale scores has been demonstrated for clinical18 and nonclinical samples.1 The 40-item PASS has been used to assess pain-

Carleton and Asmundson

31 26,36

related anxiety in community samples ; neither the instructions for the PASS-20, nor the items themselves, preclude its use in participants who do not have current pain.1 For this sample, the internal consistency was good for the total score (␣ ⫽ .92) and subscale scores (Cognitive Anxiety, ␣ ⫽ .91; Pain-related Fear ␣ ⫽ .70; Escape and Avoidance, ␣ ⫽ .87; Physiological Anxiety ␣ ⫽ .85). The ASI39 is a 16-item measure assessing the tendency to fear symptoms of anxiety based on the belief that they may have harmful consequences (eg, it scares me when I feel faint). Items are rated on a 5-point Likert scale ranging from 0 (very little) to 4 (very much). Factor analytic investigations indicate the ASI comprises 3 internally consistent lower-order factors [ie, fear of somatic sensations; somatic (eg, it scares me when my heart beats rapidly), fear of cognitive dyscontrol; cognitive (eg, when I cannot keep my mind on a task, I worry that I may be going crazy), and fear of socially observable anxiety reactions; social (eg, it is important to me not to appear nervous) that load on to a single higher-order factor.45,49 The validity and reliability of using the ASI total or subscale scores have been well documented.38,44 For this sample, the internal consistency was good for the total score (␣ ⫽ .88), somatic (␣ ⫽ .85), and cognitive subscale scores (␣ ⫽ .78) but low for the social subscale (␣ ⫽ .55).

Analyses Descriptive data, including internal consistency, were calculated using each measure and the established subscales. A series of independent t tests were conducted to check for any substantial sex differences within the FPQ-SF or the PASS-20. Thereafter, a Pearson correlation analysis was used to assess the interfactor correlations. Confirmatory factor analyses (CFA) were conducted to assess the independence of the FPQ-SF and the PASS-20. CFAs provide goodness-of-fit indices that can be used for comparing the fit of predefined model factor structures to an available data set more robustly than EFA.25 Accordingly, the goodness-of-fit indices are comparative measures of whether the FPQ-SF and the PASS-20 items load better onto 1 or 2 latent factors, and the strength of the relationship between those latent factors. Item parcels (individual items are summed to form previously identified subscales) were created to serve as observed variables. Item parcels reduce the number of indicators per latent variable, have a stronger relationship to the latent variables than individual items, and serve to restrain the model degrees of freedom.9,28 Using the item parcels, several different pairs of models were assessed (ie, with and without the Injection Pain and Dental Pain factors from the FPQ-SF4). The CFAs were performed using AMOS 6.02 with the raw data as input and the maximum likelihood estimation procedure. Models were evaluated using several goodness of fit indices and following recommendations by Hu and Bentler25: (1) ␹2 (values should not be statistically significant); (2) ␹2/df ratio (values should be ⬍2.0), (3) comparative fit index (CFI; values should be close to .95), (4) root mean square error of approximation (RMSEA; values should be equal to or less than .06), (5) the

standardized root mean square residual (SRMR; values should be equal to or less than .08), and (6) the expected cross validation index (ECVI; for comparing non-nested models). The latter 5 fit indices are emphasized because ␹2 statistics are inflated in larger samples.25 A canonical correlational analysis was performed to further assess inter-relationships between the subscales of the PASS-20, the FPQ-SF, and the ASI. Unlike a multiple regression analysis that is limited to the evaluation of multiple predictor variables against only one outcome variable, canonical correlation permits the analysis of multiple predictor and multiple outcome variables simultaneously. A combination of predictor variables is derived that has the highest correlation with a combination of the outcome variables.42 In the present study, canonical correlation analysis provided an indication of whether the ASI subscales have a different relationship to either the PASS-20 or the FPQ-SF.

Results Descriptive Statistics None of the indices of univariate skewness and kurtosis were substantially out of range (ie, had positive standardized skewness values that exceeded 2 or positive standardized kurtosis values that exceeded 7).21,42 Therefore, for the sake of parsimony, univariate descriptives, including skewness and kurtosis, are only presented for each subscale. Multivariate normality was assessed using Mardia’s coefficient of multivariate kurtosis13 for all models and the results suggested nonnormal data; however, parameter estimates and most model fit indices are robust to non-normality given maximum-likelihood estimation and a sample size of 100 or more participants.27 Nonetheless, we used the BollenStine bootstrap ␹2 and compared bootstrapped parameter estimates with estimates from a maximum-likelihood procedure.13,35 In all cases, the statistical significance value for the Bollen-Stine bootstrap ␹2 produced results comparable with those from the maximum-likelihood procedure for the CFA. Women scored statistically significantly higher than men on a majority of scales and subscales, except for the PASS-20 Fear subscale, the ASI-Cognitive subscale, and the ASI-Social subscale (Table 1). Despite the statistically significant differences, the percentage of variance accounted for was relatively small; therefore, men and women were analyzed together and the CFAs were compared for differences in pattern responding. All intermeasure and intersubscale Pearson correlations were statistically significant (P ⬍ .01; Table 2).

CFA Results Several models were tested (Figs 1 and 2), including (1) a unitary latent construct model with all subscales; (2) a unitary latent construct model with all PASS-20 and ASI subscales but only the severe and minor pain subscales from the FPQ-SF; (3) a 2-factor model with the PASS-20 and the ASI subscales loading on one factor and the

32

Table 1.

Subscale Descriptive Statistics ALL (N ⫽ 268)

WOMEN (N ⫽ 178)

WOMEN⫹

VS

MEN⫺

M

SD

S (.15)

K (.29)

␣

M

SD

S (.25)

K (.50)

␣

M

SD

S (.18)

K (.36)

␣

T

P

R2

9.94 7.22 4.78 5.34 27.28

5.13 4.30 4.28 4.50 14.98

.23 .69 1.26 1.04 .77

⫺.56 .29 1.85 .98 .74

.91 .70 .87 .85 .92

8.67 6.33 4.61 4.28 23.89

4.57 3.69 3.78 3.77 13.14

.16 .68 1.19 1.04 .83

⫺.16 .59 2.10 1.02 2.15

.90 .64 .85 .85 .92

10.59 7.67 4.87 5.87 28.99

5.29 4.52 4.52 4.75 15.58

.18 .62 1.26 .96 .70

⫺.78 .06 1.66 .71 .30

.91 .72 .88 .84 .92

3.08 2.59 .46 2.98 2.82

.00 .02 .65 .01 .00

.03 .02 ⬍.01 .03 .03

14.04 19.41 6.81 8.12 48.38 33.45

5.03 5.08 3.62 3.21 13.23 8.79

.76 ⫺.26 .77 .31 .20 .18

.16 ⫺.51 ⫺.44 ⫺.67 ⫺.49 ⫺.41

.87 .86 .91 .83 .91 .89

12.80 17.49 5.82 7.48 43.59 3.29

4.45 4.95 3.31 3.24 12.38 8.16

1.01 ⫺.33 1.17 .57 .38 .17

.41 ⫺.83 .54 ⫺.43 ⫺.43 ⫺.66

.85 .87 .90 .86 .91 .89

14.66 20.39 7.30 8.45 5.80 35.05

5.20 4.87 3.68 3.16 13.02 8.69

.63 ⫺.24 .62 .21 .11 .17

.07 ⫺.53 ⫺.68 ⫺.67 ⫺.41 ⫺.39

.88 .84 .91 .81 .91 .89

2.90 4.58 3.22 2.36 4.36 4.32

.00 .00 .00 .02 .00 .00

.03 .07 .04 .02 .07 .07

7.62 2.67 6.62 16.90

5.71 2.91 2.91 9.91

.99 1.53 .45 1.09

.94 2.47 .26 1.41

.85 .78 .55 .88

6.08 2.32 6.46 14.86

4.63 2.56 2.47 7.76

.93 1.35 .77 .97

.84 1.43 .56 .64

.81 .78 .39 .83

8.39 2.85 6.70 17.94

6.05 3.07 3.11 10.70

.88 1.53 .33 .99

.63 2.47 .08 1.05

.86 .78 .61 .90

3.48 1.40 .69 2.69

.00 .16 .49 .00

.04 .01 .00 .03

Abbreviations: S (SE) – skewness (standard error); K (SE) – kurtosis (standard error); PASS-20, Pain Anxiety Symptoms Scale-20; FPQ-SF, Fear of Pain Questionnaire-Short Form; ASI, Anxiety Sensitivity Index.

Multidimensionality of Fear of Pain

PASS-20 Cognitive Escape/Avoidance Fear Physiology Total FPQ-SF Minor Severe Needle Dental 4-Factor Total 2-Factor Total ASI Somatic Cognitive Social Total

MEN (N ⫽ 90)

Carleton and Asmundson Table 2.

33

Subscale Pearson Correlations PASS-20 ASI

FPQ-SF ESCAPE/

SUBSCALE

SOMATIC

COGNITIVE

SOCIAL

COGNITIVE

AVOIDANCE

FEAR

PHYSIOLOGY

MINOR

SEVERE

NEEDLE

Cognitive Escape/Avoidance Fear Physiology Minor Severe Needle Dental

.45 .45 .58 .55 .42 .37 .31 .32

.35 .32 .45 .39 .37 .22 .21 .21

.30 .30 .36 .36 .23 .30 .31 .26

.26 .55 .26 .17 .54 .56 .27

.33 .52 .41 .31 .22 .52

.39 .17 .53 .51 .30

.32 .35 .23 .49

.14 .14 .44

.33 .27

.17

Abbreviations: ASI, Anxiety Sensitivity Index; PASS-20, Pain Anxiety Symptoms Scale-20; FPQ-SF, Fear of Pain Questionnaire-Short Form.

FPQ-SF subscales loading on the other; (4) a 2-factor model with the PASS-20 and the ASI subscales loading on one factor and the severe and minor pain subscales from the FPQ-SF loading on the other; (5) a 2-factor model with the FPQ-SF and the ASI subscales loading on one factor and the PASS-20 subscales loading on the other; (6) a 2-factor model with the severe and minor pain subscales from the FPQ-SF and the ASI subscales loading on one factor, and the PASS-20 loading on the other; (7) a 3-factor model with the subscales from each of the 3 measures loading on a separate factor, with all 3 factors correlated; (8) a 3-factor model with the subscales from each of the 3 measures loading on a separate factor— excluding the severe and minor pain subscales from the

FPQ-SF—and all 3 factors correlated. The fit indices for each model are presented in Table 3. The correlations between factors in model 7 were, PASS-20 to FPQ-SF, r ⫽ .47; ASI to FPQ-SF, r ⫽ .55; and PASS-20 to ASI, r ⫽ .73. The correlations between factors in model 8 were, PASS-20 to FPQ-SF, r ⫽ .51; ASI to FPQ-SF, r ⫽ .59; and PASS-20 to ASI, r ⫽ .73. Model 8 provided the best fit for the data, having the highest CFI, as well as the lowest ␹2/df ratio, RMSEA, SRMR, and ECVI; moreover, the fit indices for model 8 were statistically superior to those from all of the other models. The small but statistically significant differences between men and women suggested any given model might be different across sex. In an effort to assess invariance of model 8 across men and women, measurement

Figure 1. Confirmatory factor analysis models 1 through 4. Ovals represent latent variables, whereas rectangles represent observed variables set as item parcels. Curved lines represent correlations; straight lines represent regression lines. Odd numbered models (1 and 3) included the injection and dental factors of the Fear of Pain Questionnaire-Short Form (FPQ-SF); even numbered models (2 and 4) did not; to distinguish these factors, the rectangles and regression lines for the injection and dental factors of the FPQ-SF are drawn with broken lines.

34

Multidimensionality of Fear of Pain

Figure 2. Confirmatory factor analysis models 5 through 8. Ovals represent latent variables; rectangles represent observed variables set as item parcels. Curved lines represent correlations; straight lines represent regression lines. Odd numbered models (5 and 7) included the injection and dental factors of the Fear of Pain Questionnaire Short-Form (FPQ-SF); even numbered models (6 and 8) did not; to distinguish these factors, the rectangles and regression lines for the injection and dental factors of the FPQ-SF are drawn with broken lines.

weights (the relationship between the measured variables and their latent variables) and structural covariances (the covariances among the latent variables) were assessed using a procedure in AMOS described by Byrne.13,14 This procedure involves evaluating differences between men and women data sets. In this sample, neither measurement weights, ⌬␹2(6, N ⫽ 268) ⫽ 9.19, P ⬎ .10, nor structural covariances, ⌬␹2(6, N ⫽ 268) ⫽ 8.78, P ⬎.10, differed between men and women.

Canonical Correlation Results The first set of variables included the subscales of the ASI. The second set of variables included the subscales of

the PASS-20 and the Minor Pain and Severe Pain subscales of the FPQ-SF. Given the relative superiority of CFA models using only the 2 subscales of the FPQ-SF, parsimony along with existing theory4 supported the elimination of the Dental Pain and Needle Pain subscales from this analysis. Correlations greater than r ⫽ ⫾.30 were interpreted. The first canonical correlation was .68 and explained 94.3% of the overlapping variance between the ASI and the PASS-20 and the FPQ-SF, F(18, 733.05) ⫽ 11.03, P ⬍ .001. This canonical correlation was associated with the ASI somatic (r ⫽ .99), ASI cognitive (r ⫽ .75), ASI social (r ⫽ .63), PASS-20 Fear (r ⫽ .86), PASS-20 Physiological (r ⫽ .80), PASS-20 Cognitive (r ⫽ .67), PASS-20 Escape/

Table 3.

Confirmatory Factor Analysis Fit Indices

MODEL

LATENT FACTORS

R

␹2(DF)

P

␹2/(DF)

CFI

SRMR

RMSEA

RMSEA CI (90%)

ECVI

ECVI CI (90%)

1 2 3 4 5 6 7 8

1 1* 2 2* 2 2* 3 3*

— — .55 .60 .74 .74 — —

353.56 (44) 191.89 (27) 185.52 (43) 146.69 (26) 238.16 (43) 86.54 (26) 78.86 (41) 40.11 (24)

⬍ .001 ⬍ .001 ⬍ .001 ⬍ .001 ⬍ .001 ⬍ .001 ⬍ .001 .021

8.035 7.107 4.314 5.642 5.538 3.328 1.923 1.671

.742 .831 .881 .876 .837 .938 .968 .983

.105 .079 .067 .065 .087 .059 .046 .034

.162 .151 .111 .132 .130 .093 .059 .050

.147–.178 .131–.172 .095–.128 .112–.153 .114–.147 .072–.116 .039–.078 .020–.077

1.489 .854 .867 .692 1.064 .466 .483 .308

1.278–1.728 .703–1.032 .723–1.039 .563–.849 .897–1.260 .375–.587 .403–.592 .256–.388

Abbreviations: CFI, comparative fit index; RMSEA, root mean square error of approximation; ECVI, expected cross-validation index; SRMR, standardized root mean square residual. NOTE. Higher CFI values indicate better fit, whereas lower values on all other indices indicate better fit. RMSEA CI ⫽ 90% confidence interval for RMSEA (low; high); ECVI CI ⫽ 90% confidence interval for ECVI (low; high). *Excluded Injection and Dental Factors from the FPQ-SF.

Carleton and Asmundson avoidance (r ⫽ .66), Minor Pain (r ⫽ .62), and Severe Pain (r ⫽ .55). In other words, there was substantial overlap across all of the variables; however, the correlations between the PASS-20 subscales and the ASI somatic subscale were somewhat higher. The remaining 3 canonical correlations or correlations between combinations of the measures were not different from zero.

Discussion The primary purpose of this investigation was to assess whether the construct commonly referred to as fear of pain might be better represented by distinct constructs of pain-related fear and pain-related anxiety. A secondary purpose was to assess whether fear of pain— or the underlying constructs—might be better conceptualized as a manifestation of anxiety sensitivity.6,24 These assessments were made using self-report measures as proxies for pain-related fear (ie, the FPQ-SF), pain-related anxiety (ie, the PASS-20), and anxiety sensitivity (ie, the ASI). Construct independence of the 3 measures was supported by the results of several CFAs. It appears as though there may be more than a difference in facevalidity between the FPQ-SF and the PASS-20. The results also demonstrated a notably larger correlation between the PASS-20 and the ASI, relative to the correlation between the PASS-20 and the FPQ-SF, and relative to the correlation between the FPQ-SF and the ASI. Despite the large correlation between PASS-20 and the ASI, they too appeared to remain independent. The relatively smaller correlations between the other measures also suggested the constructs overlap but may nonetheless be independent. The canonical correlation was performed specifically because it is ideal for assessing multiple predictor and multiple outcome variables simultaneously.42 Like the CFA results, the canonical correlation supported the notion that the 3 measures may represent related but independent constructs. Also in line with the CFA results, the canonical correlation suggested the PASS-20 may be, relative to the FPQ-SF, more related to the ASI. Overall, the results of this investigation support the notion that pain-related fear, pain-related anxiety, and anxiety sensitivity, may be related, but independent constructs. A precedent distinction has been made between fear and anxiety within the anxiety disorder literature.10 Paralleling that distinction, pain-related fear may be used to represent present-oriented emotive states associated with nociceptive stimulation (eg, pain from a dental procedure), whereas pain-related anxiety can represent a more general, future-oriented emotive state that does not require nociceptive stimulation but occurs in anticipation of nociception (eg, the possibility of pain from a dental procedure). There is growing explicit and implicit evidence to support this distinction. For example, unpredictable and predictable pain have been associated with hyperalgesia and hypoalgesia/analgesia, respectively,40 supporting the notion that different neuronal mechanisms may be operative depending on whether pain evokes anxiety (ie, a response to unpredictable, future threats) or fear (ie, a response to an immediate threat7).

35 Tangentially, researchers have posited a theoretical relationship been chronic pain and anxiety disorders, supported by emerging evidence that both may be ameliorated by exposure-based therapies that reduce avoidance by confronting anxiety- and fear-related stimuli.8,29,31,33 Pain-related fear and anxiety have been investigated, primarily, in relation to the development and maintenance of disabling chronic pain.8 Growing empirical support for current fear-anxiety-avoidance models of chronic pain8,47 is driving development and implementation of tailored assessment and treatment strategies for pain-related fear and anxiety in chronic pain populations. It is reasonable to expect that pain-related anxiety and fear may be interdependent; understanding that interdependency may be a key component to understanding why pain-related fear and anxiety in pain-free individuals appears qualitatively different relative to that experienced by patients with chronic pain.3,8,19,20,29 For example, a person with naturally higher pain-related anxiety, if exposed to a painful stimulus, may be more likely to react with heightened pain-related fear. If the person ruminates about the exposure (eg, during recuperation from an injury), the person’s pain-related anxiety may increase, further supporting avoidance behaviors. Over time, the activation of pain-related anxiety may become an overlearned response,16 further exacerbating pain-related avoidance and disability. Therefore, it may be important for successful therapy that the nature and interrelationship of these constructs be thoroughly understood and explored. Regarding the question of whether or not pain-related fear or anxiety are manifestations of anxiety sensitivity,6,24 the results of this investigation do not provide a definitive answer. Pain is, in part, a somatic sensation; as such, pain-related anxiety can be conceptualized as a manifestation of anxiety sensitivity. Therefore, pain-related fear would be the temporally immediate—rather than future-oriented—manifestation of pain-related anxiety, making it by proxy a manifestation of anxiety sensitivity. Conversely, it could be argued that pain-related anxiety is a fundamental fear (ie, fears of inherently noxious stimuli that are not logically reducible to other fears41,43) unto itself, related to but not dependent on anxiety sensitivity. Although somatic sensations can produce arousing and even fearful responses, they are not necessarily inherently noxious. Pain-related anxiety, unlike anxiety sensitivity, does not require a catastrophic misinterpretation of the sensation for the sensation to be noxious. For example, a rapid heartbeat can be interpreted as exhilarating as easily as it can be interpreted as terrifying, depending on the context. However, pain is almost always unpleasant, presumably by evolutionary design, and associated with specialized nociceptive pathways. Moreover, PASS-20 items (eg, “I will stop any activity as soon as I sense pain coming on”) may be less dependent on individual experience (most people will have experienced pain related to at least one activity), relative to items from the FPQ-SF (eg, “breaking your arm”; not all people will have broken an arm and some will have never

36

Multidimensionality of Fear of Pain

broken any bone). The more general construct of painrelated anxiety may share a great deal in common with the fundamental fears, including anxiety sensitivity, as described by Riess41 and Taylor.43 Specifically, pain is an inherently noxious stimuli and other more common fears can be logically reduced to pain-related anxiety (eg, fearing falling because it may cause pain). It may be that pain-related anxiety (as measured by the PASS-20), like anxiety sensitivity, is a fundamental fear; tangentially, pain-related fear (as measured by the FPQ-SF) may better represent common fears (ie, fears that can be logically reduced to fundamental fears41,43). There are several limitations of this investigation that provide directions for future research. First, despite there being several measures of pain-related fear and pain-related anxiety,4 only 1 of each were used here, precluding cross-validation, risking artificial inflation of correlations due to method invariance between measures, and artificial inflation of internal consistency within the measures. Second, the original 16-item measure of the ASI was used, rather than the new ASI-3, a revised measure that has ratified subscales.46 Third, the order of administration was not counterbalanced, increasing the probability of unaccounted for method variance; however, at this time there is no reason to believe randomizing the order of presentation would produce significantly different results. Future studies should consider randomizing the order to further ensure the robust nature of the construct independence. Fourth, the sam-

ple was a convenience sample of undergraduates, primarily women, limiting the generalizability of these results particularly to clinical populations; however, this was intentional. The more general sample used here ensured range of responses whereas responses from a clinical sample would likely have been restricted and relatively higher responses.1 Future research should further assess sex differences and consider using measures of attentional biases longitudinally with chronic pain and pain-free samples, to evaluate the hypothesized differences between groups. Finally, there is considerable debate regarding sample size requirements for CFA22; however, using item parcels with this sample size is comparable with previous pain research.15,17,18 Overall, the PASS-20, the FPQ-SF, and the ASI may be tapping distinct, albeit highly correlated, constructs. This proposition is of potential importance for understanding the development of some presentations of disabling chronic pain, and therein the tailoring of effective treatments, reemphasizing an often overlooked but critical distinction made by McCracken et al31 over a decade ago. Recent research has shown that reducing AS in undergraduate women also reduced their pain-related anxiety48; however, additional research is required to extrapolate these findings to men and chronic pain patients. In the interim, it may be useful to make a consistent distinction in the literature between pain-related fear (fear of current pain) and pain-related anxiety (fear of the possibility of pain).

References

9. Bandalos DL: The effects of item parceling on goodnessof-fit and parameter estimate bias in structural equation modeling. Structural Equation Modeling 9:78-102, 2002

1. Abrams MP, Carleton RN, Asmundson GJG: An exploration of the psychometric properties of the PASS-20 with a nonclinical sample. J Pain 8:879-886, 2007 2. Arbuckle JL: Amos 6.0.0. Spring House, PA, Amos Development Corporation, 2005 3. Asmundson GJ, Collimore KC, Bernstein A, Zvolensky MJ, Hadjistavropoulos HD: Is the latent structure of fear of pain continuous or discontinuous among pain patients? Taxometric analysis of the pain anxiety symptoms scale. J Pain 8:387-395, 2007 4. Asmundson GJG, Bovell CV, Carleton RN, McWilliams LA: The Fear of Pain Questionnaire-Short Form (FPQ-SF): Factorial validity and psychometric properties. Pain 134:51-58, 2008 5. Asmundson GJG, Carleton RN, Ekong J: Dot-probe evaluation of selective attentional processing of pain cues in patients with chronic headaches. Pain 114:250-256, 2005 6. Asmundson GJG, Norton PJ, Norton GR: Beyond pain: The role of fear and avoidance in chronicity. Clin Psychol Rev 19:97-119, 1999 7. Asmundson GJG, Norton PJ, Vlaeyen JWS: Fear-avoidance models of chronic pain: An overview, in Asmundson GJG, Vlaeyen JWS, Crombez G (eds): Understanding and Treating Fear of Pain. Oxford, Oxford University Press, 2004, pp 3-24 8. Asmundson GJG, Vlaeyen JWS, Crombez G, eds: Understanding and treating fear of pain. New York, Oxford, 2004

10. Barlow DH: Unraveling the mysteries of anxiety and its disorders from the perspective of emotion theory. Am Psychologist 55:1247-1263, 2000 11. Bernstein A, Leen-Feldner EW, Kotov R, Schmidt NB, Zvolensky MJ: A test of a panic-relevant diathesis-stress model using a taxonic index of anxiety sensitivity, in Sanfelippo AJ (ed): Panic Disorders: New Research. Hauppauge, NY, Nova Science Publishers Inc., 2006, pp 15-40 12. Bernstein A, Zvolensky MJ, Kotov R, Arrindell WA, Taylor S, Sandin B, Cox BJ, Stewart SH, Bouvard M, Cardenas SJ, Eifert GH, Schmidt NB: Taxonicity of anxiety sensitivity: A multi-national analysis. Journal of Anxiety Disorders 20:122, 2006 13. Byrne B: Structural Equation Modeling with AMOS: Basic Concepts, Applications, and Programming. Mahwah, NJ, Erlbaum, 2001 14. Byrne B: Testing for multigroup invariance using AMOS graphics: A road less traveled. Structural Equation Modeling 11:272-300, 2004 15. Cano A, Leonard MT, Franz A: The significant other version of the Pain Catastrophizing Scale (PCS-S): Preliminary validation. Pain 119:26-37, 2005 16. Carleton RN, Asmundson GJ, Collimore KC, Ellwanger J: Strategic and automatic threat processing in chronic musculoskeletal pain: A startle probe investigation. Cogn Behav Ther 35:236-247, 2006

Carleton and Asmundson 17. Compas BE, Boyer MC, Stanger C, Colletti RB, Thomsen AH, Dufton LM, Cole DA: Latent variable analysis of coping, anxiety/depression, and somatic symptoms in adolescents with chronic pain. J Consult Clin Psychol 74:1132-1142, 2006 18. Coons MJ, Hadjistavropoulos HD, Asmundson GJ: Factor structure and psychometric properties of the Pain Anxiety Symptoms Scale-20 in a community physiotherapy clinic sample. Eur J Pain 8:511-516, 2004 19. Crombez G, Eccleston C, Baeyens F, Eelen P: Attentional disruption is enhanced by the threat of pain. Behav Res Ther 36:195-204, 1998

37 34. McNeil DW, Rainwater AJ 3rd: Development of the Fear of Pain Questionnaire–III. J Behav Med 21:389-410, 1998 35. Nevitt J, Hancock GR: Performance of bootstrapping approaches to model test statistics and parameter standard error estimation in structural equation modeling. Structural Equation Modeling 8:353-377, 2001 36. Osman A, Barrios FX, Osman JR, Schneekloth R, Troutman JA: The Pain Anxiety Symptoms Scale: Psychometric properties in a community sample. J Behav Med 17:511-522, 1994

20. Crombez G, Eccleston C, Baeyens F, Eelen P: When somatic information threatens, catastrophic thinking enhances attentional interference. Pain 75:187-198, 1998

37. Osman A, Breitenstein JL, Barrios FX, Gutierrez PM, Kopper BA: The fear of pain questionnaire-III: further reliability and validity with nonclinical samples. J Behav Med 25:155173, 2002

21. Curran PJ, West SG, Finch JF: The robustness of test statistics to nonnormality and specification error in confirmatory factor analysis. Psychol Methods 1:16-29, 1996

38. Peterson RA, Plehn K: Measuring anxiety sensitivity, in Taylor S (ed): Anxiety Sensitivity: Theory, Research, and Treatment. Mahwah, NJ, Erlbaum, 1999, pp 61-81

22. Gagné P, Hancock GR: Measurement model quality, sample size, and solution propriety in confirmatory factor models. Multivariate Behav Res 41:65-83, 2006

39. Peterson RA, Reiss S: Anxiety Sensitivity Index Manual, 2nd ed. Worthington, OH, International Diagnostic Systems, 1992

23. Gosling SD, Vazire S, Srivastava S, John OP: Should we trust web-based studies? A comparative analysis of six preconceptions about internet questionnaires. Am Psychologist 59:93-104, 2004

40. Ploghaus A, Becerra L, Borras C, Borsook D: Neural circuitry underlying pain modulation: Expectation, hypnosis, placebo. Trends Cogn Sci 7:197-200, 2003

24. Greenberg J, Burns JW: Pain anxiety among chronic pain patients: Specific phobia or manifestation of anxiety sensitivity? Behav Res Ther 41:223-240, 2003 25. Hu L, Bentler PM: Fit indices in covariance structure modeling: Sensitivity to underparameterized model misspecification. Psychol Methods 3:424-453, 1999

41. Reiss S: Expectancy model of fear, anxiety, and panic. Clinl Psychol Rev 11:141-153, 1991 42. Tabachnick BG, Fidell LS: Using Multivariate Statistics, 4th ed. New York, NY, Harper and Row, 2001 43. Taylor S: The structure of fundamental fears. J Behav Ther Exp Psychiatry 24:289-299, 1993

26. Larsen DK, Taylor S, Asmundson GJ: Exploratory factor analysis of the Pain Anxiety Symptoms Scale in patients with chronic pain complaints. Pain 69:27-34, 1997

44. Taylor S (ed): Anxiety Sensitivity: Theory, Research, and Treatment of the Fear of Anxiety. Mahwah, NJ, Erlbaum, 1999

27. Lei M, Lomax RG: The effect of varying degrees of nonnormality in structural equation modeling. Structural Equation Modeling 12:1-27, 2005

45. Taylor S, Koch WJ, Woody S, McLean P: Anxiety sensitivity and depression: how are they related? J Abnorm Psychol 105:474-479, 1996

28. Little TD, Cunningham WA, Shahar G, Widaman KF: To parcel or not to parcel: Exploring the question, weighing the merits. Structural Equation Modeling 9:151-173, 2002 29. McCracken LM: “Attention” to pain in persons with chronic pain: A behavioral approach. Behav Ther 28:271284, 1997

46. Taylor S, Zvolensky MJ, Cox BJ, Deacon B, Heimberg RG, Ledley DR, Abramowitz JS, Holaway RM, Sandin B, Stewart SH, Coles M, Eng W, Daly ES, Arrindell WA, Bouvard M, Cardenas SJ: Robust dimensions of anxiety sensitivity: Development and initial validation of the Anxiety Sensitivity Index3. Psychol Assess 19:176-188, 2007

30. McCracken LM, Dhingra L: A short version of the Pain Anxiety Symptoms Scale (PASS-20): Preliminary development and validity. Pain Res Manage 7:45-50, 2002

47. Vlaeyen JW, Linton SJ: Fear-avoidance and its consequences in chronic musculoskeletal pain: A state of the art. Pain 85:317-332, 2000

31. McCracken LM, Gross RT, Aikens J, Carnrike CL Jr: The assessment of anxiety and fear in persons with chronic pain: A comparison of instruments. Behav Res Ther 34:927-933, 1996

48. Watt MC, Stewart SH, Lefaivre MJ, Uman LS: A brief cognitive-behavioral approach to reducing anxiety sensitivity decreases pain-related anxiety. Cogn Behav Ther 35:248256, 2006

32. McCracken LM, Gross RT, Sorg PJ, Edmands TA: Prediction of pain in patients with chronic low back pain: Effects of inaccurate prediction and pain-related anxiety. Behav Res Ther 31:647-652, 1993

49. Zinbarg RE, Barlow DH, Brown TA: Hierarchical structure and general factor saturation of the anxiety sensitivity index. Psychol Assess 9:277-284, 1997

33. McCracken LM, Zayfert C, Gross RT: The Pain Anxiety Symptoms Scale: Development and validation of a scale to measure fear of pain. Pain 50:67-73, 1992

50. Zvolensky MJ, Goodie JL, McNeil DW, Sperry JA, Sorrell JT: Anxiety sensitivity in the prediction of pain-related fear and anxiety in a heterogeneous chronic pain population. Behav Res Ther 39:683-696, 2001