Post-Traumatic Stress Disorder Symptoms from Multiple Stressors Predict Chronic Pain in Cancer Survivors

Abstract

Background:

Despite the association between chronic pain and post-traumatic stress disorder (PTSD), little is known about the longitudinal course of pain and PTSD during cancer treatment.

Objectives:

We examined the prevalence of PTSD and chronic pain at three time periods in veterans with a diagnosis of cancer, and the relationship between the experience of pain and PTSD.

Methods:

Participants (N = 123) with oral-digestive cancers were recruited from the Veterans Healthcare System (age M = 65.31 and SD = 9.13; 98.4% male) and completed face to face interviews at 6, 12, and 18 months post-diagnosis. Measures included the Post-traumatic Stress Disorder Checklist-Stressor-Specific version (PCL-S), Primary care PTSD (PC-PTSD), and the Patient-Reported Outcomes Measurement Information System (PROMIS) Pain Impact Scale.

Results:

About one-third (26.8%) of the sample had chronic pain, defined as elevated pain at two time periods. About one-fifth (20.3%) endorsed symptoms of combat-related PTSD at 6 months, and 22.8% endorsed symptoms of cancer-related PTSD, exceeding a clinical cutoff for older adults (12 months = 21.1%, 18 months = 23.1%). Changes over time were observed for cancer-related PTSD symptom clusters of hyperarousal (F = 3.85 and p = 0.023) and emotional numbing (F = 4.06 and p = 0.018) with a statistically significant quadratic function increasing at 18 months. In logistic regression, individuals with both combat and cancer-related PTSD symptoms at six months had 8.49 times higher odds of experiencing chronic pain (χ² = 25.91 and p < 0.001; R² = 0.28).

Conclusions:

Persisting pain may be a concern in veterans with cancer. Individuals who have experienced traumatic events with persisting PTSD symptoms may be at elevated risk for chronic pain. Veterans with PTSD symptoms from both cancer and combat are at the highest risk to experience chronic pain.

Introduction

Numerous studies document associations between pain and post-traumatic stress disorder (PTSD). For individuals diagnosed with cancer, the relationship between PTSD and pain is particularly complex. Previous trauma, such as combat, is associated with increased pain in cancer patients.¹ In a study of individuals with lung cancer undergoing thoracotomy, the pain associated with the procedure served both as a traumatic stressor and predicted symptoms of PTSD.² Conversely, cancer pain may serve as a reminder of a traumatic stressor and drive further arousal.³ For some traumatic events, pain drives arousal and avoidance.^3,4 For individuals with cancer, the potential relationship between pain and PTSD is bidirectional. Cancer pain serves as a particularly vexing potential trigger being an internal stressor, something that cannot be easily avoided.

PTSD symptoms may arise following a traumatic stressor, although most individuals do not develop PTSD following trauma exposure. Life-threatening illnesses such as cancer have been in and out of inclusion criteria as a potentially traumatic event in the Diagnostic and Statistical Manual of Mental Disorders (DSM) (e.g., DSM-III—out, DSM-IV—in, DSM-5—out).^5–7 For DSM-5, language for a qualifying traumatic event changed identifying only medical incidents that involved catastrophic events. However, despite attempts at making PTSD criteria more objective, traumatization is a subjective experience. The individual’s perception of threat to life is critical.⁸ Importantly, previous research finds the perception of cancer as a life-threatening event and the development of cancer-related PTSD independent of cancer stage and severity.^9,10

Setting aside changes to DSM inclusion criteria, nearly two decades of research document rates of cancer-related PTSD in cancer survivors from 0% to 35%, often in the range of 10%.^11–13 This range is similar to rates of PTSD in persons following other traumatic stressors, although more individuals have symptoms of PTSD that do not meet full diagnostic criteria.^14,15 The literature investigating the connection between PTSD and cancer focuses on relatively few types of malignancies. In these studies, PTSD in cancer survivors is associated with prior trauma history.^13,16

In this article, we aim to extend the literature on PTSD and pain in adult cancer survivors with two goals. First, we describe the prevalence and longitudinal course of cancer-related PTSD and chronic pain in a sample of veterans following a diagnosis of cancer across varying stages and types. Second, we examine the relationship between PTSD symptoms associated with two potentially traumatic events, a recent stressor—cancer, and an earlier life stressor—combat, and chronic pain.

Methods

Participants

Participants were from the Veterans Cancer Rehabilitation Study (Vet-Cares).¹⁷ In this study, individuals were identified from Veterans Health Administration tumor registries in Boston and Houston and screened through the electronic medical record (EMR). Inclusion criteria were clinically confirmed diagnosis of and treatment for oral (head and neck), esophageal, gastric, or colorectal cancer. These cancer types were selected as they are relatively prevalent among veterans and have variable prognoses. Exclusion criteria were not English-speaking, presence of psychotic or dementing disorders, or currently receiving hospice care. Individuals were recruited through letter and phone.

Procedure

Participants completed in-person structured interviews 6(T1), 12(T2), and 18(T3) months following their cancer diagnosis at the medical center or in their home. This analysis focuses on 123 participants with complete data at all three time points. Of the 170 participants who completed T1 interviews, at T3, 14 were deceased, 16 declined interviews, and 17 could not be reached. The Institutional Review Boards of the Boston and Houston (Michael E. DeBakey) Veterans Affairs Medical Centers approved the study. Participants completed informed consent and were compensated for their time.

Measures

Demographics.

Participants reported their age, gender, racial and ethnic identities, and level of education.

Comorbidity index.

Comorbidity ratings used EMR extraction of 22 diagnoses from the Charlson Comorbidity Index, assigning a score of 1 for each illness.

Cancer information.

Information about the cancer diagnosis, stage, and treatment (surgery, chemotherapy, and/or radiation) was obtained from EMR.

Chronic pain.

Following the ICD-11 definition of chronic pain as lasting longer than three months, individuals were classified as having chronic pain if they reported elevated pain (>1 SD above community norms) for two time periods on the Pain Impact scale from the Patient-Reported Outcomes Measurement Information System (PROMIS) 29-item version.^18,19 The Pain Impact scale is a reliable and valid measurement of chronic pain in cancer patients,²⁰ considering the extent to which pain interferes with daily activities, work, social activities, and enjoyment of life, rated 1 (not at all) to 5 (very much), in the past seven days, for a possible range of 4–20.

Post-traumatic stress disorder.

Individuals were categorized as having no PTSD symptoms, elevated symptoms of either combat-related or cancer-related PTSD at time 1 or elevated symptoms of both combat-related and cancer-related PTSD at time 1. To assess combat-related PTSD, we asked if, while serving in the military, participants were “in a combat area” defined as “firearms, shelling, and mortars.” Of those who stated yes (39.8%), we administered the 4-item Primary Care PTSD Screen (PC-PTSD²¹;). A score of ≥3 was used to indicate a potential diagnosis,²² and a score of ≥1 was used to indicate elevated symptoms. This lower score was used because of its high sensitivity and our desire to be inclusive of adults who experience PTSD symptoms.^21–23

The Post-traumatic Stress Disorder Check List-Stressor-Specific version (PCL-S)^24,25 assessed cancer-related PTSD orienting participants to cancer as a potentially traumatic event (e.g., “Repeated, disturbing memories, thoughts, or images of cancer, your diagnosis, or treatment?”), as similarly adapted for breast cancer survivors.^9,26,27 A score of ≥37 was used to indicate a potential diagnosis in older adults,²⁸ while a score of ≥23 (median split) was used to indicate elevated symptoms.

To further investigate the symptom profile of PTSD in individuals with cancer, we created subscores for four clusters²⁹ for each time point: reexperiencing (items B1–B5), effortful avoidance (C1–C2), emotional numbing (C3–C7), and hyperarousal (D1–D5). We standardized cluster subscores by dividing the raw score by the number of items in each cluster, to permit comparisons of symptom severity between clusters. Of note, the DSM-IV version of these instruments was utilized as the data collection occurred when DSM-IV was in use.

Data analysis

Data analyses addressed our two aims. First, to achieve our aim of describing the prevalence and longitudinal course of cancer-related PTSD and chronic pain, we used mean and frequency statistics and compared these over time with repeated-measure ANOVA, evaluating linear and quadratic trends, with post hoc paired t tests. Second, to achieve our aim of examining the relationship between PTSD symptoms from two traumatic events and chronic pain, we used multivariate logistic regression to determine the relationship of PTSD to chronic pain, controlling for demographic variables.

We used logistic regression rather than linear regression as our primary independent variable (PTSD) was categorical and our outcome variable (pain) was dichotomous. We controlled for potential demographic predictors by including those associated with pain at the bivariate level in the logistic regression. We had both categorical (e.g., race) and continuous (e.g., age) predictors, which we dichotomized to permit comparisons across predictors by odds ratios. Analyses were conducted in SPSS.

Results

Participants

Demographic data for participants are presented in Table 1. Participants were mostly male (98.4%). Age ranged from 48 to 88 years (M = 65.63 and SD = 8.80), with about half (48.0%) age 65+. The majority (71.5%) described their race as white or Caucasian (81%) and 28.5% as black or African American, Hispanic, or multiracial. About half (47.2%) had a high school education or less. In terms of health, participants had between 1 and 9 comorbidities at the beginning of treatment. Cancer types were colorectal (57.7%), head and neck (35.0%), and esogastric (7.3%), of AJCC Stage I (27.6%), II (28.5%), III (23.6%), and IV (20.3%), treated by surgery (78.9%), chemotherapy (56.1%), and/or radiation therapy (38.2%). All but 12% of participants were finished with treatment at 18 months post-diagnosis.

Table 1.

Demographic Characteristics and Univariate Comparisons

		Total sample demographics		Bivariate comparisons with chronic pain
Variable	Category	n	%	With, %	Without, %	χ2	p
Age	<65	64	52.0	29.7	70.3	8.80	0.003
	>65	59	48.0	8.5	91.5
Education	HS or less	58	47.2	31.0	69.0	0.99	0.320
	College or more	65	52.8	23.1	76.9
Race	African American, Hispanic, Multiracial	35	28.5	40.0	60.0	4.32	0.04
	Caucasian (non-Hispanic)	88	71.5	21.6	78.4
Comorbidity	Low	72	58.5	29.2	70.8	0.48	0.49
	High	51	41.5	23.5	76.5
Cancer stage	Early (I–II)	69	56.1	29.0	71.0	0.37	0.54
	Advanced (III–IV)	54	43.9	24.1	75.9
Treatment	Surgery	97	78.9	28.9	71.1	0.97	0.33
	Chemotherapy	69	56.1	26.1	73.9	0.04	0.83
	Radiation therapy	47	38.2	27.7	72.3	0.03	0.87
Combat PTSD	Any Current Sx	25	20.3	52.0	48.0	10.13	0.001
Cancer PTSD	Elevated Sx 6 months	64	52.0	42.2	57.8	16.03	<0.001
Combined PTSD	None	56	45.5	8.9	91.1	19.50	<0.001
	Either	45	36.6	35.6	64.4
	Both	22	17.9	54.5	45.6

PTSD, post-traumatic stress disorder.

Pain

Pain impact scores.

Pain impact scores ranged from 4 to 20 at each time period. At 6 months post-diagnosis, the mean pain impact was M = 8.11 and SD = 0.94; at 12 months M = 8.12 and SD = 4.88; and at 18 months M = 8.68 and SD = 5.33. There was no difference in the mean level of pain impact reported at 6 months versus 12 months (t = 0.05 and p = 0.958) or 12 months versus 18 months (t = 1.24 and p = 0.213).

Elevated and chronic pain compared to community norms.

At 6 months post-diagnosis, 30.1% of the sample (n = 37) had pain that was elevated more than 1 SD above the community norm; at 12 months, 30.9% (n = 38) had elevated pain; and at 18 months, 30.1% (n = 37) had elevated pain. The percentage of the sample with chronic pain—elevated pain at two time periods—was 26.8% (n = 33). In univariate analyses (Table 1), chronic pain was associated with younger age, minority race, and PTSD, but not cancer type, stage, or treatments.

Post-traumatic stress disorder

Combat-related PTSD.

Forty-nine (n = 39.8%) participants stated, while in the service, they were also in combat. Within this sample, 13.8% (n = 17) scored above the cut score of 3 for PTSD, whereas 20.3% (n = 25) had symptoms of combat-related PTSD. In contrast, 19.5% (n = 24) stated they had been in a combat area while in the military, but do not have any current combat-related PTSD symptoms, and 60.2% (n = 74) stated that, while they were in the military, they were not in a combat area.

Cancer-related PTSD.

Cancer-related PTSD scores ranged from 17 to 70 at 6 months, M = 28.49 and SD = 12.90; at 12 months, M = 27.07 and SD = 12.08 (range 17–73); and at 18 months, M = 29.05 and SD = 14.34 (range 17–82). There was no difference in the mean level of PTSD reported at 6 months versus 12 months (t = 1.68 and p = 0.095), but there was a statistically significant increase from 12 months versus 18 months (t = 1.99 and p = 0.049). At 6 months, 22.8% of the sample was above a cut score of 37 for PTSD (12 months = 21.1% and 18 months = 23.1%).

Mean cancer-related PTSD cluster scores are presented in Table 2. The overall repeated-measure analysis of change over time was not statistically significant overall or for reexperiencing and avoidance. It was statistically significant for emotional numbing (F = 4.06 and p = 0.018) and hyperarousal (F = 3.85 and p = 0.023). Furthermore, the within-subjects quadratic function for the cluster emotional numbing was statistically significant (F = 7.10 and p = 0.0009), with a statistically significant decrease from time 1 to time 2 (t = 1.99 and p = 0.049) and statistically significant increase from time 2 to time 3 (t = −2.03 and p = 0.045). Similarly, the within-subjects quadratic function for the cluster hyperarousal was significant (F = 8.06 and p = 0.0005), with a statistically significant decrease from time 1 to time 2 (t = 2.04 and p = 0.044) and statistically significant increase from time 2 to time 3 (t = −2.21 and p = 0.029). Changes over time are visualized in Figure 1, using standardized cluster scores to permit comparisons between clusters. The mean standardized cluster score, averaging across time, was highest for hyperarousal (M = 1.82 and SD = 0.79), which was significantly higher than emotional numbing (M = 1.68 and SD = 0.76; t = −3.33 and p = 0.001), reexperiencing (M = 1.55 and SD = 0.71; t = −5.77 and p < 0.001), and avoidance (M = 1.39 and SD = 0.59; t = −8.78 and p < 0.001).

Table 2.

Cancer Related Post-traumatic Stress Disorder Cluster Scores on PCL-C (DSM-IV) at Each Time Point

	Time 1 (6 months)		Time 2 (12 months)		Time 3 (18 months)
Cluster	M	SD	M	SD	M	SD	F	p
Reexperiencing	7.83	4.07	7.49	3.88	7.93	4.36	0.55	0.58
Active avoidance	2.72	1.42	2.71	1.55	2.90	1.67	0.42	0.67
Emotional numbing	8.56	4.51	7.82	3.84	8.80	4.92	2.79	0.07
Hyperarousal	9.38	4.52	8.57	4.00	9.39	5.06	2.66	0.07

DSM-IV, Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition; PCL-C, PTSD checklist-civilian version.

FIG. 1.

Cancer-related PTSD cluster subscores at 6, 12, and 18 months post-diagnosis. Subscores were standardized by dividing the number of items within each cluster, to permit comparison across clusters. PTSD, post-traumatic stress disorder.

Relationship of PTSD to chronic pain

In univariate comparisons (Table 1), having chronic pain was associated with younger age, racial or ethnic minority status, having combat-related PTSD symptoms, or having cancer-related PTSD symptoms. In multivariate logistic regression (Table 3), only PTSD symptoms were associated with chronic pain. Age and race were not predictive in the multivariate model. Individuals with both combat- and cancer-related PTSD symptoms had 8.49 times higher odds of chronic pain than those with neither combat- nor cancer-related PTSD symptoms. Individuals with either combat- or cancer-related PTSD symptoms have 4.24 times higher odds of having chronic pain (omnibus χ² = 25.91 and p < 0.001; Nagelkerke R² = 0.28).

Table 3.

Association of Demographic and Post-traumatic Stress Disorder Variables with Chronic Pain in Multivariate Analyses

	B	S.E.	Wald	p	Odds ratio	95% CI
Younger age	0.96	0.51	3.60	0.058	2.62	0.97–7.08
Minority race	0.51	0.51	1.01	0.316	1.67	0.61–4.56
Either combat or cancer PTSD Symptoms	1.44	0.59	5.94	0.015	4.24	1.33–13.53
Both combat and cancer PTSD symptoms	2.14	0.67	10.28	0.001	8.49	2.29–31.38

Analysis included only those variables significantly associated with chronic pain in bivariate analyses. Reference group for PTSD analyses is the group with neither combat- nor cancer-related PTSD symptoms.

Discussion

The main findings within this sample are threefold. First, about one-quarter of sample individuals have chronic pain in the period 18 months post-diagnosis. Second, about one-fifth have probable cancer-related PTSD (based on cut-score) and more have elevated PTSD symptoms. Furthermore, these symptoms are relatively stable in the period 18 months post-diagnosis, with hyperarousal and emotional numbing symptoms most elevated with possible reexacerbation at 18 months. Third, PTSD symptoms are associated with the experience of chronic pain, and those with PTSD symptoms associated with combat and cancer are most at risk.

Until recently, most studies of cancer-related PTSD were cross-sectional, but in recent years, a longitudinal literature has emerged examining the course of PTSD in individuals with lung cancer (N = 93, over six months),³⁰ breast cancer (N = 166, over one year),³¹ blood cancer, post-transplant (N = 102 [with complete data) over one year]³² ovarian cancer (N = 63 [with complete data] over three months), and mixed samples (N = 203 [with complete data] over four years)³³ groups. In general, these studies find low rates of diagnosed PTSD (as low as 2.0%) and high rates of PTSD symptoms (as high as 82.5%), and relative stability with modest fluctuations in the short term, with potential improvement in the longer term (e.g., four years). Similar to our findings, PTSD symptomatology is an important concern in the short term for many, and the long term for some.

The complex relationships between PTSD and pain are intriguing, and bring an added dimension to the concept of “total pain.” While integrative treatment approaches for chronic pain and PTSD are reported in populations with noncancer pain, few integrative treatment approaches have been demonstrated for individuals with cancer-related pain.³⁴ Approaches should consider treatment burden and possible community-based and telehealth approaches, as individuals move from active treatment to survivorship and are likely to seek fewer hospital-based visits.

For individuals with cancer and chronic pain, providers may consider screening for psychological risk and comorbidity factors, including PTSD. Future research is needed to understand if treatment aimed at improving symptoms of PTSD subsequently improves pain in this population. Interventions that focus on hyperarousal, the cluster most elevated in this sample, and associated with pain in other samples,^35,36 may be particularly fruitful.

Several theoretical models postulate the underlying mechanism connecting PTSD and pain such as the shared vulnerability and mutual maintenance models.⁴ In brief, both shared vulnerability and mutual maintenance models suggest a role for anxiety sensitivity as a predisposing vulnerability, and mutual maintenance elaborates on the idea that pain may trigger reminders of trauma—triggering arousal—which may trigger pain.³ Avoidance behaviors may be a natural response and thought to potentially sustain or exacerbate both PTSD and pain.³⁷ Studies are not consistent on these factors, but certainly point to a complex synergy between PTSD and pain.³⁸ These relationships have been studied primarily when cancer is not the inciting traumatic event, which may have implications on how treatment is addressed.

In our sample, avoidance symptoms were lowest in severity, perhaps because individuals undergoing treatment and/or surveillance of their cancer are to some degree, unable to avoid interaction with the health care system. Further research is needed to understand the relationship between PTSD and pain in cancer, and to create and validate treatment interventions for cancer-related PTSD, as well as measurement of chronic pain as an outcome.

Notable in our study was the increased likelihood of chronic pain when individuals reported both combat-related and cancer-related PTSD symptoms. Previous trauma appeared to make sample individuals more susceptible to development of PTSD with future stressors. For individuals diagnosed with cancer, an assessment of previous trauma and current related PTSD symptoms may help clinicians to identify individuals at risk of chronic pain. In this sample 40% were combat veterans, some of whom had concurrent combat-related PTSD symptoms. Those with symptoms may be at risk, and those without may demonstrate important resiliency.

Limitations

There are numerous limitations to our study. Given the period of data collection, our PTSD screener does not reflect the newer four cluster diagnosis as defined in DSM-5. We assessed PTSD by self-report, and only focused on either combat or cancer as potentially traumatic events. Our study is limited by the use of two different methods to assess PTSD (one for combat and one for cancer), which were self-reported and not validated by a clinician, but both measures demonstrate similar efficacy.²¹ While we clued participants to trauma “e.g. from cancer,” PTSD symptoms from one trauma may overlap with PTSD symptoms from another. Furthermore, although bivariate analyses did not suggest an association between cancer stage and pain, small sample size limited our ability to consider differences in pain among multiple subgroups or within subgroups of individuals who may have had recurrence or progression of disease.

Lacking a validated method to define chronic pain in a cancer sample, we used a statistical approach of 1 SD elevation at two or more time periods, which needs clinical validation. We did not use a cancer-specific pain measure. Our pain measure may not have fully captured pain associated with specific treatments. Furthermore, as participants were identified through the cancer tumor registry, we did not have a measure of PTSD or pain before diagnosis. In addition, we dichotomized some continuous variables (e.g., age). We find such an approach enhances clinical application by producing more information closer to clinical reasoning for patients with complex comorbidities, but such dichotomization results in a loss of statistical information.

Our study participants were mostly male and recruited from the Veterans Health Administration. Although veteran status is common in the United States with approximately one in two men older than 65 years being veterans,³⁹ and our findings on the prevalence of PTSD and pain are similar to non-veteran samples, the extent to the association between PTSD and pain in the veterans sampled are different from non-veterans is unknown. Our sample included different cancers, providing not only a good representation of a general palliative clinic or oncology practice but also making generalizing the association of pain and PTSD in specific cancer types difficult.

In conclusion, one-fifth of our sample of veterans with cancer experienced cancer-related PTSD, with prominent hyperarousal and emotionally numbing symptoms. PTSD symptoms were associated with chronic pain, and the risk of developing chronic pain increased with traumatic exposures. Further work is needed to replicate these results in other non-veteran samples, but our work deepens the understanding of “total pain.” Clinicians may consider the role of traumatic exposure and PTSD when managing chronic pain in cancer survivors.