Associations among sleep disturbance, pain catastrophizing, and pain intensity for methadone-maintained patients with opioid use disorder and chronic pain

Abstract

Objective:

This study examined the cross-sectional associations among pain intensity, pain catastrophizing, and sleep disturbance among patients receiving methadone maintenance treatment (MMT) for opioid use disorder (OUD) and reporting co-occurring chronic pain.

Methods:

Participants were 89 patients with OUD and chronic pain drawn from a larger cross-sectional study of 164 MMT patients who completed a battery of self-report measures. We conducted six mediation models to test all possible pathways (i.e., each variable tested as an independent variable, mediator, or dependent variable).

Results:

The only significant mediation effect was an indirect effect of sleep disturbance on pain intensity via pain catastrophizing. That is, greater sleep disturbance was associated with greater pain catastrophizing, which in turn was associated with greater pain intensity.

Discussion:

Altogether, findings suggest that the sleep disturbance to pain catastrophizing to pain intensity pathway may be a key mechanistic pathway exacerbating pain issues among MMT patients with OUD and chronic pain. These results suggest that interventions targeting sleep disturbance may be warranted among MMT patients with OUD and chronic pain. Future work in this area with longitudinal data is warranted.

Introduction

Chronic non-cancer pain is a biopsychosocial condition that is often accompanied by co-morbid insomnia, and other psychiatric and medical illnesses.[1–3] Recently, research has begun to explore the role of sleep in relation to pain characteristics, due to estimates suggesting that as many as 50% of chronic pain patients report a co-occurring sleep disorder.[4–7] Preliminary findings report that chronic pain and sleep disturbance have a reciprocal relationship, such that greater sleep difficulties worsen pain intensity, and greater pain intensity worsens sleep difficulties, all of which can predict new-onset pain.[8–11] This cycle is also compounded by the use of opioids, a once common treatment for chronic pain.[12] For example, opioids target the μ-opioid receptors within the ventrolateral preoptic nucleus, which is the same nucleus involved in sleep regulation.[13] Polysomnography data have revealed that acute administration of opioids can produce adverse effects on the normal cyclical pattern of sleep (sleep architecture) and disordered breathing, leading to the development of central sleep apnea, obstructive sleep apnea, ataxic breathing, and hypoxemia.[14] Chronic use of opioids can exacerbate these existing sleep issues. Specifically, studies have found robust dose-response associations between prolonged opioid therapy and the development of ataxic breathing and chronic sleep apnea.[15] Further, prolonged opioid use is known to produce hyperalgesia in some patients, thereby making these individuals more sensitive to and less tolerant of pain.[16]

Among those who develop an opioid use disorder (OUD), risk for sleep disorders, ataxic breathing, and death rise precipitously.[17] This risk is compounded by use of sleep and anxiety medications, such as benzodiazepines and sedatives, that are often taken to mitigate sleep difficulties[11]. Due to the reciprocal nature between substance use and sleep, sleep disturbance has been implicated as a universal risk factor for relapse across substance use disorders. Most of this research has been done among individuals with alcohol use disorders, wherein baseline sleep measures have predicted future relapse.[18–21] However, there is limited research on this pathway among those with chronic pain and opioid use disorder, and even less among those on methadone maintenance treatment (MMT).[22] Preliminary data suggests that rates of co-occurring sleep disorders are much higher among methadone maintained patients as compared to those with only chronic pain, with estimates as high as 75% of patients endorsing sleep disturbances.[23] Those in the initiation stage of MMT report worse sleep efficiency, shorter total sleep time, more awakenings, and shorter slow wave sleep as compared to healthy controls.[24] Further, methadone maintained patients who had been in treatment for longer than 6 months reported similar sleep difficulties, yet these did not worsen after another 6 month follow-up.[25, 26] Higher doses of methadone, worse pain severity and interference, daily tobacco use, weekly benzodiazepine use, and presence of anxiety or depression have all been associated with more sleep disturbances among MMT patients.[11, 27, 28]

There is robust evidence that pain-specific psychological processes are key contributors to worsening sleep, pain, and overall functioning. [17, 29, 30] Pain catastrophizing, which is an overappraisal of the negative aspects of experiences characterized by intrusive and negative thought patterns, has been shown to exacerbate pain intensity and sleep disturbances independent of physical disability in both chronic pain patients and MMT patients. [31–34] Poor quality of sleep the night before is also associated with higher pain severity, worse physical functioning, and more frequent pain catastrophizing the following day, further supporting the cyclical nature between sleep, pain intensity, and pain catastrophizing. [35] To date, there have only been two studies that have examined possible mediational pathways between sleep, pain intensity, and pain catastrophizing. One study found pain catastrophizing to mediate the relationship between sleep disturbance and pain intensity. [36] The other study found pain intensity to mediate the relationship between sleep disturbance and pain catastrophizing, however this study failed to test pain catastrophizing as a mediator.[37] Additionally, both of these studies were conducted among chronic pain patients, limiting our knowledge about how these relationships may change among MMT patients with OUD and chronic pain.

Given that patients with chronic pain have worse MMT outcomes, such as higher rates of relapse, frequent sleep disturbances, and worse psychiatric distress [23, 38–44], and there is preliminary evidence that pain catastrophizing may be a possible maintaining mechanism among those with chronic pain, it is important to consider how pain catastrophizing, pain intensity, and sleep disturbance may be related to each other among MMT patients with chronic pain. Specifically, this work may highlight potential treatment targets among a historically undertreated and poorly understood population. Therefore, the current study conducted a series of mediation models to examine the associations among pain intensity, pain catastrophizing, and sleep disturbances among MMT patients with OUD and chronic pain. Due to the mixed findings in chronic pain populations and a lack of research among MMT patients with OUD and chronic pain, we did not put forth any specific hypotheses about what significant mediational pathways would emerge.

Materials and Methods

Participants and Procedures

A total of 89 MMT patients with both OUD and chronic pain were included in the present study. These 89 patients were drawn from a larger cross-sectional study (n = 164) of MMT patients with OUD. The subset of patients with co-occurring chronic pain was obtained by identifying those patients in the larger study who responded “Yes” to the following item: “Are you currently experiencing physical pain that has lasted for 3 months or longer?”

The inclusion criteria for the original cross-sectional study required participants to be at least 18 years of age, English speaking, and currently receiving MMT services for OUD. Participants were recruited from the APT Foundation, Inc., a private not-for-profit community-based organization located in New Haven, CT, with a census of approximately 4,300 MMT patients when the study began. All data was collected between January 2014 and March 2015. Participants were recruited through flyers posted at three methadone clinics affiliated with the APT Foundation. The flyers stated that participants would be asked about their experiences during treatment as well as their treatment needs. Research assistants administered the questionnaire packet in-person. Participants were compensated $15 for completing the study. This study was approved by the APT Foundation Board and was presented to the Human Investigations Committees at Yale School of Medicine, which exempted it from review per United States Department of Health and Human Services (DHHS) regulation 45 CFR 6.101(b)(2).

Demographic and MMT characteristics for the sample (n = 89) included in the current study can be found in Table 1.

Table 1.

Participant demographic and treatment-related characteristics.

Demographics	Mean (SD) or N (%)
Age	45.18 (10.05)
Gender
Female	33 (37.1%)
Male	56 (62.9%)
Race
American Indian	1 (1.3%)
Asian or Pacific Islander	1 (1.2%)
Black	27 (32.1%)
White	48 (57.1%)
Other	7 (7.9%)
Marital Status
Never married	46 (51.7%)
Married or cohabitant	18 (20.2%)
Divorced, separated, or widowed	25 (28.1%)
Level of Education
< High school	26 (29.2%)
High school or GED	33 (37.1%)
Some college or vocational training	22 (24.7%)
College degree, or vocational license	8 (8.9%)
Employment Status
At least some employment	12 (13.4%)
Not working, not disabled	52 (58.4%)
Disabled	25 (28.1%)
Methadone Treatment Characteristics
Methadone dose in milligrams	81.02 (30.68)
MMTP duration in months	39.25 (47.94)
Number of MMTP treatment enrollments	2.09 (1.42)
Chronic Pain Characteristics
Duration of Current Pain Episode (months)	56.37 (73.19)
Location of Pain
Back	60 (67.4%)
Shoulder	24 (27%)
Head	15 (16.9%)
Neck	30 (33.7%)
Face	1 (1.1%)
Feet	22 (24.7%)
Pelvis	12 (13.5%)
Legs	41 (46.1%)
Arms	12 (13.5%)
Hands	19 (21.3%)
Perceived Cause of Pain
Accident-Related	49 (55.1%)
Surgery-Related	15 (16.9%)
Sports-Injury	10 (11.2%)
Disc-Related	9 (10.1%)
Nerve Pain	20 (22.5%)
Physical/Sexual Assault	6 (6.7%)
Cancer-Related	0 (0%)
Alcohol Withdrawal	2 (2.2%)
Opiate Withdrawal	9 (10.1%)
Non-Opiate Withdrawal	1 (1.1%)
Arthritis	27 (30.3%)
HIV-Related	3 (3.4%)

Notes. SD = standard deviation; GED = General Equivalency Diploma; MMTP = methadone maintenance treatment program.

Measures

Demographic Questionnaire.

Self-reported demographic information was collected, which included age, gender, race/ethnicity, marital status, employment status, and education. Methadone treatment characteristics were also collected through self-report, which included methadone dose in milligrams, number of months enrolled in MMT, and number of treatment enrollments at the APT Foundation MMT clinics.

Brief Symptom Inventory-18 (BSI-18)[45].

The BSI-18 is an 18-item abridged version of the original 53-item questionnaire, which was developed to screen for psychiatric conditions in medical and community populations. It has three subscales (anxiety, depression, and somatization) that range from 0 to 24, as well as a global severity rating scale that ranges from 0 to 72. Participants are asked to rate psychiatric symptoms on a 5-point Likert scale from 0 (not at all) to 4 (extremely). Scores from the depression subscale of the BSI are presented as standardized t-scores.

Brief Pain Inventory (BPI)[46].

The BPI is a well-validated measure of pain intensity and pain interference. This study focused on the pain intensity subscale, which includes four items. Specifically, these items ask respondents to rate their current pain intensity, as well as their worst, least, and average pain intensity in the last 7 days. All responses were measured on a scale of 0 (no pain) to 10 (extreme pain). The pain intensity score was computed as the mean of the four pain intensity items, with higher scores indicating worse pain intensity. The BPI is highly sensitive in classifying pain among individuals maintained on methadone [47].

Pain Catastrophizing Scale (PCS)[48].

The PCS is a 13-item questionnaire that evaluates three dimensions of catastrophizing, which include rumination, magnification, and helplessness. Participants are asked to reflect on past painful experiences and to indicate how often they experience each statement when they experience pain. Responses were measured on a 5-point Likert scale from 0 (not at all) to 4 (all the time). A total score was used for the current analyses, which was calculated by summing all items, and can range from 0 to 52 with higher scores indicating more frequent pain catastrophizing.

Pittsburgh Sleep Quality Index (PSQI)[49].

The PSQI is a 19-item questionnaire that assesses sleep quality and the presence of any sleep disturbances during the previous month. Responses are measured on a 4-point Likert scale from 0 to 3. The PSQI has seven subscales (subjective sleep quality, sleep latency, sleep duration, habitual sleep efficiency, sleep disturbances, use of medications for sleep, and daytime dysfunction), and subscale scores are summed to create a global sleep severity score that can range from 0 to 21. Higher scores indicate worse sleep quality, with global scores >5 indicating poor sleep quality.

Statistical Analyses

Descriptive statistics and bivariate correlations were conducted in SPSS Version 24. A series of path models were conducted using Mplus Version 8.[50] Each path model tested used bias-corrected bootstrapped confidence intervals to test the significance of the indirect and direct effects.[51] For all path models, we controlled for age, gender, ethnicity, and depression. A total of six path models were conducted, with sleep disturbance, pain intensity, and pain catastrophizing variously entered as the independent variable, the mediating variable, or the dependent variable. There were no missing data on study variables.

Results

Descriptive Results

Table 2 presents the descriptive statistics and internal consistencies (Cronbach’s alpha) of key study variables included in the path models, as well as the correlations among these variables. There were significant positive correlations among depressive symptoms, sleep disturbance, pain intensity, and pain catastrophizing – with the exception of no significant correlation between depressive symptoms and pain intensity.

Table 2.

Descriptive statistics and intercorrelations among study variables.

Variable	M	SD	α	1	2	3
1. Depression (BSI)	59.28	11.14	.90	--
2. Pain catastrophizing (PCS)	30.04	13.48	.95	.343*	--
3. Sleep disturbance (PSQI)	11.88	4.11	.78	.225*	.330*	--
4. Pain intensity (BPI)	5.64	1.50	.86	− .072	.425*	.221*

Notes. Higher scores reflect more impairment on all scales. Depression score reported in t-scores (M = 50, SD = 10). Pain catastrophizing scores can range from 0 to 52. Sleep disturbance scores can range from 0 to 21. Pain intensity scores can range from 0 to 10.

^*p < .05.

Path Models

All models demonstrated acceptable model fit based on CFI > 0.9 and RMSEA < .08. The results of all the path models are summarized in Figures 1–3. When sleep disturbance was tested as the independent variable (Figure 1), there was no significant indirect effect of sleep disturbance on pain catastrophizing via pain intensity (Model 1). However, there was a significant indirect effect of sleep disturbance on pain intensity via pain catastrophizing (Model 2). In other words, pain catastrophizing mediated the association between sleep disturbance and pain intensity, such that greater sleep disturbance was associated with greater pain catastrophizing, which in turn was associated with greater pain intensity.

Figure 1.

Path models with sleep disturbance as the independent variable. The following covariates were controlled for (not shown in figure): age, gender, ethnicity, and depression. Unstandardized regression coefficients and standardized errors are displayed for the a and b paths. For the direct and indirect effects, the unstandardized estimate is shown, as well as bias-corrected bootstrapped 95% confidence intervals. * = p < .05. β = standardized regression coefficient.

Figure 3.

Path models with pain catastrophizing as the independent variable. The following covariates were controlled for (not shown in figure): age, gender, ethnicity, and depression. Unstandardized regression coefficients and standardized errors are displayed for the a and b paths. For the direct and indirect effects, the unstandardized estimate is shown, as well as bias-corrected bootstrapped 95% confidence intervals. * = p < .05. β = standardized regression coefficient.

When pain intensity was tested as the independent variable (Figure 2), there was no significant indirect effect of pain intensity on pain catastrophizing via sleep disturbance; yet there was a significant direct effect of pain intensity on pain catastrophizing (Model 3), such that greater pain intensity was associated with greater pain catastrophizing. There was no significant indirect effect of pain intensity on sleep disturbance via pain catastrophizing, nor a significant direct effect of pain intensity on sleep disturbance (Model 4). Finally, when pain catastrophizing was tested as the independent variable (Figure 3), there were no significant indirect effects of pain catastrophizing on pain intensity via sleep disturbance (Model 5), or of pain catastrophizing on sleep disturbance via pain intensity (Model 6).

Figure 2.

Path models with pain intensity as the independent variable. The following covariates were controlled for (not shown in figure): age, gender, ethnicity, and depression. Unstandardized regression coefficients and standardized errors are displayed for the a and b paths. For the direct and indirect effects, the unstandardized estimate is shown, as well as bias-corrected bootstrapped 95% confidence intervals. * = p < .05. β = standardized regression coefficient.

Sensitivity Analyses

We conducted additional sensitivity analyses in which we also controlled for the anxiety and somatization subscales of the BSI, as well as methadone dose. The substantive pattern of results from the main analyses remained unchanged.

Discussion

The current study sought to evaluate the role of sleep disturbance among methadone maintenance treatment (MMT) patients with both opioid use disorder (OUD) and chronic pain. We tested six mediation models (i.e., pathways) with sleep disturbance, pain catastrophizing, and pain intensity variously included as independent variables, mediators, and dependent variables. Out of all the mediation models, the only significant indirect effect that emerged was an indirect effect of sleep disturbance on pain intensity via pain catastrophizing. That is, greater sleep disturbance was associated with greater pain catastrophizing, which in turn was associated with greater pain intensity. Both longitudinal and cross-sectional studies have provided support for this pathway in other chronic pain samples.[35, 36] To our knowledge, our study is the first to provide evidence for the sleep disturbance to pain catastrophizing to pain intensity pathway among MMT patients with OUD and chronic pain.

Although the current findings are cross-sectional and preliminary in nature, there are important implications that can be discussed. First, sleep quality has been understudied in both chronic pain and methadone-maintained populations, and these findings suggest that it may be an important intervention target to help reduce pain catastrophizing and pain intensity. Among methadone-maintained patients, few pharmacological and behavioral interventions for sleep have been tested. To date, only a small pilot study examining Cognitive Behavioral Therapy (CBT) for insomnia found significant reductions in sleep disturbances as compared to no intervention,[52] and two other small studies found herbal remedies to improve sleep quality.[53, 54] Given the dearth of research in this area, testing other commonly used and effective cognitive and behavioral sleep interventions are sorely needed within this population. For example, a recent meta-analysis found sleep interventions to produce a medium effect size in sleep quality and duration among adults without diagnosed sleep disorders. The most common intervention components used across studies were stress management skills, relaxation exercises, stimulus control, sleep hygiene skills, and exercise.[55] These components have also been efficacious among individuals with chronic pain.[56] Therefore, while MMT patients with OUD and chronic pain may present with worse sleep disturbance severity, future research should still pursue examining these sleep interventions among this population.

Second, pain catastrophizing and its features of rumination, magnification and helplessness, can also be a specific treatment target to help reduce pain intensity. There are multiple interventions that have successfully reduced pain catastrophizing. Among chronic pain patients, CBT, Acceptance and Commitment Therapy (ACT), and some psychoeducation programs have reduced pain catastrophizing, which in turn was associated with improved functioning outcomes.[57–59] However, among MMT patients with OUD and chronic pain, there have not been any studies that have examined the impact of an intervention on pain catastrophizing. One recent study tested a CBT intervention targeting pain interference and illicit opioid use among methadone-maintained patients with chronic pain, but did not examine pain catastrophizing.[60] Future research should examine interventions, such as CBT and ACT, among MMT patients with OUD and chronic pain.

Lastly, more complex mediation models involving pain catastrophizing and other important pain-specific processes should be tested in future work in order to better understand the relationship between sleep and pain intensity among MMT patients with OUD and chronic pain. For example, pain acceptance and self-compassion are strongly related to improved sleep quality and functioning among individuals with chronic pain.[61–63] Previous work has found changes in pain acceptance and pain catastrophizing to account for equal and unique amounts of variance in treatment outcomes over and above changes in pain intensity among a large chronic pain sample.[58] In addition, self-compassion and pain acceptance were also found to be the strongest mediators in treatment outcomes, over and above engagement in valued activity and pain coping strategies.[62] This suggests that these processes are important in promoting functioning and sustaining treatment gains, and therefore particularly important to target during treatment. However, these processes have not been tested among MMT patients with OUD and chronic pain, nor has the role of sleep disturbances been explored in relation to these processes. Future research should build from the current findings and examine how specific treatment mediators, such as pain catastrophizing, pain acceptance, and self-compassion may change in relation sleep disturbances among MMT patients with OUD and chronic pain. In addition, given that MMT patients also frequently struggle with relapse, it would be important to also explore how sleep and pain specific treatment mediators may affect substance use. This might further highlight unique pathways involving sleep, pain catastrophizing, and pain intensity that could elucidate other effective treatment targets among MMT patients with OUD and chronic pain.

One surprising finding was the lack of significant correlation between depressive symptoms and pain intensity. It is possible that this finding reflects survey selection bias, such that our sample likely consisted of more high functioning individuals. Recent data shows close to a third of patients with opioid use disorder never receive any substance use treatment[64], and that those experiencing moderate to severe pain are even less likely to receive substance use treatment[65]. Therefore, while we may have captured a chronic pain sample, it is possible that we captured those who are functioning relatively well with chronic pain.

The clinics that supplied our patient population are designed to provide methadone maintenance for patients with opioid use disorder. While primary care is integrated within the clinic and available to patients for regular medical attention, it is likely that chronic pain may be underrecognized and undertreated. We have developed specific training of substance abuse counselors at these clinics to aid in detection of chronic pain, which we discussed in another manuscript[66]. The results of the current study provide support for consideration of additional training for addiction counselors in recognizing sleep disturbances and pain catastrophizing.

Limitations

There are several limitations to our study. First, the data used for the current study was cross-sectional in nature, therefore the temporal precedence between variables in the mediation models cannot be confirmed. Studies that employ longitudinal methods are needed to instantiate these findings. Second, although this sample is relatively large and was recruited from three clinic locations, it is possible that the current findings may not generalize to other MMT populations. In particular, the current sample was recruited from one geographic location and consisted of primarily White individuals (57.1%). Studies among more racially and geographically diverse samples are needed. Our study population had, on average, been in treatment for a long time (mean of 39 months) and on a mean methadone daily dose of 81mg. This dose was titrated clinically to treat patient’s OUD rather than their chronic pain. Five previous studies have reported similar methadone doses among individuals with OUD and chronic pain, with average methadone daily doses ranging from 73mg to 90mg [38–40, 44]. Third, all assessments used in the current study were based on self-report, therefore, the measures of sleep disturbances and depression may not be accurately capturing sleeping patterns or psychiatric distress related to depression. Future studies should use other reliable and valid measurements of sleep and psychiatric diagnoses, such as polysomnography and structured clinical interviews, to further elucidate the complex relationship between sleep, pain, and pain catastrophizing. Similarly, the inclusion of patients with chronic pain based on their response to a single question may have failed to include some participants from the larger study. The high sensitivity of the BPI could have had the opposite effect, where patients without chronic pain were included in the analysis, based on their self-report. Fourth, the sample in this study (n = 89) was relatively small. However, the bias-corrected bootstrap method for mediation analyses is appropriate for samples of this size and has greater statistical power than other mediation methods (e.g., Sobell Test)[67]. Moreover, there were no missing data in the current study. Lastly, this study did not assess for use of licit or illicit substances. Therefore, it is unclear as to how substance use may also affect sleeping patterns, psychiatric distress, and pain intensity. Future research should explore how substance use may impact the mediation models presented in this study. Specifically, sedative and benzodiazepine use would be important to assess given that they are often prescribed for sleep difficulties, and methadone-maintained patients with chronic pain are more likely to misuse benzodiazepines than those without chronic pain.[27, 68, 69]

Conclusion

Despite these limitations, this study demonstrates that sleep disturbance may underlie pain catastrophizing, which in turn exacerbates pain intensity among MMT patients with OUD and chronic pain. Future studies will be necessary to evaluate effective interventions for sleep disturbances in this population. These interventions can also focus on decreasing pain catastrophizing as a way to help reduce pain intensity.