Expectancy, fear and pain in the prediction of chronic pain and disability: A prospective analysis

1. Introduction 

Back pain is a common problem that in approximately 3–10% of acute cases leads to chronic pain and (work) disability (Reid et al., 1997). Chronic back pain disability is associated with enormous societal costs and patient suffering. It is therefore important to identify pain patients that risk developing chronic disability at an early stage, so that preventative efforts can be set in motion. Risk factors for the development of persistent problems are many and cover individual, psychosocial and occupational variables (van Tulder et al., 2002). Recently, individual psychological variables such as pain-related fears have been pinpointed as especially important predictors of chronic pain disability. Cross-sectional, experimental and prospective studies show very clearly that indicators of pain-related fear such as fear avoidance beliefs and catastrophizing are related to pain and disability (Vlaeyen et al., 1995a, Klenerman et al., 1995, Crombez et al., 1996, Sullivan et al., 1998, Linton, 2000, Fritz and George, 2002).

The fear-avoidance model provides a cognitive behavioral framework for patients with high levels of pain-related fear and gives an enticing explanation of the mechanism whereby back pain patients may develop persistent disability (Lethem et al., 1983, Philips and Grant, 1991, Waddell et al., 1993, Vlaeyen et al., 1995a, Vlaeyen et al., 1995b). The model stresses the role of catastrophic appraisals of the pain experience, and subsequent fear and hypervigilance. The other prominent role is given to avoidance of activity, largely fueled by fear that activity will cause injury and will exacerbate the pain problem (Vlaeyen and Linton, 2000).

Another individual psychological variable that has been found to be an important predictor for outcome is an individual’s expectancy of outcome. Expectations about non-volitional responses to a particular stimulus or situation are referred to as ‘response expectancies’. Examples of non-volitional responses include emotional reactions, sexual arousal, and pain. These responses are considered to occur automatically but the expectation of their occurrence can a have strong impact on experience, behavior and physiological functioning; in other words, the expectation itself can be directly self-confirming (Kirsch, 1997). The influence of response expectancies has been well documented in the powerful analgesic effect of placebos, the workings of antidepressants, sexual dysfunction, pain experience, psychological treatment effects and anxiety disorders (Kirsch, 1997).

In the pain literature, the role of expectancy has been studied in various ways. Some studies have focused on the effect of treatment expectancy on the outcome of treatment (for example: “Do you expect that the treatment program will help you better cope with your pain?”, in Goossens et al., 2005). These studies show that, indeed, treatment expectancy accounts for a substantial part of the variation in treatment outcomes such as function (Kalauokalani et al., 2001, Goossens et al., 2005). Moreover, Goossens et al. showed that treatment expectancies were related to other important variables such as fear of movement/(re)injury and coping strategies.

Other studies have investigated expectancy in experimental settings where the expectation of pain is documented before participating in a pain provoking behavioral task (McCracken et al., 1993, Crombez et al., 1996). These experimental studies with chronic back pain patients highlight that anxious patients tend to have more negative expectations about the pain that they may experience during an experimental procedure and that this expectancy is related to subsequent avoidance behavior (McCracken et al., 1993, Crombez et al., 1996).

Lastly, expectancy has been studied in epidemiological studies, in relation to more distal outcomes such as recovery or return to work (Linton and Halldén, 1997, Cole et al., 2002, Schultz et al., 2004). These studies show that a negative expectancy about recovery from pain is an important risk factor for pain and disability. Prospective studies with (sub) acute back pain patients show, for example, that the expectation of persistent pain is a risk factor for pain at a one-year follow up and that expectations of recovery are independent predictors of the duration of (work) disability (Linton and Halldén, 1997, Cole et al., 2002, Schultz et al., 2004).

However, while apparently important in the development of disability, it is unclear what the mechanisms are that steer expectancies of recovery. In line with the studies investigating treatment expectancy and more immediate response expectancy, negative expectations of recovery may partly be a product of pain-related fear. One would expect negative expectations about recovery and pain-related fear to be interrelated. When patients have a catastrophic appraisal of their pain and believe that their pain signifies harm, then it is logical that they would also have negative expectations of recovery. Nevertheless, there is a lack of research studying the relationship between expectations of recovery and key risk factors such as pain-related fear.

Knowledge about the interrelationship of negative expectations of recovery and pain-related fear seems important as it would shed light on what factors could generate negative expectancies and on the possible mechanisms whereby expectancy may exert an influence on disability. A better grasp on the interrelationships between expectancies, pain, fear and disability could also be beneficial clinically, as it could improve early identification. For example, if the patient’s own expectations of recovery are a product of pain-related fears then a question on expectations of recovery could be used as a simple screening.

The objective of the current study was to investigate the interrelationships between expectations of recovery, pain-related fear, pain and disability. Given that pain-related fear and expectations of recovery are both documented risk factors for persistent pain and/or disability, a prospective analysis will be employed to ascertain how these variables are related to future pain and disability and to what degree these variables have unique predictive value above and beyond pain itself.

The specific goals of this study were:

2. Methods 

3. Results 

3.1. Description of participants 

Table 1 gives an overview of background characteristics of the participants and their scores on the included measures.

Table 1. A summary of sample characteristics at baseline

	Description of the participants at baseline	N=141
	Variable (M (SD) or %)
	Age	47.7 (8.3) (range 21–61)
	Gender (% female)	80%
	Born in Sweden (% yes)	99%
	Neck pain (% yes)	82%
	Back pain (% yes)	72%
	Sick listing (% <15 days during past year)	93%
	Number of pain sites (%)
	1	16%
	2	37%
	3	33%
	4	14%
	Pain duration (% 6–12 months)	79%
	Healthcare seeking (during past year):
	Doctor visits	28%
	Medical specialist	1%
	‘Other’ healthcare providers (unspecified)	30%
	Pain intensity last week	4.5 (2.2)
	Average pain intensity last three months	4.7 (1.7)
	Pain frequency	6.0 (2.7)
	Expectancy (0–10)a	7.3 (2.5)
	Tampa Scale for Kinesiophobia
	‘activity avoidance’ (7–28)	12.4 (4.0)
	‘somatic focus’ (3–12)	5.0 (2.2)
	Pain Discomfort Scale (0–40)	17.1 (7.3)
	Roland and Morris Disability Scale (0–18)	3.5 (4.0)

a Higher scores indicate more positive expectancies about recovery.

3.2. The interrelationships between pain, pain-related fear, expectancy and functional disability 

The relationships between expectancy, negative affect, fear avoidance beliefs, pain and functional disability are displayed in Table 2. As can been seen, pain, functional disability, fear avoidance beliefs, negative affect and expectancy of persistent pain are interrelated. However, there was no significant relationship between ‘activity avoidance’ (reflecting a belief that activity may result in (re) injury or increased pain) and expectancy of persistent pain, nor between ‘activity avoidance’ and average pain and pain frequency.

Table 2. Cross-sectional correlations between expectancy, negative affect, fear avoidance beliefs, pain and functional disability

		1.	2.	3.	4.	5.	6.	7.
	1. Expectancy	–	0.109/0.384⁎⁎	0.314⁎⁎	0.305⁎⁎	0.561⁎⁎	0.389⁎⁎	0.259⁎⁎
	2. Fear avoidance beliefs: ‘activity avoidance’/‘somatic focus’		–	0.431⁎⁎/0.482⁎⁎	0.227⁎⁎/0.295⁎⁎	0.138/0.351⁎⁎	0.139/0.284⁎⁎	0.434⁎⁎/0.388⁎⁎
	3. Negative affect			–	0.573⁎⁎	0.385⁎⁎	0.482⁎⁎	0.440⁎⁎
	4. Average pain				–	0.518⁎⁎	0.734⁎⁎	0.526⁎⁎
	5. Pain frequency					–	0.527⁎⁎	0.440⁎⁎
	6. Pain intensity						–	0.503⁎⁎
	7. Functional disability							–

⁎⁎ P<0.01.

3.3. What steers expectancy of persistent pain? 

A hierarchical regression analysis was performed to investigate which variables (pain intensity, average pain, pain frequency, negative affect and fear avoidance beliefs) explained significant variance in expectancy of persistent pain. As is shown in Table 3, 37% of the variance in the expectancy of persistent pain is significantly predicted by pain frequency (32%) and ‘somatic focus’ (reflecting a belief in underlying and serious medical problems) (5%). Note that the word predicted is here used in the statistical sense.

Table 3. Hierarchical regression analysis predicting expectancy of persistent pain

	Block	Variables	βa	R2
	1	Gender	0.015
		Age	0.003	0.01
	2	Pain frequency	0.483⁎⁎	0.32
	3	Fear avoidance beliefs ‘somatic focus’	0.242⁎⁎	0.37

a Standardized β weights are displayed from the final model. ⁎⁎ P<0.01.

3.4. The predictive value of expectancy and pain-related fear for future pain and functional disability 

Hierarchical regression analyses were performed to investigate the relative importance of expectations and different component of pain-related fear (negative affect and fear avoidance beliefs) in explaining average pain and functional disability at the 12 month follow up. Age, gender and average pain at the initial assessment were controlled for before entering expectancy, negative affect and fear avoidance beliefs in a stepwise fashion. The results of the regression analyses predicting average pain and disability at the one year follow up are presented in Table 4.

Table 4. Hierarchical regression analyses predicting pain and functional disability at 1 year follow up

	Block	Variables	βa	R2
	Hierarchical regression analysis predicting average pain at one year follow up
	1	Gender	0.134
		Age	0.054	0.01
	2	Average pain at initial assessment	0.253⁎⁎	0.24
	3	Negative affect	0.210⁎
		Fear avoidance beliefs ‘activity avoidance’	0.208⁎⁎
		Expectancy of persistent pain	0.195⁎⁎	0.38
	Hierarchical regression analysis predicting functional disability at one year follow up
	1	Gender	0.049
		Age	0.129	0.004
	2	Average pain at initial assessment	0.159	0.16
	3	Negative affect	0.235⁎
		Fear avoidance beliefs ‘activity avoidance’	0.225⁎⁎
		Expectancy of persistent pain	0.162⁎	0.31

a Standardized β weights are displayed from the final model. ⁎ P<0.05. ⁎⁎ P<0.01.

As can be seen, after controlling for age, gender and average pain at initial assessment, negative affect, expectancy and ‘activity avoidance’ (i.e., a belief that activity may result in (re) injury or increased pain) accounted for an additional 14% of the variance in average pain at the one-year follow up. In total, average pain, negative affect, expectancy and ‘activity avoidance’ predict 38% of the variance in average pain at one-year follow up. ‘Somatic focus’ (i.e., a belief in underlying and serious medical problems) is not retained in the analysis.

After controlling for age, gender and average pain at initial assessment, negative affect,‘activity avoidance’ and expectancy accounted for an additional 15% of the variance in functional disability at the one-year follow up. The total amount of variance predicted by pain, negative affect,‘activity avoidance’ and expectancy is 31%. ‘Somatic focus’ is not retained in the analysis.

4. Discussion 

The goals of this study were to explore the interrelationships between expectancy, pain-related fear, pain and functional disability; to explore which variables could possibly drive expectancy, and to investigate the relative predictive value of expectancy, pain-related fear and pain for future pain and functional disability.

As was anticipated, pain, disability, pain-related fear and negative expectations were interrelated. The more frequent and intense the pain, the more pain-related fear and the more patients expected their pain to become a persistent problem. Interestingly, the expectancy of a persistent pain problem was best accounted for by pain frequency and a belief in underlying and serious medical problems. Further, the expectation of persistent pain, as well as negative affect and fear avoidance beliefs had a small (14–15%) but unique predictive value for future pain and disability after controlling for age, gender and average pain at initial assessment.

These finding shed new light on the interrelations between pain, pain-related fear variables and expectations about recovery. Intuitively, the importance of frequent pain and a belief in underlying and serious medical problems for expectancy makes sense, as frequent pain and the conviction that this pain is caused by serious pathology may indeed logically lead to the expectation that the pain may become persistent. A possible explanation of the interrelationships could be that the negative expectations are a consequence of the intense and frequent pain experience and the belief that this pain signifies danger and harm. This could set the stage for behavioral changes in the form of for example dysfunctional avoidance behavior and vigilance to significant and threat-related stimuli such as pain itself. However, as this part of the investigation is cross-sectional it should be noted that direction of causality remains ambiguous. This analysis should therefore be seen as explorative, to generate ideas and hypotheses.

The expectation of persistent pain, negative affect as well as fear avoidance beliefs had unique predictive value for future pain and disability. These findings are consistent with the literature on the role of pain-related fear and expectancy for chronic pain and disability (Klenerman et al., 1995, Linton and Halldén, 1997, Sullivan et al., 1998, Linton, 2000, Cole et al., 2002, Fritz and George, 2002, Schultz et al., 2004). Of the fear avoidance beliefs variables it was ‘activity avoidance’ (a belief that activity may result in (re) injury or increased pain) that had unique predictive value for pain and disability one year later. The ‘somatic focus’ scale had no unique predictive value for future pain and disability when entered in the multivariate analysis. Even though this may appear puzzling as the ‘somatic focus’ scale was predictive of expectancy in the cross-sectional analysis, this finding is in line with other investigations that have found the ‘activity avoidance’ scale of the TSK to be most strongly associated with future behavior (Geisser et al., 2000, Roelofs et al., 2004). Moreover, considering that ‘somatic focus’ was related to outcome in the univariate analyses, the fact that it does not uniquely explain variance in pain and function at follow-up points to the overlap between the constructs entered.

It should be kept in mind that the selection of predictors based on statistical considerations eliminates variables that share variance. Even though these analyses fit the goal of this study (to investigate the relative predictive value of expectancy, pain-related fear and pain for future pain and functional disability), they of course do not help in theoretically clarifying the overlap between the independent variables. Clarification and further understanding of such overlaps probably has to evolve out of a combination of theory building, appropriate design, and possibly, modern analytical techniques such as structural equation modeling. The current study should be seen as a first step in pinpointing that expectancy and other highly predictive cognitive and emotional variables may best be understood in relation to one another.

Even though expectancy, negative affect and ‘activity avoidance’ might have a small unique predictive value for pain and/or disability, the strong interrelationships between these variables should caution for treating them as separate entities. In fact, expectations, beliefs and negative affect could all be seen as components of pain-related anxiety, that is, they could all be seen as signals of one and the same underlying process. Thus, negative expectancy could be seen as part of a whole set of cognitive and emotional aspects that are all related to outcome but that are shaped by anxiety. Their influence on future pain and disability may be via known behavioral and perceptual aspects of anxiety: avoidance and hyper vigilance processes. This anxiety process may be driven by frequent and unexpected pain experiences, possibly in interaction with a more general individual vulnerability factor to experience negative feelings such as negative affectivity.

The subjects in this investigation were seeking primary care and, even though most had pain durations over 6 months, they were not chronically disabled. The patients in this study had low levels of disability and sick listing. The fact that expectancy had a predictive value for negative outcome should be taken as an important signal by clinicians who meet these pain patients at time of care seeking. Obtaining additional information on the factors that might generate such negative expectancies is important as this could reveal obstacles such as anxieties about underlying pathology or other catastrophic interpretations of the pain experience. Hence, knowledge of the relationship between expectancy, pain experience, fears and disability can aid the clinician in developing a model of the patients pain problem, in identifying the relevant obstacles to recovery that the patient experiences and finally, in matching the appropriate treatment.

Several limitations to this study may restrict the conclusions that can be drawn. First, it should be kept in mind that this study is correlational in nature and that correlations are not to be confused with causal relationships. A second potential restriction is that all data are based on self-report and other sources of verification of for example disability status or health care utilization are lacking. Further research could, for example, make use of health care or sick listing records to substantiate results. Third, expectancy was measured with a single question, asking about the risk that the current pain may become persistent. It is unclear to what degree this question is a valid and reliable representation of the construct ‘expectations about recovery’, since expectancy of recovery can been seen as having many facets, such as expectations about future functioning, return to work, the clinical course, etc. (Cole et al., 2002). Different expectancies could be differentially related to pain, disability and pain-related fears. Future studies should therefore include a more extensive measure of expectations of recovery.

In conclusion, expectations, negative affect and fear avoidance beliefs are interrelated constructs that have predictive value for future pain and disability. In clinical work, inquiring about pain beliefs, expectations about the future and pain-related distress can be helpful to guide appropriate intervention and as indicators of risk for chronicity. However, from a scientific point of view, the strong interrelations between negative affect, expectations and beliefs call for precaution in treating these constructs as if they were separate entities existing in reality. Instead of conceiving them as different, expectations, beliefs and negative automatic thoughts might all be reflections of negative emotion elicited by the pain experience, that is, “cognitive correlates of pain-related anxiety” that tap into how individuals experience their pain and that may shape their perception and behavior.