Increased Experimental Pain Sensitivity in Chronic Pain Patients who developed Opioid Use Disorder

Abstract

Objective.

Although the great majority of patients who take opioids for chronic pain use them appropriately and to good effect, a certain minority will develop the problematic outcome of opioid use disorder (OUD). Patient characteristics associated with the development of OUD in patients with chronic pain have been described, however relatively unexplored is how sensitivity to pain is associated with OUD outcomes.

Methods.

We examined for differences in response to static and dynamic experimental pain stimuli between patients with chronic non-malignant pain who developed OUD after starting opioid therapy (n=20) and those on opioid therapy who did not (n=20). During a single experimental session, patients underwent cold-pressor and quantitative sensory testing pain assays, and objective and subjective responses were compared between groups; the role of pain catastrophizing in mediating pain responses was examined.

Results.

Results suggested that both groups of opioid-dependent patients were similarly hyperalgesic to the cold-pressor pain stimulus, with non-parametric testing revealing worsened central pain sensitization (temporal summation) in those who developed OUD. Significant group differences were evident on subjective ratings of experimental pain, such that those who developed OUD rated the pain as more severe than those who did not. Pain catastrophizing was unrelated to pain responses.

Discussion.

Despite the small sample size and cross-sectional design, these findings suggest that experimental pain testing may be a novel technique in identifying patients with chronic pain likely to develop OUD, in that they are likely to evidence exacerbated temporal summation and to rate the associated pain as more severe.

Introduction

The majority of patients who are prescribed opioid analgesics for the management of chronic nonmalignant pain (CNMP) use the medication responsibly and can appreciate improvements in functionality and quality of life. However, careful estimates suggest that approximately 8% to 10% of this patient population will develop the problematic and potentially life-threatening condition of opioid use disorder (OUD) (1,2), mirroring the rate of past-year substance use disorders in the US population in general (3). This adverse outcome has only been accentuated in the midst of the current opioid crisis, with prescribed opioids ascribed a role in the increasing rates of opioid abuse and overdose (4). Particularly worrisome is that patients with CNMP are uniquely resistant to medication-assisted therapy (methadone, buprenorphine) for the treatment of OUD (5–8), thus when OUD does arise in this patient population, its management is uniquely challenging.

In response to these concerns, recent clinical guidelines and professional society position papers recommend that prior to initiating opioid therapy in patients with CNMP, providers screen patients to identify those at risk for developing an OUD (9–14). Utilization of screening tools, such as the Opioid Risk Tool (15,16) or the Screener and Opioid Assessment for Patients with Pain-Revised (SOAPP-R, 17), and identification of risk factors (i.e., family history of substance use disorder, personal history of substance use disorder, depression, adverse childhood experiences) (18–20) are recommended approaches for assessing risk, however each of these require patient veracity, insight in self report and are subjective in nature. Helpful in determining a given patient’s risk for developing OUD when prescribed opioids for CNMP would be a more objective indicator or correlate of its onset during opioid therapy.

A potential predictor of OUD in this complex patient population is the patient’s sensitivity to pain. The striking overlap between brain systems implicated in pain and reward processing as demonstrated by human functional imaging and direct brain stimulation studies suggest that pain and opioid reinforcement responses are intimately linked (21). Integrating findings across studies, animal data suggest that pain and opioid reinforcement responses covary such that strains with poor tolerance for pain are also likely to find opioids highly rewarding, with the converse also being true (22–25). Edwards and colleagues (26) provided cross-sectional evidence that patients with CNMP who scored highly on the SOAPP-R (indicating increased risk for problematic drug taking behavior) rated the pain associated with repeated punctate mechanical stimuli as more severe. More recent work suggests that this positive relationship between pain sensitivity and substance use disorder liability may be mediated by affective factors referred to as “distress intolerance” (27) or “pain catastrophizing” (28,29).

In an effort to evaluate the predictive relationship between pain sensitivity and the OUD in this patient population, we conducted a pilot study. We compared responses to static and dynamic experimental thermal pain stimuli between patients with CNMP receiving opioid therapy who did not develop an OUD over the course of 18-months of opioid therapy, and patients with CNMP who had been prescribed opioid therapy and subsequently developed an OUD over the course of pain treatment. Secondly, we considered the mediating role of pain catastrophizing on OUD outcome. We hypothesized that patients with CNMP who developed an OUD would be more sensitive to noxious thermal pain (cold, heat), and that the relationship between pain sensitivity and OUD development would be mediated by pain catastrophizing.

Materials and Methods

Study Design

All study procedures (including informed consent) were approved by the Institutional Review Board of the University of Pennsylvania; informed consent was obtained, and all study data were collected in a private human laboratory setting. Utilizing a cross-sectional survey design, during a single study session, all patients underwent two randomly ordered experimental thermal pain assays: static cold pain delivered via the cold-pressor test (CPT) and dynamic heat pain delivered via quantitative sensory testing (QST) apparatus. Objective CPT responses were operationalized as threshold and tolerance to the ice bath, whereas the QST measured temporal summation to a repeated heat stimulus, thus both peripheral nociception (static) and central modulation (dynamic) of the pain experience were captured. Immediately following each assay, patients rated the perceived or subjective severity of the pain stimulus. To evaluate its mediating role between perceived pain severity and OUD, patients provided responses to a validated pain catastrophizing scale.

Sample

Patients with CNMP and no history of an OUD were identified and recruited from the University of Pennsylvania Pain Medicine Center through referrals from practice clinicians and support staff. Study eligibility of consented patients was determined using information from the electronic medical record (EMR) and self-report measures. Inclusion criteria were as follows: adult patients (age ≥18) with a history of CNMP of musculoskeletal or neuropathic origin persisting ≥6 months in duration; had been receiving opioids consistently (monthly prescriptions) for 6 months or longer and with no past history of any substance use disorder (SUD) other than nicotine use. Exclusion criteria included: Patients with pain syndromes due to cancer; gynecologic, abdominal, visceral, dental, trigeminal neuralgia, post-stroke syndrome, or migraine-related pain; neuropathic pain due to metabolic disease; pain in the upper extremities which might interfere with experimental pain assays; untreated psychiatric conditions interfering with sensory processing or preventing the provision of informed consent or questionnaire completion; and patients with uncontrolled hypertension due to the vasoconstrictive effects of the CPT.

As a validation measure to ensure that patients in this cohort did not develop OUD, the EMR was reviewed for each participant from 6 months prior to the date of study consent. Patients were classified as not having an OUD (non-OUD, n=20) if the information in the EMR indicated the following: patients were receiving opioid therapy as defined above; had no current or past history of SUD (except nicotine dependence) or evidence of aberrant drug use behaviors (determined using an expert-derived checklist) (30); and all urine drug screens in the previous six months were appropriate (presence of prescribed opioid and absence of non-prescribed opioid or illicit drug). To confirm that these patients over time did not develop an OUD, EMR records were reviewed monthly for 12 months after completing baseline assessments and each non-OUD patient completed follow-up comprehensive assessments at 6 and 12 months. Thus, this group was evaluated for a total of 18months of opioid therapy, with the assumption that if OUD were to arise, evidence of it would be noted across this extensive observation period.

Patients with CNMP and concomitant OUD were recruited from a large Suboxone® clinic in Philadelphia, Pennsylvania. Inclusion and exclusion criteria were the same as for the patients with CNMP without an OUD as outlined above. Patients classified as having an OUD (OUD, n=20) had no previous history of SUD (except nicotine dependence) as defined by DSM-IV (31) criteria prior to beginning opioid therapy; currently endorsed DSM-IV criteria for “opioid dependence” on both the MINI International Neuropsychiatric Interview (32) and DSM-IV checklist, and were actively receiving treatment for an OUD. All enrolled patients were subsequently asked to provide baseline sociodemographic data obtained by telephone interview, and were reimbursed for study participation.

Measures

Pain Sensitivity.

Sensitivity to pain was measured using two highly reliable, valid and safe pain induction techniques, the CPT (33,34) and QST (35,36), employing procedures consistent with those described in the literature. As noted, the CPT was used to evaluate static evoked pain sensitivity whereas the QST assay was administered to assess dynamic central modulation processes on perceived pain severity.

The CPT apparatus consisted of a water bath maintained at 1°C ± 0.5°C, into which patients immersed their dominant hand and forearm to elicit a nociceptive response. A stopwatch was simultaneously activated with arm immersion, and subjects instructed to indicate in seconds when the cold sensation becomes a pain sensation (threshold), and when the pain becomes subjectively intolerable or spontaneous hand removal (tolerance); tests were stopped at 5 minutes to ensure that no tissue damage occurs. The CPT was administered three times at 10-minute intervals, and the mean of these responses entered into analysis. In addition, immediately following each trial, patients were asked to rate the perceived severity of the pain on a self-report visual analogue scale (VAS) from 0 to 100, where 0 is no pain, and 100 is extreme pain.

As a component of QST, a thermal testing analyzer (TSA; Medoc TSA-2001 device) with a sized 30 x 30 mm Peltier thermode was used for the assessment of temporal summation, reflecting an increased rate of firing of ascending pre-synaptic neuron with ongoing stimulation. Specifically, four (4) phasic noxious heat pain stimuli of 47°C (starting from 37°C at an increasing and decreasing rate of 10°C/s), each lasting 3 seconds with an interval of 12 seconds, was applied to the dominant volar. With each exposure, the patient was asked to verbally report the severity of pain experienced on a VAS (0- no pain, 100-worst pain ever). For analysis, temporal summation was operationalized as the difference-in difference response to the second, third and fourth stimulation compared to the response to the first stimulation. As with the CPT, immediately following the fourth stimulation, patients were asked to rate the overall perceived severity of the pain on a VAS from 0 to 100, where 0 is no pain, and 100 is extreme pain.

Pain Catastrophizing.

For each patient, a pain catastrophizing score was derived from the Coping Strategies Questionnaire which assesses patient’s self-rated use of cognitive and behavioral strategies to cope with pain (37). It is composed of six subscales for cognitive strategies (ignoring pain, reinterpretation of pain, diverting attention, coping self-statements, catastrophizing, praying/hoping), and two subscales for behavioral strategies (increasing activity levels and increasing pain behaviors); each subscale consists of six items measured with a numerical rating scale ranging from 0 (never do that) to 6 (always do that), thus has a maximum score of 36 and a minimum score of 0. The questionnaire takes approximately 5 minutes to complete, and good internal consistency (37–40) and test-retest reliability (40–42) for the catastrophizing subscale have been reported.

Procedures

Sessions were scheduled in the morning, and patients were instructed to not smoke or ingest caffeine for one hour prior to the study session. All were encouraged to take analgesics as usually scheduled, and provided an observed urine sample to ensure that only prescribed substances appeared upon toxicology testing. Prior to the pain tests, patients completed the paper-and-pencil Pain Coping Questionnaire. During the pain assays, patients were blindfolded and provided standardized scripted instructions designed to minimize social desirability; vital signs were monitored immediately prior to and following each pain assay.

Statistical Analysis

Demographics, opioid use and chronic pain severity means and percentages were compared between the OUD and non-OUD cohorts. Opioid use was calculated in morphine milligram equivalents (MME) per day utilizing the opioid equianalgesic conversion charts from the Centers for Disease Control and Prevention (CDC) ( https://www.cdc.gov/drugoverdose/pdf/calculating_total_daily_dose-a.pdf), and the North Carolina Association of Pharmacists (https://www.ncpharmacists.org/files/MME%20Table.pdf) for formulations not on the CDC chart. At the beginning of the study session, subjects rated the average severity of their chronic pain in the past 24 hours on a self-report VAS from 0 to 100, where 0 is no pain, and 100 is extreme pain.

Group differences between cold-pressor pain threshold and tolerance were inspected via non-parametric testing, as was the perceived pain severity for both static and dynamic responses to thermal pain. To evaluate temporal summation, a difference-in-difference analysis comparing the response to each stimulus to the response of the previous stimulus was conducted using the non-parametric Mann-Whitney U statistic. The mediating effect of catastrophizing on perceived pain severity in determining group membership was examined by evaluating the bivariate correlations between perceived cold and heat pain severity and catastrophizing, and running a logistic regression on group assignment (OUD vs non-OUD) with and without catastrophizing and evaluating differences in the beta coefficients.

Results

Participants in this study were on average 46 years old, primarily White, and with high school level of education (Table 1). With respect to type of pain, 16 had pain of musculoskeletal origin, 14 of neuropathic origin and 10 classified as mixed; type of pain was equally distributed across the groups. The severity of chronic pain was rated as moderate (VAS = 56.3/100) in the past 24 hours, and they were taking a daily average of 62 MME. The only difference between patients with an OUD versus patients without an OUD was that those in the non-OUD group perceived the severity of their chronic pain to be significantly more severe.

Table 1.

Sample Characteristics

Variable		Non-OUD		OUD		Total	p-value (t-test)*
		Mean or Count	SD or %	Mean or Count	SD or %	Mean or Count
Sex	Male	12	60.0	10	50.0	22	0.525
	Female	8	40.0	10	50.0	18
Race	White	16	80.0	11	55.0	27	0.204
	Black	1	5.0	5	25.0	6
	American Indian/Alaska	0	0.0	1	5.0	1
	More than 1	3	15.0	3	15.0	6
Education	9-12 grade	9	45.0	10	50.0	19	0.534
	GED/Graduate	8	40.0	6	30.0	14
	Some College	1	5.0	3	15.0	4
	Alternative School	2	10.0	1	5.0	3
Age (years)		46.91	9.41	44.35	10.84	45.63	0.431
Daily MME (morphine mg equivalents)		42.38	45.96	82.25	97.15	62.32	0.109
Average pain rating over the past 24 hours (VAS 0-100)		62.6	15.22	50.0	15.89	56.30	0.016

^*Difference between OUD and non-OUD groups; Non-OUD – Non-opioid Use Disorder, OUD – Opioid Use Disorder, GED – General Educational Development, VAS – Visual Analogue Scale

No differences were found in catastrophizing between the OUD versus the non-OUD cohorts (Table 2), and both groups scored in the lower range of the pain catastrophizing scale. Similarly, no group differences were noted on CPT threshold or tolerance (measured in seconds), with much individual variation noted for tolerance specifically. However, the patients with an OUD rated the subjective severity of both static measures of thermal pain to be significantly greater than those who did not (Table 2). In an effort to determine how large a sample might have been needed to detect group differences in CPT responses, a post-hoc power analysis revealed that a total sample size of 11,300 subjects would be needed to achieve 80% power to detect significance at the alpha = 0.05 level (two-tailed) based on our observed pain threshold effect size (d=0.053), and a sample size of 27,500 would be needed to achieve the same for our observed pain tolerance effect size (d=0.034). These large numbers suggest that it is highly unlikely that a Type II error was made.

Table 2.

Static Pain Responses by Group

Variable	Non-OUD		OUD
	Median	min-max	Median	min-max	p-value (MW-U)
Catastrophizing	13	4-30	11	6-23	0.586
CPT Pain Threshold in seconds	11.72	0-50	14.61	2.6-58.9	0.862
CPT Pain Tolerance in seconds	21.11	0-300	28.07	10-300	0.640
Maximal Cold Pain Intensity (VAS 0-100)	60	0-90	77.50	20-100	0.030
Maximal Heat Pain Intensity (VAS 0-100)	25	0-75	50	0-95	0.018

Non-OUD – Non-opioid Use Disorder, OUD – Opioid Use Disorder, MW-U – Mann Whitney U, CPT – Cold-pressor Test, VAS – Visual Analogue Scale

Similarly, group differences were found with dynamic temporal summation responses (Table 3); patients with OUD perceived each subsequent pain stimulus as significantly more painful than the original pain stimulus and their pain scores increased over the train of stimuli, whereas the pain scores of the non-OUD group decreased or remained the same over the stimuli train (Figure 1). As presented in Table 4, the correlations across pain severity ratings were highly consistent within thermal stimuli by group; both groups evidenced high correlations amongst the static and dynamic noxious heat measures, but these were poorly correlated with the static noxious cold measures.

Table 3.

Temporal Summation by Group

DnD Comparison	Non-OUD		OUD		p-value (MW-U)
	Mean	SD	Mean	SD
Peak 1 vs 2	−2.90	9.36	5.15	9.54	0.028
Peak 1 vs 3	−2.70	11.11	5.35	14.24	0.023
Peak 1 vs 4	−3.25	12.38	6.25	16.39	0.030

Non-OUD – Non-opioid Use Disorder, OUD – Opioid Use Disorder, DnD – Difference-in-Difference, MW-U – Mann Whitney U

Figure 1. Temporal summation in subjects who did and did not develop OUD.

Change in VAS pain severity score (0-100) in response to the first, second, third and fourth stimulation over a train of 4 3-second noxious (46.5°C) stimuli, interval 12 seconds.

Table 4:

Pearson correlations (p-value) of pain measures within Non-OUD and OUD cohorts

		Non-OUD – Cohort (n=20)
		TS Peak 1	TS Peak 2	TS Peak 3	TS Peak 4	VAS Cold	VAS Heat
OUD – Cohort (n=20)	TS Peak 1	-	0.915**	0.878**	0.847**	−0.077	0.822**
	TS Peak 2	0.950**	-	0.977**	0.971**	0.046	0.912**
	TS Peak 3	0.888**	0.960**	-	0.993**	0.137	0.878**
	TS Peak 4	0.859**	0.945**	0.984**	-	0.168	0.881**
	VAS Cold	0.431	0.398	0.308	0.310	-	0.077
	VAS Heat	0.873**	0.844**	0.797**	0.744**	0.483*	-

Non-OUD – Non-opioid Use Disorder, OUD – Opioid Use Disorder, VAS – Visual Analogue Scale, TS – Temporal Summation,

^*p<0.05,

^**p<0.001

The lack of significant correlations between perceived cold pain severity and catastrophizing (r=−0.110, p=0.536) and perceived heat pain severity and catastrophizing (r=0.015, p=0.931) suggest that catastrophizing does not mediate the relationship between perceived pain severity and OUD outcomes. This lack of mediation was validated by minimal improvements in prediction in the logistic regression analysis; adding the catastrophizing to the model resulted in beta coefficient only changes of only 0.001 of a point for both cold and heat pain perception variables (not shown). In an effort to determine how large a sample might have been needed to detect group differences in catastrophizing, a post-hoc power analysis revealed that a total sample size of 4,350 subjects would be needed to achieve 80% power to detect significance at the alpha = 0.05 level (two-tailed) based on our observed effect size (d=0.085), making it highly unlikely that a Type II error was made.

Discussion

This pilot study examined differences in static and dynamic experimental pain sensitivity between patients with CNCP and concomitant OUD versus those who did not develop an OUD when receiving opioid therapy for the treatment of CNMP. In addition, the potential mediating role of pain catastrophizing on the relationship between pain sensitivity and the development of OUD was evaluated. The descriptive results suggest that patients with OUD in this cohort were similar to those who did not with respect to sex, ethnicity, age, level of education, and type of pain. The non-OUD cohort rated their chronic pain intensity as more severe than the OUD cohort, which may be related to the finding that the non-OUD patients were on lower daily doses of opioids or less likely to cope by reinterpreting the pain, although this difference was not statistically significant.

With respect to static pain sensitivity, there were no group differences related to evoked CPT pain. Notably, relative to normative cold-pressor data (33,43), all patients appeared comparatively pain sensitive on the CPT assay, tolerating the ice bath for approximately an average of one minute or less. It is likely that their relative pain intolerance is a reflection of opioid-induced hyperalgesia (OIH) (44,45). Decreased tolerance for cold-pressor pain has been reliably demonstrated in persons receiving opioid therapy for the treatment of OUD (46–56) and patients receiving opioid therapy for the treatment of CNMP (57–59). Similarly, patients with CNMP and OUD are hyperalgesic to the CPT in comparison to patients with CNMP who are not on opioid therapy (60), and the degree of CPT hyperalgesia in chronic pain patients on opioid therapy is similar to those of patients without CNMP but prescribed opioids for OUD (61). The lack of group differences in CPT pain tolerance in this sample validates the findings of Fishbain, Lewis and Gao (62) who demonstrated that patients with chronic pain receiving opioid therapy with and without OUD were similarly hyperalgesic in comparison to matched controls on a mechanical pressure pain assay. Further, Walsh and colleagues (43) provide evidence that the presence of chronic pain alone (in the absence of opioid therapy) does not alter CPT responses, suggesting that opioid exposure in these patients with chronic pain results in equivalent levels of CPT hyperalgesia, irrespective of risk for OUD. The trend for somewhat lower CPT pain tolerance in the OUD group may be explained by the relatively higher dose of opioids to which they are exposed.

The groups did significantly differ on both the dynamic measure of temporal summation and static measures of perceived severity of the experimental pain stimuli. Notably, in both cases, these pain responses are based on a subjective rating of pain rather than a more objective indicator (i.e., removal of hand from ice bath); each stimulus in the temporal summation assay is rated on a scale of 0–100, as is the overall severity of the overall pain induced by the CPT and QST. This tendency to rate experimental pain as more severe was the best discriminator of group membership, is suggestive that patients with chronic pain who perceive a standardized CPT pain stimulus as severe (i.e., VAS ≥ 70) and standardized QST-induce heat pain stimulus as moderate to severe (i.e., VAS ≥ 50) maybe at higher risk of developing OUD. These findings support those of Edwards and colleagues (26) using a mechanical punctate pain stimulus, who revealed that patients with CNMP who scored highly on an opioid misuse risk tool rated the experimental pain as more severe than those with lower scores.

The results showing differential temporal summation response in the OUD patients is a relatively novel finding; not only did those with OUD rate the initial pain stimulus as more painful than non-OUDs, but each subsequent stimulus was rated as significantly more painful than the first. In cross-sectional studies in the laboratories of Chen (63) and Zhang (64), an exacerbated temporal summation response was similarly noted in patients with CNMP on opioid therapy in comparison to those not receiving opioid therapy and to matched healthy controls. However, no measures of OUD propensity were included in the design. A single urine for toxicology demonstrating the absence of illicit drugs was required for inclusion, however this relatively imprecise indicator of current or future OUD suggests that both patients at risk and not at risk could have been included in the sample. In the current study, when temporal summation was compared between the OUD cohort and the non-OUD cohort, little evidence of temporal summation was noted in the non-OUD group; clear temporal summation was evident in OUD group. This suggests that testing for temporal summation prior to initiating or during opioid therapy may provide an estimate of risk for developing OUD. Taken together, the pain response data indicate that central processes which contribute to or amplify the experience of pain may be more closely associated with OUD than is simple peripheral nociception.

These findings are at odds with the many studies suggesting a decline in temporal summation with opioid administration, however the majority of studies showing that opioids reduce temporal summation utilized an acute opioid dosing design in healthy controls or pain-free individuals (oxycodone [65], morphine [66,67], codeine [68], fentanyl [69, 70]) or in patients with chronic pain and not on opioid therapy (remifentanil [71], fentanyl [72]). Using the same acute dosing design, morphine was also shown to have no effect on temporal summation in healthy controls (73) or patients with chronic pain (74, 70); similarly, acute intravenous injection of tramadol in healthy controls had no effect on temporal summation (75). Conversely, the studies referenced above (63, 64) and others (76) which suggest enhanced temporal summation were more likely under chronic opioid dosing conditions in samples of persons with chronic pain, suggesting that the duration or chronicity of opioid exposure predicts the effects of the medications on temporal summation. Notably, the magnitude of change between first and last noxious stimuli for chronic pain patients on opioid therapy reported are within the range of change found in this analysis (77, 78).

In this sample, subjects reporting lower severity of chronic pain ratings demonstrated greater temporal summation. These findings are inconsistent with the 2016 work of Suzan and colleagues (71), who found that self-reported severity of chronic pain was positively and significantly correlated to the severity of temporal summation. These conflicting findings may be explained by differences in sample characteristics in that those in the referenced study were not currently on opioid therapy and were excluded if they had a history of opioid use disorder. Further, the reported positive correlation between clinical pain and temporal summation severity was only noted during remifentanil and saline conditions; this correlation did not hold in the control (baseline) condition, which more closely maps on the nature of the subjects in this work.

Surprisingly, pain catastrophizing was not associated with perceived experimental pain severity nor did it play a role in mediating the relationships among perceived pain severity and group membership. An accumulating literature suggests that the tendency to have a pain-specific anxiety, fear or affective distress, broadly conceptualized as being a “pain worrier”, (79,80) is associated with opioid misuse as well as problematic use of other substances. McHugh and colleagues (27) showed that a measure of distress intolerance is positively associated with misuse of prescribed opioids in patients with CNMP (and not associated with mechanical punctate pain threshold, tolerance or severity). Similarly, the investigative group lead by Martel found that pain catastrophizing was associated with higher levels of opioid craving (29) and risk for opioid misuse (81), to a greater degree than various demographic (i.e., age, sex), psychological (i.e., depressive symptoms), and pain-related (i.e., pain intensity, pain duration, opioid analgesic dose) variables. Using the same measure of pain catastrophizing used in the current study, Lutz and colleagues (28) likewise demonstrated an association with risk for opioid misuse. Conceptualizing this tendency to worry about pain as anxiety related to physical sensations, Rogers and colleagues (82) found that patients with chronic pain and greater anxiety sensitivity were more likely to report misusing their prescribed opioids and higher levels of opioid dependence. Zale and colleagues (83) found that male patients with chronic pain who had a high score on a measure of pain-related anxiety engaged in more problematic alcohol consumption and report higher levels of alcohol dependence, suggesting that the catastrophizing construct may be related to misuse of substances other than opioids. However, this body of literature evaluates the relationship between catastrophizing and potential correlates of OUD (i.e., misuse, craving, dependence), not OUD. Further, all of this evidence is cross-sectional in nature; patients were not followed to determine if OUD developed. It may be that catastrophizing results in misuse of substances to manage the anxiety or affective distress associated with pain, but is not associated with the development OUD.

Limitations

This pilot study was primarily limited by its small sample size as there were inadequate cell sizes to attempt any multivariable examinations, and analyses were primarily descriptive and nonparametric; notably a mediating effect of pain catastrophizing may have been appreciated in a larger sample. The measure of pain catastrophizing utilized in this study was a subscale of a larger coping scale, thus a mediating effect may also have been appreciated with the use of a more targeted measure, such as the Pain Catastrophizing Scale (84). In addition, the effects of other adjuvant pain medications (i.e., anticonvulsants, antidepressants) on pain responses were not considered in these analyses related to notable individual variation in their patterns of use. Findings were additionally limited by the reliance on equianalgesic conversion tables to enable comparison of opioid use by group; the pharmacology of different opioid formulations differ in properties other than analgesia (i.e., receptor selectivity, onset/offset of action) and there is great variability between equianalgesic conversion tables with regard to converting buprenorphine to morphine which were not considered in these analyses.

Importantly, the study was cross-sectional as opposed to prospective in design and we cannot assert predicting the risk of developing OUD and other explanations for group differences in pain responses may be possible. For example, the effects of differences in medications currently used (i.e., suboxone vs. other opioid analgesics), how recently they had taken the opioid, and the effects of baseline medications used (i.e., opioids or other adjunctive analgesics) cannot be accounted for. In addition, the overall length of time subjects were taking opioids was not considered.

Further, only a single measure of central sensitization was obtained. The work of Zhang and colleagues (64) and Ram and colleagues (85) suggest that diffuse noxious inhibitory control (DNIC) responses differ between patients with CNMP receiving as compared to patients not receiving opioid therapy. Although these data do not refer to risk for OUD, DNIC represents a descending form of pain modulation as opposed to temporal summation, which is an ascending form of pain modulation, thus a fuller understanding of the role of central sensitization in risk for developing OUD would be appreciated if included as a measure. Finally, the sample was overwhelmingly White. Racial differences in CPT responses (42) could not be evaluated, thereby limiting interpretation of the evoked pain data, and generalizability of the findings to populations of patients with CNMP and ethnic diversity.

Conclusions

The results of this pilot study suggest that subjective static and dynamic responses to experimental pain assays is associated with an OUD in patients with CNMP being considered or receiving opioid therapy. Potentially, patients who demonstrate temporal summation to a standardized heat stimulus, and who rate experimental cold and heat pain as more severe maybe more likely to develop OUD over the course of opioid treatment than patients who do not. These relationships do not appear to be mediated by pain catastrophizing, nor are they evident in threshold and tolerance responses to evoked CPT pain, which may be better explained by opioid-induced hyperalgesia. Replication of these findings in a larger, prospective and more ethnically diverse sample of patients with CNMP on opioid therapy is necessary to support the use of experimental pain testing to gauge a patient’s risk for developing OUD.