Health Reports
Understanding the socioeconomic profile of people who experienced opioid overdoses in British Columbia, 2014 to 2016

by Gisèle Carrière, Claudia Sanmartin, and Rochelle Garner

DOI: https://www.doi.org/10.25318/82-003-x202100200003-eng

In 2016, British Columbia’s (B.C.) Provincial Health Officer declared a public health emergency in response to increases in illicit-drug overdoses and related deaths. Between 2011 and 2016, the number of illicit-drug toxicity deaths in B.C. totalled 2,788, having increased from 295 in 2011, to 991 in 2016.^{Note 1} These deaths were largely related to the use of illicit fentanyl and its analogues. In 2019, these substances were detected in 87% of illicit-drug toxicity deaths. In 2012, this proportion was 5%.^{Note 1} Although the opioid crisis is national in scope, B.C. continues to experience a disproportionately higher rate of deaths attributed to illicit drug overdoses, at 20.7 per 100,000 population (age-adjusted), relative to 8.4 for the whole of Canada.^{Note 2Note 3}

In response, the B.C. government created a public health surveillance infrastructure to monitor overdoses in near real time. The B.C. Centre for Disease Control (BCCDC), various provincial departments, the B.C. First Nations Health Authority and other agencies collaborated to develop record linkages that meet information needs to help develop policy and intervention strategies. This collaboration led to the creation of B.C.’s Provincial Overdose Cohort (hereafter, the Overdose Cohort) that focused initially on basic demographics, health services use and prescription patterns of people who had experienced opioid overdoses in B.C.^{Note 4}

The Overdose Cohort revealed important information pertinent to the crisis; i.e., males comprised two-thirds (66%) of overdose cases, with close to half (49%) aged 20 to 39 years old.^{Note 4Note 5Note 6} Furthermore, it revealed that people who overdosed had high engagement with health services,^{Note 5} with 75% having visited community-based physicians, 54% having visited an emergency department and 26% having been admitted to hospital in the year before the overdose episode. By comparison, 17% and 9% of matched B.C. population controls (no overdoses) visited an emergency department and/or were admitted to hospital, respectively. Substance use and mental health-related concerns were the most common diagnoses among people who were hospitalized and who subsequently overdosed. Given the context of a widely held view that medical prescribing contributed to the crisis, Smolina et al. used the Overdose Cohort to examine this and showed that most cohort members did not have a prescription for a pain-relieving opioid when they overdosed, and half had not had a prescription in the five years prior to their overdose.^{Note 7} Other researchers have shown that the number of overdose deaths was 40% higher during the weeks when social assistance cheques were distributed than during the non-cheque weeks.^{Note 8}

While provincial interventions to prevent overdose deaths showed some success,^{Note 9Note 10Note 11} deaths from illicit-drug toxicity remained disproportionately high in B.C.,^{Note 12} warranting further integration of other relevant information.

In 2018, Statistics Canada worked with B.C.’s existing inter-sectoral overdose response partnership to provide a more fulsome depiction of the economic and social circumstances of individuals who had experienced an opioid overdose. Through this collaboration, Statistics Canada generated the B.C. Opioid Overdose Analytical File (BC-OOAF), a mirror version of B.C.’s Overdose Cohort, but augmented through linkages to other federal data holdings, e.g., on tax, social assistance, justice and immigration.

The objective of the present study is to use the BC-OOAF to extend and update information previously published by Statistics Canada for individuals who experienced illicit-drug toxicity deaths^{Note 13Note 14} by including people who experienced non-fatal overdose events.

Methods

Statistics Canada B.C. Opioid Overdose Analytical File (BC-OOAF)

With assistance from the BCCDC, Statistics Canada applied its case-finding algorithms^{Note 4} to define opioid overdose cases in B.C. from January 1, 2014, through to December 31, 2016, and created the BC-OOAF, which contains integrated federal data that characterize individuals’ contact with the health, justice and economic systems. The approach is detailed in a companion report.^{Note 15}

Data sources for case ascertainment and linkage

Opioid events were identified in each of the following provincial data sources:

(i) the Medical Services Plan (MSP)
(ii) B.C. Emergency Health Services (EHS), i.e., paramedic-attended overdoses
(iii) B.C. coroner, i.e., confirmed illicit-drug toxicity deaths (hereafter referred to as fatal overdoses)
(iv) Discharge Abstract Database (DAD), i.e., acute-care hospitalizations
(v) National Ambulatory Care Reporting System (NACRS), i.e., emergency department visits.

Ministries and agencies in the province of B.C. are responsible for collecting the data for the MSP, EHS and the coroner, and they shared these data with Statistics Canada for this project. The Canadian Institute for Health Information shares the DAD and NACRS data with Statistics Canada on an annual basis. More information about these data sources is available elsewhere.^{Note 16Note 17Note 18Note 19} These five data sources (MSP, EHS, B.C. coroner, DAD and NACRS) were linked at Statistics Canada. Two data sources, originally used by B.C. in its construction of the Overdose Cohort—the B.C. Drug and Poison Information Centre, and case-based reporting by emergency departments in three of the five B.C. Health Authorities—were not available to Statistics Canada and, therefore, were not included in the construction of the BC-OOAF. However, those two excluded sources contributed less than 1% of unique person records and 0.5% of event records to B.C.’s Overdose Cohort.^{Note 20} The BC-OOAF represents individuals who experienced overdoses between January 1, 2014, and December 31, 2016, and uses the first overdose episode identified within the observation period as the index overdose.

BC-OOAF members were linked to the following Statistics Canada databases: (i) the Longitudinal Worker File,^{Note 21} (ii) T5007 Statement of Benefits File,^{Note 22Note 23} (iii) the Longitudinal Immigration Database^{Note 24} and (iv) the Uniform Crime Reporting Survey^{Note 25} to provide information on employment and social assistance, immigration status, and contacts with police, respectively. BC-OOAF members were also linked to B.C.’s PharmaNet data,^{Note 26} shared with Statistics Canada by the B.C. Ministry of Health, which included information on prescription drug dispensations at the time of, and in the years prior to, the index overdose episode.

All linkages were conducted in Statistics Canada’s Social Data Linkage Environment.^{Note 27} Survey and administrative data were linked to the Derived Record Repository using G-Link, a SAS-based generalized record linkage software that supports deterministic and probabilistic linkage developed at Statistics Canada.^{Note 28} The linkage was approved by Statistics Canada’s Strategic Management Committee,^{Note 29} and the use of the data is governed by the Directive on Microdata Linkage.^{Note 30} More information about these data, data governance, this record linkage method, and rates is reported elsewhere.^{Note 15}

Measures

Age and sex: Age (in years) was derived at Statistics Canada as the difference between the date of the index overdose and the person’s date of birth (DOB). Information regarding DOB and sex was primarily obtained from the B.C. Ministry of Health Client Registry System, the Enterprise Master Patient Index (2011 to 2017).^{Note 31} Of the 13,318 individuals described in the data file, 43 (0.3%) were missing both DOB and sex information in the B.C. Ministry of Health Client Registry System. For these individuals, DOB, age and sex were obtained from other data sources, in the following order of priority: EHS, B.C. Coroners Service, NACRS and DAD. Ultimately, there were 10 people for whom sex could not be determined (0.08%).

Immigrant and temporary resident status: Linkage to the Longitudinal Immigration Database identified people who held immigrant or temporary resident status in Canada at the time of the index overdose. This includes records for all landed immigrants and temporary residents, from 1980 through to 2017 (latest year available at the time of linkage). Landed immigrants were categorized according to their admission category: family-class immigrants, economic-class immigrants, refugees, and other immigrants. Years between the landing date (for landed immigrants) or the most recent permit (for temporary residents) and the index overdose were calculated for the BC-OOAF members linked to the Longitudinal Immigration Database.

Employment and social assistance: Information about employment and receipt of social assistance in the calendar year of, and in each of the five calendar years prior to, the index overdose episode was available for those with a valid and identifiable social insurance number (n=13,184, 99%). Employment characteristics were based on T4 information, the Government of Canada (Canada Revenue Agency) summary of employment earnings and deductions for a given year, while social assistance information was based on T5007 information, the form that reports compensation benefits paid to a person during a calendar year, e.g., disability benefits.

People were considered to be employed if their T4 earnings were $500 or more in the calendar year. For employed people, the industry of their main job (i.e., the job with the highest income) in the calendar year is reported according to the North American Industry Classification System (NAICS). This classifies industry as follows: construction (NAICS 23); administrative and support, waste management and remediation services (NAICS 56); accommodation and food services (NAICS 72); retail trade (NAICS 44-45); manufacturing (NAICS 31-33); and all other industry codes. People were considered to have received social assistance if they were successfully linked to a T5007 form, regardless of the amount of social assistance received in the calendar year. Summary measures were also created to describe the number of years, in the five years prior to the index overdose episode, in which people were employed or received social assistance, ranging from none to all five years.

Health care use: Use of the health care system, in the year prior to the index overdose, took account of the number of visits to an emergency department, for any reason, and for selected main reasons, based on NACRS records, and the number of in-patient acute-care hospitalizations overall, for any cause, and for selected most responsible diagnoses, as identified by DAD records. The main reasons for emergency department visits and hospitalizations were based on the following diagnostic codes from the International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10): opioid poisonings (T40.0 to T40.4, T40.6), other substance use-related conditions (F10 to F19), mental health-related conditions other than substance use (F00 to F09, F20 to F99), and injuries and poisonings other than opioid poisonings (S00 to T98, excluding T40.0 to T40.4 and T40.6). It should be noted that not all NACRS records provide an ICD-10 diagnostic code: 24.4% of NACRS records linked to the BC-OOAF for the two years prior to the index overdose were missing an ICD-10 code.

Prescription medications: Information on whether individuals had community dispensations for prescription medications was derived from B.C.’s PharmaNet database, which excludes medications dispensed in hospital. The following dichotomous variables were created to reflect the receipt of prescribed medications in the year (365 days) prior to the index overdose: (i) at least one dispensed prescription of any kind and (ii) at least one opioid prescription. Prescribed opioids were also further distinguished as either opioid agonist treatment or pain-related opioid medications, as per the algorithm provided by the BCCDC.

Police contacts: Measures of police contacts in the 24 months prior to the index overdose were created, and they include the number of contacts and violation types (e.g., property offences, shoplifting, offences against the administration of justice). The elapsed time between the most recent police contact prior to the index overdose was reported as a cumulative frequency of members categorized as follows: 0 to 29 days (one month), 30 to 90 days (three months), 91 to182 days (six months), 183 to 364 days (12 months), 365 to 547 days (18 months), or 548 to 730 days (24 months) prior to the index overdose.

Results

Overall, 13,318 people who experienced one or more opioid overdose events were identified between January 1, 2014, and December 31, 2016 (Table 1). These people were primarily male (65%), with the incidence of fatal overdoses more prevalent among males than females. Half (52%) of the index overdose episodes occurred among people younger than 40 years old. Individuals who experienced fatal overdoses were slightly younger than people in the non-fatal cohort (mean age: 40 vs. 42 years old). Most people (78%) experienced one overdose episode, rather than several, during the observation period. The proportion of individuals who experienced only one overdose episode, rather than more than one, was higher among the fatal cohort (83.5%) than among individuals in the non-fatal cohort (77.3%; Table 1).

Immigration status

Seven percent of individuals were identified as landed immigrants at the time of their index overdose and, at the same time, another 0.4% as temporary residents (Table 2). Among immigrants and temporary residents described in the BC-OOAF, 41.0% arrived in Canada more than 20 years before their index overdose, and 11.3% arrived less than 5 years before their index overdose. Overall, 29.4% of immigrants or temporary residents who experienced opioid overdoses were born in Southern Asia (Table 2).

Employment, income and social assistance

Almost two-thirds (66.2%) of people were not employed in the calendar year of their index overdose (Table 3), although that figure was slightly lower among people who experienced a fatal overdose (63.9%). Among people employed during the year of their index overdose, one-fifth (21.4%) were employed in construction, followed by 12.2% in administrative and support, waste management and remediation services and 11.7%, in accommodation and food services. Over the five years preceding people’s index overdose, 41% did not have paid employment, while one-fifth (20.4%) were employed in all five years (Table 3).

Approximately half (50.4%) of people received social assistance in the calendar year of their index overdose. Over the five years preceding their index overdose, 44% of people had no record of having received social assistance, while one-third (29.4%) received social assistance in all five years prior to their index overdose (Table 3).

Health care use

In the year prior to their index overdose episode, almost 62% of people visited an emergency department, with 32% having visited three or more times in that year (Table 4). The proportion of people who visited the emergency department was lower among people who experienced a fatal overdose (57%) than among people in the non-fatal overdose subgroup (63%). Among people admitted to an emergency department, 37.2% were admitted for an injury or poisoning (other than an opioid overdose), 17.1% were seen for a substance use-related mental health condition, and 15.6% were seen for a mental health condition not related to substance use (Table 4).

Overall, about one-third (29.3%) of people were admitted to an acute-care hospital as in-patients in the year prior to their index overdose. Fifteen percent were seen once during the year, while 7.8% were admitted three or more times. Among people who were hospitalized, more than one-quarter (26.6%) were hospitalized for substance use-related conditions, 20.9% for non-substance use-related mental health conditions, and 20.6% for injuries other than opioid poisonings (Table 4).

Most people (87.2%) were dispensed a prescription in the year prior to their index overdose; 45.9% of people were dispensed a prescription opioid product. These opioid products were classified as opioid agonist treatment (23.1%) or pain treatment (22.7%). The proportion of people who did not have a prescribed opioid product dispensed within the year before their index overdose was higher among people who experienced a fatal overdose (61.0%) than among those who did not (53.3%).

Police contacts

Almost two-thirds (61.2%) of people (Table 5) had no formal contact with police in the two years prior to their index overdose, and there was no difference in formal police contact between those who experienced fatal overdoses and those who did not (both 61.2%). Among the 38.8% of people who had formal police contact in the two years before their index overdose, 34% had only one contact, while 50% had three or more formal contacts. The leading reasons for the offences charged were largely non-violent: 16.5% of charges were for shoplifting $5,000 or under, followed by disturbing the peace (10.9%), failure to comply with order (8.7%), and breach of probation (7.0%). The ranking and distribution of offence types did not vary according to whether or not people experienced a fatal overdose (Table 5). Furthermore, among people with formal police contact, about one-quarter (24.3%) had at least one police contact less than 30 days prior to their index overdose (Table 5). A smaller proportion of people who experienced a fatal overdose (15.4%) had police contact within that timeframe, compared with 25.4% among those who did not have a fatal overdose.

Discussion

Working collaboratively with partners in B.C., such as the BCCDC and provincial ministries and agencies, Statistics Canada created the BC-OOAF, a near-replication of the B.C. Overdose Cohort that contains additional federal information on people who experienced fatal and non-fatal opioid overdoses in B.C. between January 1, 2014, and December 31, 2016. The data enhancements focused on information about immigration status, employment, income assistance and police contacts. The linkage rates between members of the BC-OOAF and federal data sources were high,^{Note 15} and comparisons between the analytical file used in this study and the B.C. Overdose Cohort revealed similar distributions of demographics and the use of health care services. However, the BC-OOAF provided new information about employment, social assistance and police contacts during the time that led up to these people’s index overdose.

Results of this study indicated that immigrants and temporary residents were underrepresented among people who experienced an opioid overdose: 7% of BC-OOAF members were immigrants or temporary residents compared with 28.3% of B.C.’s provincial population in 2016.^{Note 32} Results also indicated that immigrants who overdosed were primarily established immigrants who had arrived two or more decades before their index overdose and were primarily from Southern Asia. This reflects the primary source countries for immigrants in B.C., in general. The immigrant population in B.C. is largely composed of people born in Asia (61%) and primarily from China (15.5%) and India (12.6%).^{Note 32}

Results indicated that, overall, people in B.C. who experienced an overdose had also experienced income and employment instability in the years that led up to their index event. Approximately 34% of this population were employed in the year of their index overdose, compared with 59.6% of the overall B.C. population, in 2016.^{Note 33} Previous research at Statistics Canada has shown that, in general, people in B.C. who experienced a fatal overdose had earnings that were far lower than the median employment incomes for employed people in the province.^{Note 13} Similarly, the results of the present study showed signs of economic marginalization among people who had fatal and non-fatal opioid overdoses. Associations between unemployment, marginalization and illicit substance use,^{Note 34Note 35Note 36} and, more specifically, between socioeconomic deprivation and opioid overdose, have previously been recognized.^{Note 37Note 38Note 39Note 40}

In a review article, Henkel reported that problematic substance use among people increased their likelihood of unemployment and decreased their chance of finding and holding down a job, and that unemployment was a significant risk factor for their subsequent development of substance use disorders.^{Note 34} Similar findings on economic disadvantage have been made among opioid users in B.C. and Ontario.^{Note 3Note 35Note 41} In future work, the integration of additional federal data holdings, including information on social housing stock,^{Note 42} could further illuminate intersections between people’s socioeconomic circumstances and their overdose risk.

Among the employed people described in this analytical file, workers in the construction industry were overrepresented (21.4% of those employed) relative to the overall rates for B.C.’s population working in this industry. In B.C., workers in this industry represented 8% overall, in 2015, and 13% for males, specifically.^{Note 33} This is consistent with narrative evidence collected by the B.C. Coroners Service, which reported that 44% of people who had died of illicit drug toxicity were employed at the time of their deaths, largely (55%) in trades and transport industries (which include construction).^{Note 43}

Almost two-thirds of people who experienced an overdose had no formal contact with the police. While the proportion of those who had at least one contact with the police in the two years prior to their index event (38.8%) was higher when compared with the provincial population in general (0.3% of B.C.’s population had been accused of a criminal incident in 2016),^{Note 14} their offences were primarily non-violent. The majority of offences among the people described in the analytical file who had police contacts were related to shoplifting. These police contact and employment results underscored the recognized relationship between greater economic marginalization and more likely contact with the justice system,^{Note 44} as well as potential relationships between substance dependence, lost employment and property crimes committed to secure basic survival.^{Note 45Note 46Note 47Note 48} Further research using the BC-OOAF is needed to fully understand the relationship of these factors to opioid overdoses.

While this study provided new information regarding the characteristics of people who experienced an opioid overdose, the following limitations are noted. First, results represented univariate associations between cohort characteristics and outcomes that should not be interpreted as having direct effects. Second, the analysis did not include a comparative group of people among whom no overdose was detected. Third, although sources that covered much of the spectrum of health care services were used for overdose case ascertainment, and the observation period covered two years, overdoses were likely undercounted, since some people who overdosed possibly did not have any contact with medical health services.^{Note 49} Fourth, although the index overdose event was the first observed within the study period, it may not have been people’s first overdose. Fifth, while fatal overdose events were ascertained by the BC Coroners Service and defined here as those that occurred within the observation period, other people in the non-fatal overdose subgroup described in the analytical file did die from other causes during this time. Under other study conditions, the fatal and non-fatal cohort designations could be applied differently. Further study of other sources of fatality information, such as death registries, could yield important additional information. Sixth, other work has shown that members of B.C. First Nations communities are overrepresented among people experiencing overdoses and overdose deaths.^{Note 50} Information used to identify Information used to identify Indigenous peoples was not available in the present study. Finally, since the end of this study’s observation period, there may have been changes to overdose circumstances in the overdose crisis. As such, exercising caution is warranted when generalizing the current results to years beyond those reported here.

Conclusion

Understanding socioeconomic determinants associated with opioid overdose is critical to informing actionable efforts that could reduce or prevent illicit-drug harm. Partnerships and collaborative use of integrated data to inform these efforts partially addressed existing information gaps. This demonstrated that, going forward, existing data that now include federal data could be used to monitor trends over time and to inform preventive efforts. This could possibly be a first step towards using federal data holdings to integrate more sectors, such as housing.