‘I crave not to feel uncomfortable’ – investigating craving for opioids and cannabis among individuals with chronic pain

Abstract

Background:

The assessment of drug craving is common in survey studies, including those using real-time data collection methods, such as Ecological Momentary Assessment (EMA). However, few studies investigate how participants with chronic pain interpret the word ‘craving’ and how interpretations impact survey responses.

Methods:

We conducted a mixed-methods study among 12 individuals with chronic pain who were using prescription opioids and cannabis. Participants completed baseline surveys, cognitive interviews, and 14-day smartphone-based EMA data collection. Analyses included deductive and inductive coding of interviews and t-tests of EMA self-reports of opioid and cannabis craving.

Results:

Four participants had negative reactions to the word ‘craving’, including mentions that these questions offended them. The remaining eight participants mentioned no negative connotation. EMA data showed that participants without negative reactions reported a greater range (opioids), higher standard deviation (opioids), and higher maximum (opioids, cannabis) on Likert-type EMA craving items, compared to those with a negative reaction.

Conclusions:

Some individuals with chronic pain may have a negative reaction to the word ‘craving’ related to opioid and cannabis use and this reaction may impact survey responses. Alternative wording of survey items is recommended, for example focused on ‘wanting’ or ‘needing’.

Introduction

Chronic pain is a major contributor to reduced quality of life and burden of disease worldwide (GBD 2017 Disease and Injury Incidence and Prevalence Collaborators, 2018). Global prevalence estimates are as high as 30% and more (Cohen et al., 2021) and in the US chronic pain affects more than 50 million individuals (Dahlhamer et al., 2018). Despite recent policy changes that have reduced opioid prescribing in the US, opioid pain medications remain a major treatment modality for chronic pain, with use reported by 29–41% of individuals with chronic pain (Zajacova et al., 2023). Paradoxically, opioids have limited efficacy for chronic pain (Ballantyne, 2017) and their use is associated with a variety of negative physical and mental health consequences, including overdose (Adewumi et al., 2018). Thus, alternative treatments for chronic pain are needed. A substantial number of patients suffering from chronic pain are experimenting with cannabis for pain symptom management. In fact, chronic pain is among the most frequently endorsed reasons for medical cannabis use (Kosiba et al., 2019). However, only a few studies have started investigating opioid and cannabis co-use among those with chronic pain in real time and in their natural environment (Anderson Goodell et al., 2021; Mun et al., 2022).

Ecological Momentary Assessment (EMA) is a data collection method that uses repeated administration of brief surveys that participants complete multiple times per day as they go about their daily lives (Stone & Broderick, 2007). The advantages of EMA over other survey methods include low recall bias and high real-world validity of collected data (Shiffman et al., 2008). In our previous study, we demonstrated the feasibility of collecting EMA data in individuals with chronic pain who are using both opioids and cannabis (Anderson Goodell et al., 2021). However, our previous work did not involve qualitative cognitive interviews with participants around their understanding and interpretation of items used in the study’s EMA surveys. Cognitive interviews are a method to assess participant understanding and interpretations of survey questions and response options (Willis, 2004). As such, cognitive interviews provide a unique opportunity in a research study to conduct pilot testing of surveys and improve items based on participant feedback.

Assessment of craving for substances with potential for addiction is common in survey studies, including EMA studies that assess substance use (Nguyen et al., 2018; Thrul et al., 2014), and, more specifically, those that investigate opioid or cannabis use (Buckner et al., 2011; Mun et al., 2021; Preston et al., 2018). Craving to use substances is an indicator for problematic substance use, including opioid use in treatment studies for opioid use disorder (OUD) (McHugh et al., 2014), is cited by patients as a primary treatment goal and unmet need in their OUD treatments (FDA Center for Drug Evaluation and Research, 2019), and has been included in the DSM-5 as one of the primary criteria for diagnosis of a substance use disorder (American Psychiatric Association, 2013). EMA typically administers brief surveys to reduce participant burden, relegating assessment of complex constructs, such as craving, to single-item questions that can be asked repeatedly (Mun et al., 2021; Nguyen et al., 2018; Thrul et al., 2014). Few studies, however, have rigorously pilot tested conventional EMA survey items with the target population to investigate study participants’ interpretation and understanding of questions related to craving. This is an important shortcoming in the existing literature that the current study sought to address.

The assessment of substance use craving may be especially complicated among individuals with chronic pain who are using multiple types of medications, such as opioids and cannabis, for pain management, as both products can engender and impact craving in different ways. In the context of the ongoing opioid epidemic (Seth et al., 2018), many providers are reluctant to prescribe opioids and may screen their patients for signs of nonprescribed opioid use (Jamison et al., 2014; Pearson et al., 2017; Wilson et al., 2022). Moreover, providers also report a lack of knowledge on how to counsel patients on cannabis use (Rønne et al., 2021). Many individuals with chronic pain report feeling guilt, fear, and stigma around their opioid use (Brooks et al., 2015). Therefore, these patients may have negative reactions when asked questions assessing opioid craving, since they may interpret these questions as implicit accusations that they have developed OUD or are using opioid medications outside of how they were prescribed. Studies are needed to investigate how individuals with chronic pain react to assessment of drug cravings and how these reactions may be associated with self-reported craving in repeated surveys. EMA data repeatedly collected from the same participants provide an opportunity to investigate how participant interpretations and reactions to drug craving questions may impact response behavior. Compared to a single survey, EMA data provide the advantage that participant variability in response behavior can be investigated. For example, analyses can investigate if participants are using the range of response options on a Likert scale, which would not be possible with single surveys or even longitudinal surveys with relatively few assessment waves.

The current study used a mixed methods design, combining cognitive interviews with EMA data collection, to evaluate participants’ perceptions about the term ‘craving’ and how their perceptions were directly related to endorsements of craving reported over a 2-week period of EMA.

Methods

Procedure

Participants were recruited between April and August of 2021 from 11 U.S. states (Alaska, California, Colorado, Illinois, Maine, Massachusetts, Michigan, Nevada, Oregon, Vermont, and Washington) and Washington D.C., which all had legalized non-medicinal cannabis use and had implemented retail sales to adults. Recruitment was conducted using Facebook and Instagram ads (targeted to location and interests in topics including ‘personal care’, ‘health and wellness’, and ‘research’). Recruitment was also supported by a Colorado-based non-profit organization (Realm of Caring Foundation), with which the investigators have an ongoing collaboration, and which shared recruitment information on social media to their over 100,000 followers. Ads contained a link that directed interested individuals to a Qualtrics survey with more information about the study, the eligibility questionnaire, and the online informed consent form.

After screening into the study and consenting to participate, eligible participants were required to send study staff a picture of a government-issued identification card (e.g. driver’s license) to validate their age, identity, and place of residence. Enrolled participants completed a baseline survey to report demographic characteristics, substance use history and current behavior, and pain. Participants were also asked to identify any long-acting and/or short-acting oral and non-oral opioid medication(s) they had used in the past 30 days. They were similarly asked to select from a list of medical cannabis products (e.g. flower, oil, concentrates, edibles, topicals, and prescription medications) that they used in the past 30 days. Baseline surveys were hosted on the Qualtrics (Provo, UT, USA) platform.

After completing the baseline survey, participants were invited to complete an initial cognitive interview on the phone with one study investigator (JT or CDN). These interviews were semi-structured and focused on specific survey items to be included in the EMA data collection phase.

After the cognitive interview, participants moved to the EMA phase of the study, which was conducted using an EMA data collection app (MetricWire; Kitchener, Ontario, Canada), on their personal smartphones. Before responding to surveys, participants received an email containing written instructions on how to install and use the study app and information related to survey timing, frequency, and incentives. Study staff conducted follow-up telephone calls to answer questions and confirm that participants understood how to use the app. Participants then began a 14-day EMA phase during which they responded to prompted surveys about opioid use, cannabis use, and pain symptoms. On each of the 14 days, participants were prompted to complete five surveys, four of which were prompted at random times between 8 am and 11 pm of their device’s local time. Participants had a one-hour time window to complete the EMA survey and each question inquired about activities in the hour that preceded survey completion. The 1-hour time interval for the randomly prompted surveys was selected to maximize the probability of observing situations in which opioid medications and cannabis products were used by participants, while limiting the recall window for time-coverage in a way that would minimize recall bias. The fifth survey was a daily diary that was prompted between 10 and 11 am, pertained to the entire previous day, and had a 12-hour time window for completion. Each EMA survey took a maximum of two minutes to complete, for an expected maximum time commitment of no more than 10 minutes per day to complete all five prompted surveys.

During the EMA phase, study staff sent participants SMS text message check-ins about their current compliance rate on Days 3, 4, and 14 and communicated via email and text message if participants had questions or comments. The study team monitored participant EMA survey activity and conducted regular check-ins to understand and troubleshoot issues and work with the survey app programming team to address any identified technical problems.

Participants received $2 for each day they completed at least one survey, for a maximum possible $28, and an additional $30 was given if they completed multiple surveys per day and reached a total compliance rate of at least 75% over the 14-day EMA study period. Participation in the cognitive interview was incentivized with an additional $25. Incentives were provided in the form of electronic gift cards, which were emailed to participants at the end of their participation in the study.

All study procedures were approved by the Johns Hopkins Bloomberg School of Public Health Institutional Review Board (IRB00009327).

Participants

In order to qualify for the study, participants had to meet the following self-reported inclusion criteria: be at least 18 years of age, have an opioid medication prescription for treating pain symptoms, self-report use of opioid medication in the past 7 days, received a medical recommendation for or started using cannabis for medicinal purposes in the past 30 days, have a chronic pain disorder, self-reported pain severity rating of at least a 3 on a scale of 0 to 10 on at least 10 days of each month for the past 3 months, have an iPhone or Android smartphone, and currently living in a state that had legalized recreational cannabis use. Individuals who reported having a serious mental illness, such as schizophrenia, psychosis, or dementia, were excluded from the study, as these conditions were deemed to potentially interfere with participants’ abilities to complete study procedures.

Among 1,703 individuals who completed the screening questionnaire, a total of 49 participants were eligible for the study. The most frequent reasons for being ineligible (not mutually exclusive) were not having used prescription opioids in the past 7 days (66.5%), no current opioid prescription (64.0%), or not having a medical cannabis recommendation (59.1%). Of those eligible, 28 people provided identification and were verified for study participation. Among these 28 individuals, 19 completed the baseline survey, and 17 completed the cognitive interview and were invited to the EMA phase. Finally, 12 participants completed the 14-day EMA and were included in the current analyses.

Measures

Baseline survey

Participant characteristics were measured at baseline and included: gender (male = 0, female = 1); age (younger than 40 years = 0, 40 years or older = 1); race/ethnicity (Non-Hispanic White = 0, Other = 1); education (less than college degree = 0, college degree or higher = 1); number of days of opioid medication use in the past 30 days, short- and long-acting opioids; number of days of cannabis use in the past 30 days; average, least, and worst pain in the past 30 days (rated 0 to 10); and Graded Chronic Pain Scale-revised (GCPS-R) category (Grade 0: Chronic pain absent, Grade 1: mild chronic pain, Grade 2: bothersome chronic pain, and Grade 3: high impact chronic pain). The GCPS-R assesses pain frequency and severity, as well as interference with work, enjoyment of life, and general activity. The scale has been validated with adults and has demonstrated strong concurrent validity with health status indicators, including negative pain coping beliefs, health status, activity limitations, and pain medication use (Von Korff et al., 2020). The baseline survey also assessed chronic pain conditions for each patient, including assignments of primary condition, which were classified into 2 categories (i.e., neuropathic or nociceptive, other/multiple). Prescription opioid medication adherence over the past month was assessed with the 8-item version of the Opioid Compliance Checklist (OCC) (Jamison et al., 2016). The NIDA-Modified ASSIST V2.0 (National Institute on Drug Abuse, 2011) was used to screen for substance use disorders.

EMA surveys

EMA surveys asked two separate questions about craving for opioid medications and cannabis products (‘In the past hour: Craved opioid pain medications?’; ‘In the past hour: Craved marijuana or cannabis?’) and responses were recorded on a 7-point Likert scale from 1 ‘Very low’ to 7 ‘Very high’.

Cognitive interviews

Cognitive interviews were conducted on the phone, semi-structured, and presented the two EMA craving questions (among others). During scheduling of the interview, participants also received all EMA items the interview would cover in writing. The interviewer first asked about how the participant would respond to this question and used follow-up probing questions, querying areas of comprehension (‘What do you think the question refers to?’), response (‘How did you pick an answer to that question?’), potential misunderstandings (‘Could there be any misunderstanding about this question?’), and suggestions for improvement (‘How could we improve this question?’).

Data analyses

Data analysis was conducted using a convergent mixed methods design with parallel analysis of qualitative and quantitative data (Fetters et al., 2013). Coding of cognitive interviews was conducted at first deductive, based on interview guide, then inductive and identified sub-themes in participant responses. The first and second author (JT and CDN) coded all interviews, with one author conducting the initial coding, and the other author verifying accuracy. The sample size for this study was based on user experience design recommendations that a sample size as small as 5 users will detect 85% of usability errors in a given system (Nielsen, 2000; Nielsen & Landauer, 1993). Studies investigating sample size needs for qualitative interviews suggest that saturation of codes occurs withing the first 12 interviews, with basic elements for themes being present after 6 interviews (Guest et al., 2006).

EMA data were initially analyzed descriptively and plotted. The first author (JT) extracted participant responses to craving questions from the qualitative interviews. The first author (JT) then conducted mixed methods integration of qualitative themes reported by individual participants and quantitative responses on EMA craving items, with the second author (CDN) reviewing and verifying the integration. Based on the qualitative responses, participants were classified as having a negative reaction to craving questions (e.g. craving as addiction, mentioned feeling offended) or no negative reaction (e.g. craving as wanting or needing). Based on this grouping of participants (negative vs. no negative reaction), we then statistically compared these groups at baseline using Fisher’s exact tests and t-tests, and compared their response patterns (e.g. mean, standard deviation, maximum, and range) to EMA craving questions during the 14-day data collection period using t-tests.

Results

Sample description

Participants were on average 48.0 (SD = 11.0) years old, and the majority were female (83%) and Non-Hispanic White (75%). Overall, 92% of participants reported short-acting opioid medication use, with an average of 16.8 (SD = 12.3) days out of the past 30 days among those with any use. Long-acting opioid medication use was reported by 42% of participants, with an average of 8.1 (SD = 12.9) days out of the past 30 days among those with any use. Use of cannabis products was reported by 100% of participants, with an average of use on 27.4 (SD = 7.5) out of the past 30 days. Participants reported an average pain severity of 5.2 (SD = 1.5) on a scale from 0 to 10. Based on the Graded Chronic Pain Scale-Revised, the majority of participants (75%) reported high impact chronic pain. Participants reported a range of different types of chronic pain, with most individuals (83%) reporting chronic pain other than neuropathic or nociceptive or multiple types of chronic pain (Table 1). Non-compliance with using opioids medications as prescribed in the past month was reported by 2 participants in the OCC (17%). Lifetime use of nonprescribed opioids in the NIDA ASSIST V2.0 was reported by 1 participant (8%), and no participant reported recent (past 3-month) nonprescribed opioid use.

Table 1.

Participant characteristics at baseline (N = 12).

Characteristic	N (%) or mean (sD)			t-test/Fisher’s exact test
	Entire sample (N = 12)	Participants with neg. reaction to craving question (N = 4)	Participants without neg. reaction to craving question (N = 8)
Demographics
Age (M, SD)	48.0 (11.0)	52.0 (7.4)	46.8 (11.9)	p = 0.4
Gender				p = 0.5
Female	10 (83%)	4 (100%)	6 (75%)
Male	2 (17%)	–	2 (25%)
Race/ethnicity				p = 0.5
Non-Hispanic White	9 (75%)	4 (100%)	5 (62%)
Other	3 (25%)	–	3 (38%)
Education				p = 0.5
Less than bachelor’s degree	6 (50%)	1 (25%)	5 (62%)
Bachelor’s degree or higher	6 (50%)	3 (75%)	3 (38%)
Use of short acting opioid medication
Participants using short acting opioid medications	11 (92%)	4 (100%)	7 (88%)	1
Number of days short acting medication use (M, SD) (out of past 30 days; among those with any use)	16.8 (12.3)	26.0 (7.4)	12.2 (11.9)	p < 0.05*
Use of long-acting opioid medication
Participants using long-acting opioid medications	4 (33.3%)	1 (25%)	3 (38%)	1
Number of days long-acting medication use (M, SD) (out of past 30 days; among those with any use)	8.1 (12.3)	7.5 (15.0)	8.4 (12.9)	p = 0.9
Opioid medication use other than prescribed in past month	2 (16.7%)	0 (0%)	2 (25.0%)	p = 0.5
Lifetime use of nonprescribed opioids	1 (8.3%)	0 (0%)	1 (12.5%)	p = 1.0
Use of cannabis products
Participants using cannabis products	12 (100%)	4 (100%)	8 (100%)	–
Number of days cannabis use (out of past 30 days)	27.4 (7.5)	22.2 (12.4)	30.0 (0.0)	p = 0.3
Chronic pain
Average Pain severity past month (M, SD)	5.2 (1.5)	4.5 (1.3)	5.5 (1.5)	p = 0.3
Least Pain severity past month (M, SD)	3.5 (1.7)	3.0 (1.6)	3.8 (1.8)	p = 0.5
Worst Pain severity past month (M, SD)	7.8 (1.6)	7.3 (1.9)	8.1 (1.6)	p = 0.5
Graded Chronic Pain Scale-Revised Category				1
Less than high impact chronic pain	3 (25%)	1 (25%)	2 (25%)
High impact chronic pain	9 (75%)	3 (75%)	6 (75%)
Chronic pain condition type				p = 0.5
Neuropathic or nociceptive	2 (17%)	-	2 (25%)
Other/multiple	10 (83%)	4 (100%)	6 (75%)

Cognitive interviews

Four participants had negative reactions to the word ‘craving’ and associated it with addiction, including mentions that these questions offended them.

For example, one participant reported on their experience interacting with treatment providers and screening questions for problematic opioid use:

P0046: ‘I could understand how for some other people, if they read the word craving, I could see how they would get offended because when I was in the worst of the worst of the worst of my pain, which was pretty severe, I know that I got easily upset when some of my doctors would say to me, “Well, you know, this could be getting worse as time goes on,” or, “Are you taking these as prescribed?” you know, “We really shouldn’t go up in milligrams.” you know, so I could understand how some people on the survey would be sensitive to that word.’

Another participant mentioned that a question like this one may suggest that a study using this terminology may not fit their specific situation:

P0016: ‘The implication of the way the words were used is if there’s an addiction element, and not everyone on opioids has an addictive element, so to me I don’t feel I crave them. I more crave not feeling uncomfortable. <laughs > To me it could make me feel offended and not want to participate, because I may not feel as if that fits me in my situation as it would maybe other people.’

The same participant also added that the terminology of craving for them was associated with recreational use and substance use disorders for both opioids and cannabis:

P0016: ‘Well, the word “crave,” that’s definitely more on the recreational or substance-abuse side. Absolutely.’

And

P0016: ‘Okay, I crave potato chips. When I’m on a carb-free diet I will crave missing those comfort foods. However, I don’t crave drug use or recreation use.’

This participant also suggested a different way to assess this construct as use of medication outside of how it was prescribed, which could be similar to the concept of nonprescribed medication use:

P0016: ‘Would the question be more appropriate to say “Do you have a tendency to need or want medication outside of your normal prescribed dosage?”‘

The theme of addiction and loss of control was also mentioned by two other participants who had a negative reaction to the wording of the craving item. For example, one participant mentioned:

P0042: ‘I know that some people who take opioids have a great deal of trouble controlling their access to opioids because they’re addicts. I’ve been taking opioid medication for almost 15 years, I’ve been completely compliant the entire time.’

Another participant had a similar interpretation with focus on addiction in response to the question what the craving item assesses:

P0049: ‘[The item assesses] your addiction to the pain meds. I don’t feel that I’m addicted to it. I haven’t felt a need for it.’

In sum, these participants had a negative reaction to an assessment of craving of their medication use, especially regarding the use of opioids. Participants mentioned that this wording could make them or others in similar situations feel offended or like the study was not meant for them. Moreover, some participants made a distinction between their own opioid use for pain management and use that either was non-medicinal or that deviated from prescribed dosing, which may be associated with loss of control, substance misuse, use disorders, or addiction.

On the other hand, the remainder of the sample (8 participants) mentioned no negative connotation with the word craving. Prominent examples included mentions of ‘wanting’ or ‘needing’ a medication or substance. For example, one participant mentioned craving in the context of the body or mind ‘wanting’ something that is currently absent:

P0028: ‘Whether it’s your body or your mind wanting or thinking that it needs a substance.’

A similar point was made by another participant, who asked a clarification question when presented with the item assessing craving:

P0032: ‘My body is physically craving it, physically wanting it, right?’

Other participants associated the term craving with ‘needing’ pain medication or cannabis. For example, this participant who responded in a short and direct way to the question of how he interpreted the craving questions:

P0014: ‘Just the way I need or don’t need pain medication.’

And

P0014: ‘Usage of marijuana and the need for it.’

Similarly, another participant also asked a clarification question when presented with the craving question:

P0025: ‘you’re asking me how much have I needed to use pain medication in the past hour?’

Taken together, participants who did not have a negative reaction to the wording of the craving question associated it with the body or mind wanting or needing the medication or substance.

Group differences in baseline data

Participants with a negative reaction to craving cognitive interview assessments reported more days of short-acting opioid medication use in the past 30 days (M = 26.0; SD = 7.4) compared to participants that had no negative reaction (M = 12.2; SD = 11.9; t=−2.0; p = 0.04). There were no other significant baseline differences between participants with a negative reaction to the craving assessment and those participants who did not have a negative reaction (Table 1).

Group differences in EMA data

In total, the 12 participants included in this analysis completed 459 of 672 possible EMA surveys (68% compliance) (Figures 1 and 2). Based on the grouping of participants (negative vs. no negative reaction to craving question in cognitive interviews), we statistically compared their response patterns with regard to total number of EMA surveys submitted, mean, standard deviation, maximum, and range to EMA craving items in the 14-day data collection period using t-tests (Table 2).

Figure 1.

Opioid craving EMA reports throughout the study period (rows are participants, columns are days, assessments within days, and participants with negative reaction to craving items are highlighted in red).

Figure 2.

Cannabis craving EMA reports throughout the study period (rows are participants, columns are days, assessments within days, and participants with negative reaction to craving items are highlighted in red).

Table 2.

Craving summary EMA data.

			Opioid craving summary (scale range 1–7)				Cannabis craving summary (scale range 1–7)
Participant	Reaction to craving question	Count	Min	Max	M	SD	Min	Max	M	SD
P0016	Neg	39	1	3	1.36	0.54	1	5	2.92	1.13
P0042	Neg	17	1	2	1.29	0.47	1	2	1.29	0.47
P0046	Neg	2	1	1	1.00	0.00	1	1	1.00	0.00
P0049	Neg	53	1	2	1.02	0.14	1	3	1.06	0.30
P0003	Not neg	46	1	6	2.26	1.31	1	1	1.00	0.00
P0014	Not neg	39	1	7	2.18	2.01	1	7	2.85	1.91
P0021	Not neg	49	1	7	3.65	2.41	1	7	3.78	2.62
P0025	Not neg	53	1	2	1.17	0.38	5	7	6.85	0.46
P0028	Not neg	28	1	6	2.89	1.40	1	6	3.00	1.41
P0032	Not neg	49	1	3	1.12	0.48	1	6	1.31	1.02
P0033	Not neg	44	1	7	1.82	1.78	1	7	2.93	2.32
P0037	Not neg	40	1	3	1.10	0.38	1	4	1.18	0.68

Compared to participants with a negative reaction to the craving item (Neg), those without a negative reaction (Not Neg) reported a greater range in the EMA assessment on craving opioids (Neg: M = 1.00; Not Neg: M = 4.13; t=−2.8; p < 0.05), a higher standard deviation for opioids (Neg: M = 0.29; Not Neg: M = 1.27; t=−2.4; p < 0.05), and a higher maximum for opioids (Neg: M = 2.00; Not Neg: M = 5.13; t=−2.8; p < 0.05) and cannabis (Neg: M = 2.75; Not Neg: M = 5.63; t=−2.3; p < 0.05). There were no significant differences in total number of EMA surveys submitted (Neg: M = 27.75; Not Neg: M = 43.50; t=−1.8; p = 0.10) and mean EMA craving scores for opioids (Neg: M = 1.17; Not Neg: M = 2.02; t=−1.8; p = 0.10) or cannabis (Neg: M = 1.57; Not Neg: M = 2.86; t=−1.3; p = 0.24) between participants with a negative reaction to the craving item, and those without a negative reaction.

Discussion

The current study conducted a mixed methods data collection, combining cognitive interviews with EMA data, to investigate assessments of opioid and cannabis cravings among individuals with chronic pain. Four participants had strong negative reactions to the wording of ‘craving’ and associated it with addiction, also mentioning that these questions offended them. In addition, some of these participants mentioned that items like these make them not want to participate in a study. EMA data showed that participants without a negative reaction reported a greater range (opioids), higher standard deviation (opioids), and higher maximum (opioids and cannabis) on Likert-scale EMA craving items, compared to those with a negative reaction. These findings indicate that individuals with a negative reaction to assessment of these substance use craving experiences may not use the entire response scale of assessment items and report less variability of cravings. Interestingly, we did not find significant differences between groups in overall EMA survey completion rates as well as mean scores on opioid and cannabis craving items.

EMA data provide an opportunity to investigate participant response behavior with the advantage that multiple assessments are available for each individual participant. In our analyses of EMA data of craving items, we found three significant differences for opioid use craving (i.e., range, SD, and maximum), but only one for cannabis use craving. Future studies with larger samples are needed to confirm if persons with chronic pain truly are more sensitive to assessments using ‘craving’ language for their opioid use compared to cannabis use. This may be suggested by some of our qualitative findings in which participants talked about challenging experiences they had with healthcare providers on the topic of opioid use for pain management. These findings align with previous qualitative research demonstrating that many patients with chronic pain report feeling guilt, fear, and stigma regarding their prescription opioid use (Brooks et al., 2015). Our findings suggest that some of these patient experiences may lead them to have negative reactions to wording of items that assess opioid and cannabis use cravings and subsequently impact their response behavior.

The assessment of craving for opioids and cannabis is common and is strongly associated with drug use in EMA studies (Buckner et al., 2011; Mun et al., 2021; Preston et al., 2018; Vafaie & Kober, 2022). Craving is a core criterion in the assessment of substance use disorders (American Psychiatric Association, 2013), with demonstrated validity across different substances (Shmulewitz et al., 2021). It is currently unclear if the impact on response behavior we found in the current study that included individuals with chronic pain is unique to this sample or extends to other individuals who are using substances or are diagnosed with substance use disorders. It is an open question if other patient groups take similar offense with craving items and in what way, if any, past studies using similar assessment strategies have been impacted by this issue. We can speculate that provider reluctance to prescribe opioids and screening of patients for opioid use outside of a prescription (Jamison et al., 2014; Pearson et al., 2017; Wilson et al., 2022) may contribute to the development of negative reactions toward these questions among some patients. For example, in a previous qualitative study patients with chronic pain perceived a lack of care from providers, reported doubts that providers believed them, and expressed difficulties communicating a subjective condition like chronic pain (Buchman et al., 2016). Patients may experience stigma and pressure from their healthcare providers to reduce their opioid use and hence be sensitive to the wording of craving items in survey studies.

Participants who were sensitive to the term craving seemed to make a distinction between themselves, who were using opioids for pain management, and others who may be using opioids recreationally, which some participants felt was associated with loss of control over use, addiction, and substance use disorder. In line with our findings, previous qualitative research has demonstrated that individuals with chronic pain who are using opioids are trying to differentiate themselves from others who may be using recreationally and struggle with OUD (Dassieu et al., 2021), which may reinforce stigma towards individuals with OUD. Taken together, these findings may demonstrate prevalent stigma against individuals with substance use disorders (Adams & Volkow, 2020) and a perceived need to differentiate oneself from this group. Our results and those of previous studies highlight the ongoing need to destigmatize substance use disorders in society more broadly.

Even among individuals with substance use disorders, there is substantial variability in how craving is described. Further, there appears to be no single unifying dimension of craving and descriptions of craving differ as a function of treatment status (i.e., in treatment vs. seeking treatment) (Bergeria et al., 2021; Kleykamp et al., 2019). More in-depth craving assessments typically include the following constructs: (1) feelings of anticipation related to drug effects, (2) preoccupation or pervasive thoughts of drug effects, (3) lack of control related to drug use, and (4) drive or intention to use drugs. These unique dimensions may be differentially relevant as a function of treatment needs. Therefore, craving assessments which incorporates one or more of these dimensions – that do not mention craving specifically – may be more acceptable and/or relevant to individuals who use opioids or cannabis to treat pain, but do not have a use disorder.

Limitations

The current study was based on a self-selected sample of 12 participants who reported chronic pain, as well as cannabis and opioid use, and generalizability of findings may be limited, especially considering that the sample was 83% female and 75% Non-Hispanic White. Future studies with larger and more diverse samples are needed to replicate our findings and improve their generalizability. Additionally, a large number of individuals who completed the screening questionnaire were excluded due to our specific and narrow study inclusion criteria. Our small sample size may have limited power to detect between-group differences, for example related to overall EMA compliance and mean craving scores, which did not reach statistical significance. Additional factors other than participants’ reaction to the ‘craving’ question may also have impacted their craving experience and response behavior. For example, participants who had a negative reaction to the assessment of craving also reported more frequent short-acting opioid medication use compared to those that had no negative reaction. More generally, craving is not a well-defined and operationalized concept, so differences in understanding of craving may also be responsible for differences in participant response behaviors in EMA items. Qualitative responses showed that some individuals perceive cravings as physical, others as mental/psychological, and yet others as reflective of problematic drug use and addiction. Future studies are needed to explore potential differences in understanding and interpretation of craving more broadly. Moreover, we initially conducted the cognitive interview asking participants about their interpretation of the craving items, but subsequently had all participants respond to these same unchanged items in the EMA data collection phase. Based on this order of data collection, participants who had responded with negative reactions to craving items in the interviews may have been offended by the fact that they shared their opinion, but surveys were not changed in response, which may have further impacted their responses to EMA items. It remains unclear if this had a significant impact on our results.

Despite these limitations, the current study provides preliminary evidence that individuals’ reactions to wording of items to assess craving of opioid and cannabis use has the potential to impact their survey response behavior.

Conclusions

A subset of individuals with chronic pain may have a negative reaction to wording of ‘craving’ related to opioid and cannabis use in survey items and this reaction may impact survey responses. Alternative wording of survey items is recommended. For example, studies assessing this or similar constructs may want to reword survey questions to assess ‘wanting’ or ‘needing’ of medications and substances instead of ‘craving’. For example, the cue-induced craving subscale of the Autonomy Over Smoking Scale (AUTOS) assesses ‘wanting’ of cigarettes (e.g. ‘When I feel stressed, I want a cigarette’) (DiFranza et al., 2009). Based on our findings from cognitive interviews it seems that some individuals with chronic pain who are using opioids and cannabis are already associating ‘craving’ with ‘wanting’ or ‘needing’, therefore wording survey items in this way may be an intuitive approach that could be well understood and accepted by study participants and individuals with chronic pain more broadly.