Associations between Neuroticism and Depression in Relation to Catastrophizing and Pain-Related Anxiety in Chronic Pain Patients

Several cognitive-affective constructs, including pain catastrophizing and pain-related anxiety, have been implicated in the onset and progression of chronic pain, and both constructs have been identified as key targets for multidisciplinary pain treatment. Both neuroticism and depression have been linked to these constructs (and to each other), but how each may contribute to the pain experience is unknown. This study tested associations between neuroticism, depression, and indices of catastrophizing and pain-related anxiety among persons seeking treatment for chronic non-malignant pain. We hypothesized, as a higher-order personality trait, neuroticism would remain uniquely associated with both pain catastrophizing and pain-related anxiety, even after accounting for current symptoms of depression. A retrospective study design assessed depression (as measured by the Centers for Epidemiologic Studies-Depression scale), neuroticism (measured with the Neuroticism-Extraversion-Openness Personality Inventory), the Pain Catastrophizing Scale, and the Pain Anxiety Symptom Score in a consecutive series of patients (n=595) admitted to a 3-week outpatient pain treatment program from March 2009 through January 2011. Hierarchical regression indicated that neuroticism was independently associated with greater pain catastrophizing and pain-related anxiety, above-and-beyond the contributions of sociodemographic characteristics, pain severity, and depression. A depression by neuroticism interaction was not observed, suggesting that associations between neuroticism and cognitive-affective pain constructs remained stable across varying levels of current depression. These findings represent an early but important step towards the clarification of complex associations between trait neuroticism, current depression, and tendencies toward catastrophic and anxiety-provoking appraisals of pain among persons seeking treatment for chronic pain.

Introduction three-week outpatient program is based on a cognitive-behavioral framework, and primary goals include the restoration of physical and emotional functioning (see Hooten, 2011 for a complete description). Patients admitted to the program had undergone prior treatment for their pain without relief. All adults consecutively admitted to the Center from March 2009 to January 2011 were eligible for analysis (N = 629). Data were collected as an integral part of the pain treatment program on the first day of the program (i.e., prior to treatment). All patients provided authorization for use of their medical records for research. Excluding patients for whom at least one study measure was missing (n = 34), the final sample included 595 participants.

Study Measures
Clinical and demographic data. Demographic information, including age, sex, race, ethnicity, employment history, marital status, and BMI was collected at baseline, as were clinical data including source, duration, and current therapy for pain.
Depression. The Center for Epidemiologic Studies-Depression scale (CES-D) provides a measure of depressive symptoms within the past week. The 20-item self-administered questionnaire has established reliability and validity among adults with chronic pain [31,32]. The CES-D is scored on a 4-point Likert scale, and total scores range from 0 to 60, with scores of above 19 indicative of major depressive symptoms in chronic pain population [33,34].
Neuroticism. The Neuroticism-Extraversion-Openness Personality Inventory (NEO-PI) is a 240-item questionnaire that has been extensively validated [35], and is the most comprehensive self-report instrument that operationalizes the five-factor model of personality [36,37]. The NEO-PI yields a total neuroticism score, which represents a tendency or predisposition to experience negative affective states. The neuroticism scale is comprised of six component or facet scores, with each representing a tendency to experience anxiety, depression, angry hostility, self-consciousness, impulsiveness, or vulnerability.
Multidimensional Pain Inventory Pain Subscale (MPIPS). The MPIPS is a valid measure of clinical pain severity that is calculated from three individual items [38,39,40,41,42,43,44]. Raw scores were converted to standardized T-scores using means and standard deviations from a heterogeneous group of over 700 chronic pain patients [45]. The standardized subscale has a mean of 50 (range 0 to 100) and a standard deviation of 10, where higher scores indicate greater pain.
Pain Catastrophizing Scale (PCS). The PCS assesses cognitions and emotions associated with actual or anticipated pain experiences [3], and higher scores indicate a tendency to focus on and exaggerate the threat of pain, and to hold negative appraisals of capacity to cope with pain. The PCS is a reliable and valid measure that has been used extensively in chronic pain populations [46,47].
Pain Anxiety Symptom Scale-20 Items (PASS-20). The PASS-20 is a valid measure of cognitive and physical anxiety in response to pain, fearful appraisals of pain, and the use of escape and avoidance behaviors among persons with chronic pain [48]. The PASS-20 yields a composite score from 0 to 100, with higher scores indicating greater levels of pain-related anxiety.

Data Analysis
First, pair-wise associations between socio-demographic variables, pain severity, depression, neuroticism, catastrophizing, and pain-related anxiety were assessed using Pearson's product moment correlation and summarized using a correlation matrix. Socio-demographic characteristics that were associated with the predictor or outcome variables were retained as covariates in subsequent analyses.
Hierarchal linear regressions were conducted to test associations between neuroticism, depression, and both of the cognitive-affective pain constructs. Hierarchical regression was selected because it allows for examination of the incremental, relative contributions of predictor variables (or groups of predictor variables) as they are sequentially entered into the model. Hierarchical linear regression models were constructed in JMP version 10 (SAS institute, Gary, North Carolina, USA). Separate hierarchical models were constructed with pain catastrophizing and pain-related anxiety used as the respective dependent variables.
The primary aim of the analysis was to assess the contribution of neuroticism above and beyond that of depression. Therefore, in each of the hierarchical regression models, sociodemographic factors (i.e., age, gender, marital status, education) and pain severity were entered at Step 1, depression was entered at Step 2, and neuroticism was entered at Step 3. The relative contribution of neuroticism, above-and-beyond sociodemographic factors, pain severity, and depression, was determined by examining change in R 2 at Step 3. In order to test whether associations between neuroticism and the cognitive-affective pain variables were potentially moderated by depression, we constructed a depression-by-neuroticism interaction term, and included it as an additional explanatory variable (Step 4). Given that no significant neuroticism-bydepression interactions were observed (interaction term p>0.05 in all models), results summaries and subsequent analyses include only main effects. Finally, we conducted exploratory analyses to test associations between each facet of neuroticism (i.e., anxiety, hostility, depression, and impulsiveness) and the cognitive-affective pain variables. To do so, separate hierarchical regression models were constructed with socio-demographic factors (i.e., age, gender, marital status, education) and pain severity entered at Step 1, depression entered at Step 2, and each respective neuroticism facet entered at Step 3.

Participant Characteristics
The study sample (N = 595) included predominantly middle-aged Caucasian females with pain of relatively long duration. Scores on measures of depression and pain indicated that the sample was experiencing significant pain and distress at the time of assessment. Complete participant characteristics and average scores for all study variables are presented in Table 1.

Bivariate Associations
As seen in Table 2

Hierarchical Regression Analyses
Pain catastrophizing. Results of the hierarchical regression indicated that neuroticism was independently associated with greater pain catastrophizing, above-and-beyond the contributions of sociodemographic characteristics, pain severity, and depression (Table 3). At Step 1, pain severity was observed to account for a significant portion of the variance in pain catastrophizing scores (p < .0001), while none of the sociodemographic factors contributed independently (all ps > .11). As hypothesized, the addition of depression to the model at Step 2 increased the predictive utility (ΔR 2 = 0.34, p < .0001). Finally, the addition of neuroticism at Step 3 revealed that neuroticism accounted for a significant portion of the variance in pain catastrophizing, even after controlling for sociodemographic characteristics, pain severity, and depression (ΔR 2 = 0.16, p < .004).
Pain-related anxiety. Similar results were observed for the hierarchical regression that included neuroticism, depression and pain-related anxiety. At Step 1, pain severity (p < .0001) and age (p < .05) were observed to account for a significant portion of the variance in painrelated anxiety scores (see Table 4). The addition of depression at Step 2 increased the predictive utility of the model (ΔR 2 = .31, p < .0001). As hypothesized, at Step 3, neuroticism  accounted for a significant portion of the variance above-and-beyond the contributions made by pain severity, sociodemographic factors, and depression (ΔR 2 = .13, p < .0001). Exploratory Analyses. Separate hierarchal regression analyses were conducted to examine associations between pain catastrophizing, pain-related anxiety, and each of the individual neuroticism facets. As seen in Table 5, the neuroticism facets anxiety (p < .001), hostility (p = .016), depression (p = .033), and impulsiveness (p < .001) were positively associated with greater painrelated anxiety scores (after controlling for sociodemographic factors, pain severity, and depression). With regard to pain catastrophizing, positive associations were observed for the neuroticism facets anxiety (p < .001), hostility (p = .006), self-consciousness (p = .03), impulsivity (p = .006), and vulnerability (p < .001).

Discussion
Because previous research has tested associations between cognitive-affective pain constructs and each of neuroticism [13] and depression [22,23,24,25] in isolation, the goal of this study was to examine the relative contribution of both neuroticism and depression to indices of paincatastrophizing and pain-related anxiety among persons with chronic pain. As hypothesized, we found neuroticism to be uniquely and positively associated with both pain catastrophizing and pain-related anxiety, above-and-beyond the contribution of depression, even after controlling for pain severity and relevant socio demographic factors. We did not observe a depression by neuroticism interaction, which suggests that associations between neuroticism and our cognitive-affective pain constructs remained stable across varying levels of current depression.   Finally, exploratory analyses revealed that most individual neuroticism facets were related to both pain catastrophizing and pain-related anxiety. These findings are consistent with prior research that demonstrated positive associations between neuroticism and maladaptive cognitive processes (e.g., ruminative thinking) that may contribute to the onset and progression of chronic pain [49], and evidence that persons high in neuroticism may be more likely to perceive a pain stimulus as threatening [50,51,52]. These findings are also consistent with known relations between neuroticism and depression [53], and previous findings that depression may be associated with greater catastrophizing and painrelated anxiety [28,29,30]. Indeed, researchers have noted the importance of accounting for negative mood states when assessing cognitive-affective pain constructs [54], and the authors of a seminal meta-analysis concluded that all models of psychopathology should account for the influence of higher-order personality factors [53]. Results of the current study support the utility of measuring both trait neuroticism and state depression among persons with chronic pain. These data also suggest that, despite evidence of conceptual and statistical overlap between measures of neuroticism and depression [53], assessing either construct in lieu of the other may result in the loss of important information that may be of predictive clinical utility.
Strengths of the current study include the recruitment of a large sample of treatmentseeking chronic pain patients, the simultaneous testing of several constructs that have been shown to influence responses to pain and treatment outcomes [2], and the utilization of wellestablished measures that have demonstrated excellent reliability and validity in the population of interest. Several limitations also bear noting. First, these cross-sectional data preclude inferences of causality or temporal precedence. Second, some of the constructs (e.g., depression) may have been influenced by the onset and course of chronic pain, and we were not able to account for the characteristics of patients prior to the onset of their pain disorder. Third, the study sample was largely comprised of Caucasian females (70%) who were motivated to engage an intensive multidisciplinary pain rehabilitation treatment program. Thus, although the current sample was comprised of an important subgroup of pain patients that had not previously responded to treatment [55], these results may not generalize to all persons with chronic pain. Finally, we note that although depression and neuroticism were correlated in our sample, we believe multicollinearity did not adversely impact our analyses, which were focused on change in R 2 . Indeed, it is important to retain both depression and neuroticism in the models to examine the relative contribution of neuroticism above-and-beyond current negative affective states, the purpose of the study.
In summary, these findings represent an early but important step towards the clarification of complex associations between trait neuroticism, current depression, and tendencies toward catastrophic and anxiety-provoking appraisals of pain among persons seeking treatment for chronic pain. Future research would benefit from testing the differential effects of neuroticism and depression on cognitive-affective processes (e.g., changes in catastrophizing and pain-related anxiety), as they relate to the course of treatment for chronic pain. Future research is also needed to determine the extent to which associations between neuroticism and catastrophizing/pain-related anxiety may influence the outcomes of psychosocial interventions aimed at cognitive restructuring and increasing attentional control [16].