Correction of Pain Expectancies Following Exposure to Movement in Chronic Back Pain. A thesis presented to. the faculty of

Transcription

1 Correction of Pain Expectancies Following Exposure to Movement in Chronic Back Pain 1 A thesis presented to the faculty of the College of Arts and Sciences of Ohio University In partial fulfillment of the requirements for the degree Master of Science Zina Trost November Zina Trost. All Rights Reserved.

2 2 This thesis titled Correction of Pain Expectancies Following Exposure to Movement in Chronic Back Pain by ZINA TROST has been approved for the Department of Psychology and the College of Arts and Sciences by Christopher R. France Professor of Psychology Benjamin M. Ogles Dean, College of Arts and Sciences

3 3 ABSTRACT TROST, ZINA, M.S., November 2008, Psychology Correction of Pain Expectancies Following Exposure to Movement in Chronic Back Pain (119 pp.) Director ofthesis: Christopher R. France According to the fear-avoidance model, kinesiophobia (pain-related fear) is an important factor in the development of chronic pain and disability. The current research project investigated the extent to which expectancy corrections (i.e., the tendency to bring expected pain/harm in line with experienced pain/harm) following exposure to one movement generalize to identical movements of increased intensity in chronic low back pain patients with high versus low levels of kinesiophobia. Additionally, the study addressed the impact of increased exposure on generalization. Participants were asked to consecutively perform four adaptations of a reaching task, each introducing a discrete element of increased intensity. Expected and experienced pain and harm ratings were collected for each trial. It was predicted that highly kinesiophobic participants will demonstrate greater overprediction of pain and harm relative to their low-kinesiophobic counterparts. It is also predicted that in highly kinesiophobic participants, expectancy corrections evidenced in the later trials of a given movement will show less generalization to the subsequent movement performed. Approved: Christopher R. France Professor of Psychology

4 4 TABLE OF CONTENTS Page Abstract... 3 List of Tables... 7 List of Figures... 8 Introduction General Background The Fear-Avoidance Model of Low Back Pain Constructs Associated with the Fear-Avoidance Model Pain Catastrophizing Fear Avoidance Beliefs Pain Related Anxiety Kinesiophobia (Movement-related fear of pain/ (re)injury) Relationship Among the Pain-Related Fear Measures Measurement of Pain-Related Fear in the Prediction of Pain, Avoidance, and Disability Pain-related fear and self-reported disability Assessment of pain-related fear in acute pain populations Pain-related fear in the prediction of future pain Pain-related fear and prediction of physical performance The Role of Hypervigilance Clinical Management of Pain-Related Fear and Avoidance Behavior Experimental Exposure and the Restricted Effect of Generalization... 48

5 5 The Present Study Methods Design Participants Power Analysis Questionnaires Tampa Scale of Kinesiophobia Visual Analog Scale Short-Form McGill Pain Questionnaire Roland and Morris Disability Questionnaire Behavioral Test Procedure Data Analyses Results Characteristics of Participants Tests of Main Hypotheses Main effects Test of Hypothesis Test of Hypothesis Hypothesis Supplemental Findings Correction of Overprediction... 83

6 6 Discussion Limitations to the Current Study Directions for Future Research Conclusion References

7 7 LIST OF TABLES Page Table 1: Correlation Coefficients (r) between the TSK and Pain-Related Fear Measures...24 Table 2: Correlations of Pain-Related Fear and Self-Reported Disability Measures in Chronic Low Back Pain Samples...26 Table 3: Correlations of Pain-Related Fear and Self-Reported Disability Measures in Acute Pain Samples...32 Table 4: Correlations of Pain-Related Fear and Behavioral Performance Measures...36 Table 5: Characteristics of participant sample...69 Table 6: Characteristics of participants in High and Low Kinesiophobia categories...71 Table 7: Results of Separate 2 Group (high, low kinesiophobia) x 2 Sex (male, female) x 4 Movement (A through D) x 2 Rating (predicted, experienced) repeated measures MANOVAs on Trial 1 Pain and Harm Ratings...72 Table 8: Predicted and experienced pain and harm scores across movements for participants in the high and low kinesiophobic category...74 Table 9: Results of Separate 2 Condition (2, 4 trials) x 2 Sex (male, female) x 4 Movement (A through D) x 2 Rating (predicted, experienced) repeated-measures MANOVAs on Trial 1 Pain and Harm Ratings...80

8 8 LIST OF FIGURES Page Figure 1. Fear-Avoidance Model of Low Back Pain (Vlaeyen & Linton, 2000) Figure 2. The four reaching tasks comprising the behavioral test Figure 3. Proposed study protocol Figure 4. Mean ( ± SEM) Trial 1 harm overprediction (predicted-experienced) ratings across Movements A D for male and female participants in the High and Low Kinesiophobia groups Figure 5. Mean (± SEM) Trial 1 pain overprediction ratings (predicted experienced pain) across Movements A through D for participants in the High vs. Low Kinesiophobia groups Figure 6. Mean (± SEM) Trial 1 harm overprediction ratings (predicted experienced harm) across Movements A through D for participants in the High vs. Low Kinesiophobia groups Figure 7. Mean (± SEM) Trial 1 pain ratings across Movements A D for High Kinesiophobic participants in the Two- and Four-Trial conditions Figure 8. Mean (± SEM) Trial 1 pain overprediction ratings (predicted experienced) across Movements A D for High Kinesiophobic participants in the Two- and Four-Trial conditions Figure 9. Mean ( ± SEM) Trial 1 harm overprediction ratings (predicted experienced) across Movements A D for High Kinesiophobic participants in the Two- and Four-Trial conditions... 82

9 9 Figure 10. Mean (± SEM) predicted and experienced pain ratings on Trial 1 of Movements A through D Figure 11. Mean (± SEM) predicted and experienced pain ratings on Trial 2 of Movements A through D Figure 12. Mean (± SEM) predicted and experienced harm ratings on Trial 1 of Movements A through D Figure 13. Mean (± SEM) predicted and experienced harm ratings on Trial 2 of Movements A through D... 86

10 10 INTRODUCTION Back pain is one of the most frequent health problems in the general population and one of the most costly (Nachemson, 2000). Previous research has established kinesiophobia (i.e., fear of pain and injury due to movement, Kori et al., 1990) as an important predictor of disability in individuals with back pain independent of actual pain intensity. Highly kinesiophobic individuals are particularly vulnerable to developing chronic disability, possibly through the maintenance of avoidance behavior resistant to common exposure treatment strategies. Research has likewise demonstrated that exposure to movement may be an effective treatment strategy in counteracting avoidance behavior. The increasingly influential fear-avoidance model incorporates pain-related fear as a vital component in a process of thought and behavior that may mire some individuals in a cycle of debilitating chronic pain and disability (Vlaeyen et al., 1995a,b). A subgroup of the chronic pain population may thus require a more extensive or modified version of the customary exposure treatment paradigm. To this end, the current project investigated the extent to which the effects of exposure to one movement generalize to identical movements of increased intensity in individuals with chronic low back pain who report high versus low levels of kinesiophobia. By simulating a graded exposure paradigm and focusing on the relationship between pain expectations and pain reports, the present study observed whether expectancy of harm and injury is maintained across similar physical tasks. The results should bear on the approach to disability prevention/rehabilitation for this particular population. A review of the concepts and theoretical underpinnings of the fear

11 11 avoidance model as well as existing evidence for predictions regarding pain-related fear originating from this model appear below. General Background Epidemiological surveys in western industrialized countries reveal that low back pain is among the most frequent health problems in the general population, affecting approximately 58 84% of adults at least once in their lives (Boersma & Linton, 2005). Most acute back pain episodes are self-limited, with symptoms remitting within a few weeks with little or no intervention. However, it is estimated that up to 10% of low back pain sufferers develop a chronic pain condition, experiencing long-term pain and associated disability. This minority of the population accrues great healthcare and societal costs, consuming more than 50% of resources allocated toward back pain (Nachemson, 2000; Boersma & Linton, 2005). The traditional biomedical approach which views pain as a purely sensory experience in response to noxious stimulation is insufficient to explain the condition of these chronic patients, as pain persists after identifiable organic pathology has healed. Characterizations of acute versus chronic pain commonly invoke both temporal and organic distinctions (Kaplan & Tanner, 1989). Acute pain is typically considered secondary to specific tissue damage or another medical condition subject to treatment. As pain persists beyond three to six months and (in some situations) the relationship between organic pathology and the pain complaint becomes less readily apparent, the pain condition is likely to be considered chronic. Importantly, among people suffering benign chronic pain, the pain itself as opposed an underlying medical condition -- is often seen as the primary problem (Kaplan & Tanner, 1989).

12 12 With invigorated discussion stemming from the Gate-Control theory of pain (Melzack & Wall, 1965), over the past several decades the idea of pain has evolved from a simple sensory experience to a complex multidimensional subjective and perceptual phenomenon (Keefe & France, 1999). A number of models have attempted to incorporate multiple aspects of the pain experience, including psychological (e.g., perception, cognition, affect), behavioral (e.g., avoidance), and psychosocial factors to comprehensively account for pain behavior (Vlaeyen & Linton, 2000; Asmundson, Norton, & Norton, 1999). The concept of avoidance, long recognized as a spontaneous adaptive response to acute injury, has emerged at the center of several models attempting to account for the development of chronic dysfunction. Thus, avoidance, naturally adopted in acute pain situations, has been posited as a crucial mechanism in the development of long-term pain problems. First applied to the field of pain in an article by Lethem (1983), avoidance refers to a pattern of learned behavior that delays or averts an undesirable/aversive situation or experience (Asmundson et al., 1999). Several avoidance-based models rest on the basic premise that avoidance of the pain experience in general and painful activities in particular actually leads to the perpetuation of pain and related behaviors. Similar models of avoidance behavior have been associated with phobic states (Philips, 1987). Both instrumental (Fordyce, 1976) and cognitive (Philips, 1987) approaches have underpinned influential models that purport to explain how pain behaviors can be maintained in chronic musculoskeletal pain. The instrumental/operant understanding of chronic pain distinguishes private pain experiences from observable and quantifiable pain behaviors, proposing that initially adaptive responses to acute pain

13 13 (e.g., medication taking, complaining, inactivity) may come under the control of external contingencies (e.g., the positive reinforcement of a spouse s attention) and thus develop into a chronic pain problem (Fordyce, 1976). Cognitive perspectives highlight individual perception and interpretation of the pain experience, rather than the presence or elimination of the pain per se. The cognitive approach refers to such processes as cognitive appraisal, coping (Folkman, Lazarus, Gruen, & DeLongis, 1986), and control/self-efficacy beliefs (Bandura, 1977). The basic activity avoidance model (Linton et al., 1984) combines classical and operant conditioning processes wherein a threatening and pain-producing situation elicits a conditioned response of sympathetic activation (i.e., fear) which leads to avoidance of the situation. The avoidance behavior is reinforced by a reduction in unpleasant stimuli (i.e., pain, fear, anxiety). As successful avoidance inherently prevents contact with the consequences of the threatening situation (i.e., movement), established avoidant behavior is extremely resistant to extinction. Fear returns whenever the avoidance behavior cannot be carried out (Vlaeyen & Linton, 2000). The Fear-Avoidance Model of Low Back Pain (See Figure 1) Building on this previous model and on the work of Lethem et al. (1983), Philips (1987), and Waddell et al. (1993), the fear avoidance-model of low back pain and disability (Vlaeyen et al., 1995a,b) has evolved over the past two decades as a cognitivebehavioral account of how some individuals with an acute musculoskeletal condition go on to develop chronic pain and disability whereas others do not. Vlaeyen and Linton

14 14 " Avoiders " " Confronters " Musculoskeletal Injury Disuse, Depression, Disability Recovery Avoidance & Hypervigilance Fear of Movement and Reinjury Pain Confrontation No Movement & Reinjury Fear Catastrophizing No Catastrophizing Figure 1. Fear-Avoidance Model of Low Back Pain (Vlaeyen & Linton, 2000). (2000) recently revised the model to incorporate a greater focus on individual differences in perception and appraisal of painful stimuli. The model suggests that, following acute musculoskeletal injury, individuals are differentially susceptible to movement-related fear (i.e., kinesiophobia) based on their cognitive appraisal of the painful injury. The model posits two opposing appraisal/behavioral response types: avoidance and confrontation. These are two extremes on a continuum of response to fear. Avoiders are high on scales of kinesiophobia (Crombez et al., 1998) and are prone to catastrophic thought (Lethem et al., 1983), defined as an excessively negative orientation toward pain, its consequences, and pessimistic beliefs about ability to cope with pain. Thus, catastrophic appraisal is a potential precursor to pain-related fear. In turn, painrelated fear promotes hypervigilance towards pain sensations and initiates behaviors designed to escape/avoid potentially painful situations. Thus, fear leads to avoidance, which results in functional disability, broadly referring to disturbances or limitations in

15 15 physical functioning relevant to activities of daily living (Ghelof et al., 2005). Finally, longstanding avoidance of motoric activity may reinforce the pain experience through its detrimental impact on the musculoskeletal and cardiovascular system and the adoption of precarious postural strategies the so-called disuse syndrome (Bortz, 1984; Mosely et al., 2004), detailed in a following section. Pain might be further exacerbated by increased frustration, depression, or irritability accompanying the loss of essential reinforcers as the individual withdraws from daily activities. Both depression and disuse are known to be associated with decreased tolerance for pain (e.g., Romano & Turner, 1985; McQuade et al., 1988; Gormsen et al., 2004; Campbell et al., 2003). In contrast, confrontation is posited as an adaptive response. Confronters, low on scales of kinesiophobia, are less likely to engage in catastrophic thought and experience fear of movement and (re)injury. They are thus more likely to confront potentially painful situations and return to daily activities. This behavioral alternative is associated with faster rehabilitation and recovery. It is important to note that while the Fear-Avoidance Model presents an enticing explanatory sequence for the evolution of chronic back pain and disability, researchers emphasize its primarily heuristic structure, organizing hypothesized relationships that have emerged in existing data, rather than an inflexible causal framework. Accordingly, any temporal insinuations (in which constructs are posited as predictors or precursors ) are tempered with the recognition of the circular and, in a sense, necessarily incomplete nature of the model. In addition, the model does not address or discount the possibility of a sub-group of chronic low back pain sufferers whose condition may be exacerbated by a maladaptive

16 16 over-exertion propelled by a denial of functional limitations. These individuals display excessive behavioral endurance in the face of increasing pain sensations, leading to possibly greater physical damage and increased pain. Recent investigations have started to further explore this avoidance-endurance model of pain chronicity (Hasenbring 1993, 2000). Constructs Associated with the Fear-Avoidance Model The literature on low back pain introduces a number of related constructs which, although presenting a degree of conceptual overlap, are nevertheless discrete contributors to physical and psychological disability and recovery, and to the fear-avoidance sequence. These include Catastrophizing, Fear-avoidance beliefs, Pain-related anxiety, and Movement-related fear of pain/(re)injury (or, Kinesiophobia). The following section is intended to further explicate the discrete (yet not-mutually exclusive) nature of the above constructs, and to address the issue of construct redundancy. Each is addressed in its own historical context and with regard to its unique contribution to the understanding of the perseverance of back pain. Pain Catastrophizing. Prior to its emergence in the pain literature, catastrophizing had primarily been described as a cognitive distortion (Beck, 1967) in the context of cognitive theories of depression, contributing to the precipitation and maintenance of depressive symptomatology (Sullivan et al., 2001a; Turner & Aaron, 2001). The role of catastrophic cognition has likewise been recognized in anxiety disorders and hypochondriasis, where a catastrophic cognitive style signals the tendency to misinterpret and exaggerate the threat value of specific situations for example, bodily

17 17 sensations (Van Damme et al., 2002). In the pain literature, catastrophizing has been broadly conceived as an exaggerated mental set brought to bear during actual or anticipated pain experience (Sullivan et al., 2001a, p.53). Within the field of pain, the understanding of catastrophic thought continues to evolve from initially somewhat diverse conceptualizations (Turner & Aaron, 2001; Van Damme et al., 2002). Developed by Sullivan et al. (1995), the Pain Catastrophizing Scale (PCS) drew on and unified the existing literature. The PCS is a 13-item self-report measure which asks participants (clinical and nonclinical, chronic, acute, and pain-free populations) to indicate the degree to which a number of thoughts and feelings have contributed to past painful experiences (Sullivan et al., 1995). Reflecting the (muchreplicated) three-factor structure to emerge from analyses of the PCS, catastrophizing is currently viewed as a multidimensional construct comprising elements of rumination ( I can t stop thinking about how much it hurts ), magnification ( I m afraid something serious may happen ), and helplessness ( There is nothing I can do to reduce the intensity of the pain ) (Sullivan et al., 1995; Sullivan et al., 2001b). Subsequent studies (e.g., Osman et al., 2000; Sullivan et al., 2000) have confirmed that the PCS taps a single construct characterized by three related but theoretically distinct dimensions. The validity of each dimension the attentional, cognitive, and coping-relevant components of catastrophizing likewise rests on empirical abundance. Finally, research has confirmed that the relation between catastrophizing and pain (described below) is independent of other distress-related variables, such as anxiety, depression, and fear of pain (for review, see Sullivan et al., 2001a).

18 18 Reinforcing the theoretical integrity of the fear-avoidance model, several studies provide evidence that catastrophic thoughts may be considered (statistical) precursors to pain-related fear. Vlaeyen et al. (1995b) specifically investigated the predictive power of proposed determinants of kinesiophobia. In a sample of 33 chronic low back pain patients (mean age 37.4 years, mean pain duration 7.6 years), pain catastrophizing, as measured by the Catastrophizing subscale of the Pain Cognition List (PCL-e; Vlaeyen et al., 1990) was found to be superior in predicting pain/movement-related fear (assessed by the Tampa Scale of Kinesiophobia) compared to medical status or pain severity. The PCL-e is a 77-item questionnaire aimed at the assessment of distorted pain cognitions and experienced self control. Current pain severity was measured by a visual analog scale (VAS; Jensen & Karoly, 1992) anchored by no pain at all and the worst pain ever experienced. Catastrophizing added an additional 17% to 15% to the variance simultaneously accounted for by gender and pain duration. Pain intensity and physicianassessed biomedical findings added no predictive value to these accumulated variables. In a similar study (Vlaeyen et al., 1995a), 103 chronic low back pain patients (mean age 40.9 years, mean duration of pain 10.1 years) completed a comprehensive battery of pain and mood related instruments. Additionally controlling for compensation status, a multiple regression analysis replicated the earlier findings. Catastrophic thought has likewise been linked to greater fear intensity in samples of pain-free (Crombez et al., 1998a) and persistently suffering (McCracken & Gross, 1992) volunteers. Fear-Avoidance Beliefs. The term fear-avoidance was first coined by Lethem et al. (1983) in the context of the Fear-Avoidance Model of Exaggerated Pain Perception

19 19 (Buer & Linton, 2002). The model first posited that the two polar responses to pain -- avoidance and confrontation-- determine an individual s prognosis in the face of acute injury. The characteristic adaptive (confrontational) or maladaptive (avoidant) reaction is likely to impact the course of recovery; again, while confronters view pain as a temporary nuisance and are strongly motivated to return to work and normal activity, avoiders pursue a cycle of diminished physical burden and subsequent increasing physical and psychological distress. At what point on this continuum a particular person will exist is determined by his or her fear of pain. These early theoretical propositions have initiated a number of studies aiming to delineate the role that fear-avoidance beliefs play in the development and maintenance of pain and disability in chronic pain populations (Fritz & George, 2002). While the rest of this paper details the relevant results of these studies, it is useful to provide a general overview of the discrete significance of fear-avoidance beliefs in the chronic pain literature. Virtually all such investigations employ the Fear-Avoidance Beliefs Questionnaire (FABQ), an instrument developed by Waddell et al. (1993) on the basis of the Fear-Avoidance Model (Lethem et al., 1983; Fritz et al., 2001). The FABQ, designed to quantify an individual s beliefs about how physical activity and work may affect his or her pain and risk of re-injury, consists of 16 items and is divided into two subscales; fearavoidance beliefs for work (FABQ-W) and fear-avoidance beliefs for physical activity (FABQ-PA). Respective items include the statements I should not do physical activities that (might) make my pain worse and I should not do my normal work with my present

20 20 pain. Items in both subscales are rated on a 7-point scale, with higher numbers indicating elevated levels of fear-avoidance beliefs (Crombez et al., 1999a; Fritz et al., 2001). A number of studies have confirmed the FABQ as a reliable and valid instrument, (e.g., Waddell et al., 1993; Crombez et al., 1999a). Various investigators have demonstrated a strong relationship between higher levels of fear-avoidance beliefs and chronic disability due to low back pain. Correlation coefficients between disability scores and measures of fear-avoidance beliefs in studies involving patients with chronic low back pain have ranged between.37 and.55 (for review, see Fritz & George 2002), with some disagreement over the relative predictive strength of the two component subscales (for review, see Woby et al., 2004a, b). Consistently, FABQ measures have accounted for more functional disability and work loss than such variables as anatomical and temporal patterns of pain or pain severity (e.g., Waddell et al., 1993; Woby et al., 2004a, b; Crombez et al., 1999a). Moreover, studies focusing on acute pain populations have posited fear-avoidance beliefs as an important factor influencing the transition from acute to chronic pain (e.g., Fritz & George, 2002; Fritz et al., 2001; Grotle et al., 2004; Buer & Linton, 2002) and even playing a role in the inception of low back pain episodes (Linton et al., 2000). As a result of these empiricallyreiterated relationships, several authors have advocated the prognostic value of the FABQ in identifying acute low back pain patients at risk for fear-mediated chronic disability. Pain-Related Anxiety. Theory and research suggest that anxiety is an important emotional and physiological concomitant of chronic pain (Burns et al., 2000; Vowels et al., 2004), as well as an integral component of fear-avoidance models. Heightened pain-

21 21 related anxiety is associated with maladaptive coping responses, medication use, and occupational disability (McCracken et al., 1992). While the FABQ taps beliefs about the necessity of avoidance (e.g., Physical activity might harm my back ; McCracken et al., 1996), the Pain Anxiety Symptoms Scale (PASS, McCracken et al., 1992) assesses several dimensions of pain-related anxiety in adherence to more holistic conceptions of pain-related fear. In particular, Lang s Three-System Model asserts that fear is minimally comprised of the subjective cognitive experience, behaviors, and physiological arousal (Vowels et al., 2004). The PASS (McCracken et al., 1992) is a 40-item self-report measure in which respondents rate anxiety related to pain on a 6-point Liker-type scale ranging from 0 (never) to 5 (always). Summation of individual items allows the derivation of a total and four subscale scores. As suggested above, the four subscales describe (a) cognitive anxiety symptoms related to the experience of pain ( I find it hard to concentrate when I hurt ), (b) escape and avoidance responses intended to reduce pain ( I try to avoid activities that cause pain ), (c) fearful appraisals of pain ( Pain sensations are terrifying ), and (d) pain-related physiological anxiety symptoms ( Pain makes me nauseous ; Larsen et al., 1997; Vowels et al; 2004). The measure has demonstrated good validity and reliability across items and administrations (McCracken et al., 1996). Consistent with fear-avoidance models of chronic pain, research employing the PASS with chronic pain samples has demonstrated a pattern of greater pain anxiety relative to matched comparison groups, overprediction of pain intensity, poorer coping responses (e.g., avoidance), positive correlation with measures of general anxiety and

22 22 self-reported disability, and evidence of greater somatic reactivity in anticipation of paineliciting physical movement (Crombez et al., 1999a; McCracken et al., 1996, 1998). Recently, the PASS has also been linked with physical capacity variables (Burns et al., 2000). Kinesiophobia (Movement-related fear of pain/ (re)injury). At the center of the present study, the Tampa Scale for Kinesiophobia (Kori et al., 1990) is a more specific measure of pain-related fear. The 17 items of the TSK are intended to assess fear of movement and (re)injury, and include such statements as, It s really not safe for a person with a condition like mine to be physically active. The TSK has established validity and reliability with chronic pain patients, and has shown to be strongly associated with measures of functional disability (Swinkels-Meewisse et al, 2003). Studies have successfully replicated the proposed two-factor solution for the items of the TSK (Clark et al., 1996; Geisser et al., 2000; Roelfos et al., 2003) across both chronic low back pain and fibromyalgia samples (Goubert et al., 2003). These two factors have been labeled somatic focus (reflecting a belief in underlying serious medical problems) and activity avoidance, which reflects the belief that activity may result in (re)injury or increased pain. Relationship among the pain-related fear measures. The described constructs and their instruments of measurement are often used interchangeably in reference to painrelated fear. Indeed, each maps a different descriptive level of pain-related fear, an umbrella term encompassing very specific (e.g., fear that movement will trigger reinjury) to very general (e.g., globally dysfunctional ) responses. While conceptual

23 23 overlap is apparent, the patterns of correlations to emerge in recent literature have further distinguished the boundaries of each construct. For example, while the measures show significant (but imperfect) intercorrelation (see Table 1), in terms of predictive validity, the TSK and FABQ subscales appear to be more strongly related to self-reported and functional disability than the PASS (Crombez et al. 1999a; Burns et al., 2000). In a multiple regression analysis, Crombez et al. (1999a) found that only the TSK and the FABQ Physical Activity subscale were superior to biomedical variables, negative affect, and catastrophizing in predicting poor behavioral performance in a chronic low back pain cohort. Beside these more-specific measures, the PASS showed no unique contribution in predicting behavioral performance or disability. Furthermore, while the correlations between the TSK and FABQ subscales were higher than correlations between these measures and negative affect, the PASS was more highly correlated with negative affect and pain catastrophizing. As noted by the authors, the pattern of results supports the unique validity of painrelated fear measures; although there were reliable associations between pain-related fear scales and negative affect, only pain-related fear contributed in explaining self-reported disability and behavioral performance. Moreover, while the FABQ and TSK were specifically developed to measure fear of physical movement and injury in patients with musculoskeletal pain, the PASS may reflect a more general apprehension applicable to all forms of pain. Measurement of Pain-Related Fear in the Prediction of Pain, Avoidance, and Disability Pain-related fear and self-reported disability. A number of studies have

24 24 Table 1 Correlation Coefficients (r) between the TSK and Pain-Related Fear Measures Measure Correlation FABQ-Physical 0.57** a 0.76** b FABQ-Work 0.56** a 0.53** b PASS 0.34* c 0.60** d PCS 0.53** c 0.35** d FABQ, Fear Avoidance Beliefs Questionnaire (Waddell et al., 1993); PASS, Pain Anxiety Symptoms Scale (McCracken et al., 1992); PCS, Pain Catastrophizing Scale (Sullivan et al., 1995) a Crombez et al., 1999a (Study 1) b Crombez et al., 1999a (Study 2) c Crombez et al., 1999a (Study 3) d Roelofs et al., 2004

25 25 investigated how pain-related fear and associated escape/avoidance relate to self-reported disability in work and daily life (see Table 2). Waddell et al. (1993) found that fearavoidance beliefs about work and physical activity (as measured by the FABQ) predicted a substantially larger portion of the variance in disability both in work loss and activities of daily living than all available biomedical measures combined. Clinical assessment of pain included anatomical pattern, time pattern, and current severity (assed via visual analog scale). Disability was assessed using the Roland and Morris Disability Questionnaire (RDQ; Roland & Morris, 1983), a 24-item 2-point scale measuring the difficulty of performing daily tasks. In a sample of 184 patients suffering from chronic low back pain and/or sciatica (mean age 39.7 years, mean duration of pain 7.4 years) fear-avoidance beliefs correlated strongly with both daily and work-relevant self-reported disability (r =.23 to.55, p <.01 to.001). Regression analyses confirmed this relationship, as fear avoidance beliefs about work accounted for 23% of the variance in daily living disability and 26% of the variance in work loss, even after allowing for severe pain; fear avoidance beliefs about physical activity explained an additional 9% of the variance in disability. In comparison, pain severity accounted for 14% of disability variance. Strikingly, all of the available biomedical measures combined could only explain 5% of the variance in work loss. As part of a study cited above, using linear regression analysis Vlaeyen et al. (1995b) found that fear of movement/(re)injury as measured by the TSK was a better predictor of self-reported disability levels (RDQ) than patients biomedical status and current pain level. Disability measures correlated significantly with TSK scores (r =.49,

26 26 Table 2 Correlations of Pain-Related Fear and Self-Reported Disability Measures in Chronic Low Back Pain Samples Investigators N Disability measure Pain-related fear measure Correlation Waddell et al., RDQ (self-report) FABQ-Physical FABQ-Work Present/Past year work loss (months) 0.51** 0.55** 0.13/0.23** Vlaeyen et al., 1995b 33 RDQ TSK 0.49** Crombez et al., 1999a 35 RDQ FABQ-Physical FABQ-Work TSK 0.51** ** McCracken et al., PDI PASS Total Cognitive Escape/avoidance Fear Physiological FABQ Total FABQ-Physical FABQ-Work 0.61** 0.51** 0.66** 0.38** 0.51** 0.52** 0.37* 0.48** Grotle et al., ODI FABQ-Physical FABQ-Work 0.21** 0.30** RDQ, Roland Disability Questionnaire (Roland and Morris, 1983); PDI, Pain Disability Index (Pollard, 1984); ODI. Oswestry Disability Index (Fairbank et al., 1980). ** p < 0.01; * p < 0.05

27 27 p <.01); controlling for pain duration and gender, fear of movement accounted for an additional 17% of the variance in disability. In a sample of 35 chronic back pain patients (mean age 36.1 years, mean pain duration 6.7 years) Crombez et al. (1999a) likewise found the TSK and FABQ superior to current pain intensity (VAS) and duration in predicting self-reported disability. The pain-related fear instruments exhibited a stronger correlation with RDQ-rated disability (r =.51 to.56, p <.001) than did pain intensity and a measure of general negative affect, which showed no significant association. Again, controlling for sociodemographic variables, regression analyses corroborated these relationships. Applying the Multiaxial Assessment of Pain taxonomy (MAP; Turk & Rudy, 1988), an empirical approach to the identification of chronic pain subgroups, Asmundson et al. (1997) found that chronic pain patients classified as globally dysfunctional (and thus most disabled) using the Multidimensional Pain Inventory clustering procedure (MPI; Kerns et al, 1985) reported more escape/avoidance behavior, elevated fearful appraisals of pain, and pain-specific cognitive and physiological anxiety compared to those classified as interpersonally distressed or adaptive copers. The MPI is a 52-item self-report measure that taps a wide range of physical, psychological, and social factors involved in individuals pain experience. Of the chronic back pain patients comprising the n=200 sample (mean age 38.3 years, mean duration of pain 2.3 years) globally dysfunctional participants were likewise characterized as experiencing higher than average pain severity, pain interference and affective distress, and lower levels of selfefficacy and general activity. The PASS (McCracken et al., 1992), and the State-Trait

28 28 Anxiety state form (Spielberger et al., 1970) constituted relevant fear/anxiety measures. In a comparison of instruments quantifying anxiety and fear in 45 chronic pain patients, McCracken et al. (1996) showed that pain-specific anxiety measures (i.e., FABQ, PASS) were more predictive of pain, disability, and pain behavior than instruments tapping more general anxiety responses. These more generic tools included the trait form of the Spielberger Trait Anxiety Inventory (STAI; Spielberger et al; 1983) and the Fear of Pain Questionnaire (FPQ; McNeil et al., 1986), a 30-item measure of fear of painful stimuli which includes three subscale scores for minor, severe, and medical pain; a scale of 1 (not at all) to 5 (extreme) qualifies each item. Disability was assessed via the Pain Disability Index (PDI; Pollard et al., 1984), a 7-item self-report measure which describes participants perceived disability in seven areas of daily functioning. Consistent with the preceding studies, conclusions were drawn from an unambiguous pattern of zero-order correlations and hierarchical regression analyses. Assessment of pain-related fear in acute pain populations. Prospective investigations of patients with acute low back pain likewise reveal pain-related fear as a significant predictor of pain, disability, and work status, supporting the idea that fear is a precursor to disability, rather than a consequence. Using a large sample of acute low back pain patients in the primary care setting, Klenerman et al. (1995) concluded that fearavoidance variables were the most relevant in predicting chronicity one year later. By assessing participants at the acute (no more than one-week post onset of pain), two, and twelve month stage following onset of back pain, the principal aim of the study was to map the natural history of back pain, with emphasis on the contribution of fear-of-pain

29 29 and avoidance factors to the development of dysfunctional chronicity. Underlying the study was the question of whether psychological variables/ comorbidity are a consequence of back pain or potential contributors to it. Of the three hundred patients initially recruited for the study, information from a core group of 123 participants was included at each point of data collection. Fear-avoidance contextual measures were based on the four psychosocial factors hypothesized by the Fear Avoidance Model (Lethem et al., 1983) to determine a confrontational or avoidant coping style. Thus, assessment tapped stressful life events (Social Readjustment Rating Scale; Holmes & Rahe, 1967), characteristic personality patterns (Modified Somatic Perception Questionnaire MSPQ, Main, a measure in which the participant must qualify the frequency and severity of 13 somatic symptoms), use of active vs. passive coping strategies in the context of internally originating pains, and previous pain history along three categories of pain experience (externally-produced, internally-produced, and accidental pain). At the two month follow-up (n=162), fear-avoidance variables collected at the acute stage accounted for 25% of the variance in outcome; importantly, analyses revealed that those who had not recovered by 2 months (7.3% of the core sample) went on to become chronic low back pain patients, with unremitting and even increasing back pain. Results of multiple discriminant function analyses suggested that the future course of back pain can be correctly classified in 66% of patients from fear-avoidance variables alone and in 88% when all variables are employed these included various physical measures and psychosocial (demographic and pain history) variables. In a study comparing early-stage (less than three weeks; n=123) and chronic

30 30 (exceeding one year; n=233) low back pain patients, Grotle et al. (2004) found that fearavoidance beliefs were significantly related to disability in both groups after adjusting for sociodemographic, pain, and clinical variables. Current pain intensity in lower back and limb was scored on a visual analogue scales ranging from 0 (no pain) to 100 (worst possible pain). Version 2.0 of the Oswestry Disability Index (ODI; Fairbank et al., 1980) assessed activity limitations by gauging 10 different items (e.g., pain intensity, personal hygiene, sleeping, social and sexual activity). The FABQ accounted for fear-avoidance beliefs. All assessments were made prior to consultation with a physician. While patients with chronic low back pain reported significantly more fear-avoidance beliefs, distress, and activity limitation due to pain than the acute sample, regression analyses within each group and in the merged sample revealed that the associations between avoidance beliefs and disability were equally strong in both groups. Importantly, the results suggest that fear-avoidance beliefs are already present in patients prior to their first consultation with a doctor. Focusing on acute work-related low back pain, Fritz et al., (2001, 2002) found that, in a sample of 78 participants with less than 3 weeks since pain onset (mean age 37.5 years, mean pain duration 5.5 days), higher levels of fear-avoidance beliefs were associated with more persistent occupational disability (i.e., returning to work full status). In the first study (Fritz et al., 2001), the FABQ work and physical activity subscales were significantly correlated with initial pain and disability scores (r =.40 to.53, p <.01, r =.34 to.40, p <.05 for pain and disability, respectively). Disability was again assessed by the Modified ODI (Fairbank et al., 1980). Following 4 weeks of standardized

31 31 rehabilitation treatment, the FABQ work subscale explained the greatest amount of variability in disability scores (r 2 = 21%), exceeding even the initial disability measures. Fear-avoidance beliefs about work likewise significantly improved regression models accounting for disability and work status outcome, even after controlling for initial levels of pain, physical impairment, disability, and type of treatment received. Applying receiver operator characteristics and logistic regression analysis to the data collected in the above study, Fritz and George (2002) again identified FABQ-W as the strongest predictor variable of restricted work status after four weeks of physical therapy. Swinkels-Meewisse et al. (2003) found that fear of movement/(re)injury as measured by the TSK, together with pain intensity, significantly predicted disability and daily activity participation in a sample of 615 acute low back pain patients. Acute low back pain was defined as less than 4 weeks duration. Disability was measured using the RDQ while the five-item participation measure tapped work/housekeeping, sport, leisure, and social or family activities. Both constructs comprising the TSK were found significantly associated with pain and disability (r =.18 to.38, p <.01). The results of acute low back pain studies are summarized in Table 3. Pain-related fear in the prediction of future pain. Notably, recent studies provide evidence that pain-related fear may be an important predictor of future pain episodes and disability in the general population. Linton et al. (2000) found that, in a sample of painfree individuals, elevated fear avoidance beliefs (as measured by the FABQ) doubled the risk of experiencing spinal pain in the following year. In order to examine the prevalence and prospective effects of fear avoidance beliefs in a non-clinical population, 415 people

32 32 Table 3 Correlations of Pain-Related Fear and Self-Reported Disability Measures in Acute Pain Samples Investigators N Disability Pain-related Correlation measure fear measure Grotle et al., ODI FABQ-Work 0.25** Fritz et al., modi Initial ratings FABQ-Work FABQ- Physical 4-week ratings FABQ-Work FABQ-Physical 0.40** 0.34* 0.46** 0.37* Swinkels- Meewisse et al., RDQ TSK Harm TSK Activity avoidance 0.30** 0.38** RDQ, Roland Disability Questionnaire (Roland and Morris, 1983); ODI, Oswestry Disability Index (Fairbank et al., 1980).

33 33 (mean age 43 years) who reported no spinal symptoms in the previous year underwent initial assessment and a 12 month follow-up. At the follow-up, 19% (78 of 413) of the sample reported experiencing an episode of back pain. Moreover, while the initial correlation between physical functioning and the psychosocial measures was weak (r = -.08, p >.05 and r =.13, p <.05, for the FABQ and PCS scales, respectively), logistic regression analyses on the median split of the FABQ scores demonstrated a significant two-fold increase in the risk of developing back pain in participants with elevated fear-avoidance belief scores. In a similar fashion, risk of lowered physical functioning increased 1.7 times above the FABQ median. Risk analysis of catastrophizing revealed a less significant effect for both pain occurrence and physical function (adjusted odds ratio (OR) = 1.5, with less than adequate confidence intervals). Importantly, for those who went on to experience back pain, the correlation between the FABQ and catastrophizing measure rose from.26 (p <.05) to.42 (p<.01) at posttest; these values remained relatively low for those who remained pain-free (p <.01) at pretest compared to.25 at the posttest (p <.01). Because the study could not control for the effects of prior (non-spinal) pain experience, the authors suggest that fear-avoidance beliefs may be an important element in the interactive learning process implicated in the development of a pain problem, rather than a directly causal factor. While the study demonstrates that fear-avoidance beliefs do occur among the general healthy population, the strengthened association between catastrophizing and fear-avoidance in an initially pain-free sample (as compared with more powerful associations exhibited in chronic pain populations) likely reflects a similar dynamic sequence.

34 34 Using a sample of the general Dutch population, Picavet et al. (2002) found that a high level of catastrophizing or kinesiophobia increased the risk of future chronic low back pain and disability. The examined cohort included 1, 571 citizens aged 25 to 65 who did not suffer from severe forms of illness (e.g., serious heart disease, cancer, rheumatoid arthritis). Questionnaires administered at baseline included the PCS and TSK; several features low back pain were assessed both at baseline and at a 6-month follow-up. Results followed the striking pattern of associations displayed in the previous study. Among the 1,160 participants without low back pain at baseline, a high level of kinesiophobia (highest TSK tertile; n=361) predicted low back pain and disability at follow-up (OR = 3.4, 95% CI: 1.6, 8.7). Elevated catastrophizing scores (highest PCS tertile; n=345) were predictive of severe low back pain (rated greater than 5 on a scale of 1 to 10; OR = 2.2, 95% CI: 1.0, 5.0), chronic low back pain (exceeding 3 months; OR = 2.1, 95% CI: 1.1, 3.9), and low back pain with disability (OR = 3.1, 95% CI: 1.1, 8.7), as determined by the Quebec Back Pain Disability Scale; Kopec et al., 1996 a questionnaire consisting of 20 activities such as walking, carrying a suitcase etc. For the 411 participants experiencing low back pain at baseline, a high level of kinesiophobia (highest tertile TSK score; n=161) predicted existing low back pain at follow up (OR = 1.6, 95% CI: 1.0, 2.7) and chronic low back pain (OR = 1.6, 95% CI: 1.0, 2.7). Elevated TSK scores were especially predictive of low back pain limitation in daily activities (OR = 3.6, 95% CI: 1.9, 6.7), severe low back pain (OR = 3.0, 95% CI: 1.8, 5.1), and low back pain with disability (OR = 4.4, 95% CI: 2.5, 7.9). High catastrophizing showed a similar pattern of associations. These significant associations remained after adjustment for pain

35 35 duration, pain severity, or for disability at baseline, and were observed to be in the same direction across gender, age, and education-sorted SES subgroups. Pain-related fear and prediction of physical performance. As posited by the Fear- Avoidance Model (Vlaeyen et al., 1995a, b) and identified empirically in studies reviewed thus far, the factor mediating pain and disability, or acute and chronic pain, is the behavioral option of avoidance. Accordingly, research conducted with individuals experiencing chronic low back pain explicitly links pain-related fear with escape and avoidance of physical activities resulting in impaired behavioral performance. Detailed studies are summarized in Table 4. Vlaeyen et al. (1995a) asked thirty-three chronic low back pain patients (mean age 42.4 years, mean duration of pain 10.3 years) to lift and hold a 5.5 kg weight with their dominant arm for as long as possible. They found a substantial negative correlation between fear of movement/(re)injury as measured by the TSK and behavioral performance as measured by lifting time (r = -.44, p <.01). When the sample was divided into high and low fear responders (according to median TSK score of 37), the difference in mean lifting scores between the two groups was likewise significant (t = 2.63, p <.05). Patients reporting more kinesiophobia avoided the motoric activity to a greater extent than patients with lower TSK scores. The study also found a significant moderate correlation (r =.52, p <.001) between the TSK and a visual analog scale measuring fear of (re)injury administered immediately after the lifting task. The fear VAS was anchored with the terms I am not afraid to re-injure myself and I have never been so afraid to reinjure myself. Finally, significant positive correlations were identified between the

36 36 Table 4 Correlations of Pain-Related Fear and Behavioral Performance Measures Investigators N Behavioral performance Pain-related Correlation measure fear measure Vlaeyen et al., 33 Lifting a 5.5 kg weight TSK ** 1995 Crombez et al., 1998a LBPQ Fear of pain 49 Knee-extension- flexion unit (Cybex 350 system) Peak torque Performance variability Work ratio Peak toque Performance variability Work ratio Fear of reinjury * 0.38** * 0.33* 0.39** Crombez et al., 1999a Al-Obaidi et al., Trunk-extension-flexion unit (Cybex 350 system) Peak torque Lifting a 5.5 kg weight 63 Medx lumbar extension machine measure of isometric torque of back muscles TSK FABQphysical FABQ-work TSK PASS FABQphysical (FABQ-work all NS) ** ** ** ** ** ** ** ** ** TSK, Tampa Scale for Kinesiophobia (Kori et al., 1990); LBPQ, Leuven Back Pain Questionnaire (Crombez et al., 1998); FABQ, Fear Avoidance Beliefs Questionnaire (Waddell et al., 1993); PASS, Pain Anxiety Symptoms Scale (McCracken et al., 1992).

37 37 TSK and the State Trait Anxiety Inventory (STAI, Van der Ploeg et al., 1980), with a moderate negative difference between high and low TSK responders. Importantly, pain intensity ratings were not predictive of fear of movement/(re)injury, highlighting that such fear occurs independently of current nociception. Besides the small sample size, one essential limitation of the study was the potential confounding of fear scores with pain duration (and, by extension, pain intensity); high TSK responders reported a significantly longer duration of pain complaints than lower responders (t=2.51, p =.017). As a consequence, it cannot be ruled out that the inhibited performance of the highly fearful patient was due to more extensive experience with severe pain rather than to avoidance. In order to clarify this issue, and to generalize avoidance tendencies beyond backstraining activity, a similar study by Crombez et al. (1998) employed a behavioral task (knee-extension-flexion) believed by participants to be minimally back straining. Based on response to a survey designed for the study, forty-nine patients suffering from nonspecific chronic musculoskeletal back pain (mean age years, mean pain duration 6.2 years) were classified as avoiders or confronters. From a seated position, participants were required to push a bar attached to the thigh until the knee was extended (extension) and then pull the bar back downward (flexion) until he or she could no longer continue. The protocol repeated for each leg consisted of a maximum 20 consecutive extensions and flexions.. Using a verbal graphical rating scale (GRS; Jensen & Karoly, 1992) which offers the adjectives not, weak, moderate, strong, and very strong at equal distances along a 100 mm line, participants rated their back pain following a practice

38 38 session, as well as immediately prior to and following the behavioral test session. GRS ratings of fear of (re)injury were also obtained following the test. Behavioral performance was reflected in peak torque, variability of muscle strength within a series, and a calculated work ratio. A significant association between behavioral performance and pain-related fear but not between performance and pain intensity supported the idea that poor performance was due to escape/ avoidance rather than increased back pain. Moreover, in further support of the fear-avoidance model, avoiders reported higher fear of pain (t=5.05, p =.001) and (re)injury (t=4.32, p =.001), more frequent pain (t=1.96, p =.05), more daily disability (t=3.21, p =.05), fatigue, and longer recovery periods (t=2.55, p =.01) than confronters. However, no significant differences in pain intensity were found between the two groups. In relevance to the design of the present study, the pattern of pain reports during the behavioral test reflected an initial overprediction of pain followed by swift correction during the next series. Crombez et al. (1999a) subsequently replicated and extended these findings. The first study employed an isokinetic trunk-extension-and-flexion test, which requires participants to push a bar placed across the chest until the back is maximally flexed and then pull the bar back until maximum extension; this is assumed to assess the functional capacity of the trunk flexors and extensors. Thirty-eight chronic back pain patients (mean age years, mean pain duration 6.35 years) were requested to consecutively flex and extend three times with maximal speed and force. Measures of pain-related fear the TSK and the physical subscale of the FABQ were the most consistent predictors of behavioral performance (R 2 =.56, p <.05; R 2 =.57, p <.05, respectively) from among a

39 39 number of variables, including pain duration and intensity. A second study using 31 chronic back pain patients (mean age years, mean pain duration 10.1 years) and the same lifting paradigm as Vlaeyen et al. (1995) demonstrated as well that pain related fear (TSK) was superior in predicting behavioral performance in comparison with other painrelevant variables, such as pain intensity and experienced pain increase (measured by visual analog scale; R 2 =.79, p <.001). In a sample of 104 outpatients at a multidisciplinary pain management center (mean age 45.0 years, mean pain duration 5.2 years), the majority of whom presented with back pain, McCracken et al. (1992) found a significant negative correlation between pain-related fear as measured by the PASS and range of motion as measured by a flexometer during a passive straight leg raise test. During this routine physical examination, a physical therapists lifted the supine patient s leg until the patient indicated that the maximum tolerated raise had been achieved. Three such trials were performed on each leg. Al-Obaidi et al. (2000) found that fear-avoidance beliefs about physical activity, rather than actual experienced pain, significantly accounted for variation in spinal isometric strength deficit observed during a lumber extension test. Sixty-three patients experiencing a minimum 7 weeks chronic low back pain participated in a behavioral protocol that included measurement of maximal voluntary isometric contraction of the lumbar extensor muscles at 0, 12, 24, 36, 60, and 72 degrees of lumbar flexion. VAS ratings of anticipated and experienced pain were collected prior to and following the test at each angle. Anticipation of pain and fear-avoidance beliefs (assessed by the FABQ-

40 40 physical activity subscale) showed a significant inverse relation to isometric strength tested at all angles (r = to -.520; p <.05 to.01). Following pain-anticipation in a stepwise regression analysis, fear-avoidance beliefs about physical activity explained an additional 12% of the variance at the various tested angles. In sum, a strong correlation has been observed between psychosocial variables indicative of fear-avoidance beliefs, behavioral performance, and disability. While the results appear persuasive, a number of factors qualify the tide of data. The majority of the studies are cross-sectional in design (e.g., Crombez et al., 1998, 1999a; McCracken & Gross, 1993); it is thus impossible to draw conclusions about the direction of the observed associations. As always, strong correlations or regression weights cannot be confused with causal effects (Waddell et al., 1993; Vlaeyen et al., 1995; Swinkels- Meewisse et al., 2003; Grotle et al., 2004). Logically, it is possible that pain-related fear occurs secondary to the experience of low back pain, or that it is a prime determinant of becoming a chronic back pain patient. Equally likely is the possibility that pain catastrophizing and pain-related fear are markers for other variables that have not (adequately) been assessed, such as the all-important history of back pain (or other type of pain) at baselines, which may increase the likelihood of avoidance and disability. Similarly, the supposed predictor and outcome variables may be related to a third overarching factor (e.g., traumatic experience). Some researchers (e.g., Roland and Morris, 1983) have adamantly maintained that emotional distress in pain patients is more likely a consequence rather than an antecedent of chronic pain problems. Research suggests that it is premature to draw such firm theoretical distinctions. Patterns of

41 41 correlations tend to support a dynamic evolving relationship between one s pain and one s (possibly dispositionally-guided) response to it, leading some authors to note that psychological variables can act both as antecedents and consequences, reinforcing the pain problem in a complex chain of events with feedback loops (Vlaeyen et al., 1995, p. 248). Despite these limitations, including sometimes a modest sample size, the results of independent studies support the same conclusions with both univariate (correlations) and multivariate analyses (regression) pointing to the central importance of pain-related fears. Moreover, recent prospective/longitudinal designs such as those by Klenerman et al. (1995), Picavet et al., (2002), and Linton et al. (2000) support the idea that painrelated fear is a precursor to and powerful predictor of disability, rather than its consequence. Discussed later in this paper, the success of treatments based on the hypothesized sequence of factors laid out in the Fear Avoidance Model also adds support to its validity. Additional potential limitations include a lack of homogeneity across studies regarding the characteristics of the sample involved. In particular, some studies included individuals who have sustained work-related injuries and thus qualify for disability compensation. Research has shown that compensation status may affect outcome in the form of pain reporting, depression, and disability in chronic low back pain patients (e.g., Rainville et al., 1997). Authors caution about generalizing results to non work-related back injury (e.g., Fritz & George, 2002), particularly regarding the predictive power of work versus physical-activity subscales on measures of pain-related fear such as the

42 42 FABQ. Moreover, some studies were conducted in countries that lack workers compensation and litigation for chronic back pain (Al-Obaidi et al., 2000, 2001), further highlighting the population-specific validity of certain instruments. Finally, the preceding analyses were based almost exclusively on self-report data which, ideally, should be checked or corroborated by externally validated measures, such as physiological or kinetic parameters. It is also worth noting that the majority of the cited studies, especially those focusing on acute low back pain, explicitly advocate the early assessment and screening of psychosocial issues relevant to the development of longterm pain problems. The Role of Hypervigilance Hypervigilance is defined as an unintentional and goal-dependent attentional process, emerging when one s current goal is related to avoidance and escape from pain (Crombez et al., 2005; Eccleston & Crombez, 1999). Charting the evolution of pain and disability, the fear-avoidance model posits hypervigilance as a process mediating the mutually-reinforcing tendencies toward catastrophic and kinesiophobic thought and similarly reciprocal disuse and avoidance patterns. In that somatic rumination occurs at the expense of other cognitive (and consequently physical and social) demands, hypervigilance is considered a dysfunctional attentional process. As a signal of potential tissue damage, acute pain serves an essential adaptive function, urging cessation of activity and escape (Crombez et al., 1999b). The basic attentional demand of intense pain has been well-documented in both clinical and nonclinical populations (for review see Eccleston & Crombez, 1999; Pincus & Morely,

43 ). Fundamentally, such studies demonstrate that painful stimuli intrude onto ongoing attentional engagement; this disruption is particularly resistant to displacement and relegation (e.g., Crombez et al., 1996a, 1997; Eccleston et al. 1995a, b). Studies manipulating the threat value of stimuli have shown that anticipation of imminent pain appears to effect preparatory responses; specifically, priority access into focal awareness is assigned to stimuli signaling the object of threat (Eccleston & Crombez, 1999; Crombez et al., 2005; Ohman, 1979). Moreover, such prioritization of pain is exaggerated by the presence of catastrophic thought (Crombez, 1998b; Vlaeyen & Linton, 2000; Eccleston et al; 1997); for example, high catastrophizers report significant difficulty in suppressing or diverting attention from pain-related thoughts (Sullivan et al., 1995). With the advent of the fear-avoidance hypothesis, empirical focus has increasingly been directed at the association between hypervigilance and pain-related fear. Crombez (1999b) reported that, in a sample of forty chronic pain patients, attentional intrusion on a numerical interference test was best predicted by the interaction between pain severity and pain-related fear, as assessed by the TSK, with fear exacerbating attentional interference produced by the chronic pain condition. In a dotprobe paradigm, healthy individuals scoring high on the Fear of Pain Questionnaire (FPQ; McNeil & Rainwater, 1998) showed greater attentional bias toward pain-related stimuli than their low-fear counterparts (Keogh et al., 2001). No group differences were found for socially-threatening or positive stimuli, leading the authors to suggest that the observed biases were indeed pain-specific, rather than due to general negativity or emotional predisposition. Citing the fear-avoidance model, the study concluded that the

44 44 preferential attentional processing of pain-related materials is not just a product of chronic pain states (Keogh et al., 2001, p. 97) but an important vulnerability underpinned by more fundamental cognitive mechanisms. Finally, analyses of a successful pain-management program specifically at maladaptive beliefs and behaviors in forty-two chronic pain patients, led Delghani et al. (2004) to conclude that, in the clinical setting, modification of attentional biases is most likely facilitated by reductions in movement-related fear. Particularly relevant to the rationale underlying the current research, recent investigations have identified a finer dimension of hypervigilance in highly fearful chronic pain patients, the inability to disengage from pain cues interpreted as threatening. In a cross-modal cueing paradigm, Van Damme et al. (2002) instructed participants to detect targets (auditory tone or electrocutaneous pain stimuli) as quickly as possible. Each target was preceded with the cue words PAIN or TONE. (Following the manipulation, self-report assessments indicated that participants believed the cues to be predictive of the corresponding target.) Irrespective of stimulus type, detection reaction time was facilitated by a stimulus-congruent cue (operationally defined as engagement); incongruent information slowed down detection (disengagement). Importantly, when pain was cued and did not occur, difficulty in disengagement from the pain cue was particularly pronounced and extended across time in participants high in catastrophic thinking about pain. Moreover, the effect of catastrophic thought on the retardation of disengagement from the pain cue was mediated by a strong and unsubstantiated belief that the pain target was contingent on the pain cue. This covariation bias appeared to be

45 45 pain specific, in that it was related positively to catastrophic thinking about pain, and not to more generic (trait and state) measures of anxiety. Catastrophic thought was likewise unrelated to the belief that tone cues predicted tone targets. Equally important is the authors interpretation of these findings. Specifically, Van Damme et al. (2002) suggested that catastrophic thinking about pain may operate via protection of the belief that the cue for pain is a valid one, despite repeated evidence to the contrary. Put another way, catastrophizers appear to be less sensitive to disconfirming evidence regarding the predictive value of pain cues. As will be elaborated, this resistance to learning forms the basis of the rationale for the present project. Clinical Management of Pain-Related Fear and Avoidance Behavior How do chronic low back pain patients recover from avoidance behavior? As avoidance behavior occurs in anticipation of, rather than in response to pain, opportunities to receive corrective feedback regarding maladaptive fears are limited. Drawing on work conducted in the treatment of phobias, Philips (1987) suggested that chronic pain patients should be exposed to fear-producing activity in order to disconfirm inaccurate expectations regarding the pain experience. As a combination of classical conditioning and cognitive-behavioral treatment, graded exposure (in-vivo) involves the establishment of an individual hierarchy of pain related fear stimuli; subsequently, fear is activated and catastrophic expectations are challenged and disconfirmed, reducing the threat value of the originally fearful stimulus (Roelofs et al., 2002). Participants in exposure-oriented treatment are explicitly provided with an explanation of the fearavoidance model; individual symptoms, beliefs, and behaviors are used to illustrate the

46 46 self-perpetuating cycle of pain catastrophic thought pain/ movement related fear avoidance/ disability/ depression pain. An increasing number of experimental and clinical studies demonstrate the utility of exposure in reducing pain-related fear, catastrophizing, and disability in fearful chronic low back pain patients. Using a replicated single-case cross-over design with four highly fearful chronic low back pain participants (TSK > 40), Vlaeyen et al. (2001) demonstrated that gradual exposure to idiosyncratically selected physical activities was more effective than a graded activity paradigm based on operant shaping principles in extinguishing fear-avoidance beliefs. Using the Photograph series of Daily Activities (PHODA; Kugler et al., 1999) a standardized procedure in which patients are requested to judge the threat value of 98 common physical movements, participants concretely represented treatment goals with a hierarchy of fear-eliciting situations. After a notreatment baseline, participants were randomly assigned to one of two interventions. In intervention A, they received exposure followed by graded activity; in intervention B, the sequence of interventions was reversed. Time series analysis of daily measures regarding pain-related cognitions and fears revealed improvements only during exposure treatment, irrespective of treatment order. Analysis of pre- and post-treatment differences showed the same pattern of decline in pain/movement-related fear (TSK, PHODA), with concurrent decreases in catastrophizing (PCL) and disability (RDQ). Similar studies (Vlaeyen 2002a, b) provide further evidence for the effectiveness of exposure in-vivo with this subgroup of back pain sufferers. Using an identical AB, BA paradigm with six chronic low back pain patients reporting high levels of kinesiophobia

47 47 (TSK >40), Vlaeyen et al. (2002a) supplemented the standardized measures of painrelated fear, catastrophizing, and disability with assessment of pain vigilance (Pain Vigilance and Awareness Questionnaire; PVAQ) and ambulatory monitoring. Time series analysis again supported the unique and order-irrelevant effectiveness of exposure treatment; moreover, pre- to post-treatment differences in pain-related fear were accompanied by a decline in vigilance and disability, and an increase in physical activity levels. Self-reported improvements (TSK, RDQ, PVAQ) remained at a one year followup. A study employing exposure/ behavioral experiments in a randomized AB design with two highly fearful chronic pain patients (Vlaeyen et al., 2002b) supported a similar pattern of results. In both cases, dramatic changes in pain-related fear and catastrophizing, and clinically-significant improvements in vigilance and disability, coincided with a slower-paced decrease in pain intensity. A similar study by Linton et al., (2002) in which two chronic low back pain patients exhibited marked decrease in painrelated fear and functional improvement following exposure in-vivo treatment, extended the generality of these optimistic results to a different setting and treating therapist. Promoting generality across study design, Boersma et al. (2004) employed a single subject, multiple baseline paradigm to investigate the short-term effects of exposure in six patients with high levels of pain-related fear and avoidance (TSK, PHODA). Again, the results demonstrated clear decreases in rated fear and avoidance beliefs while function increased substantially; gains were maintained at a three-month follow-up.

48 48 Importantly, the exposure intervention administered in three of the above studies (Vlaeyen et al., 2001, 2002a, b) was embedded in a comprehensive treatment program which followed an operant treatment regimen and included a multidisciplinary treatment staff. The supplemental value of this context and its many constituent elements cannot fully be ruled out as contributing to the study effects. Despite the relevance of the crossover design and appropriate analyses, the studies described here are limited by their small sample size and need to be supplemented with larger randomized controlled trials. Moreover, while ambulatory monitoring (Vlaeyen 2002a) supported an immediate generalization to the home environment, the one long-term follow-up outside of the potential safety net afforded by study participation (Boersma et al., 2004; Vlaeyen et al., 2002) relied on self-report data subject to a host of confounding variables (e.g., demand effects). Some indication of this limited generalizability is provided by additional analyses performed by Linton et al. (2001) on participants PHODA scores. For each participant, larger reductions in fear occurred for those movements directly selected for exposure training from the constructed fear-hierarchies, relative to those which were not directly included in treatment. Finally, all the studies purposefully recruited participants with very high levels of self-reported kinesiophobia; while the observed results are impressive, it remains unclear what effects exposure might have on individuals with less extreme, but nevertheless significantly incapacitating, levels of pain-related fear. Experimental Exposure and the Restricted Effects of Generalization Outside the clinical setting, a related line of experimental research has found that exposing back pain patients to stressing physical activities leads to a swift correction of

49 49 overprediction regarding pain and fear of (re)injury. In a sample of 42 chronic low back pain patients (mean age years; mean duration of pain 1.6 years) McCracken et al. (1993) found that participants high in pain-related anxiety (PASS) tended to overpredict pain and anxiety on early trials of a passive straight leg-raise test but that predictions became more accurate with experience. A total of six exposure trials (three trials per each leg) were administered; each trial was preceded by expected and experienced pain and anxiety ratings. Crombez et al. (1996b) exposed 29 chronic back pain patients (mean age 38 years, mean pain duration 5.8 years) to a physical performance test consisting of three trials flexing and extending the knee with maximal force over four exercise bouts (two with each leg). Prior to the first exercise bout, participants rated their expected pain using a 100 mm verbal graphical rating scale (GRS) employing the equidistant adjectives not, weak, moderate, strong, and very strong ; after each exercise bout, participants rated the maximal back pain they had experienced as well as the experienced fear of movement and (re)injury. Importantly, the study did not allow comparison of responses at different levels of pain-related fear, measured as an individual difference. Moreover, as no subjective expectations of harm or (re)injury were collected prior to each exercise session, it is impossible to examine concurrent changes in this variable as a function of exposure; the recalled experience of fear is irreparably biased by the experience of exposure. Overpredictions of pain during the first exercise bout were corrected during the second exercise bout with the same leg, while experienced pain remained unchanged. However, this correction did not completely generalize to the trial with the other leg, as small increases in pain expectancy returned.

50 50 This latter result is in line with similar studies raising doubts about the generalizability of exposure effects. Crombez et al. (2002) had 37 individuals with chronic low back pain (mean age 43.2 years) perform two variants of movements selected from the Movement and Pain Prediction Scale (MAPPS, Council et al., 1988): touching toes from a standing position and straight leg raise from a supine position. Participants executed both movements twice; the order of the movements (AABB vs. BBAA) was counterbalanced across participants. Prior to the performance of each movement, a measure of baseline and expected pain was collected using an 11-point numerical rating scale; the same scale was used to assess experienced pain immediately after the performance of each movement. Self-reports regarding expected harm (baseline, expected, and experienced harm) were collected in tandem in a similar manner. The TSK, PCS, the Negative Affect (NA) subscale of the Positive Affect-Negative Affect Scale (PANAS; Watson, Clark, & Tellegen, 1988) were administered to measured the relevant constructs. Results indicated that, for high catastrophizers (scoring above the PCS median of 20.5), an initial overprediction of pain on the first trial of the first movement, t(26)=3.30, p<.01, was readily corrected on the second trial of the same movement. This exposure effect did not generalize toward the second movement, as participants again exhibited a significant overprediction during the first trial of Movement 2, t(26)=2.64 p<.01. Unexpectedly, the study did not find a significant association between overprediction of pain and TSK scores, despite the significant positive correlation observed between the PCS and TSK (r =.58, p <.01). Moreover, none of the individual difference variables were significantly related to overprediction of perceived harm.

51 51 Employing the same movements and paradigm but focusing exclusively on pain assessments, a similar study by Goubert et al. (2002) likewise found that high catastrophizers correction of inaccurate pain expectancies did not generalize from the first movement to the second. The sample of 39 low back pain patients (mean age years, mean pain duration 7.8 years) completed the TSK, PASS, PCS, the Trait version of the State-Trait Anxiety Inventory (STAI; Spielberger et al., 1983), as well as the Multidimensional Pain Inventory Part 1, which assesses pain severity, interference in daily life due to pain, perceived life control, affective distress, and social support. (MPI- DV; Louseberg et al., 1999).The pattern of results reiterated those obtained by Crombez et al. (2002): as indexed by the difference between mean expected and experienced pain, high catastrophizers displayed a significant overprediction of pain expectancies during the first trial of the first movement, t(12)=2.72, p<.001. This difference was no longer significant in the second trial of the first movement, t(12)=0.71, ns., indicating a correction of overprediction. Lack of generalization was again evidenced by a significant overprediction on the first trial of the second movement, t(12)=2.52, p<.05, followed by accurate expectancy (correction) on the second trial of Movement 2 (i.e., a nonsignificant difference between expected and experienced pain, t(12)=1.72, ns.). In this study, a similar pattern of associations was found for the effect of fear of movement/(re) injury (TSK); the failure of these effects to reach significance was attributed to a shortfall in statistical power. TSK and catastrophizing scores were significantly correlated in both studies; in both studies, additional individual difference measures were nearly unrelated to the overprediction of pain. Neither study found evidence of overprediction in the low catastrophizing group.

52 52 While the above-described studies have potentially far-reaching clinical implications, methodological limitations qualify interpretation. As acknowledged by both investigations (Crombez et al. 2002; Goubert et al., 2002), small sample sizes dampen statistical power. Moreover, both studies excluded participants unable to carry out the exercises due to especially high levels of pain-related fear and possible injury. For reasons not explicitly defined, Crombez et al. (2002) excluded four participants while Goubert et al. (2002) excluded eight, reporting their exact fear complaints. Since both studies administered questionnaires after completion of the movement protocol (to be completed at home and mailed back), no TSK or PCS scores were available for analysis. In a small sample, the omission of this input could have impacted the observed relevance of TSK scores. Moreover, while completion of pain and injury - related measures prior to engaging in the prescribed movements might prime participants responses toward fearful wariness, home-based completion implies bias toward the other extreme. For example, it is possible that in the safety of the home environment and with no impediments to familiar avoidance behavior, TSK and PCS scores may not as accurately represent their intended constructs as they would in an impending threatening situation i.e., movement. Performance of the movement itself may bias answers, possibly in a manner reactive to underlying anxieties. Finally, as there was no standardization of the time when questionnaires were to be filled out, a host of situational factors remains uncontrolled. For example, a participant s response to the TSK on the way home from successfully fulfilling a challenging movement task may differ from responses provided after a backstressing day at work. Daily changes in pain-related fear measures following exposure

53 53 applied during in-vivo studies testify to this possibility. With the limited information derived from descriptions of these two experiments, these possibilities remain speculative. Importantly, both studies employ highly standardized movement tasks, further standardized by devices constructed to urge movements toward given parameters. Unlike the PHODA-derived fear hierarchies employed in exposure treatment studies, Goubert et al. (2002) and Crombez et al. (2002) recognize the potential lack of ecological validity for their employed paradigm. Participants may not have associated these movements with the threatening motions avoided in everyday routine. No validation test for personal relevance was conducted. Moreover, as compared to the in-vivo studies --which selected participants from rehabilitation settings-- Goubert et al. (2002) note the relatively low pain severity and distress levels (MPI) of their current sample. In light of these limitations, the inevitable lack of contextual control in the clinical exposure studies makes it difficult to ascertain the specific contribution and generalizability of exposure therapy. Despite the methodological shortcomings of these studies, a recent investigation employing an innovative vignette methodology found support for the relationship between pain-related fear and the non-generalization of corrective experiences relevant to lower back pain. Goubert, Crombez, and Danneels (2005) had 85 patients suffering from nonspecific chronic low back pain (mean age 40.4 years, mean pain duration 6.8 years) reply to a series of vignettes based on the Cognitive Errors Questionnaire (CEQ; Lefebvre, 1981), tapping dysfunctional catastrophizing, overgeneralization, and selective

54 54 abstraction related to both general life and low back pain-specific experiences. Additionally, three vignettes assessing the reluctance to generalize corrective low back pain experiences were developed and confirmed for appropriate validity and reliability. Participants also completed a battery of questionnaires, including the PCS, TSK, and the MPI-Part 1 as a measure of pain severity and interference in daily life activities due to pain. Patterns of correlations supported a significant relationship between lack-ofgeneralization and PCS and TSK scores (r =.38, p <.0005, and r =.47, p <.0005, respectively). Regression analyses with TSK scores entered in the last step revealed that pain severity and fear of movement significantly contributed to the prediction of nongeneralization. Fear of movement explained an additional 27% of the variance, F (1, 73) = 28.62, p < Separate analyses for pain catastrophizing showed the PCS measure accounting for an additional 25% of the variance in lack of generalization, F (1, 73) = 25.95, p < A high frequency of catastrophic thinking about pain and high fear of movement/(re)injury thus explained the lack of corrective generalization over and above the contribution of pain severity. An intriguing finding was that pain catastrophizing and kinesiophobia were significantly related to low back pain overgeneralization (r =.49, p <.0005, and r =.42, p <.0005), indicating that once participants establish that a movement hurts, they remain convinced that it will remain painful in the future. Goubert et al. (2005) refer to this as an asymmetry in the generalization effect: patients who catastrophize about pain or who are fearful of movement and reinjury are liberal in assuming that a painful experience during one movement applies to even slightly similar movements but are conservative in

55 55 assuming that the experience that a movement is less painful than expected applies to slightly dissimilar movements (p. 1064, italics added). Moreover, the authors find evidence of this asymmetry in the positive correlation between overgeneralization and the interference of pain in daily activities, r =.31, p <.005; a liberal avoidance of an increasing number of comparable activities or, activities assumed to be comparable expands the range of perceived limitations and consequent disability. While the crosssectional nature of the data and the potential bias inherent in self-report instruments preclude causal inferences, these corroborative results are nevertheless disturbing. In sum, previous research has shown that the effects of exposure to one movement do not generalize toward a second, dissimilar movement in highly fearful chronic low back pain patients. These patients overpredict expected pain for the second movement even after correcting their overpredictions regarding the first movement. These findings are reinforced by evidence from participants self-reported reactions to probable life events. One possible explanation for the absence of carry-over correction from one movement to the next may be patients reluctance to change their general belief about the painful potential of all physical movements. Current theorizing about the nature of extinction in classical conditioning paradigms suggests that exposure (often posited as a clinical analogue to extinction) cannot simply be equated with unlearning. In a series of classical conditioning studies with animals, Bouton (2000) demonstrated that extinction effects do not easily generalize across different contexts but are largely restricted to the context of the original acquisition. Analogously, Bouton (2000, 2002) has argued that

56 56 exposure applied in the clinical setting with chronic pain patients results in learning an exception to the rule (e.g., lifting my leg isn t painful or dangerous ), rather than a fundamental change in the rule itself (the belief that all stressful movements are painful and dangerous ). The Present Study While previous studies have only examined generalization across identical or dissimilar movements, the present study hoped to tap a finer aspect of generalization by having the patients consecutively perform four versions of the same task, with each version introducing a discrete element of increased intensity. These tasks were neither identical nor explicitly different. The four adaptations required the participant to (1) reach for a target located at a comfortable height, at a comfortable pace, (2) reach for a target located at a comfortable height, but at a fast-pace, (3) reach for a target located at a lowered height, at a comfortable pace, and (4) reach for a comfortably-placed target (at a comfortable pace) while fitted with a forearm-weight. This design more closely approximates situations participants are likely to encounter in everyday life and expands the repertoire of movements which have hitherto been examined. Consistent with previous findings, it was predicted that, relative to low-kinesiophobic participants, highly kinesiophobic participants would demonstrate greater overprediction of pain and harm; they would also be less likely to generalize corrections of overprediction across the 4 types of movement. Previous studies reliance on expectancy changes across only two trials may obscure gradations in the course of participants pain appraisal that would emerge if more

57 57 trials were available to document evolving pain expectancies in response to greater exposure. Moreover, while the two-trial paradigm reveals significant differences in generalization between high and low catastrophizers, greater exposure (i.e., more practice) is likely to impact the generalization response itself. It is possible that opportunity for more exposure trials per movement enhances generalization across movements. For the purpose of this study, generalization refers to the maintenance of corrected overpredictions across movements. Analyses of generalization focused on the first trial of Movements A through D. In sum, three primary hypotheses were proposed: (1) Highly kinesiophobic participants (i.e., those scoring above the sample median on the TSK) would demonstrate greater overprediction of pain and harm relative to low-kinesiophobic participants. (2) In comparison to their low-kinesiophobic counterparts, highly kinesiophobic participants would demonstrate less generalization of corrected overpredictions across the four movements. That is, there would be a greater difference in overprediction (predicted minus experienced ratings on Trial 1) between high and low TSK groups as they proceed from Movement A to Movement D. This analysis was conducted separately for (a) pain ratings and (b) harm ratings. (3) In comparison to high-kinesiophobic participants engaging in two trials per movement, high-kinesiophobic participants engaging in four trials would show more generalization of corrected overpredictions across movements. That is, there would be a greater difference in overprediction (predicted minus experienced ratings on Trial 1) between the 2-trial and 4-trial groups as they proceed from

58 58 Movement A to Movement D. This analysis was conducted separately for (a) pain ratings and (b) harm ratings. METHODS Design The study employed a 2 (Condition: two trials, four trials) 2 (Group: high kinesiophobia, low kinesiophobia) 4 (Movement type: A: Comfortable height/comfortable speed; B: Comfortable height/ fast speed; C: Low height/comfortable speed; D: Comfortable height/comfortable speed/forearm weight) mixed factorial design. Movement type is a within participants factor while Condition and Group are between participants variables. Participants Participants included individuals aged years old experiencing chronic low back pain. This included individuals who report persistent lower back pain for a minimum of six months. Women who were pregnant or suspect that they may be pregnant (i.e., have missed their most recent menstrual period) were excluded from the study. Additional exclusion criteria for all participants included: 1) a history of spinal surgeries, 2) orthopedic injuries to the upper or lower extremities that restrict motions of these segments, 3) lower extremity weakness or frank neurological signs, 4) taking prescription narcotics for symptom relief, and 5) other chronic diseases that may restrict movement (e.g., arthritis). Participants were recruited from the local community through advertisements in local newspapers and posted advertisements on the university campus. As expected, the same included Ohio University students as well as individuals not associated with the school. Eligible individuals received a $20 remuneration for their

59 59 time. Participants scoring above and below the median on the Tampa Scale for Kinesiophobia (described below) were classified into the high kinesiophobia group and low kinesiophobia group, respectively. Contact with the potential participants was established by phone. An interview, including a review of their medical history, determined their eligibility for participation in the study (see Appendix A for interview protocol). Participants deemed eligible based on the above-described exclusion criteria were scheduled for a laboratory testing session. Power Analysis Previous studies have demonstrated large effect sizes when comparing high vs. low kinesiophobic participants on various movement tasks (e.g., d =.98, d =.92, Vlaeyen et al., 1995a; d = 1.12, Crombez et al., 1999), including reaching paradigms (d =.98, Thomas & France, 2003). For the current study, power analyses were conducted to ensure a sample size sufficient to detect the hypothesized effects, using the smallest of the prior effect sizes (d =.92) \\as a conservative estimate. To test the main effect of Group (Hypothesis 1), power analysis on a 2 (Group) x 4 (Movement) ANOVA indicated that, assuming an effect size d =.92 and α =.05, a total of 40 participants would be necessary to achieve a power of To test the interactions predicted in Hypotheses 2 and 3, a power analysis conducted on the 2 x 4 ANOVA likewise called for 40 participants to achieve a power of 1.00, with d =.92, and.05. Questionnaires Tampa Scale for Kinesiophobia. The Tampa Scale for Kinesiophobia (TSK; see Appendix B) was designed to assess the fear of injury due to physical movement (Kori et

60 60 al., 1990). Respondents rate 17 items on a four-point scale ranging from 1 (strongly disagree) to 4 (strongly agree). The items includes such statements as I wouldn t have this much pain if there weren t something potential dangerous going on in my body ). These responses are summed to create a total score ranging between 17 and 68. Validation studies conducted on patients with chronic low back pain indicate that scores on the TSK are normally distributed, with mean scores ranging from 35.8 to 38.4 and standard deviations ranging from 7.4 to 7.8 (Kori et al., 1990; Vlaeyen et al., 1995). The TSK has demonstrated acceptable internal consistency (Cronbach s alpha = 0.77), and its validity has been supported by evidence that TSK scores (1) are moderately correlated with other measures of pain related fear such as the Fear-Avoidance Beliefs Questionnaire (r = 0.57) (Crombez et al., 1999), (2) discriminate between chronic low back pain patients who are avoiders and confronters on behavioral performance tests (Vlaeyen et al., 1995; Crombez et al., 1999) and (3) are better predictors of disability in patients with chronic low back pain than general negative affect or general negative pain beliefs such as catastrophizing (Crombez et al., 1999). Visual Analog Scale. The Visual Analog Scale (VAS; see Appendix B) was employed for ratings of expected and experienced pain, and ratings of expected and experienced harm. Each VAS is a 10-cm horizontal line with no numbers, marks, or descriptive vocabulary along its length. For perceived pain, the VAS is anchored with the descriptors No pain and Worst pain imaginable, respectively, at each end of the line. For perceived harm, the VAS is anchored with the descriptors Not at all concerned to Extremely concerned regarding harming or having harmed the back during task

61 61 performance. VAS word anchorings that most clearly delineate extremes and are cm in length have been shown to have the greatest sensitivity and to be the least vulnerable to distortions or biases in rating (Price & Harkins, 1992). The VAS has demonstrated reliability, validity, construct validity, and minimal inherent bias in a variety of clinical and experimental pain studies (Kendall, 2000). The Short-Form McGill Pain Questionnaire. Developed from the most frequently used descriptors on the long-form MPQ, the short-form McGill Pain Questionnaire (SF- MPQ; Melzack, 1987; see Appendix B) is widely used to assess dimensions of the pain experience in both clinical and experimental research settings (Wright, Asmundson, & McCreary, 2001). The Pain Rating Index (PRI) of the SF-MPQ comprises the summed ranking of 15 adjectives that describe the sensory (11 words) and affective (4 words) dimensions of pain. The adjectives are ranked on a four-point scale from 0 (none) to 3 (severe). Pain intensity is measured using a 100 mm visual analogue scale and a present pain intensity rating (0-5), which comprises the Present Pain index (PPI). In adult patients, the SF-MPQ demonstrates high reliability, validity, and sensitivity to change across a variety of pain and illness categories (Burkhart & Jones, 2002). Roland and Morris Disability Questionnaire (RMDQ; Roland & Morris, 1983). The Roland and Morris Disability Questionnaire served as the primary measure of functional status (or disability) related to back pain symptoms. This questionnaire is widely used in low back pain research, having demonstrated good reliability and validity in pain patients.

62 62 Behavioral Test The performance test consisted of four variants of a reaching movement currently employed in studies of motor control and low back pain (Thomas, Corcos, & Hasan, 2003). Figure 2 depicts the four movement conditions. The first movement required the participant to reach for a target located at a comfortable height, at a self-selected speed (Movement A). Target location was normalized to the participant s hip-to-shoulder length, arm-length (humerus + forearm), and hip height. Comfortable target location was calculated so that the subject (with the elbow extended and the shoulder flexed at 90 degrees) could, in theory, reach the target by flexing the pelvis 15 degrees. Previous studies have established comfortable speed duration-- the time from movement initiation until the participant makes hand-to-target contact-- as approximately one second. The second movement required the participant to reach for the comfortably - placed target at approximately twice their self-selected pace (Movement B). The third movement required the participant to reach for a target located at a low height relative to their physical proportion, at a self-selected speed (Movement C). The low target required hip flexion of 60 degrees, thus challenging the participants with larger amounts of forward bending. The fourth movement required the participant to reach for a comfortably placed target at a self-selected pace while fitted with a 2 kg. forearm-weight (Movement D). Targets were positioned on a vertical bar placed in front of the participant at a

63 63 TASK DESCRIPTION: Movement A Participant reaches for target placed at a comfortable height (15 0 ), at a self-selected/comfortable speed. Movement B Participant reaches for target placed at a comfortable height (15 0 ), at approximately twice the comfortable speed. Movement C Participant reaches for a target placed at a low height (60 0 ), at a comfortable speed. Movement D Participant reaches for target placed at a comfortable height (15 0 ), at a comfortable speed, while fitted with a 2 kg. forearm weight. Figure 2. The four reaching tasks comprising the behavioral test. Participants performed two or four trials of each movement.

64 64 distance calculated to their physical proportion. Participant reached for the target with both hands. Procedure Figure 3 provides an overview of the laboratory testing protocol that was used for each testing session. Upon arrival to the testing laboratory, each participant was asked to complete a number of questionnaires, including the TSK, RMDQ and the Short-Form McGill Pain Questionnaire. Following completion of the questionnaires, anthropometric data were collected in order to determine comfortable and low target locations for the individual participant. The experimenter measured the participant s hip-to-shoulder length, arm length (humerus + forearm), and hip height. These measures were then used to determine the respective target locations relative to the participant s unique proportions. Participants were then randomly assigned to the two-trial condition or the four-trial condition by means of coin toss. Questionnaires were be administered by means of computerized online application. The experimenter conducting the behavioral test had no knowledge of the participants TSK score or any other measures at the time of testing. Participants then engaged in the performance test, composed of four reaching motions (Movements A-D). Participants randomly assigned to the two-trial condition were asked to perform each movement two times, resulting in a total of eight performance trials. For participants assigned to the four-trial condition, this resulted in a total of sixteen performances. Trials of a given movement were performed in succession. Immediately prior the performance of the first trial of the first movement, the participant was asked to rate the

65 Figure 3. Proposed study protocol. 65