Health Reports
Significant factors associated with problematic use of opioid pain relief medications among the household population, Canada, 2018

by Gisèle Carrière, Rochelle Garner and Claudia Sanmartin

DOI: https://www.doi.org/10.25318/82-003-x202101200002-eng

Abstract

Background

Reliance on the use of opioids to manage pain has increased over time, as have opioid-related morbidity and deaths. In 2018, 12.7% of Canadians reported having used opioid pain relief medications (OPRMs) in the previous year. Among these people, 9.6% had engaged in problematic use that could cause harm to their health. Though socioeconomic characteristics associated with opioid-related harms have previously been reported, population-level evidence based on administrative health data lacks important behavioural and psychosocial information. This analysis extends previous research by using modelling to report factors related to the problematic use of OPRMs for the household population aged 15 and older in Canada.

Data and methods

This analysis uses responses to the 2018 Canadian Community Health Survey to identify factors that are significantly associated, after adjustment using multivariate logistic regression models, with elevated odds of problematic use of OPRMs.

Results

The fully adjusted model confirmed that being male, being younger (ages 20 to 24), having fair or poor mental health, having unmet needs for help with mental or emotional health or substance problems, being a smoker, or being unattached and living with others were significantly related to problematic OPRM use.

Interpretation

Subjective perceptions significantly related to problematic OPRM use, independent of socioeconomic circumstances, were examined in this study. While previous research based on administrative health data has contributed much to knowledge about factors associated with opioid harms, modelled results revealed that self-reported experiential factors also warrant consideration as they are significantly associated with problematic use. Having fair or poor mental health, having unmet perceived needs for help, and being unattached in terms of household arrangement relationship were related to problematic use of OPRMs, even after adjustment for socioeconomic and other health covariates. This study suggests risk profiles that could be used to inform health care providers, and strategies to support safe pain management.

Keywords

Opioids, opioid overdose crisis, pain, problematic use, misuse

Authors

Gisèle Carrière (gisele.carriere@statcan.gc.ca) and Rochelle Garner are with the Health Analysis Division, Analytical Studies and Modelling Branch, Statistics Canada, Ottawa. Claudia Sanmartin is with the Strategic Analysis, Publication and Training Division, Analytical Studies and Modelling Branch, Statistics Canada, Ottawa.

End of text box

Introduction

Pain and its management are fundamental to the opioid overdose crisis, leading some to call for a national pain strategy.^{Note 1} This persisting public health emergency has been exacerbated by the COVID-19 pandemic in Canada,^{Note 2Note 3Note 4} which has disrupted opioid harm reduction initiatives and health care services.^{Note 5} Pain management has primarily relied on prescription opioids as treatment, and this has contributed to the development of the opioid crisis in Canada.^{Note 6Note 7} Prescription opioid use has increased over the past two decades in North America,^{Note 6Note 8} as have overdoses, poisonings and deaths related to both prescription and non-prescription (illicit) opioid use.^{Note 9Note 10}

Opioid harms have been experienced at a greater rate in some regions of the country^{Note 10Note 11} and among some specific population groups. These groups include people who experience lower income,^{Note 12Note 13} or have periods of employment instability,^{Note 14} or are employed in the construction industry.^{Note 14} First Nations people, particularly on-reserve First Nations, also face greater rates of opioid harms.^{Note 15Note 16Note 17Note 18} Inequalities in the socioeconomic determinants of health (e.g., personal or household income) and racially discriminatory health services stemming from the historical and ongoing impact of colonization^{Note 19Note 20} have reportedly resulted in, for example,  Indigenous people being undertreated, being treated without their consent or having experienced poor-quality treatment.^{Note 20} Discriminatory services and health determinants have resulted in trauma, intergenerational trauma and increased susceptibility to disabling pain among Indigenous people.^{Note 4Note 21}

Previous administrative health data-based analyses of the characteristics of individuals experiencing opioid-related poisoning revealed an increasing trend among youth aged 15 to 24.^{Note 22} However, information about people’s broader sociodemographics was lacking in those analyses. More fulsome descriptors of people who experienced opioid-related overdoses, whether fatal or non-fatal, extended the research to fill certain information gaps using person-level record linkages of population administrative data. For example, a linkage documenting an individual’s economic circumstances in the years prior to an opioid overdose thereby provides evidence to inform planned interventions.^{Note 14Note 23}

The opioid emergency prompted the implementation of Canada’s opioid-prescribing practice guidelines in 2017.^{Note 24} These guidelines aimed to reduce prescription opioid exposure in favour of alternate pain treatment modes in future populations, thereby to reduce future harms. However, experts have deemed the new guidelines unlikely to resolve all harm-related issues for all people experiencing chronic pain and already receiving opioid treatment, and instead called for a national pain strategy.^{Note 1} In particular, pain treatment outcomes represent an important remaining gap pertinent to strategizing on pain.^{Note 4Note 21} This study seeks to identify significant factors relating to one outcome of opioid-based pain management, namely problematic use.

Presently in Canada, opioid pain relief medications (OPRMs), such as codeine, remain available both with and without (i.e., over the counter) a prescription. In 2018, 12.7% of Canadians aged 15 years and older reported OPRM use in the past year, 9.7% (3.7 million) of whom engaged in behaviour patterns indicative of problematic use.^{Note 25}

Significant information gaps related to pain prevalence and problematic use of OPRMs for pain management remain.^{Note 4Note 21Note 26} For example, Statistics Canada has reported on the national prevalence of chronic pain,^{Note 27} and others who conducted meta-reviews concluded that a substantial proportion (i.e., approximately one in five) of Canadians experience chronic pain.^{Note 25Note 26Note 27Note 28Note 29} Yet comprehensively capturing pain prevalence is challenging because of the lack of an international definitional consensus and inherent idiosyncrasies of experiencing pain.^{Note 21Note 26} Furthermore, while Statistics Canada has reported elevated pain prevalence among specific populations—for example, the Indigenous population^{Note 30}—this information is not available for every year.

Additionally, existing evidence regarding the circumstances of people experiencing opioid-related harms is largely based on administrative data, which lack important information about pertinent health behaviours (e.g., smoking, alcohol consumption),^{Note 8Note 31Note 32} psychosocial factors, identity or sociodemographics. Some of this information may relate to cultural differences in the ways pain is experienced, understood and treated.^{Note 33Note 34} Psychosocial factors play a role in pain perception processes,^{Note 35Note 36} including historical and ongoing trauma^{Note 37} that may catalyze pain experiences. Despite great diversity in circumstances and culture, Indigenous people generally experience a higher prevalence for experiencing pain than the general population, with a rate of 17.7 among First Nations people and 20.6 among Métis, versus 13.8 among non-Indigenous people.^{Note 30} Given the general reliance on prescription opioids to manage pain,^{Note 6Note 7} Indigenous people may thereby also experience greater opportunity for opioid-related harm. It has been reported that in 2018 12.8% of all overdose deaths in British Columbia occurred among First Nations people. This rate is 4.2 times higher relative to other residents and, furthermore, had increased from a rate of 3.4 in 2017.^{Note 16} However, information gaps remain for both the general population and among specific population groups in regard to such issues as psychosocial risk factors for pain and the effectiveness of pain management. These were some gaps identified by the Canadian Pain Task Force as priorities requiring additional research.^{Note 4Note 21Note 26}

In 2019, Statistics Canada reported descriptive associations between demographic and geographic descriptors, certain mental health disorders, and problematic OPRM use among Canada’s OPRM-using population aged 15 years or older.^{Note 25}  The goal of this analysis is to extend that previous research by examining the associations for a broader range of characteristics. It strives to clarify which socioeconomic, health behaviour and psychosocial factors are independently and significantly associated with a greater likelihood of problematic OPRM use after accounting for other descriptors.

Methods

Data source

The Canadian Community Health Survey (CCHS) is a cross-sectional survey that collects information related to health status, determinants and health care use of the Canadian household population 12 years of age or older. The 2018 CCHS included an optional module regarding the use of OPRMs and was collected from January to December 2018. Its overall response rate was 58.8%, resulting in a final sample of 55,600 respondents.^{Note 38} In 2018, Statistics Canada used the term “Aboriginal” in its data collection and dissemination products. Since that time, the term “Indigenous” has been adopted as the preferred terminology. This analysis will hereafter refer to respondents to the 2018 CCHS who self-identified on the survey as being First Nations, Métis or Inuit as having Indigenous identity. First Nations people can include people who are status and non-status self-identifying First Nations. A limitation of the CCHS is that people living on reserves and other Indigenous settlements in the provinces are not part of the CCHS target population and are not included. Nor are full-time members of the Canadian Armed Forces, youth aged 12 and older living in foster homes, institutionalized people, or those living in the Quebec health regions of Région du Nunavik and Région des Terres-Cries-de-la-Baie-James. These exclusions represent less than 3% of the target population, the Canadian household population aged 12 and older. Given that Indigenous people are recognized as one of the key subpopulations affected by pain,^{Note 4Note 30} and given this noted limitation, Indigenous people likely were not adequately covered in this study.

Furthermore, because the present study uses only one year of CCHS-collected data, and the CCHS is considered to be representative of the territories only when a two-year file is used   (i.e., CCHS 2017 and 2018 cycles combined), residents of the territories were not included in the CCHS 2018 annual data file. This limitation is particularly relevant to the reported Indigenous identity covariate results, since Inuit primarily comprise Nunavut’s population, and elsewhere, 50.7% of the Northwest Territories’ population reported having Indigenous identity. Therefore, excluding on-reserve populations and the territories resulted in a non-representative Indigenous population.

The analytic sample was further restricted to respondents aged 15 years or older, and model analyses were also restricted to only those respondents who reported having used OPRMs in the previous 12 months.

Definitions

Problematic use of opioid pain relief medications

Respondents were asked whether they had used any OPRMs in the previous 12 months; respondents were told to include codeine-containing medications, but to exclude pain relievers that do not contain opioids (e.g., Advil). Problematic use of OPRMs was defined as the reporting of any of the following behaviours among OPRM users: (1) taking the medication in greater amounts than prescribed, or more often than directed; (2) intentionally using OPRMs for the experience, for the feeling they caused, or to get high; (3) using OPRMs for problems other than pain relief, for example, to feel better (improve mood) or to cope with stress or problems; (4) tampering with a product before taking it (e.g., crushing tablets to swallow, snort or inject).

Covariates

Covariate selection was guided by published research regarding factors known to be related to problematic OPRM use and opioid adverse events.^{Note 13Note 14Note 22Note 23Note 25} Smoking behaviour relates to pain.^{Note 31Note 32} Neurological mechanisms have been invoked in relating tobacco smoking-heaviness or nicotine-dependence and pain to increased risk for prescription opioid use dependence.^{Note 31} Social connectedness has been shown to be health protective.^{Note 39} Instability of connections, including unemployment (i.e., a period without usual contact with a workplace, employer, co-workers or clients), has been associated with substance dependence, misuse and harms.^{Note 23Note 40} Other research shows how the risk of problematic opioid use is related to low income and unemployment^{Note 41} and to poor social networks or support when examined distinctly from financial resources.^{Note 39} Thus, in addition to household income and labour force status, information about perceived degree of community belonging and household living arrangements were used as indicators of social connectedness.

Demographics and geography

Respondent age was categorized as follows: 15 to 19 years, 20 to 24 years, 25 to 29 years, 30 to 39 years, 40 to 49 years, 50 to 59 years, 60 to 64 years, and 65 years and older.

Respondents were classified according to their province of residence at the time of the survey. Subprovincially, respondents were classified according to their residential location’s population centre size, according to the 2016 Census: small population centres (1,000 to 29,999 people), medium population centres (30,000 to 99,999 people) and large urban population centres (100,000 or more people).^{Note 42} Areas outside population centres were classified as rural areas.

Socioeconomic

Respondents’ highest educational attainmentwas categorized as less than secondary school graduation; secondary school graduation, no post-secondary education; or postsecondary certificate, diploma or university degree.

Respondents’ working status in the week prior to the survey was categorized as employed (working) at a job or business, employed but absent from a job or usual business (e.g., vacation), did not have a usual job or business, or aged 75 or older (i.e., out of scope for this item).

Total household income quintile represents the adjusted ratio of the respondent’s household income to the low-income cut-off for their household and community size, divided into quintiles relative to provincial distributions. When missing (approximately 10% of respondents), total household income was imputed during survey processing.^{Note 38}

Social connectedness

Respondents were categorized according to their household living arrangements as follows: unattached individual living alone, unattached individual living with others, individual living with spouse or partner (with or without children), single parent living with children or child living with a single parent (with or without siblings), child living with two parents (with or without siblings), and other.

Respondents were asked to rate their perceived level of community belonging on a four-point scale. Responses were dichotomized to distinguish respondents who perceived their community belonging as very strong or somewhat strong from those whose sense of belonging was somewhat weak or very weak.

Health measures

Respondents rated both their general health and their mental health on five-point Likert scales. Responses were regrouped to reflect three levels of self-perceived general and mental health: excellent or very good, good, and fair or poor.

Respondents were asked to rate their life stress on a five-point Likert scale. Responses were dichotomized to reflect those who perceived their life as not at all stressful, not very stressful or a bit stressful, compared with those who said that life was quite a bit or extremely stressful.

Respondents were asked whether they had been diagnosed by a health care professional with any of the following chronic health conditions: suffers effects of a stroke, arthritis, heart disease, high blood pressure, diabetes, cancer (ever in their lifetime), cancer (presently), asthma, inflammatory bowel disorders (e.g., colitis, irritable bowel syndrome), mood disorder and anxiety disorder. The number of chronic conditions reported by a respondent was also summed and examined categorically: none of the chronic conditions, one chronic condition, and two or more chronic conditions.

Respondents were asked whether they had sustained an injury in the past 12 months that was serious enough to limit usual activity, excluding repetitive strain. This variable was examined dichotomously (yes or no).

Respondents’ smoking status during the 12 months prior to the interview was categorized as ever-smoker (i.e., former or current daily or occasional smokers); experimental smoker (i.e., has smoked at least one cigarette in their lifetime but non-smoker at the time of the survey); and never-smoker (i.e., never smoked a whole cigarette).

Similarly, respondents’ alcohol consumption in the 12 months prior to the survey was used to describe the type of drinker that they were: regular drinker (consumed alcohol at least once a month), occasional drinker (consumed alcohol less than once a month) and non-drinker (including both former drinkers and people who have never been drinkers).

Respondents were asked a series of questions to determine unmet health care needs regarding mental health or use of alcohol or drugs. Responses to those questions classified respondents into one of the following categories: no perceived need for mental health care and no problems with use of alcohol or drugs, all perceived needs were met, perceived needs were partially met, and perceived needs were not met.

Analytical techniques

The prevalence of two outcomes—use of OPRMs and problematic use among those using OPRMs—was examined by selected population characteristics, including testing for significance across covariates (Tables 1 to 3). Characteristics found to be significantly related to problematic OPRM use after adjusting for age and sex (Table 4) were included in a multivariate logistic model (Table 5). A multivariate logistic regression model was used to determine which factors were independently associated with problematic use among individuals using OPRMs after full-adjustment.

Sample weights were applied to make results generalizable to the Canadian household population aged 15 and older, while bootstrap weights were applied in SAS EG 7.1 with SAS-callable SUDAAN 11.0 to account for underestimation of standard errors resulting from the complex survey design.^{Note 43} Confidence intervals for each estimate point are reported, and coefficient of variation notations are presented in the tables wherever interpretation warranted caution because of a greater degree of unreliability owing to small sample numbers.

Results

Tables 1 to 3 report the prevalence of OPRM use and problematic use of OPRMs among OPRM users in the 12 months prior to the survey by selected respondent covariates. Among OPRM users, problematic use was significantly more common in males (11.3%) versus females (8.2%); in respondents aged 20 to 24 years (16.1%) relative to those aged 50 to 59 (7.7%); in respondents with less than secondary school graduation (12.7%) versus those with postsecondary education (8.3%); and in those living in low-income households (11.2%) relative to those in the highest-income households (7.6%). Conversely, problematic use was significantly lower among people residing in small population centres (6.9%) relative to those in large population centres (9.6%).

Regarding social connectedness characteristics, the prevalence of problematic use was significantly higher in unattached individuals living alone (10.6%) or with others (18.6%) relative to people living with a spouse or partner (8.0%), and in people with a somewhat weak or very weak sense of community belonging (13.2%) relative to those who perceived somewhat or very strong community belonging (7.5%).

According to health-related circumstances and conditions, the prevalence of problematic OPRM use was higher for those who perceived their life as quite a bit or extremely stressful (14.2%) versus those who perceived it as a bit, not very or not at all stressful (7.5%); whose general health was fair or poor (13.5%) relative to those rating it as very good or excellent (7.4%); whose mental health was fair or poor (21.2%) or good (9.4%) relative to people with very good or excellent mental health (6.6%); and who reported two or more of the examined chronic conditions (12.9%) relative to people reporting none (7.0%) (Table 3). The prevalence of problematic OPRM use was more than twice as high for people who reported mood (19.2%) or anxiety (17.1%) disorders relative to people who did not (respectively 7.3% and 7.7%). In addition, ever-smokers (11.8%) and experimental smokers (10.8%) had almost twice the prevalence of problematic OPRM use as never-smokers (5.5%).

In terms of health service use and access, those using OPRMs and who had unmet or partially met needs for help with emotional or mental health concerns or use of alcohol or drugs (19.6% and 23.9% respectively) had about three times higher prevalence of problematic use compared with those who did not perceive these needs (6.6%).

Of note, while use of OPRMs varied across some of the examined characteristics and by Indigenous identity group, problematic OPRM use was not associated with each of: Indigenous identity, work status, having sustained an activity-limiting injury in the past 12 months, and drinker type (Tables 1 to 3).

Using identifying covariates that were significantly related to problematic OPRM use after age and sex adjustment (Table 4), the fully adjusted model (Table 5) confirmed significantly elevated odds of problematic OPRM use in the previous 12 months for people with the following characteristics: being aged 20 to 24 years (odds ratio [OR]=2.66), being male (OR=1.55), having fair or poor mental health (OR=1.90), being an unattached individual living with others (OR=2.25), being an ever-smoker (OR=2.06) or experimental smoker (OR=2.02), or having an unmet (OR=2.61) or partially met (OR=2.18) need for mental health care (i.e., problems with emotional or mental health, or use of alcohol and drugs).

Discussion

This study used multivariate adjustment to report new population-level information for Canada about key psychosocial, health and socioeconomic descriptors that were related to an elevated likelihood of reporting problematic use of OPRMs in 2018 among people aged 15 and older who used OPRMs in the previous 12 months. Although the prevalence of OPRM use varied by Indigenous identity and all selected characteristics that were examined in the present study, the prevalence of problematic OPRM use did not always differ across all factors. A fully adjusted logistic regression model confirmed that being male, being younger (20 to 24 years), being unattached and living with others, being an ever-smoker or experimental smoker, having fair or poor mental health, and—particularly—having unmet needs for help regarding problems with emotional or mental health or use of drugs or alcohol were significantly related to an elevated risk of problematic OPRM use.

 In contrast to previous research using administrative health data to examine factors associated with opioid harms, a strength of the present study was the inclusion of psychosocial and behavioural health determinants as measured by the CCHS.

Modelled results showed independent, significant relationships between subjective measures, such as perceived unmet needs, that suggested psychosocial determinants are important factors beyond the impact of income, employment and education. Furthermore, this study yielded new information about the degree to which being disconnected from others—measured as unattached people living alone—fair or poor mental health, and unmet needs for help are independently related to an elevated likelihood of problematic OPRM use.

The significant association between being an unattached person living with others and elevated problematic OPRM use found in this study aligns with previous work that showed this living arrangement was related to a rate of opioid-poisoning-related hospitalizations more than twice as high relative to that of people living with a spouse or partner.^{Note 13} While a U.S. study found that living alone was not associated with opioid or diazepine misuse, misuse was three times more likely among unmarried adults.^{Note 44} Furthermore, a systematic review found a strong inverse association between having social support in the form of being in a relationship and opioid overdose.^{Note 12}

One explanation for the significant relationship between smoking behaviour and problematic OPRM use found in this study is that cigarette smoking-induced nicotine exposure can be analgesic, although these effects may differ by sex.^{Note 31} Perhaps smoking may represent another pain relief-seeking behaviour.

Having an unmet or a partially met perceived need for help for problems with emotional or mental health or the use of alcohol or drugs (including medications) was associated with more than twice as high odds of problematic OPRM use compared with people without those needs, after accounting for socioeconomic confounders. Furthermore, the model showed that the likelihood of problematic OPRM use was no different between people without these needs and people who had these needs but whose needs were fully met. Future study might examine mechanisms behind these associations by more closely examining the types of help that were needed, or why help was not sought out by some when needed. Although the survey captured some of this information, an investigation of associations between specific types of help needed was beyond the scope of this study.

The finding that having partially met needs was also associated with an elevated likelihood of problematic OPRM use raises the possibility that pain relief needs persisted among people using pain relievers who experienced mental health issues together with other comorbidities. Many of the specific chronic conditions examined in this study, such as arthritis, are known to be associated with physical pain,^{Note 45} mental anguish or disability.^{Note 46Note 47} The co-occurrence of mental health issues, chronic pain and substance use is a well-recognized intersection of health conditions and behaviours.^{Note 48} For example, having a history of substance use or mental health issues, especially anxiety disorders, has been shown to be prospectively associated with opioid misuse among adults living with chronic non-cancer pain.^{Note 49} Associations between anxiety disorders and chronic pain have also been reported by others.^{Note 46Note 49Note 50Note 51} Therefore, people suffering mental health issues may concomitantly have a greater need for pain relief owing to other comorbidities not fully accounted for here. It should be noted that 35% of OPRM users, and 25% of problematic OPRM-using respondents, reported none of the 10 chronic conditions examined. Such respondents may have used OPRMs for acute conditions or for chronic conditions not examined by the present study. Although the 2018 CCHS did not evaluate respondents’ pain levels, future research could examine the reasons for which individuals said they used OPRMs and determine whether these were differentially associated with problematic use.

In this study, 13% of OPRM-using respondents, and 30% of problematic OPRM users, reported having partially or fully unmet needs for help. Other work has indicated that opioids can have limited efficacy to manage pain entirely or for extended durations^{Note 6Note 7} or that evidence is insufficient to determine effectiveness.^{Note 52} Future research could explore these people’s reasons for using pain relievers, types of help that were needed but not received (e.g., emotional counselling), why some individuals did not try to access needed care in greater detail and how these factors relate to problematic OPRM use. Such investigations could partly respond to some of the priorities identified by the Canadian Pain Task Force, such as evidentiary gaps to patient-centred outcomes.^{Note 4Note 21}

Despite this study’s strengths, there were several limitations. Firstly, the stigma surrounding opioid use, particularly its problematic use, may have resulted in underreporting in this study. Next, because the CCHS is a cross-sectional survey and the temporal order of reported events could not be examined, reported associations should not be interpreted as causal. The CCHS used Likert scale response categories to measure most of the examined covariates (i.e., rather than continuous measures). These scales were further collapsed from five-point scales to two- or three-point scales for this analysis. Results also may not be generalizable to people younger than 15 or to populations excluded from the 2018 CCHS, such as the population living on-reserve or residents of the territories. While Indigenous identity was not related to greater prevalence of problematic OPRM use in the present study, it is recognized that Indigenous people have disproportionally experienced greater opioid-related harm.^{Note 15Note 16Note 17} The discordance between those facts and the presented results may be partly owing to the limiting scope of the CCHS. Data regarding culturally safe pain management among Indigenous population groups continue to present critical gaps that need to be addressed to comprehensively and accurately depict pain-related topics.^{Note 8Note 15Note 26}

Conclusion

This study provides new information about significant, subjective psychosocial, health and socioeconomic factors associated with the problematic use of OPRMs. Beyond the effects of income and education on problematic OPRM use, perceived unmet needs for help regarding problems with emotional or mental health or substances were significantly and independently associated with elevated risk of problematic use. This was also the case with having fair or poor mental health, being unattached and living with others, being a young person, being a smoker, and being male. Future research is warranted to examine types of unmet needs for help, perceived barriers to seeking help, and different reasons for using OPRMs. This study’s findings could serve to inform health care providers and to support safe prescribing and pain management strategies going forward.