Post Traumatic Stress Symptoms to Depression and Pain in patients with Accident-Related Chronic Pain.
- By Randy S. Roth, * Michael E. Geisser*, and Rachel Bates.
* To see the tables that go along with this article, please go to, Category: Journal, Subcategory, Table(s)*
Abstract: Symptoms of post-traumatic stress (PTSD) are a common comorbidity in patients with a history of accident-related chronic pain and depression. However, little is known regarding the influence of PTSD in contributing to the affective distress, pain experience, and disability associated with chronic pain in this population. This study used structural equation modeling to examine 3 models that assess these relations in a sample of chronic pain patients with accident-related pain. Subjects were assessed for pain experience,depressive symptoms, anxiety, PTSD symptoms, pain disability, and relevant demographic variables. Pearson correlations indicated that symptoms of depression were significantly associated with higher disability. The model of best fit indicated that after controlling for the influence of anxiety on the dependent measures, PTSD symptoms have a direct influence on severity of depressive symptoms, whereas depressive symptoms have a direct influence on pain intensity and an indirect impact on pain intensity by way of their effect on disability. These data point to the importance of unresolved PTSD symptoms in contributing to the level of depression, pain, and disability exhibited by chronic pain patients and highlight the need to consider directed and primary treatment of PTSD in pain rehabilitation programs.
Perspective: This study highlights the impact of symptoms of PTSD on levels of depression, disability, and pain in patients with pain secondary to physical injury. Our results suggest that pain rehabilitation programs provide directed interventions for PTSD symptoms among this population to improve pain treatment outcomes.
Studies of chronic pain patients with a history of accidental injury and physical trauma attest to the influence of post-traumatic stress in enhancing the pain experience and dysphoria commonly observed in this population. Traumatic events are independently associated with the development of both major depressive disorder and post-traumatic stress disorder (PTSD). Importantly, symptoms of both depression and PTSD are risk factors for more severe pain experience and pain morbidity. For example, chronic pain patients with depressive symptoms report more severe pain intensity, greater pain-related disability, and relatively poor treatment outcomes compared with chronic pain patients who are not depressed. Similarly, chronic pain patients who describe their pain as accident-related with or without symptoms of PTSD, report higher levels of pain, affective distress, and greater functional disability when compared with patients with pain of insidious onset. Despite the consistent association of post-traumatic stress symptoms with both depressive symptoms and pain, little is known regarding the mechanisms by which post-traumatic stress may contribute to the course and severity of mood disturbance in this population. PTSD may be associated with depressive symptoms through the presence of neurobiological abnormalities that are common to both disorders. PTSD also may have an indirect effect on depression through its impact on behavioral disability. Rudy et al describes a cognitive behavioral model of chronic pain that in part draws attention to the effect of pain interference with life activities in mediating the relation between chronic pain and depression. For example; PTSD can contribute to functional impairment as the result of its association with panic disorders and agoraphobia, which promote generalized patters of behavioral avoidance and psychosocial disability. Thus, it is possible that the functional disability exhibited by a depressed chronic pain patient, attributed to either depression or pain or their combination, may actually derive from an embedded PTSD.
Alternatively, a history of physical trauma with related symptoms of post-traumatic stress may influence pain morbidity through their effect on the development of maladaptive pain beliefs and coping strategies. Turk and Okifuju and Vlaeyan et al have proposed that when pain is acquired as a result of physical injury, some individuals adopt maladaptive beliefs about the origin of their pain and incorrectly interpret their pain to reflect underlying and progressive structural damage. These beliefs are hypothesized to foster a fear of of any movement that is associated with pain, catastrophic worry about the uncontrollability and inevitable enhancement of pain experience, leading to the avoidance of activities that are associated with pain. In this cognitive-behavioral formulation of chronic pain disability, the spiral of pain catastrophizing, and fear of (re)injury result in progressive disability, deconditioning and depression. Of note, the model proposed by Vlaeyen et al postulates somewhat contrary to clinical intuition, that pain intensity results from both functional disability and depression rather than the converse. Cook et al have recently reported data to support Vlaeyen’s proposed relation of disability and depressive symptoms to pain experience. This study will examine the contribution of PTSD to pain experience, functional disability, and frequency of depressive symptoms through the use of structural equation modeling. We will specifically examine 3 possible models that describe these interrelations for a cohort of chronic pain patients with accident-related pain. For the first model, following the pathway proposed by Rudy et al, it is proposed that pain influences disability, which in turn predicts depression. In this model, PTSD symptoms are hypothesized to have a simultaneous influence on levels of depression and disability. Based on the pain-related fear model suggested by Vlaeyen et al, the second model examined the seperate influence of depression and disability on pain, with symptoms of PTSD hypothesized to independently predict levels of depression. Finally, the third model examines the severity of depressive symptoms in relation to both pain and disability, as prior research has reported that depression influences functional activity independent of pain. For this model, PTSD symptoms are examined in association with depressive symptoms, and the influence of disability on pain are examined on the basis of the model proposed by Vlaeyen et al. Because PTSD is known to share collinearity with a number of psychosocial variables (eg, anxiety) for which PTSD may serve as proxy in an observed statistical association with depressive symptoms, additional analyses will be conducted to examine the independent influence of PTSD on depression, disability, and pain by controlling for the influence of anxiety on the variables of interest.
Materials and Methods
Sample Selection
Subjects in the present study comprised 241 consecutive patients with chronic pain (defined as duration of 3 months or greater) referred to a university hospital pain rehabilitation program and who reported that their pain began after a traumatic injury. The mean age of the sample was 39.1 years ( SD=9.4). One hundred forty-two subjects were female, and 99 were male. Participants had a mean duration of pain of 39.0 months ( SD=49.3). Fifty-five percent of the subjects indicated that their pain began after an accident at work, 35% reported that their pain was caused by a car accident, 8% indicated their pain began as a result of- an ” other” accident, and the remainder reported that their pain was caused by an accident at home. THe most common site of pain was low back pain ( N=102), followed by pain in 3 or more sites ( n=53), neck and upper back pain( n=51), and thoracic pain ( n=12). Two hundred sixteen subjects were Caucasian, 16 were African-American, 3 were Asian-American, 1 was Native-American, and 6 were Hispanic . Educational achievement was as follows: One subject reported ahving less than a 6th grade education, 16% reported that they did not complete high school, 22% reported having a high school education, 41% reported having some college or technical school training, 11% completed college, 9% reported having a graduate or professional degree. Education data were missing for 1 subject. Maritial status indicated that 60% reported being married, 15% had never married, 17% indicated they were divorced, 3% were seperated, 5% lived with a significant other, and 1 subject was widowed.
Instruments
Pain Assessment
Subjects completed the McGill Pain Questionnaire (MPQ). the MPQ measures the subjective pain experience in a quantitative form. This measure consists of 20 groups of single-word descriptors describing a wide range of sensory,emotional, and cognitive features of pain experience. The MPQ provides a measure of the sensory, affective, and evaluative aspects of pain experience, as well as an overall summed total of pain experience intensity. ( PRI; Pain Rating Index). The PRI was used in this analysis to measure self-reported pain intensity. Repeat administration of the MPQ has revealed a 70.3% rate of consistency in the PRI score.
Disability Assessment
To measure disability due to pain, subjects were administered the Pain Disability Index ( PDI). The PDI measures the degree that pain interferes with various activities of daily living. Areas of disability asses include ( 1) Family/home responsibilities, (2) recreation, (3) Social activity, (4) occupation, (5) Sexual behavior, (6) Self-care, and (7) life-support activity. Ratings of disability for each functional area are quantified along a 10-point scale from ” no disability” to ” total disability.” A total disability score is calculated by averaging the scores from all the responses . For this study, the total PDI score was used as a measure of activity interference of pain disability. Test-retest reliability for the total score is moderately high (r=0.44), and some of the subscales significantly correlate with levels of patient pain behavior.
PTSD Assessment
To measure symptoms of PTSD related to pain, the Post-traumatic Chronic pain Test ( PCPT) was issued in a modified form. THe PCPT was developed specifically to assess post-traumatic stress symptoms among a cohort of chronic pain patients who report the onset of their pain to coincide in time with a traumatic event. THe PCPT measures the presence of 6 symptoms of PTSD as it related to the accident or injury that is associated with the onset of pain. In its original form, the PCPT used a true/ false response set but what modified for this study to solicit subject responses along a 7-point Likert scale. Subjects were asked to rate the frequency with which they experience specific symptoms of PTSD for the following items: ( 1) I feel upset or nervous when exposed to events that caused my pain; ( 2) Since the Injury, I find myself avoiding places or activities that remind me of the accident or injury; (3) I have recurrent and intrusive memories of the events surrounding my accident or injury; (4) I have experienced recurrent dreams about the events surrounding my accident or injury; (5) Since the accident or injury I have become “wound up” and startle easily; and (6) At times, I have suddenly felt or acted as if the accident or injury were actually happening again. Subject responses range from 0 indicating ” not at all” to 6 indicating ” very much.” The summary score used was the average of the responses to all 6 items. The PCPT has demonstrated good test-retest reliability. (0.90), split-half reliability is reported to be 0.59, and inter-rater reliability reported to be 0.93. Using the revised format of the PCPT as described above, Geisser et al found that the scale has good internal consistency as measured by Cronbach’s man-Brown method. In the p resent sample,, Cronbach was 0.88 and split-half reliability using the Spearman-Broen method was 0.87. Scores on the PCPT are known to be predictive of PTSD, when assessed by structured diagnostic interview, with a criterion-based validity Spearman correlation coefficient of 0.75.
Depression Assessment
The severity of depressive symptoms was assessed by the Center for Epidemiological Studies-Depression Scale (CES-D). The CES-D is a 20-item questionaire that assesses the frequency of depressive symptoms experienced by the subject during the previous week. These symptoms are measured on a 0-3 scale from ” none of the time” to ” all of the time,” and a total score is obtained by totaling all of the item responses. Recent studies among patients with chronic pain indicated that the CES-D has good predictive validity for predicting major depression.
Anxiety Assessment
Anxiety was assessed by using the Anxiety subscale of the Brief Symptom Inventory (BSI). The BSI consists of 53 items that are purported to measure 9 symptoms dimensions including somatization, obsessive-compulsivity, interpersonal sensitivity, depression, anxiety, hostility, phobic anxiety, paranoid ideation, and psychoticism. Scores for each of the subscales are computed by summing the responses and dividing by the total number of items on that subscale. Persons endorse the relevance of each item to their experience during the past 7 days on a 5-point scale, from ” not at all” ( 0) to ” extremely” ( 4). Internal consistency of the Anxiety subscale is reported to be .81 and test-retest reliability is .79.
Pain Questionaire
Last, Patients completed a general questionaire that included information regarding their pain duration, compensation and litigation status, and other demographic information.
Procedure
As part of their clinical evaluation, subjects were mailed the questionaires and asked to complete them before their initial evaluation appointment. This information was collected during the subject’s first appointment. The information was then examined for accuracy and thoroughness of completion. Data analysis was performed after approval from the University of Michigan Medical School Institutional Review Board. Retrospective data was collected after clinical assessment of patients referred to our center over a period of years; participant consent was therefore bypassed, as approved by the institutional review board.
Statistical Analysis
To examine the hypothesis that PTSD affects the experience of pain and disability through its association with us, Pearson correlations were examined first, then path models were tested, using Structural Equation Modeling (SEM). AMOS 5.0 was used to calculate the statistics for the models examined. Maximum likelihood estimation with bootstrapping was used to calculate the various parameter estimates. The fit of the path models was examined using the comparative fit index (CFI), adjusted goodness of fit index (AGFI), and root mean square error of aproximation (RMSEA). Acceptable levels of fit for these measures of CFI > .95, AGFI >.80, and RMSEA < .06. Path coefficients (p) for each model were calculated based on standardized regression coefficients. Error estimates for endogenous variables in the model were set at one. ?² values were also calculated for each model but were not examined as measures of fit because this statistic is influenced by factors such as sample size. The first model tested was based on results from Rudy et al, who reported that the impact of pain on depression in younger adults is mediated, in part, by one’s evaluation of how much pain interferes with daily activity. A model was constructed in which pain influences disability, which in turn predicts depression. PTSD was examined as a seperate influence on depression and disability. The second model examined was similar to the model presented by Cook et al, which examined the tenability of Vlaeyen’s fear-avoidance model of chronic pain. In the current dataset, we examined the seperate influence of depression and disability on pain. PTSD was added as a predictor for depression. Finally, a third model was examined whereby depression was allowed to be related to both pain and disability. PTSD was allowed to be associated with depression in this model, and disability is hypothesized to influence pain.
Results
Sample Means and Standard Deviations
The sample means and standard deviations for pain intensity, depression, PTSD symptoms, and disability are presented in table one. ( see other post category entitled: Journals: Tables) Because the data suggested that scores on the PCPT were positively skewed (0.91), these scores were log-transformed. This transformation reduced the skewness to 0.19. The remaining analyses examining scores on the PCPT were conducted on the log-transformed values.
Pearson Correlation Analysis
Pearson correlations between age, gender, pain intensity, disability, PTSD symptoms, and depression are presented in Table 2. Age and gender were not significantly correlated with any of the other variables of interest. Depression symptoms were significantly associated with higher pain intensity, greater PTSD symptoms, and greater disability. Also, greater endorsement of PTSD symptoms was significantly associated with higher disability and greater depressive symptoms.
Structural Equation Modeling Analysis
The first SEM is presented in Fig. 1. The ?² value for the model was statistically significant ( ?²(2)=8.6, p=.01), suggesting that the model did not fit the data. In addition, 2 of the goodness-of-fit statistics did not satisfy the a priori criteria for a good model fit ( CFI = .946; RMSEA = .117), but the AGFI was acceptable (.913).
The second model is presented in Fig. 2. Again, the ?² value was significant ( ?²(3)=45.5, p<.001), suggesting that the model was not a good fit. None of the goodness-of - fit statistics satisfied that a priori critera for a good model fit ( CFI= .655 ;AGFI =.735, RMSEA = .243).
The third model tested is presented in Fig. 3. ?² value for this model was not statistically significant ( ?²(2)=2.51, P= .29), suggesting that the model fit the data well. All of the a priori criteria for a good model fit were also satisfied. (CFI = .996 ; AGFI = .974, RMSEA =.033). This model indicated that PTSD symptoms have a direct influence on severity of depressive symptoms, whereas depressive symptoms have both a direct influence on pain intensity and indirect impact on pain intensity by way of their effect on disability. To determine whether the relationships observed in model 3 where due to an association of PTSD symptoms with anxiety, the model was recalculated while simultaneously examining or controlling for anxiety as measured by the Anxiety subscale of the BSI. First, anxiety was added as an additional variable in model 3 that could influence depression and was allowed to covary with PTSD. The ?² for this model was not statistically significant (?²(3) - 7.56, p = .11). CFI (.987) and AGFI (.954) were acceptable, but the RMSEA fell short of the acceptable limit (.061). Alternatively, another model was constructed that was identical to model 3, except that the transformed measure of PTSD was replaced with the standardized residuals of the transformed PTSD measure that excluded the influence of the BSI Anxiety subscale. These results are presented in Figure 4. This model had an acceptable fit, based on all 4 criteria ( ?²(2)=2.97, P=.23; CFI=.989; AGFI =.969; RMSEA = .045). Although the path coefficient from PTSD to depression is reduced compared with model 3, which did not control for anxiety, the influence of PTSD on depression remains statistically significant.
Discussion
Borrowing from 2 cognitive-behavioral models of chronic pain, the present stud used SEM to explore 3 models that examine possible contributions of PTSD symptoms to severity of pain experience, depressive symptoms, and disability among patients with chronic pain of traumatic onset. Consistent with previous studies, the correlational analysis found that symptoms of PTSD and depression are significantly correlated and further that both disorders correlate with perceived disability attributed to pain. For the SEM analysis, only 1 model demonstrated a significant fit for the data. In this model ( model 3), symptoms of PTSD were found to directly influence depressive symptoms, which in turn exerted both a direct influence on pain and an indirect influence on pain through its impact on disability. These relations remained significant even when the influence of anxiety on PTSD was statistically controlled. These data are interpreted to provide partial support for Vlaeyen’s chronic pain model of pain-related fear while also highlighting the critical importance of ongoing symptoms of PTSD in understanding the persistent affective disturbance and disability demonstrated by patients with accident-related chronic pain.
Influence of PTST on Depression
A main finding for this study suggests that in case of disabling accident-related chronic pain with co-morbid depression, PTSD symptoms may be critical to understanding the chronicity of both disorders and provide insight into intervention strategies for their remediation. Our data indicate that PTSD symptoms exert a significant influence on depressive symptoms, which then enhance disability and pain experience in multiple ways. Several studies have suggested that PTSD may be related to depression inc chronic pain patients with traumatic -onset chronic pain. For example, Toomey et al compared pain patients of idiopathic and post-traumatic onset. Patients with pain of post-traumatic onset reported higher levels of depression compared with pain patients whose pain was not associated with a physical trauma. Turk and of the idiopathic pain group, exhibited evidence for a depressive disorder. Importantly, our data suggest that reduction of symptoms of PTSD can have a significant influence on reducing depressive symptoms in chronic pain patients. Because depression is known to bear a negative influence on pain morbidity and pain treatment outcome, successful amelioration of PTSD symptoms in this population can be an important avenue for enhancing pain rehabilitation outcomes.
Treatment of PTSD in Chronic Pain
Increased attention to the treatment of PTSD as a primary focus in the rehabilitation of individuals with chronic pain and co- morbid depression is particularly important when prior treatment efforts for pain and depression have been disappointing, implicating the potential for PTSD symptoms to be an ” anchor” for prevailing mood disturbance and intractable pain disability. This concern is supported by evidence that patients with depression or chronic pain, with a co-morbid history of physical trauma, demonstrate a diminished response to treatment compared with a cohort of patients without a history of trauma. Moreover, recent clinical reports have described the indirect and successful treatment of intractable depression and chronic pain in patients with co-morbid PTSD only after instituting behavioral therapy ( eg, progressive exposure therapy) targeting the PTSD symptoms. As cognitive-behaviroal therapies with proven efficacy for the treatment of PTSD are now available to pain practitioners, it is noteworthy that these interventions are now being tailoreid within comprehensive pain rehabilitation programs. The mechanisms by which PTSD symptoms influence depressive symptoms in chronic pain remain unclear. Biological and psychological vulnerabilities may predispose some individuals who experience a physically traumatic event to develop both PTSD and major depression ( see Asmundson et al and Keane and Barlow for related discussion). PTSD symptoms may also contribute to depressive symptoms in some chronic pain patients by way of its association with intrusive memories of negative past experiences, a maladaptive cognitive style that is also present in major depression. From a treatment standpoint, a pain patient with PTSD may respond with a heightened psychophysiologic responce ( eg, muscle tension and spasm) to trauma-related and more general environmental stimuli, contributing to incsreased pain and interference with physical physical therapy efforts to restore muscle elasticity and function, leading to treatment failure and increased hopelessness and depression. Finally, PTSD symptoms may interfere with the psychological adjustment necessary for successful pain coping. For example, PTSD symptoms may mire a chronic pain patient in a whirl of fear, anger, feelings of victimization related to the trauma, and the experience of recurrent, trauma-related intrusions. Such emotional upheaval may delay or prevent a decision by the patient to accept and learn to live with his or hear pain and ” move on” to a more functional life despite persistent pain, portending continued affective distress as pain-related disability persists.
Implications for Cognitive-Behavioral Models of Chronic Pain
The present findings also have implications for cognitive-behavioral formulations of chronic pain disability. The SEM methodology did not replicate the previous report by Rudy et al, which found that, among a heterogeneous group of chronic pain patients, the relation of pain and depression was mediated by the perceived effect of pain to interfere with activity. The discrepancy may be due to differences in the sample composition of the 2 studies, particularly given that our sample consisted only of chronic pain patients who had sustained their pain as a result of injury. Furthermore, it is possible that the presence of PTSD symptoms may alter the interrelations of pain, disability, and depression and mitigate the mediational role of disability in explaining depressive symptoms among chronic pain patients. In addition, our study used the PDI as a measure of disability , whereas Rude et al assessed pain interference from the Multidimensional Pain Inventory. Although, the PDI asks patients to rate the degree to which pain interferes with their activities of daily living, which would appear to be comparable to a measure of pain interference, it is possible that these alternative measures of functional impairment are not comparable to a measure of pain interference , it is possible that these alternative measures of functional impairment are not comparable either psycho-metrically or conceptually. For example, pain interference implies the negative impact of pain on daily functioning, whereas pain disability may reflect, for many patients, a more global and stable level of decreased ability to function. Further research will need to determine the psychometric similarity of these factors as measures of pain-related functional impairment. The model of best fit for our SEM was drawn from a cognitive-behavioral paradigm that posits a central role for pain-related fear in explaining persistent and intractable pain disability. Vlaeyen et al, and more recently, Cook et al, have hypothesized that for patients who have chronic pain, after a physical injury the patient acquires a catastrophic cognitive style in experiencing and coping with pain, which leads sequentially to fear of movement and activity avoidance that results in multiple comorbidities including depression, disability, and disuse. Our SEM suggests a refinement of the Vlaeyen model. We replicated previous evidence that depression may have a direct influence of pain. However, our analysis further indicated a sequential influence of depressive symptoms on disability, which in turn influences pain experience. In addition, the present findings argue for a formalized inclusion of PTSD in the fear-avoidance for model of chronic pain proposed by Vlaeyen et al. As the model posits fear-avoidance for chronic pain patients who receive a physical injury , it is likely that many of these persons have post-traumatic stress symptoms or actual PTSD. Future studies should consider a formal assessment of PTSD in investigations of fear avoidance approaches to chronic pain, given the potential for PTSD symptoms to contribute to depression, anxiety, and avoidance. More specifically, the presence of PTSD may be associated with an enhanced risk for pain catastrophizing, which may lead to greater pain-related fear, activity avoidance, affective disturbance, and pain.
Study limitations
This study has several methodological constraints. First, although SEM permits an influence of causality among variables examined, the study design is cross-sectional and correlational, and, as such, no causal direction for the obtained associations can be definitively determined. In addition, the subject pool was composed of chronic pain population that derived from a tertiary care, university hospital locale. As a result, the clinical sample probably represents pain patients with relatively severe pain morbidity. Thus, our findings may not generalize to pain patients with comorbid depression and PTSD who present with acute pain or within community clinical care settings. Third, Our measures of depression, PTSD, and pain disability may constrain the validity of our findings. These diagnostic measures were obtained by self-report inventories and are susceptible to methodological problems associated with factors that distort self-report estimations of psychological functioning. Moreover, the assessments of depression and PTSD represent a measure of symptom severity for each disorder rather than a categorical diagnosis of major depression and PTSD such as would be determined by the strict adoption of DSM-IV diagnostic criteria. Future studies will nedd to replicate our findings using empirically validated interview schedules to determine categorical assignments of pain patients to major depression and PTSD diagnostic groups.
With, regard to the assessment of PTSD used in this study, the PCPT has preliminary empirical support for its psychometric validity as a measure of PTSD, but further investigation is warranted to ensure its divergent validity apart from the measures of affective distress and anxiety. In our sample, the average PTSD score was modest ( 1.7 on a 0-6 scale), suggesting that a majority of our subjects did not have severe PTSD symptoms. As our study design is cross sectional, it is possible that a third variable, for which the PCPT scores sever as proxy, may actually account for the relation between PTSD and depressive symptoms observed in our analysis. For example, PTSD is known to have a consistent and significant collinearity with anxiety disorders and affective distress. This is attributable, in part, to the diagnostic overlap between PTSD and somatic symptoms that are commonly associated with anxiety and affective distress such as emotion distress, physiologic reactivity, hyper-vigilance, difficulty concentrating, and exaggerated startle response. When we controlled for the influence of a measure of general anxiety in our analysis, the variance was substantially reduced although continued to have a significant influence on depressive symptoms. Although these findings support a unique association of PTSD symptoms with depressive symptoms that is independent of more general psychological disturbance, it raises concern that measure of PTSD may serve as a marker for anxiety or global affective distress in studies examining the relation of PTSD symptoms to depression among a chronic pain population. Taken together, future pain studies are encouraged to assess PTSD with psychometric tools with established and sound psychometric validity and to examine the effet of PTSD on depression in chronic pain patients while controlling for potential psychological comorbidities that are known to share covariation with PTSD, depression and pain.
Finally, our results and data from other centers encourage the consideration of prospective, controlled, investigations that isolate PTSD as a significant influence on depression in chronic pain patients and that examine the effect of targeted treatment of PTSD in ameliorating co-morbid symptoms of depression and disability.
