CPMT Journal

Introduction

Maladaptive responses to illness are described as inappropriate thoughts, evaluations and/or actions [1,2], with this conceptualization informed by elements of the sick role [3]. The term maladaptive (or abnormal) is applied to illness behaviors (IB) when self-reported somatic symptoms do not correspond to observed biomedical abnormality [4], suggesting a psychological rather than physical focus. Much original research reflected Pilowsky’s interest in chronic pain, [5-9] with many other researchers also exploring IB in this context [10-16]. Contemporary theoretical and empirical developments, detailed below, make it timely to revisit the relevance of IB to understanding chronic pain treatment outcomes.

Pilowsky’s Illness Behavior Questionnaire (IBQ) [17] comprised seven primary scales (General Hypochondriasis, Disease Conviction, Psychological versus Somatic Focusing, Denial, Affective Inhibition, Affective Disturbance, Irritability), and two second order scales (Disease Affirmation, Affective State). This complexity has led to the selective use of both items and scales [9,12,18-20] which are rarely accompanied by evidence of validity or reliability [21]. However, more rigorous analyses of the IBQ have resulted in the presentation of a brief 31-item version (IBQ31) assessing only three dimensions [22,23]. Reflecting their thematic content, they have been termed Affirmation of Illness (AI; the insistence of the presence of an illness), Concern for Health (CH; health-related worry), and General Affective State (GAS; emotional distress irrespective of illness concerns). Reliability of the IBQ-31 has been demonstrated using both healthy community members [22-24] and multiple illness groups (as a measure of reaction and adaptation to chronic illness) [25-27].

Related to the complex structure of the original IBQ has been the difficulty determining the response profile that might constitute ‘maladaptive’. However, a unique perspective [28,29] grouped the IBQ responses of stroke rehabilitation patients statistically using cluster analysis. Inspection of IBQ means revealed evidence of elevated IB in one of two clusters, comprising nearly 30% of patients in one study [28] and over 20% in a validation study [29]. Importantly, elevated IB was a key determinant of long-term disability. That is, maladaptive IB proved to be a valuable index of chronic illness management. The potential benefit of this finding is the additional evidence it provides to clinicians in tailoring interventions to assist adaptation and/or recovery.

The aims of the current study were to use a chronic pain sample and matched sample of healthy community members to directly compare IBQ-31 profiles to establish whether elevated scores were generally evident in the pain sample. Further, the patterns of association between cognate psychological constructs and IBQ-31 dimensions provided insight into the profile of those with chronic pain compared with community participants. For example, the measurement of coping offered a preliminary exploration of potential behavioral concomitants of IB, acknowledging the lack of items in the IBQ with an overt behavioral focus [30-32]. Conversely, symptom attributions allowed the consideration of further cognitive and emotional aspects of symptom interpretation, with the quantification of recent somatic symptoms reflecting the degree to which individuals are attuned to physical variations in their bodies [33]. Cluster analysis was then applied to IBQ-31 responses from the pain sample to determine whether a subgroup of participants with elevated means could be identified. The obtained solution was tested by comparing membership across demographic, psychological, health- and pain-related variables. The goal was to identify variables associated significantly with the cluster solution and to use them to build a predictive equation for maladaptive responses. The results have the potential to highlight psychosocial indicators of rehabilitation outcome. That is, motivation for successful rehabilitation or disease management may be either enhanced or compromised depending on specific cognitive and/or affective interpretations of health and illness.

Materials and Methods

Participants

Target participants, who were assured their decision to participate would have no influence on current or future medical care, were at least 18 years of age with sufficient command of receptive and expressive English to allow completion of a self-report questionnaire. Participation was voluntary and unpaid.

The pain sample (n = 79) were required to have a confirmed diagnosis of chronic benign pain of at least 12 months’ standing, allowing for initial psychological adjustment and the acquisition of disease-specific knowledge and self-management skills. The mean age was 52.1 years (SD = 10.9, range 23-83), with 47 women (60%). Post-secondary education had been attained by 58%. Participants were matched according to sex, age and education with a pain-free community sample (n = 79) extracted from a parent file [22]. Again, 60% were female and 58% had attained post-secondary education. The mean age was 52.4 years (SD = 10.8, range 23-83). There were no significant group differences in these matching variables.

Questionnaires

Standard demographic details (e.g., sex, age), health-related information, and published scales to measure AIB and related constructs were included.

Pain-specific Health Information

Years since diagnosis, site(s) affected, and whether compensation was being sought or received previously was provided by the pain sample. Two questions assessed current pain, ranging from 0 (‘no pain’) to 10 (‘as bad as could be’), and functional interference with day-to-day activities (during the past six months), ranging from 0 (‘no interference’) to 10 (‘unable to carry on any activities’) [34].

General Health Information

Both samples provided two measures of health status. Self-rated health was measured on a 5-point scale (‘poor’ to ‘excellent’) with higher scores indicating better health. This global measure has demonstrated strong and consistent associations with adverse health outcomes such as mortality, institutionalization, and hospitalization [35,36]. Health trajectory taps any perceived change in health over a one-year period on a 5-point scale (‘much worse’ to ‘much better’) with higher scores indicating improvement. Evidence suggests that health trajectory has an association with adverse health outcomes independent of self-rated health [36]. In accord with common practice, these two measures were dichotomized to reflect poor self-rated health and declining health trajectory, respectively [35]. The frequency of visits to a General Practitioner (GP) during the past two years was also sought, using a 7-point scale ranging from extremely frequent (‘at least once a week’) to extremely infrequent (‘once or less’). The number of common medical conditions with which participants had been diagnosed (cancer, high cholesterol, diabetes, etc.; max. = 18) was also collected.

Illness Behavior

Paralleling the work of Pilowsky [17] two versions of the IBQ-31 have been validated. Form A (which addresses the presence of a chronic condition) was administered to the pain sample whereas the community sample completed Form B which questions whether participants believe they have a compromising medical condition. Three dimensions of IB were calculated from each version [22,23]. Affirmation of Illness (AI; 12 items, range 0-12) reflects the endorsement by respondents that they have a disease (e.g., “Do you think there is something seriously wrong with your body?”, “Does your illness interfere (Do you have an illness which interferes [for Form B]) with your life a great deal?”). Concern for Health (CH; 8 items, range 0-8) places emphasis on healthrelated worry (e.g., “Do you worry a lot about your health?”, “Do you think that you worry about your health more than most people?”), whereas General Affective State (GAS; 11 items, range 0-11) focuses on emotional distress irrespective of illness concerns (e.g., “Do you find that you get sad easily?”, “Do you find that you get angry easily?”). Participants indicated the items true of them (‘yes’/‘no’), scoring a point for each response characteristic of a maladaptive approach to illness interpretation.

Coping Strategies

The Brief COPE [37] comprises 28 items accompanied by a 4-point scale (‘I usually don’t do this at all’ to ‘I usually do this a lot’) indicating the extent to which each strategy is typical for participants. Higher responses (0 to 3) denote greater endorsement of each strategy. In the present study items reflected the general way in which people deal with illness (i.e., dispositional). For example, ‘I try to come up with a strategy about what to do’ and ‘I accept the reality of the fact that it has happened.’ While 14 unique subscales are available, in accord with recommendations only two scores were calculated [38,39]. Adaptive Coping (0 to 48) comprised Active Coping, Planning, Instrumental Support, Emotional Support, Religion, Positive Reframing, Acceptance, and Humor. Maladaptive Coping (0 to 36) comprised Venting, Disengagement, Self-Distraction, Self-Blame, Substance Use, and Denial.

Symptom Attributions and Symptom Experience

Symptom attributions refer to the causal interpretations of common, albeit hypothetical, bodily sensations, or physical symptoms (e.g., headache, dizziness, sleep difficulties). The Symptom Interpretation Questionnaire (SIQ) comprises thirteen such items which are accompanied by somatic (i.e., attributed to underlying physical disorder), psychological (i.e., triggered by emotional distress or similar stressor), and normalizing (i.e., attributed to a situational/ environmental factor such as transient fatigue, overexertion, or lack of sleep) explanations [40]. Each explanation is endorsed on a 4-point scale (‘not at all’ to ‘a great deal’) with responses for each explanatory style summed (range 13-52). Participants also indicate (‘yes/no’) whether they have experienced each of the 13 symptoms during the past three months. Responses are summed to create a Symptom Experience score (range 0-13).

Procedure

Community-based support groups were the primary participant source, with suitable individuals also recruited through the practices of two pain/musculoskeletal specialists. For both sources, permission was granted for the first author to give a brief verbal outline of the research followed by the provision of a written description and questionnaire booklet to those who expressed interest. Completed questionnaires were either collected at a subsequent session or returned using a reply-paid envelope. The goal for the community sample was to recruit participants who were not current health-care seekers. They were sourced through non-illness-related community groups in a manner parallel to the procedures outlined above for pain participants recruited from community-based self-help groups. A more detailed description of recruitment procedures is provided by Prior and Bond [22].

Statistical Analyses

All comparisons were repeated measures analyses due to the oneto-one matching of pain and community participants. Analyses were routinely non-parametric for nominal and ordinal measures and parametric for scale data. Two-tailed probabilities are reported. IBQ-31 data from pain participants were subjected to a 2-step cluster analysis. This procedure supports auto-clustering to enable the identification of the optimal number of clusters [41]. The obtained solution was then treated as the outcome variable in a discriminant function analysis [42] to assess the predictive utility of pain-specific, general health and psychological variables in the determination of AIB in the pain sample.

Results

Profile of Pain Participants

Time since diagnosis was 9.3 years (SD = 7.4, range 1-52). Multiple sites of pain were common, with lower back (73%), neck (44%), upper back (32%), and limbs (30%) the most prevalent. The mean level of current pain was 5.44 (SD = 2.01, range 1-10), and functional interference was 6.46 (SD = 2.00, range 2-10). Compensation had either been received, was being received, or was being sought, by 33%.

Sample Comparisons

Table 1 summarizes those demographic and health variables that may be compared across the two samples. Significantly more of the healthy sample were married and conversely more of the pain sample were separated or divorced. The samples also differed in terms of employment, with substantially more of the pain sample not currently in the work force, predominantly due to their condition.

Table 1: Demographic and health comparisons between pain and community participants.

A significant difference was also evident in participants’ health ratings, with the pain sample more likely to report poor health. This was echoed by a declining health trajectory and more medical diagnoses for the pain sample. Finally, on average, members of the pain sample were significantly more frequent visitors to a GP. Table 2 compares the psychosocial profiles of the samples. Scores for the pain participants were significantly elevated for all variables other than Adaptive Coping and Psychological Attributions for illness.

Table 2: Psychosocial characteristics of pain and community participants.

Classification of Maladaptive IB among Pain Participants

The cluster analysis was optimized by a two-group solution, providing a cohesion and separation index (goodness-of-fit) of 0.6 (defined as ‘good’) [43]. The smaller cluster comprised 27 participants (34.2%), with of IBQ-31 scale scores (Table 3) suggesting elevated IB in this group. That is, significantly higher means for all scores, particularly CH and GAS.

Table 3: A comparison of cluster means for IBQ-31 scales.

The nature of this classification was further tested using discriminant function analysis. All variables demonstrating the bivariate potential to discriminate between the two obtained clusters were tested. However, only data pertaining to the final predictive model are reported (following removal of non-significant indicators from the multivariate model due to multicollinearity). Log determinants suggested covariance matrices were equivalent (Box’s M = 35.61, p = .293), while the discriminant function itself was significant (χ²(7) = 51.00, p ≤ .001). Correct group classification was achieved for 77.8% of those with elevated IB, and 94.2% for non-elevated IB cases). There were seven significant predictor variables (Table 4) indicating that members of the pain sample were more likely to report maladaptive IB if they also reported higher levels of Maladaptive Coping, Somatic Attributions for illness, current pain, functional interference, and more GP visits. They were also likely to have poor self-rated health and a declining health trajectory over the past 12 months.

Table 4: Variables contributing significantly to the prediction of cluster membership.