Self-reported cumulative medical opioid exposure and subjective responses on first use of opioids predict analgesic and subjective responses to placebo-controlled opioid administration

Abstract

Background and objectives

To expand the evidence base needed to enable personalized pain medicine, we evaluated whether self-reported cumulative exposure to medical opioids and subjective responses on first opioid use predicted responses to placebo-controlled opioid administration.

Methods

In study 1, a survey assessing cumulative medical opioid exposure and subjective responses on first opioid use was created (History of Opioid Medical Exposure (HOME)) and psychometric features documented in a general sample of 307 working adults. In study 2, 49 patients with chronic low back pain completed the HOME and subsequently rated back pain intensity and subjective opioid effects four times after receiving saline placebo or intravenous morphine (four incremental doses) in two separate double-blinded laboratory sessions. Placebo-controlled morphine effects were derived for all outcomes.

Results

Two HOME subscales were supported: cumulative opioid exposure and euphoric response, both demonstrating high test–retest reliability (Intraclass Correlation Coefficients > 0.93) and adequate internal consistency (Revelle’s Omega Total = 0.73–0.77). In study 2, higher cumulative opioid exposure scores were associated with significantly greater morphine-related reductions in back pain intensity (p=0.02), but not with subjective drug effects. Higher euphoric response subscale scores were associated with significantly lower overall perceived morphine effect (p=0.003), less sedation (p=0.04), greater euphoria (p=0.03) and greater desire to take morphine again (p=0.02).

Discussion

Self-reports of past exposure and responses to medical opioid analgesics may have utility for predicting subsequent analgesic responses and subjective effects. Further research is needed to establish the potential clinical and research utility of the HOME.

Trial registration number

NCT02469077.

INTRODUCTION

The potential negative consequences of managing chronic pain using opioids are increasingly recognized.^{1 2} This has spurred interest in identifying individual difference factors that predict opioid-related benefits versus risks for use in clinical decision-making in patients for whom opioid therapy is being considered.^{3 4} To facilitate more appropriate initiation of opioid therapy, well-controlled studies are required to identify variables predicting opioid effectiveness.^{3 4}

Prior work suggests several phenotypic factors that may predict opioid analgesic responses. For example, studies suggest that negative affect (eg, depression),^5–10 ‘fibromyalgia-ness’^{11 12} and evoked pain sensitivity^13–15 all may predict opioid analgesic efficacy. Other work has focused on identifying factors associated with elevated risk for opioid misuse, such as medication-related attitudes and behavior.^{16 17}

Given the known risks associated with medical opioid use, more predictive information is needed to guide appropriate initiation of opioid therapy. This work focuses on two understudied potential predictors of opioid response. Specifically, greater cumulative medical opioid exposure over the life span might potentially, via enhanced opioid-induced glial activation and associated inflammation, be related to diminished opioid analgesic effectiveness.¹⁸ Alternatively, individuals who, based on their experiences, know that they obtain excellent analgesia from opioids may be more likely to seek out opioids when experiencing a painful condition (ie, self-selection). Individuals who experience positive subjective responses on first exposure to an opioid (eg, feeling euphoric) might also be more likely to fill future opioid prescriptions and seek more opioid refills. This behavior could be driven by the operant reinforcing properties of opioids (reduced negative feelings, increased positive feelings), which are also regarded as contributors to opioid abuse liability.^19–22 Consistent with this latter idea, one prior study found that retrospective reports of euphoric opioid responses on initiating therapy for chronic pain were higher in ‘prescription opioid addicted’ patients than in patients with chronic pain using opioids without signs of opioid misuse.²³ Taken together, evidence suggests that self-reported subjective responses on first use of opioid analgesics might predict the degree to which reinforcing opioid effects (and misuse risk) are experienced on future opioid exposure.

To our knowledge, there are no previously validated self-report measures specifically indexing cumulative medical opioid exposure or subjective responses on first exposure to an opioid analgesic. The measure of subjective opioid responses used in the Bieber et al²³ study described above evaluated responses when initiating opioid therapy for chronic pain and was a modification with unknown psychometric properties of a previously published lengthy instrument (the Addiction Center Research Inventory).²⁴ Given the absence of proven measures to assess the two understudied aspects of prior medical opioid exposure targeted in the current work, we first (study 1) sought to develop and demonstrate the psychometric properties of a brief measure assessing cumulative exposure to medical opioid analgesics and subjective responses to first opioid use. Then, in an entirely independent sample using laboratory assessment methodology (study 2), we examined whether these two aspects of prior medical opioid use were predictive of actual analgesic and subjective responses when patients with chronic low back pain (not using daily opioids) were subsequently administered an opioid analgesic under placebo-controlled conditions.

METHODS AND RESULTS

We describe below a sequence of two independent studies, targeting first the development of a new survey, the History of Opioid Medication Exposure (HOME; study 1), and then evaluating the predictive validity of the HOME in terms of responses to controlled morphine administration (study 2). Due to the different samples and methodologies, these studies are presented in series, with separate descriptions of the methods and results of each. All procedures in both studies were approved by the relevant institutional review boards, and all participants provided informed consent prior to participation.

Study 1

Sample

Participants in study 1 were working adults (n=308; age≥18 years) responding anonymously to an online survey invitation sent via the employee email system at a large academic medical center. Study 1 participants completed the HOME questionnaire twice, once at the beginning of the online survey and then again after completing a series of questionnaires including a validated measure of opioid misuse risk. Participants did not receive any remuneration for participation in study 1.

Measures

HOME questionnaire

Items for the HOME were generated rationally to assess the two theoretically driven aspects of opioid exposure and responses described previously: (1) estimated cumulative medical opioid exposure and (2) recalled subjective responses to first use of an opioid analgesic (see online supplementary appendix 1/Supplemental Digital Content 1 for a copy of the measure). The cumulative opioid exposure items consisted of seven items designed to assess use of opioid analgesic medications for medical reasons across the life span. Participants indicated (1) whether they had ever taken opioid analgesic medications for any reason, (2) their estimate of the number of separate opioid prescriptions filled in their lifetime, (3) whether they had ever received general anesthesia for a surgical procedure, (4) whether they had ever been hospitalized overnight, (5) longest duration of continuous daily opioid analgesic use, (6) whether they would categorize themselves as ever having used opioid analgesics on a regular basis and (7) duration of time elapsed since their last use of an opioid analgesic. Due to the likelihood of increased reporting error with attempts at more fine-grained assessment of specific opioid dosages, dosing frequencies and opioid agents over the lifespan, items assessing these more detailed aspects of past medical opioid use were not included in the HOME.

The first subjective response to opioid analgesic items contained a checklist of 12 adjectives that individuals were asked to use (by checking any items applying) to describe how they felt the very first time they recall receiving an opioid analgesic medication. Adjectives selected were similar to those previously used by Zacny and colleagues to measure subjective effects of opioids.^25–27 Adjectives were selected to reflect both positively (eg, ‘energized’, ‘euphoric’, ‘comfortable’) and negatively (eg, ‘sleepy’, ‘nauseous’, ‘uncomfortable’) valenced items.

Opioid misuse risk

In an attempt to provide initial evidence of convergent validity of the HOME regarding its relevance to opioid misuse risk (ie, that the HOME is associated in the expected direction with scores on an independent validated measure of opioid misuse risk), the Screener and Opioid Assessment for Patients with Pain-Revised (SOAPP-R) was administered. The SOAPP-R is a validated self-report questionnaire designed to assess risk of opioid misuse in chronic pain populations.^{28 29} It comprises 24 items that are rated on a 5-point scale indicating frequency ranging from 0 (never) to 4 (very often). The SOAPP-R score is the sum of all 24-item responses. Higher SOAPP-R scores indicate higher opioid misuse risk.

Data analysis

With the exception of measures of internal consistency, all statistical analyses were performed in IBM SPSS Statistics for Windows V.24.

First, basic descriptive statistics (eg, mean, SD, frequencies) were used to examine item psychometric properties on the HOME. Because the cumulative opioid exposure items and first response to opioid analgesic items were designed to be conceptually distinct, empirical support for subscales within each set of items was examined separately. Within both sets of items, two-step cluster analysis procedures examined the pattern of participant responses to items in order to assess conceptual coherence and determine whether single or multiple subscales were supported by the data. Internal consistency of items comprising each proposed subscale was evaluated using Revelle’s Omega Total, a measure of internal consistency (ranging from 0 to 1.0) that is considered superior to the traditional alpha index.³⁰ Revelle’s Omega Total was computed for each subscale with the psych package in R V.3.4.0 (R Foundation for Statistical Computing, Vienna, Austria).^{31 32} Any potential subscales with poor internal consistency reliability were dropped from further analyses. Pearson correlations (r) were used to evaluate the relation between HOME subscales and an existing validated measure of opioid misuse risk (SOAPP-R). Finally, the short-term test–retest reliability of the HOME items and subscales were evaluated with intraclass correlations (ICCs; absolute agreement, single measures).

Results

Of the 308 individuals who started the questionnaire battery, one participant did not complete the first iteration of the HOME and was excluded from all analyses. Of the 307 individuals who completed the first administration of the HOME, 13.0% (n=40) reported they had never taken an opioid analgesic medication. Thus, primary analyses of HOME subscales (including test–retest reliability) were conducted on the 267 participants who reported having previously taken an opioid analgesic medication. Item responses indicated that 28.8% of the sample had previously used opioid analgesics continuously for at least a few weeks or more, with 5.2% describing themselves as a regular opioid user. Of the full sample, 26.2% reported using opioid analgesics within the past year.

Cumulative opioid exposure subscale

Results from two-step clustering procedures on items 2–7 designed to measure prior medical opioid exposure indicated a two-cluster solution was optimal (Silhouette measure of cohesion and separation=0.4). Participant responses clustered in a pattern supportive of items representing a single cumulative opioid exposure construct. Based on item response patterns (see table 1), one cluster was characterized by individuals with more lifetime opioid exposure across all items (n=136, 50.9%), while the other cluster was characterized by individuals with less lifetime opioid exposure across all items (n=125, 46.8%).

Table 1.

Cumulative opioid exposure items by cluster derived using two-step cluster analysis

HOME item	More exposure (n=136)	Less exposure (n=125)	P value
2. Number of opioid prescriptions ever? M (SD)	3.5 (1.32)	2.4 (0.76)	<0.001
3. Ever surgery where unconscious? (%)	99.3	68.0	<0.001
4. Ever hospitalized overnight? (%)	97.8	4.0	<0.001
5. Longest continuous opioids? M (SD)	2.2 (1.21)	1.5 (0.76)	<0.001
6. Ever used opioids regularly? (%)	10.3	0.0	<0.001
7. Last opioid use? M (SD)	3.4 (0.94)	3.8 (0.55)	<0.001

HOME, History of Opioid Medication Exposure survey.

Because cluster analyses supported an underlying cumulative opioid exposure construct, a total cumulative opioid exposure subscale score was created based on items 2–7. Continuous items were dichotomized based on being below (coded as 0) or above (coded as 1) the median of item distributions for the full sample to facilitate the creation of a total score. Specifically, number of opioid prescriptions (item 2) received a score of 1 if the participant endorsed receiving six or more opioid prescriptions in their lifetime. Longest continuous use (item 5) received a score of 1 if the participant endorsed ever using opioids longer than ‘a few weeks or more’. Last opioid use (item 7) received a score of 1 if the participant endorsed any option indicating opioid use within the past year. Items 3, 4, 6 (all coded 0 for ‘no’ and 1 for ‘yes’) and recoded dichotomous items 2, 5 and 7 were summed to yield a total score for the cumulative opioid exposure subscale ranging from 0 to 6, mean (SD)=2.1 (1.44). Internal consistency reliability for the six items comprising the cumulative opioid exposure subscale was 0.77, indicating adequate internal consistency.

First response to opioid analgesic subscales

Results from two-step clustering procedures on the 12 items assessing first subjective response to an opioid analgesic indicated a three-cluster solution were optimal (Silhouette measure of cohesion and separation=0.3). Frequencies at which each item was endorsed across the three clusters are detailed in table 2. Inspection of item frequencies across clusters suggested that two non-overlapping empirically-based subscale scores were appropriate to consider. One cluster, labeled low sedation, was distinguished from the other clusters by low endorsement of items reflecting feeling ‘sleepy’, ‘sedated’, and ‘numb’, and higher endorsement of feeling ‘energized’ on first receiving an opioid analgesic. Another cluster, labeled euphoric response, was distinguished by more frequent endorsement of items reflecting feeling ‘euphoric’, ‘comfortable’, ‘happy’ and ‘powerful’, and less frequent endorsement of items reflecting feeling ‘nauseous’, ‘uncomfortable’ and ‘drunk’ in response to first exposure to an opioid analgesic. A low sedation subscale (mean (SD)=1.7 (0.92), range=0–4) was created summing the items noted above, coded as 1 if endorsed and 0 if not endorsed, with ‘sleepy’, ‘sedated’ and ‘numb’ reverse-scored. Similarly, a euphoric response subscale (mean (SD)=2.9 (1.29), range=0–7) was created, with ‘nauseous’, ‘uncomfortable’ and ‘drunk’ items reverse-scored. For both empirically derived subscales, higher scores reflected greater presence of the construct indicated. The four items comprising the low sedation subscale exhibited poor internal consistency (0.35), so this subscale was not further examined. The seven items comprising the euphoric response subscale exhibited adequate internal consistency reliability (0.73). A third subscale corresponding to the third empirically identified cluster was not considered in order to avoid item overlap between subscales, and given the low number of items remaining that were not included in the other two subscales.

Table 2.

First subjective response to opioid exposure items by cluster derived using two-step cluster analysis

	Cluster
Item	Negative response (n=127)	Low sedation (n=71)	Euphoric response (n=69)	Omnibus test P value
Sleepy	99.2%	0.0%	84.1%	<0.001
Energized	0.8	4.2	2.9	0.27
Nauseous	44.9	35.2	15.9	<0.001
Euphoric	3.9	2.8	27.5	<0.001
Sedated	56.7	23.9	52.2	<0.001
Comfortable	0.0	38.0	95.7	<0.001
Numb	20.5	5.6	20.3	0.02
Uncomfortable	24.4	22.5	0.0	<0.001
Agitated	7.9	4.2	5.8	0.59
Happy	0.4	0.7	6.7	<0.001
Drunk	13.4	12.7	7.2	0.42
Powerful	0.0	0.0	4.3	0.01

Note: Items that appeared most unique between the three clusters are in bold font.

Convergent validity

In an effort to provide preliminary evidence of convergent validity, we examined correlations between the cumulative opioid exposure subscale and the euphoric response subscale of the HOME, and opioid misuse risk as measured by the SOAPP-R. As expected if the HOME possessed convergent validity regarding its relevance to opioid misuse risk, higher cumulative opioid exposure subscale scores displayed a significant association with greater opioid misuse risk as indexed by the SOAPP-R (r(265)=0.26, p<0.01). Euphoric response subscale scores showed a similar direction of association with SOAPP-R scores, with the association smaller in magnitude but significant (r(265)=0.18, p<0.05).

Secondary analyses included participants who reported no prior opioid exposure by assigning these participants a score of ‘0’ for the cumulative opioid exposure subscale. This change did not substantially alter the correlations between the cumulative opioid exposure subscale and the SOAPP-R. It therefore appeared appropriate going forward to assign individuals not reporting any previous exposure to medical opioids a score of ‘0’ on the cumulative opioid exposure subscale. The relatively small correlation observed between the two HOME subscales (r(265)=0.16) suggested that these subscales were adequately independent.

Short-term test–retest reliability

Intraclass correlations (absolute agreement, single measures) indicated excellent short-term test–retest reliability for the two identified HOME subscales (cumulative opioid exposure: ICC=0.98, euphoric response: ICC=0.94). Furthermore, all individual items on the HOME, including item one regarding ever having received opioid analgesics, exhibited a high level of test-retest reliability (all ICC’s ranged from 0.85 to 1.00).

Study 2

Design

Study 2 was part of a separate, larger, ongoing study evaluating the effect of a structured aerobic exercise training program on chronic low back pain and opioid analgesic responsiveness, and the role of endogenous opioid mechanisms in observed effects (NCT02469077). The study used a double-blinded within-subject design, with study drugs (placebo, morphine and naloxone) administered in randomized, counterbalanced order across three separate identical sessions (conducted over a 10-day period), with this protocol carried out both before and again after an 18-session (6-week) aerobic exercise training program. Data presented herein are based on the pre-exercise sessions only, to avoid confounding with intervention effects. Details and results for the naloxone condition are beyond the focus of the current paper, and therefore are not reported in the interests of space. Due to the design of the primary study, morphine and placebo sessions were conducted between 5 and 10 days apart (depending on drug administration order), and therefore drug washout period was not a study confounder. The study was conducted at two separate study locations using identical procedures in parallel in a closely coordinated fashion.

Sample

The sample for study 2 included 49 individuals with chronic low back pain who were not using any opioid analgesics on a daily basis. As needed opioid analgesic use was permitted, with instructions to abstain from any opioid use within the 3 days prior to each laboratory session (confirmed via urine opioid screen). Participants were recruited through an informatics-based targeted recruiting system mining electronic medical records to identify potentially eligible patients previously indicating a willingness to participate in research studies (‘My Research at Vanderbilt’), online advertisements on the Vanderbilt employee email recruitment system, the Rush Pain Clinic, advertisements in local print media and posted flyers. General criteria for participation included age between 18 and 55; no self-reported history of liver or kidney disorders, post-traumatic stress disorder, bipolar disorder, psychotic disorder, diabetes, seizure disorder, or alcohol or drug dependence; no daily use of opioid analgesics; and engaged in moderate or vigorous exercise <2 days/week and<60 min/week (due to the primary study requirements). Chronic low back pain was defined as daily low back pain of at least 3 months duration, and an average past month severity of at least 3 on a 0–10 verbal numeric pain intensity scale. All participants were required to be able to provide documentation of a previous medical provider diagnosis consistent with chronic back pain. Imaging results and pain diagnoses based on comprehensive medical examinations were not available. Individuals self-reporting chronic pain related to malignancy or autoimmune disorders were excluded. Potential participants who were pregnant (determined by urine pregnancy screens) were excluded. All participants were compensated for their time ($75 for initial screening and $100 for each lab visit).

Study drug

The opioid analgesic examined in this study was morphine sulfate, the prototypic mu opioid receptor agonist. Because of the hypotheses of the primary study, morphine was infused intravenously over a 2 min period through an intravenous cannula placed in the non-dominant arm in four incremental doses: 0.03 mg/kg initially, followed by 0.02 mg/kg × 3. After each dose, there was a 10 min seated rest for drug effects to peak followed by assessment of pain and subjective drug effects, with incremental drug doses provided at 25 min intervals. The total morphine dosage used was equal to 6.75 mg for a 165 pound individual. Normal saline was infused in the same incremental manner during the placebo condition. Drug order randomization was determined using the Proc Plan procedure in SAS V.9.2.

Measures

HOME

The HOME was administered to all participants at baseline and scored as described in study 1, with individuals reporting no prior exposure to opioids scored as ‘0’ on the cumulative opioid exposure subscale as noted previously. Individuals reporting no prior opioid exposure (n=4) did not contribute data on the euphoric response subscale.

Pain measures

Current back pain intensity was assessed pre-drug and repeatedly 10 min after each incremental drug dose during both laboratory sessions using the Short Form-McGill Pain Questionnaire-2 (MPQ-2)^{33 34} to rate the low back pain being experienced ‘at this moment’. The MPQ-2 is a validated measure containing 22 items rated using a numeric rating scale format (0=none and 10=worst possible) that allows assessment of the intensity of multiple qualitative features of pain.³³ It contains four subscales (continuous, intermittent, neuropathic and affective) as well as a total score.

At baseline screening (prior to the first laboratory session), three items adapted from the Brief Pain Inventory³⁵ were used to assess recent chronic pain intensity using a 0 (no pain) to 10 (worst possible pain) numeric rating scale. Item stems asked for ratings of back pain reflecting the ‘worst pain’, ‘average pain’ and ‘best pain’ experienced in the previous 24 hours. This measure was included to capture the level of chronic back pain during a broader time period than the ratings of current pain intensity obtained in the laboratory setting.

Chronic pain-related life interference was assessed using the Patient-Reported Outcomes Measurement Information System (PROMIS) Pain Interference scale (V.8a). This scale has been well-validated.³⁶ This measure was included to characterize the degree of functional impairment related to chronic pain over the past week in the sample.

Subjective opioid effect measures

Non-analgesic subjective effects of acute opioid administration were assessed as in several prior opioid studies by Zacny and colleagues using the Drug Effect Liking/Take Again (DELTA) questionnaire and a 26-item visual analog scale (VAS) opioid effects rating scale.^37–39 This assessment approach has been used in our prior work as well.^{6 17} The DELTA consists of three items asking about overall perceived effect of the drug (1–5 scale anchored with no effect and very strong effect), degree to which the effects of the drug were liked (100 mm VAS anchored with dislike a lot and like a lot), and the level of desire to take the drug again (100 mm VAS anchored with definitely would not and definitely would). Items on the VAS opioid effects scale (anchored with not at all and extremely) tap into both the positive and negative subjective (eg, cognitive, emotional) effects of opioids, including effects relevant to abuse potential (eg, feeling good, euphoric, comfortable). The latter type of positive subjective opioid effects as well as drug liking (assessed by the DELTA) are believed to contribute to abuse liability of opioids.^{21 22}

To reduce the number of measures examined and mitigate against elevated type I error rate, VAS opioid effect scale responses were used to create three opioid response factors identified based on principal components analysis in our prior work in a large sample.⁶ Each factor score in the current work reflected the sum of observed values for the top three items loading on each factor in this prior work.⁶ The three factor scores (and items contained within them) were labeled sedation (dreamy, coasting, floating), unpleasantness (down, anxious, feeling bad) and euphoria (stimulated, elated, having pleasant thoughts).

Opioid side effects measure

Common opioid-related side effects were assessed following morphine administration using the Opioid Adjective Rating Scale (OARS).^{37 40} The OARS is a 12-item questionnaire describing the extent to which common somatic and sensory effects of opioids are experienced. The OARS items are rated on a 0 to −4 scale, ranging from not at all to extremely.

Procedure

All procedures were conducted at the Vanderbilt General Clinical Research Center or a dedicated research room at the Rush University Pain Center. After providing informed consent, participants completed a packet of questionnaires, including the HOME, chronic pain-related measures and information regarding demographics. Individuals then participated in identical experimental procedures across both drug conditions (morphine vs saline placebo), with sessions scheduled at the same time of day to control for variance due to circadian rhythms. Participants were instructed not to use any opioid medications for at least 3 days prior to each study session (confirmed via urine opioid screen) and were also instructed not to use any non-steroidal anti-inflammatory drugs or over-the-counter analgesics for at least 12 hours prior to each session. The laboratory protocol for study 2 is summarized in figure 1.

Figure 1.

Summary of study 2 laboratory protocol.

Participants remained seated upright in a chair throughout all laboratory procedures. The investigational pharmacy at each institution prepared and provided the study drugs in blinded fashion to the study nurses. At the beginning of each session, an indwelling venous cannula was inserted into the non-dominant arm by a trained research nurse under physician supervision. Participants next completed the MPQ-2 to describe their current low back pain intensity. Participants then received (via the cannula) their first dose of either saline placebo or morphine per the randomization protocol. After a 10 min rest period to allow drug activity to peak, participants again described their current level of low back pain using the MPQ-2, followed by completion of the VAS opioid effects measure and the DELTA. Fifteen minutes later, the second assigned drug dose was given, followed again by ratings of chronic pain and drug effects, with the same procedure followed through the fourth and final drug dose.

All participants remained in the lab under observation for 1 hour after the final drug dose to allow drug effects to remit, after which they were released to a responsible adult.

Statistical analysis

All analyses were conducted using IBM SPSS Statistics for Windows V.24. In preparation for conducting analyses, changes in back pain intensity from pre-drug baseline to the assessment following each drug administration were derived (as post-drug minus pre-drug value) separately within the morphine and placebo conditions. The four obtained change values within each drug condition were then separately averaged as an overall index of within-session changes in back pain, a strategy potentially increasing the accuracy of these change measures by averaging out random measurement error associated with the individual measures. Next, morphine effects (as a placebo-controlled index of opioid analgesia) were derived for changes in low back pain intensity based on the MPQ-2. These morphine effects were derived by subtracting pre-drug to post-drug changes under placebo from comparable changes after morphine had been administered, such that more negative morphine effect values indicated greater morphine-induced reductions in back pain intensity. For example, if VAS intensity decreased from 50 to 45 pre-drug to post-drug in the placebo condition (change of −5), but decreased from 50 to 30 after morphine administration (change of −20), the morphine effect value would be −15, indicating that morphine produced analgesia relative to the placebo condition. An identical approach was used for the subjective non-pain opioid response measures.

Primary analyses used Pearson correlations (r) to examine associations between the two HOME subscales and placebo-controlled morphine effects as derived above. All analyses used the maximum number of available cases and a two-tailed probability value of p<0.05 as the criterion for significance. Given the sample size of 49 participants, the study was powered to detect an effect size (Pearson correlation) as low as r=0.29. Thus, it was sufficiently powered to detect a moderate or larger effect size (ie, r=0.30 or greater as per Cohen),⁴¹ a magnitude of effect likely to be required for the findings to be clinically meaningful.

Results

Preliminary analyses

Sample characteristics are summarized in table 3. Participants in study 2 were primarily female, of white non-Hispanic race/ethnicity, with moderate intensity chronic back pain of extended duration. They reported pain-related life interference (equivalent T-score=63.5) that was more than 1 SD above the mean (T=50) of the original PROMIS validation sample. No participants used opioids on a daily basis per inclusion criteria, and only a small number (n=7) used opioids on an as needed basis (none within 3 days of each study session).

Table 3.

Baseline characteristics of the study 2 sample (n=49)

Characteristic	Descriptive statistic
Sex (% female)	67.3
Race (%)
White	61.2
African-American	26.5
Asian	6.1
Ethnicity (% non-Hispanic)	97.9
Age (MN±SD)	39.7±9.96
CLBP duration (median, in months)	78.7
Worst pain past 24 hours (mean±SD)	6.0±2.36
Average pain past 24 hours (mean±SD)	4.6±2.49
Best pain past 24 hours (mean±SD)	3.1±2.59
PROMIS pain interference 8a (mean±SD)	23.7±9.12
Medications (%)
As-needed opioids	14.3
Antidepressants	49.0
Neuroleptics	38.7
As-needed NSAIDs	42.9
HOME: opioid exposure (mean±SD)	2.2±1.54
HOME: euphoric response (mean±SD)	3.2±1.25

CLBP, chronic low back pain; HOME, History of Opioid Medical Exposure; NSAID, non-steroidal anti-inflammatory drug; PROMIS, Patient-Reported Outcomes Measurement Information System.

Based on side effects rated after the final morphine dose (using the OARS), levels of adverse drug effects were low. Mean±SD for levels of typical opioid side effects were flushing=0.4±0.82 (range: 0–4), itching=0.3±0.65 (range: 0–3), sweating=0.0±0.55 (range: 0–3), nausea=0.4±1.00 (range: 0–4) and vomiting=0.1±0.28 (range: 0–1). Not surprisingly given the effects of morphine, 72.9% of study participants correctly identified the session during which they had received morphine.

Associations between HOME subscales and baseline chronic pain characteristics

The cumulative opioid exposure and euphoric response subscales of the HOME were not correlated (r(43)=−0.01, p=0.97). A significant positive association was noted between best pain intensity in the past 24 hours and scores on the cumulative opioid exposure subscale of the HOME (r(47)=0.31, p=0.03), with directionally similar but marginally significant associations for average pain (r(47)=0.26, p=0.08) and worst pain (r(47)=0.24, p=0.10). No significant association was noted with the pain interference measure (r(47)=0.09, p=0.52). Euphoric response subscale scores were not significantly associated with any baseline chronic pain characteristic examined (all rs<0.17, ps>0.28).

Associations between HOME subscales and morphine responses

Correlations between HOME subscale scores and placebo-controlled morphine effect measures are summarized in table 4. Higher cumulative opioid exposure scores were significantly related to greater observed magnitude of morphine analgesia (for MPQ-2 continuous, neuropathic and total scores). It is notable that these associations were present despite the absence of obvious confounding effects related to recent opioid use (ie, none of the study participants were using opioids on a daily basis and only seven were using as needed opioids, none within the 3 days prior to laboratory sessions). Partial correlation analysis adjusting for as-needed opioid use revealed that the associations described above were virtually unchanged and remained significant. Cumulative opioid exposure scores were not related to any of the non-pain subjective opioid effect measures.

Table 4.

Pearson correlations (r) between HOME subscales and placebo-controlled analgesic and subjective responses to morphine

Morphine Response Outcome	Home subscale
	Cumulative opioid exposure		Euphoric response
	Pearson r	P value	Pearson r	P value
DELTA drug effect	−0.01	0.98	−0.44	0.003
DELTA drug liking	−0.09	0.56	0.27	0.08
DELTA desire to take again	−0.03	0.85	0.34	0.02
Sedation factor	−0.18	0.21	−0.30	0.04
Unpleasantness factor	0.15	0.31	−0.21	0.17
Euphoria factor	−0.08	0.59	0.32	0.03
MPQ-2 continuous	−0.35	0.02	−0.08	0.59
MPQ-2 intermittent	−0.17	0.24	0.25	0.11
MPQ-2 neuropathic	−0.32	0.02	0.00	0.98
MPQ-2 affective	−0.24	0.09	0.06	0.70
MPQ-2 total	−0.33	0.02	0.06	0.69

Significant correlations are highlighted in bold text.

DELTA, Drug Effect, Drug Liking, and Desire to Take Again scale; HOME, History of Opioid Medication Exposure survey; MPQ-2, Short Form McGill Pain Questionnaire-2.

Scores on the euphoric response subscale of the HOME, in contrast, were not associated significantly with degree of morphine analgesia observed. However, these scores were associated significantly with several non-pain subjective opioid effects. Higher HOME euphoric response scores, reflecting retrospective recall of euphoria-related responses to opioid analgesics the first time they were ever used, were significantly associated with lower overall perceived effect of morphine (DELTA), less sedation (sedation factor), greater euphoria (euphoria factor) and greater desire to take morphine again (DELTA). Associations between higher euphoric response scores and greater drug liking (DELTA) also approached significance. The magnitude of all of these associations increased slightly in partial correlation analyses adjusting for the use of as-needed opioids.

DISCUSSION

Reducing unnecessary opioid prescribing has been suggested as a means of mitigating against opioid risks and maximizing analgesic efficacy.^{3 4} This study focused on two understudied aspects of past medical opioid use which might be expected to predict opioid responses for theoretical reasons: cumulative exposure to medical opioids and non-pain subjective responses on first exposure to opioid analgesics.

Results of study 1 revealed two empirically derived subscales of the HOME questionnaire: cumulative opioid exposure and euphoric response (to first use of opioids). Results indicated that both demonstrated adequate internal consistency reliability and strong short-term test–retest reliability. Study 2 examined the predictive validity of the HOME in a totally independent sample of patients with chronic low back pain. Higher cumulative opioid exposure scores were associated with significantly greater acute reductions in back pain in response to placebo-controlled morphine administration. A possible explanation for this finding is that individuals who find opioid analgesics to be more effective for pain management may over time self-select and be more likely to seek out opioids when experiencing pain, leading to greater cumulative opioid exposure. The observed pattern of findings would, however, argue against the glial-mediated hyperalgesic effects of prior opioid exposure suggested by some researchers,¹⁸ at least in individuals not currently using opioids regularly as in the present sample. The HOME index of cumulative opioid exposure was not related significantly to non-pain subjective response to opioids.

Results of study 2 also indicated that individuals reporting greater euphoric-type responses on their first exposure to opioid analgesics (ie, higher euphoric response scores on the HOME) displayed a pattern of associations suggesting they are potentially at increased risk of problematic opioid use. Specifically, higher euphoric response scores were subsequently associated in the lab with reports of less sedation, greater euphoria and greater desire to take morphine again after receiving morphine under placebo-controlled conditions. Past work suggests that positive subjective responses to opioids like those noted above may enhance the reinforcing qualities of opioid use, and therefore predict increased likelihood of opioid misuse or abuse.^19–22 More importantly, one prior study had examined individuals’ retrospective accounts of responses on initiation of opioid analgesic therapy for chronic pain management.²³ Consistent with findings of the current work, patients receiving substance abuse treatment for prescription opioid use disorder reported experiencing significantly more euphoric and stimulating effects in response to first use of an opioid for chronic pain management compared with chronic pain controls without opioid use disorder.²³ Taken together with this prior work, our findings suggest that euphoric response scores on the HOME may index individual differences in responses to medical opioids that may be relevant to understanding risk for opioid use disorder.

Observed effect sizes for significant associations between both HOME subscales and responses to controlled opioid administration were all in the moderate range per the accepted criteria of Cohen.⁴¹ The HOME predicted approximately 10% of the variance in morphine responses. It seems unlikely that this moderate level of prediction using the HOME, by itself, would be sufficient to accurately guide opioid therapy in clinical settings. It is more likely that a predictive algorithm incorporating multiple measures predictive of opioid responses (eg, the HOME, negative affect measures, evoked pain responsiveness, genetic markers) may be necessary to predict opioid responses to a more clinically meaningful extent.⁴ This predictive approach using a phenotypic/genotypic algorithm avoids the undesirable alternative of directly evaluating analgesic and subjective responsiveness to a test dose of opioids in clinical settings that might otherwise be required to guide treatment decisions.

A number of study limitations should be noted. First, by design, the HOME does not address the impact of cumulative exposure to or subjective responses to past illicit opioid use. Second, items on the cumulative opioid exposure subscale did not attempt to capture highly detailed individual differences in specific opioid dosages, frequency or agents over the life span. While potentially important, it was felt that asking participants to recall such detailed information over the life span would likely inflate reporting error relative to the more global items included in the HOME. Third, the sample was primarily female, white and of non-Hispanic ethnicity. Results may not generalize beyond this population. Finally, the impact of retrospective recall bias cannot be ruled out. Long-term recall of pain conditions is known to be somewhat inaccurate,⁴² and recall of past opioid use may be similarly inaccurate. It was not feasible to determine the correspondence of cumulative opioid exposure subscale scores with objective medical and pharmacy records. It is likewise possible that euphoric response subscale responses might have been biased in favor of describing participants’ responses to more recent opioid use. Current results therefore indicate only that participants’ perceptions of their past opioid medication use and retrospective recall of their past subjective responses to opioid analgesics significantly predicted their subsequent responses to morphine administration.

In summary, the HOME appears to be a brief and reliable self-report measure for assessing cumulative medical opioid exposure and individual differences in subjective non-pain responses on first exposure to an opioid analgesic. The two empirically derived subscales of the HOME, cumulative opioid exposure and euphoric response, predicted degree of analgesia (the former) and subjective non-pain responses (the latter) experienced following placebo-controlled morphine administration in a sample of patients with chronic low back pain not using daily opioids. Additional work is required to replicate the current findings and evaluate the potential clinical relevance of the HOME.