Analytical Studies: Methods and References
Statistics Canada British Columbia Opioid Overdose Analytical File: Technical Report
by Claudia Sanmartin, Rochelle Garner, Gisèle Carrière, Anthony Matarazzo, Shannon Brennan, Jillian Boyce, Jennifer Thomas, Benjamin Mazowita, Lindsay Porter, Lindsay J. Dorder, Grant Schellenberg, Yan Zhang, Chris Schimmele, and Richard Trudeau
Acknowledgements
Acknowledgements to the Government of British Columbia data steward representatives who were provided the opportunity to review the appropriateness of use of these data for this report: Dr. Brian Emerson, Acting Deputy Provincial Health Officer; Dr. Réka Gustafson, Deputy Provincial Health Officer; Elenore Arend, Assistant Deputy Minister; Andrew Elderfield, Executive Director and Chief Data Steward, Data Management and Stewardship, Ministry of Health; Lisa Lapointe, Chief Coroner, Office of the Chief Coroner; Tej Sidhu, Office of the Chief Coroner Burnaby; Sandra Jenneson, Medical Director and Practice Lower Mainland Paramedic, BC Emergency Health Services; Louise Meilleur, Director of Health Surveillance, BC First Nations Health Authority; Dr. Shannon McDonald, Acting Chief Medical Officer, First Nations Health Authority; Sonia Isaac-Mann, Vice-President Programs and Services, First Nations Health Authority; Mary-Ellen Johnson, Data Access Audits and Agreements, Data Management and Stewardship, Ministry of Health; Stephanie MacPherson, Provincial Director, Adult Custody, Ministry of Public Safety and Solicitor General; Leigh Greiner, Director of Research and Strategic Planning, BC Corrections.
“All inferences, opinions and conclusions drawn in this report are those of the authors and do not reflect the opinions or policies of the data steward(s).”
Acknowledgements to the experts consulted at the BC Centre for Disease Control for providing expert guidance on the algorithm used to define the British Columbia Opioid Overdose Analytical File cohort at Statistics Canada, for pharmaceutical coding to identify opioid prescriptions and therapeutic agonist treatments, and for reviewing this report: Michael Otterstatter, Amanda Slaunwhite and Margot Kuo.
Acknowledgements to partners in the Government of British Columbia who provided data and expert advice regarding these data to Statistics Canada: British Columbia Ministry of Health, Elizabeth Vickery (BC Stats), Martin Monkman (BC Stats), Lisa Lapointe (BC Office of the Chief Coroner),Tej Sidhu (BC Office of the Chief Coroner).
Acknowledgements to federal policing and municipal partners who provided data and expert advice regarding these data to Statistics Canada: the City of Surrey, the Surrey Fire Service, the Surrey Royal Canadian Mounted Police Detachment and Public Safety Canada.
Acknowledgements to Len Garis, Senior Advisor, Centre for Social Data Insights and Innovation at Statistics Canada; Fire Chief (ret) for the City of Surrey B.C.; Adjunct Professor, School of Criminology and Criminal Justice & Associate to the Centre for Social Research at the University of the Fraser Valley (UFV); member of the Affiliated Research Faculty at John Jay College of Criminal Justice in New York; faculty member of the Institute of Canadian Urban Research Studies at Simon Fraser University.
Executive summary
Canada continues to experience an opioid crisis. While there is solid information on the demographic and geographic characteristics of people experiencing fatal and non-fatal opioid overdoses in Canada, there is limited information on the social and economic conditions of those who experience these events. To fill this information gap, Statistics Canada collaborated with existing partnerships in British Columbia, including the BC Coroners Service, BC Stats, the BC Centre for Disease Control and the British Columbia Ministry of Health, to create the Statistics Canada British Columbia Opioid Overdose Analytical File (BC-OOAF). This analytical file uses data and algorithms developed in British Columbia to identify illicit drug toxicity deaths (henceforth referred to as “fatal overdoses”) in order to compile fatal and non-fatal opioid overdoses that occurred between January 1, 2014, and December 31, 2016. Using data linkage capabilities at Statistics Canada, this information was integrated with data sources available at the agency (including tax, health care, immigration and justice data) to provide socioeconomic information regarding people experiencing overdoses. This technical report provides methodological details on how the BC-OOAF was created. Data integration activities were conducted at Statistics Canada using the Social Data Linkage Environment (SDLE)—a highly secure linkage environment established to support the creation of linked population data files.
The following administrative health data sources were used to identify fatal and non-fatal opioid overdoses: the British Columbia Medical Services Plan, BC Emergency Health Services (BCEHS), the BC Coroners Service database, the Discharge Abstract Database (DAD) and the National Ambulatory Care Reporting System (NACRS). Linkage rates using the SDLE environment and the British Columbia Ministry of Health’s Client Registry exceeded 97% for all databases except the BCEHS for which linkage rates were between 88% and 92% for the BCEHS data. Overall, 13,318 people who experienced 19,125 opioid overdose events between January 1, 2014, and December 31, 2016, are represented in the BC-OOAF. Linkage to a range of Statistics Canada data holdings was conducted to provide information on health service use (DAD, NACRS, BC PharmaNet), employment earnings and social assistance (T4, T5007 tax files), encounters with the justice system (i.e., via the incident-based Uniform Crime Reporting Survey, Integrated Criminal Court Survey and Integrated Correctional Services Survey), immigration status (Longitudinal Immigration Database) prior to the opioid overdose event, and death (Canadian Vital Statistics-Death Database) during the time period.
1 Introduction
Canada is currently experiencing an opioid crisis. The number of opioid-related deaths is rising steadily, with over 13,900 deaths occurring across the country between January 2016 and June 2019, with the highest rate occurring in British Columbia (Government of Canada 2018, 2020). While there is solid information on the demographic and geographic characteristics of people who experience fatal and non-fatal opioid overdoses in Canada, there is limited information on the social and economic conditions of those who experience these events. While policy makers, health practitioners and public safety practitioners have focused on addressing the immediate need for prevention and treatment in an effort to reduce harm and save lives, attention is turning toward better understanding the socioeconomic determinants of this crisis.
To fill this information gap, Statistics Canada partnered with the following organizations in British Columbia: the BC Coroners Service, BC Stats, the BC Centre for Disease Control (BCCDC) and the British Columbia Ministry of Health. This project was initiated originally by key stakeholders in the city of Surrey to address information gaps related to their efforts to address the opioid crisis. As such, the following partners were also involved: the City of Surrey, Surrey Fire Service, the Surrey Royal Canadian Mounted Police Detachment and the Fraser Health Authority. Administrative health and emergency response data provided to Statistics Canada have been integrated with national data, including hospital discharge, emergency room discharge, employment, income, social assistance, justice and immigration data. The following technical report provides details on the data, the linkage process and the methodology used to create the Statistics Canada British Columbia Opioid Overdose Analytical File (BC-OOAF). This analytical file is intended to be used to better understand the social determinants of opioid overdoses—both fatal and non-fatal—and is not intended to be used to generate official surveillance counts of overdoses.
2 Data sources
The following section describes the administrative databases provided by the Government of British Columbia and Statistics Canada to create the BC-OOAF. Some datasets were used to define opioid overdose episodes, while others were used to provide additional socioeconomic, justice and immigration information on those who experienced an opioid overdose during the observation period, which was defined as occurring between January 1, 2014, and December 31, 2016.
2.1 British Columbia data sources
British Columbia Ministry of Health’s Client Registry (2011 to 2017)
The British Columbia Ministry of Health’s Client Registry (“the Client Registry”) is the central registry that holds a Personal Health Number (PHN) for every person that receives a health care service in the province, including persons who do not have active health insurance coverage or who are not residents of British Columbia. A PHN is a unique, numerical identifier used to identify an individual client who has had any interaction with the British Columbia health care system. The Client Registry contains patient identifiers, including given name and surname, date of birth, sex, street address, postal code, city, census metropolitan area, census subdivision, and province of residence (British Columbia Ministry of Health 2011a) (Government of British Columbia n.d.a). The Client Registry was used by Statistics Canada to facilitate linkage, to obtain cohort members’ date of birth to derive age at first overdose, and to assign sex (if available).
Medical Services Plan
The Medical Services Plan (MSP) represents medical fee-for-service and alternate payment services provided by general practitioners and specialists in the province of British Columbia (British Columbia Ministry of Health 2011b, 2011c, 2011d). Also included in the file are MSP-insured services provided by other health practitioners, such as chiropractors, naturopaths, physical therapists, oral surgeons, podiatrists, optometrists, dental surgeons, oral medicine practitioners, orthodontists, massage practitioners, acupuncturists and midwives. MSP records include patient PHN, age, sex, place of residence, diagnostic codes, expenditure information (i.e., fee item, number of services provided) and service date. Practitioner type and identifying codes are recorded for service records in the MSP, as well as the practitioner’s geographical location and the type of facility in which the service was rendered or the procedure was performed. More information on British Columbia’s MSP can be found elsewhere (Government of British Columbia n.d.b).
BC Emergency Health Services
The BC Emergency Health Services (BCEHS) data file provides information on emergency dispatch centre details, including date and time of dispatch, type of incident, location of incident, paramedic impressions (e.g., difficulty breathing), interventions performed (including whether Naloxone was administered), number of times, and dosage, and—if the patient was transported—details about the destination (e.g., hospital name and location). Records also include PHNs, age, sex and place of residence. More information on the BCEHS can be found elsewhere (Government of British Columbia n.d.c).
For the purposes of this project, BCEHS records shared with Statistics Canada were used to identify opioid overdoses using the same protocol as MacDougall et al. 2019 to define these: [Naloxone_Used] = ‘9042’.
PharmaNet
PharmaNet is a province-wide network that links all pharmacies in British Columbia to a central set of real-time data systems (British Columbia Ministry of Health 2011e, 2011f). It supports drug dispensing, monitoring and claims processing. This system captures all prescribed drugs and medical supplies dispensed from community pharmacies in British Columbia. It does not cover prescriptions dispensed in inpatient settings. It does include dispensations to federally insured clients. These data provide information about all dispensed drugs, including unique drug identification numbers, the quantity of each drug dispensed (days-supply), date of dispensation and directions for drug use. PharmaNet also records patient identifiers (i.e., PHN) and demographic information including address, name, date of birth and reported drug allergies. These data exclude antiretroviral medications dispensed from the Centre of Excellence for HIV/AIDS at St. Paul’s Hospital and medications purchased without a prescription (i.e., over-the-counter medications).
BC Coroners Service database
The BC Coroners Service database is an extract from the BC Coroners Service Database (Tosca) for confirmed illicit drug overdose deaths in the province of British Columbia between January 1, 2007, and December 31, 2017, that was provided to Statistics Canada. The file contained name variables, dates of birth and death, PHNs, and geographic information (i.e., postal code, census subdivision, city of residence, census metropolitan area and province of residence).
2.2 Statistics Canada data holdings
Health services
Health administrative databases used in the development of the BC-OOAF include the Discharge Abstract Database (DAD) and the National Ambulatory Care Reporting System (NACRS). The DAD and NACRS are shared with Statistics Canada by the Canadian Institute for Health Information (CIHI) on an annual basis. The DAD is a census of all acute care discharges from public hospitals in Canada (excluding Quebec) and contains demographic, administrative and clinical data for approximately 3 million discharges annually. More information on the DAD can be found on the CIHI website (Canadian Institute for Health Information 2011a, 2011b).
The NACRS contains data about visits to health care facilities for ambulatory care, including community-based services, day surgery procedures, emergency department visits, diagnostic imaging and selected clinic visits (e.g., oncology care). Information provided in NACRS includes patient demographics, clinical information (e.g., diagnoses, surgical interventions), and administrative, financial and service-specific data. More information on the NACRS can be found on the CIHI website (Canadian Institute for Health information 2011c). Considerable variation across jurisdictions and by year for numbers of ambulatory facilities reporting to the NACRS that impact coverage were documented by the CIHI (Canadian Institute for Health information 2015, 2017).
Canadian Vital Statistics Death Database
The Canadian Vital Statistics Death Database contains information on all deaths in Canada, with data obtained annually from all provincial and territorial vital statistics registries (Statistics Canada 2018a). The death records contain patient characteristics, as well as date and major cause of death.
Justice
Information on justice contacts was derived from the following databases:
The incident-based Uniform Crime Reporting Survey (UCR) collects detailed information on criminal incidents that have been reported to and substantiated by Canadian police services. This information includes characteristics pertaining to criminal incidents, victims and accused persons.
The Integrated Criminal Court Survey (ICCS) collects information on adult and youth court cases involving the Criminal Code and other federal statute offences. The primary unit of analysis in the ICCS is a case, which is defined as one or more charges against an accused person or company that were processed by the courts at the same time and received a final decision. A case combines all charges against the same person having one or more key overlapping dates (e.g., date of offence, date of initiation, date of first appearance, date of decision or date of sentencing) into a single case.
The Integrated Correctional Services Survey (ICSS) collects microdata on adults and youth under the responsibility of the federal, provincial and territorial correctional systems. Data include sociodemographic characteristics (e.g., age, sex, Indigenous identity), as well as information pertaining to correctional supervision, including admissions and releases by legal hold status (e.g., remand, sentenced, probation). More information on the justice data is provided elsewhere (Statistics Canada n.d.a).
Employment and social assistance
Information on employment income and social assistance was derived from the following administrative data sources:
The Longitudinal Worker File (LWF) is an administrative database that integrates tax file data provided by the Canada Revenue Agency (CRA) to Statistics Canada, including the T1 Personal Master File, T4 Statement of Remuneration Paid files, and T4E Statements of Employment Insurance and Other Benefits, as well as the Record of Employment file provided by Employment and Social Development Canada and business-level data from Statistics Canada’s Longitudinal Employment Analysis Program (LEAP). Information derived from the LWF data for the BC-OOAF includes T4 employment earnings, employment history and industry of employment. More information on the LWF and LEAP files is provided elsewhere (Statistics Canada n.d.b, n.d.c).
Social assistance income data come from the CRA’s T5007 Statement of Benefits file. Social assistance refers to “payments made to beneficiaries or third parties based on a means, needs, or income test and include payments for food, clothing, and shelter requirements” (Canada Revenue Agency n.d.). The T5007 form contains basic personal information about the recipient (e.g., name, address and social insurance number [SIN]), the name and address of the payer, and the total amount of social assistance income received in a year. Information on the type of support is not provided. The T5007 form is issued for people who received $500 or more in assistance payments in a tax year. The measure of social assistance used excludes disability assistance payments from the Government of British Columbia and Canada Pension Plan disability benefits.
Immigration
The Longitudinal Immigration Database (IMDB) is a national database that represents a census of immigrants and temporary residents who have arrived in Canada since 1980. The IMDB combines linked administrative immigration and tax data files obtained from Immigration, Refugees and Citizenship Canada and the CRA, respectively. Information available in the IMDB includes date of entry, country of birth, admission category (i.e., economic class, family class, refugee) and principal applicant status. More information on the IMDB is provided elsewhere (Statistics Canada n.d.d).
3 Record linkage
3.1 Methodology
The linkage was conducted at Statistics Canada using the Social Data Linkage Environment (SDLE). The SDLE is a highly secure linkage environment established to support the creation of linked population data files for analysis through linkage to a central depository called the Derived Record Depository (DRD)—a dynamic relational database containing only basic personal identifiers (Statistics Canada n.d.e). Data are linked to the DRD using G-Link, a SAS-based generalized record linkage software that supports deterministic and probabilistic linkage and that was developed at Statistics Canada (Statistics Canada n.d.f).
The linkages were conducted using a range of methods (i.e., probabilistic, hierarchical deterministic and deterministic) based on the availability of unique identifiers in each database. Table 1 summarizes the details of the linkage methodology for each database. The databases were linked using a multi-phase approach, with an aim to append a derived key that uniquely identifies a given individual (i.e., DRD_ID) to all database sources to facilitate the full integration of the data at the person level:
- Phase 1: Linkage of the Client Registry—An internal linkage of the Client Registry was conducted to identify duplicate entities. Next, the unduplicated Client Registry was linked to the DRD in the SDLE to associate the unique identifier (DRD_ID) used within Statistics Canada to each entity in the Client Registry to facilitate linkages to other databases.
- Phase 2: Linkage of British Columbia’s administrative health databases (i.e., MSP, BCEHS, PharmaNet)—British Columbia’s three administrative health databases were linked to the DRD in the SDLE to associate each record in the data files with a DRD_ID.
- Phase 3: Linkage of other administrative health databases held at Statistics Canada—The CVSD, DAD and the NACRS were linked to the DRD directly to obtain the DRD_ID. Preliminary validation results revealed lower-than-expected linkage rates for opioid-related records in the DAD and NACRS. As a result, the remaining unlinked DAD and NACRS records were linked to the Client Register and subsequently re-linked to the DRD, which resulted in improved linkage rates.
- Phase 4: Linkage of Statistics Canada’s justice databases—Policing (UCR) and corrections (ICSS) data were linked deterministically to the Client Registry using a combination of Soundex, sex and date of birth. The courts (ICCS) data were linked to the DRD directly in the SDLE. All linkages resulted in appending a DRD_ID to the justice records.
- Phase 5: Linkage of employment and social assistance records—The T4 and T5007 files were linked using SINs derived from the DRD to associate a DRD_ID with each SIN.
- Phase 6: Linkage of the immigration data—The IMDB data had been linked previously to the DRD at Statistics Canada (Statistics Canada 2018c).
| Data files—Health files | Linkage variables | Record linkage methodology |
|---|---|---|
| British Columbia Ministry of Health Client RegistryTable 1 Note 1 | Date of birth, names (given names and surnames), PHN, sex, surnames of parents, postal code, city, CMA, CSD, province | Probabilistic linkage to the DRD |
| British Columbia Coroners ServiceTable 1 Note 2 | Dates of birth and death, names (given names and surnames), surnames of parents, PHN, postal code, CSD, city, CMA, province | Probabilistic linkage to the DRD |
| British Columbia Emergency Health ServicesTable 1 Note 3 | Date of birth, names (given names and surnames), PHN, sex, surnames of parents, postal code, city, CMA, CSD, province | Probabilistic linkage to the DRD |
| Medical Services Plan (MSP)Table 1 Note 4 | First step: Date of birth, names (given names and surnames), PHN, sex, surnames of parents, postal code, city, CMA, CSD, province. Second step: Deterministic process of linkage to physician claim file used CLNT_LABEL, MRG_CLIENT_LABEL |
Probabilistic linkage of BC Ministry of Health Client Registry system roster to DRD, then deterministic linkage to MSP payment information for provider services |
| PharmaNetTable 1 Note 5 | Date of birth, names (given names and surnames), Personal Health Number (PHN), sex, surnames of parents, postal code, city, CMA, CSD, and province. Second step: Deterministic process of linkage to physcian claim file used CLNT_LABEL, MRG_CLIENT_LABEL | Probabilistic linkage of BC Ministry of Health Client Registry system roster to DRD, then deterministic linkage to MSP payment information for provider services |
| Discharge Abstract Database (2014 to 2016)Table 1 Note 6 | Date of birth, sex, postal code, PHN | Deterministic linkage to the DRD; if nonlinking, then linkage to DRD via the BC Ministry of Health system roster |
| National Ambulatory Care Reporting System (2014 to 2016)Table 1 Note 7 | Date of birth, sex, postal code, PHN | Deterministic linkage to the DRD; if nonlinking, then linkage to DRD via the BC Ministry of Health system roster |
|
||
3.2 Protecting privacy
The linkage was approved by Statistics Canada’s Strategic Management Committee (Statistics Canada 2018b) and the use of the data is governed by the Directive on Microdata Linkage (Statistics Canada n.d.g). Participants’ privacy during record linkage and the use of the linked files are ensured by Statistics Canada. Access to the unique identifying information (e.g., names) was limited to employees directly involved in linking the databases, and these individuals did not access the complete data files with information on person-level characteristics. After record linkage, all identifying information was removed and a de-identified analytical file was created for subsequent use and analysis.
4 Opioid overdose case definition
Case ascertainment of fatal and non-fatal opioid overdose events was conducted based on the protocol developed by MacDougall et al. (2019). Administrative health records related to coroner-confirmed fatal and non-fatal opioid events were identified based on data and the criteria defined below. Select databases used by the Government of British Columbia to define the official British Columbia Provincial Overdose Cohort that were not available to Statistics Canada included the BC Drug and Poison Information Centre information and case-based reporting by emergency departments used in three of British Columbia’s five health authorities.
- BCEHS: Ambulance-attended events in which patients were coded as receiving naloxone by paramedics.
- BC Coroners Service: Deaths attributable to overdoses involving illicit drugs (e.g., heroin, cocaine, MDMA, methamphetamine), medications that were not prescribed to the deceased, combinations of the above (except prescribed medications) and those overdoses where the origin of the drug is not known. Only closed, confirmed cases were included.
- MSP: Records with the following International Classification of Diseases, Version 9 (ICD-9) codes in the primary diagnostic field: 965.0 – poisoning by opiates and related narcotics, E850.0 – accidental poisoning by opiates and related narcotics.
- DAD: Records with the following ICD-10 codes in the primary discharge diagnosis field: T40.0 – poisoning by opium, T40.1 – poisoning by heroin, T40.2 – poisoning by other opioids, T40.3 – poisoning by methadone, T40.4 – poisoning by other synthetic narcotics, T40.6 – poisoning by other and unspecified narcotics.
- NACRS: Emergency department records with the following ICD-10 codes in the emergency department discharge diagnosis field: T40.1 – poisoning by heroin, T40.6 – poisoning by other and unspecified narcotics.
All overdose records occurring between January 1, 2014, and December 31, 2016, were identified in the separate databases. Records were then organized at the individual level using the unique DRD_ID. Multiple records from different data sources could exist for a single opioid overdose event (e.g., ambulance attendance followed by an emergency room visit are related to the same overdose event). To avoid double counting, all records for a given person occurring within a 24-hour period were considered to be associated with a single opioid overdose event.
5 Results
5.1 Linkage results
Results of the linkage of health databases used to identify fatal and non-fatal opioid overdoses are provided in Table 2. Overall, the linkage rates of each database to the DRD exceeded 97% for the MSP, DAD, NACRS and BC Coroners Service databases. Linkage rates were between 88% and 92% for the BCEHS data. Linkage rates were similar for records related specifically to opioid overdose events indicating no signs of bias. More information on the methodological details and results of the record linkage can be provided upon request (Statistics Canada 2018d, 2018e, 2018f, 2018g, 2018h).
| Data source and year | Number of records in source | Percent linked to DRD | Number of OD records | Number of OD records linked to DRD | Percent of OD records linked to DRD |
|---|---|---|---|---|---|
| BCEHSTable 2 Note 1 | |||||
| 2014 | 25,985 | 91.9 | 2,720 | 2,458 | 90.4 |
| 2015 | 29,424 | 91.5 | 3,922 | 3,548 | 90.5 |
| 2016 | 38,121 | 87.8 | 6,601 | 5,881 | 89.1 |
| Total | 93,530 | 90.1 | 13,243 | 11,887 | 89.5 |
| MSPTable 2 Note 2 | |||||
| 2014 | 91,878,799 | 99.8 | 616 | 614 | 99.7 |
| 2015 | 94,572,650 | 99.7 | 1,145 | 1,138 | 99.4 |
| 2016 | 97,586,790 | 99.7 | 3,134 | 3,104 | 99.0 |
| Total | 284,038,239 | 99.7 | 4,895 | 4,856 | 99.2 |
| DADTable 2 Note 3 | |||||
| 2014 | 1,074,220 | 99.3 | 549 | 542 | 98.4 |
| 2015 | 889,342 | 99.3 | 545 | 536 | 98.3 |
| 2016 | 677,917 | 99.2 | 461 | 452 | 98.0 |
| Total | 2,641,479 | 99.3 | 1,555 | 1,530 | 98.4 |
| NACRSTable 2 Note 4 | |||||
| 2014 | 1,833,673 | 98.2 | 1,441 | 1,376 | 95.5 |
| 2015 | 1,570,180 | 98.1 | 2,143 | 2,034 | 94.9 |
| 2016 | 1,188,981 | 97.8 | 3,011 | 2,888 | 95.9 |
| Total | 4,592,834 | 98.1 | 6,595 | 6,298 | 95.5 |
| BC Coroners ServiceTable 2 Note 5 | |||||
| 2014 | 362 | 99.2 | 362 | 359 | 99.2 |
| 2015 | 490 | 97.1 | 490 | 476 | 97.1 |
| 2016 | 658 | 97.3 | 658 | 640 | 97.3 |
| Total | 1,510 | 97.7 | 1,510 | 1,475 | 97.7 |
|
|||||
5.2 Cohort results
Results of the cohort creation are presented in Table 3 and Figure 1. Overall, 13,318 people experienced 19,125 opioid overdose events between January 1, 2014, and December 31, 2016. Only 13.8% of first overdose events were identified in more than one database. The BCEHS (ambulance) data were the greatest contributor to identifying opioid cases (see Figure 1).
Description for Figure 1
The title of Figure 1 is “Distribution of individuals in the Statistics Canada British Columbia Opioid Overdose Analytical File, by data source used to identify their index overdose, January 2014 to December 2016.”
There is a note marker in the title associated with the index overdose which states: “The total Statistics Canada British Columbia Opioid Overdose Analytical File included N=13,318 individuals, each having one index (first) opioid overdose within the study reference period of January 1, 2014, to December 31, 2016. For these individuals, a total of 19,125 overdose events were recorded over the reference period.”
The background of the figure is white.
Figure 1 shows a series of five spheres (oval shapes) of different sizes and colours. Each of these five spheres represents the health data source within which each cohort member’s first opioid overdose (the index overdose) was recorded during this study’s reference period. The person’s first opioid overdose is reported in only one of each of these five data sources that are represented in this figure as one of the spheres.
The spheres are placed in a row. The higher the proportion of cases (people) who had their first overdose recorded in a given data source (and thus entered the cohort from that specific data source), the larger the sphere. The placement of the spheres from left side to the right side can be thought of as the continuum of health services—from paramedic to emergency department, to consult with a mental health specialist physician and then an overnight stay in an acute care hospital.
Each sphere slightly overlaps the sphere to its right or to its left. This overlap is depicted by oblong shapes in a darker colour, and represents opioid overdose cases recorded in two adjacent data sources (types of services) within the same 24 hours. The spheres and overlaps are described from left to right.
Two dark oblong shapes partly overlap the left side of the first sphere, the Coroners’ data source. The proportions relating to each of these oblong shapes are shown in black colour and placed one above the other, outside and to the left of the small oblong shapes. The top proportion is “.01% NACRS,” representing the proportion of the cohort who had a hospital Emergency Department visit (NACRS) recorded and their death recorded by the Coroner. The bottom proportion is “.01% DAD,” representing an inpatient hospital visit record of their first opioid overdose (i.e., index overdose) and death record by the BC Coroners Service.
The first sphere represents cases found in the BC Coroners Service data (i.e., deaths from opioid overdose). This data source sphere is the second smallest of all five spheres and represents the proportion of people who had overdoses and entered the cohort form this data source. The proportion “8%” appears in black in the centre of this sphere. Below this first sphere, “BC Coroners Service (9%)” appears in black. The sphere is the second smallest of all five spheres since the second smallest proportion of the cohort (9%) entered the cohort from this Coroner data source.
The right side of the first sphere slightly overlaps the left side of the second sphere (representing the Emergency Health Service data source [ambulance paramedic]). The overlap is shown by the third dark oblong shape. “1.1%” appears in white in this third oblong shape. It represents the proportion of people who had a first opioid overdose and were attended by a paramedic (recorded in the Emergency Health Service data source), but died of their opioid overdose the same day. Their death was recorded in the Coroner’s data source, so their first opioid overdose entered the cohort via the coroner’s records data source.
The second sphere (at the right of the first sphere), the Emergency Health Services (i.e., paramedic, ambulance records), contains the cases for people whose first opioid overdose was recorded by paramedics in the Emergency Health Services data source (ambulance). In the centre of this second sphere, “42%” appears in black. This is the proportion of the cohort who entered the cohort by having their first opioid overdose recorded only in this data source. This second sphere is the largest of all five spheres, as 42% of people in the cohort had their first opioid overdose recorded in the Emergency Health Services data source only. Below this second sphere, “Ambulance (53%)” appears in black to indicate the proportion of people in the cohort whose first overdose was recorded in the Emergency Health Services data source (ambulance, paramedic) alone, or in the Emergency Health Services data source and also in one or more of the other four data sources as occurring on the same date.
The right side of this second sphere (the Emergency Health Services [Ambulance] sphere), slightly overlaps the third sphere (the hospital Emergency Department visit cases from the National Ambulatory Care Reporting Services data source [the NACRS]). This fourth darker shaded oblong shape represents people whose first opioid overdose was recorded in the Emergency Health Services (ambulance, paramedic) data source and also recorded in the hospital Emergency Department data source on the same day. In this oblong shape, “7%” is written in white, showing the proportion of people in the cohort that entered the cohort this way.
The third sphere in the row represent cases (people) who visited the hospital Emergency Department using records from the National Ambulatory Care Services data source (the NACRS). In the centre of this sphere, “19%” appears in black, indicating the proportion of people in the cohort who entered by the hospital Emergency Department (NACRS) data source only. This third sphere is therefore smaller than the Ambulance sphere (the second sphere), but larger than the Coroner’s sphere (the first sphere).
Below the third sphere, “NACRS (29%)” appears in black. This shows the proportion of people in the cohort who had their first opioid overdose recorded in this NACRS data source only, or in this source as well as in one or more of the other four data sources.
A portion of the right side of this third sphere overlaps with the next sphere, which represents the Medical Services Pan (MSP) data source. This fifth darker coloured oblong shape, between the NACRS and the MSP sphere, shows “4%” in white in its centre. This is the proportion of the cohort whose first opioid overdose was recorded in the NACRS data source and the MSP data source on the same day.
The MSP claims sphere (the fourth sphere) is the next medium-sized sphere, and it appears to the right of the hospital Emergency Department NACRS sphere. This fourth sphere represents people whose first opioid overdose was recorded in the MSP claims for physician database. These physician claims were recorded in this source as a diagnosis of an opioid overdose. This fourth sphere is slightly smaller than the sphere for the NACRS but larger than the Coroners sphere. “12%” appears in black the centre of this fourth sphere. This indicates the proportion of the cohort whose first overdose was recorded only in this MSP data source. Below this MSP claims sphere, “MSP (16%)” appears in black. This is the proportion of the cohort whose index overdose was recorded in the MSP data source alone, or in combination with one or more of the other four data sources.
A portion of the right side of this fourth sphere overlaps with the fifth sphere, which represents the hospital Discharge Abstract Database (DAD) data source. This sixth darker coloured oblong shape, between the MSP and the DAD sphere, shows “.2%” in white in its centre. This is the proportion of the cohort whose first opioid overdose was recorded in the MSP data source and the DAD data source.
The fifth and last sphere appears at the rightmost side of the row of five data source spheres. This fifth data source sphere represents the proportion of people whose first opioid overdose was identified in the acute care hospital records, the hospital Discharge Abstract Database (DAD). This sphere is the smallest of all five data source spheres. “6%” appears in black in the centre of this sphere. This indicates the proportion of the cohort whose first opioid overdose was recorded in this data source only. Below this DAD sphere, “DAD (8%)” appears in black, indicating the proportion of the cohort whose first opioid overdose was recorded in the DAD data source alone, or in combination with one or more of the other data sources depicted.
Above the row of five spheres, starting at the outside of the largest sphere (Emergency Health Services Data source [Ambulance]), there is a black horizontal bar that spans above the NACRS sphere and bends down to touch the MSP sphere. This points to the information that reports the proportion of people in the cohort whose first opioid overdose was recorded for the same day in both the Emergency Health Service (ambulance) data source and the MSP (physicians’ claims) data source. “Ambulance–MSP 7%” appear in black font above the bar, indicating the proportion of the study cohort who entered the cohort this way.
Another black horizontal bar spans above the DAD sphere and bends downwards to touch the DAD sphere. It represents the people whose first opioid overdose was recorded on the same day in both the Emergency Health Services database (ambulance) and the DAD database. “Ambulance–DAD 2%” appear in black above this portion of the bar, indicating the proportion of the cohort whose first overdose was recorded in the Emergency Health Services and DAD databases on the same day, and thus entered the cohort this way.
The notes for Figure 1 are as follows:
Notes: MSP: Medical Services Plan, DAD: Discharge Abstract Database, NACRS: National Ambulatory Care Reporting System. Percentages inside the ovals indicate the percentage of the cohort that has only the source (ovals) as the source that entered the person into the cohort. Percentages under the ovals indicate the percentage of the cohort that has the source (oval) as either only the source or in combination with other sources for first overdose. At the overlapping edge of each oval is the percentage of people who have two overlapping sources only.
The source for Figure 1 is as follows:
Source: Statistics Canada British Columbia Opioid Overdose Analytical File, Statistics Canada.
Overall, 1,475 events were fatal as identified in the BC Coroners Service database, representing 11.1% of people and 7.7% of events. Overall, males represented 65% of persons who experienced an opioid overdose and 79% of fatal events. Approximately 47% of individuals were aged between 20 and 39 at the time of their first opioid overdose in the observation period. The majority of people (78%) experienced only one event, rather than several during the study period (see Table 3).
| Total overdose cohort | Non-fatal overdoses | Fatal overdoses | ||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Number of individuals | Percent | Number of events | Percent | Number of individuals | Percent | Number of events | Percent | Number of individuals | Percent | Number of events | Percent | |
| Total for British Columbia | 13,318 | 100 | 19,125 | 100 | 11,843 | 100 | 17,249 | 100 | 1,475 | 100 | 1,876 | 100 |
| Sex | ||||||||||||
| Male | 8,682 | 65 | 12,714 | 67 | 7,515 | 64 | 11,254 | 65 | 1,167 | 79 | 1,460 | 78 |
| female | 4,626 | 35 | 6,388 | 33 | 4,318 | 36 | 5,972 | 35 | 308 | 21 | 416 | 22 |
| Not available | 10 | 0.1 | 23 | 0.1 | 10 | 0.1 | 23 | 0.1 | 0 | 0 | 0 | 0 |
| Age group | ||||||||||||
| 0 to 14 years | 71 | 0.5 | 85 | 0 | 71 | 1 | 85 | 0 | 0 | 0 | 0 | 0 |
| 15 to 19 years | 511 | 4 | 675 | 4 | 473 | 4 | 624 | 4 | 38 | 3 | 51 | 3 |
| 20 to 29 years | 3,092 | 23 | 4,773 | 25 | 2,778 | 23 | 4,330 | 25 | 314 | 21 | 443 | 24 |
| 30 to 39 years | 3,180 | 24 | 4,755 | 25 | 2,780 | 23 | 4,258 | 25 | 400 | 27 | 497 | 26 |
| 40 to 49 years | 2,490 | 19 | 3,634 | 19 | 2,154 | 18 | 3,224 | 19 | 336 | 23 | 410 | 22 |
| 50 to 59 years | 2,196 | 16 | 3,006 | 16 | 1,885 | 16 | 2,621 | 15 | 311 | 21 | 385 | 21 |
| 60 years and older | 1,778 | 13 | 2,197 | 11 | 1,702 | 14 | 2,107 | 12 | 76 | 5 | 90 | 5 |
| Number of overdose episodes per person | ||||||||||||
| 1 | 10,389 | 78 | 10,389 | 54 | 9,158 | 77 | 9,158 | 53 | 1,231 | 83 | 1,231 | 66 |
| 2 | 1,706 | 13 | 3,412 | 18 | 1,544 | 13 | 3,088 | 18 | 162 | 11 | 324 | 17 |
| 3 | 587 | 4 | 1,761 | 9 | 540 | 5 | 1,620 | 9 | 47 | 3 | 141 | 8 |
| 4 | 270 | 2 | 1,080 | 6 | 253 | 2 | 1,012 | 6 | 17 | 1 | 68 | 4 |
| 5 | 143 | 1 | 715 | 4 | 134 | 1 | 670 | 4 | 9 | 1 | 45 | 2 |
| 6 or moreTable 3 Note 1 | 223 | 2 | 1,768 | 9 | 214 | 2 | 1,701 | 10 | 9 | 1 | 67 | 4 |
|
||||||||||||
5.3 Linkage of ancillary files to the cohort
Results of the linkage of individuals experiencing an opioid overdose to the ancillary health, tax and justice files are provided in Tables 4, 5 and 6, respectively. Over 80% of people who experienced an opioid overdose were linked to at least one MSP record in each year between 2011 and 2016, resulting in over 3.65 million MSP records. Between 20% and 33% of people who experienced an opioid overdose were linked to a DAD record in each year between 2011 and 2016, for a total of 46,100 hospital discharges. Between 11.7% and 62.5% of people who experienced an opioid overdose were linked to a NACRS record over this same time period, for a total of 31,400 records. Approximately 80% of BC-OOAF members per year were linked to a PharmaNet record, representing over 9.46 million prescriptions (see Table 4).
| Data source and year | Number of records in raw source | Number of records linked to DRD | Number of records linked to OD cohort | Number of OD cohort members linked to given source year | Percent of overall OD cohort members linked to given source, given year |
|---|---|---|---|---|---|
| MSPTable 4 Note 1 | |||||
| 2011 | 84,787,868 | 84,588,424 | 497,379 | 10,967 | 82.3 |
| 2012 | 87,466,435 | 87,265,260 | 546,613 | 11,117 | 83.5 |
| 2013 | 90,140,985 | 89,918,957 | 592,146 | 11,315 | 85.0 |
| 2014 | 91,878,799 | 91,638,249 | 635,613 | 11,609 | 87.2 |
| 2015 | 94,572,650 | 94,316,239 | 676,754 | 11,496 | 86.3 |
| 2016 | 97,586,790 | 97,297,229 | 697,490 | 11,033 | 82.8 |
| Total | 546,433,527 | 545,024,358 | 3,645,995 | 12,869 | 96.6 |
| DADTable 4 Note 2 | |||||
| 2011 | 817,199 | 812,294 | 5,160 | 2,794 | 21.0 |
| 2012 | 830,450 | 825,408 | 5,710 | 2,898 | 21.8 |
| 2013 | 849,868 | 844,467 | 6,861 | 3,348 | 25.1 |
| 2014 | 853,938 | 847,853 | 9,043 | 4,256 | 32.0 |
| 2015 | 875,018 | 868,801 | 9,554 | 4,382 | 32.9 |
| 2016 | 895,208 | 887,868 | 9,822 | 4,444 | 33.4 |
| Total | 5,121,681 | 5,086,691 | 46,150 | 9,533 | 71.6 |
| NACRSTable 4 Note 3 | |||||
| 2011 | 209,196 | 208,362 | 5,139 | 1,554 | 11.7 |
| 2012 | 732,961 | 731,058 | 13,214 | 4,024 | 30.2 |
| 2013 | 887,951 | 885,727 | 17,715 | 5,144 | 38.6 |
| 2014 | 1,833,673 | 1,800,976 | 42,847 | 8,329 | 62.5 |
| 2015 | 1,570,180 | 1,540,013 | 40,405 | 8,305 | 62.4 |
| 2016 | 1,188,981 | 1,163,057 | 31,391 | 7,343 | 55.1 |
| Total | 6,422,942 | 6,329,193 | 150,711 | 11,768 | 88.4 |
| PharmaNetTable 4 Note 4 | |||||
| 2011 | 62,375,180 | 62,007,390 | 1,192,015 | 10,359 | 77.8 |
| 2012 | 65,171,004 | 64,793,898 | 1,396,992 | 10,576 | 79.4 |
| 2013 | 67,927,746 | 67,547,444 | 1,596,150 | 10,842 | 81.4 |
| 2014 | 69,564,838 | 69,186,524 | 1,713,849 | 11,114 | 83.4 |
| 2015 | 71,468,190 | 71,057,538 | 1,781,392 | 10,950 | 82.2 |
| 2016 | 73,911,802 | 73,447,722 | 1,781,201 | 10,373 | 77.9 |
| Total | 410,418,760 | 408,040,516 | 9,461,599 | 12,937 | 97.1 |
|
|||||
Individuals who experienced an opioid overdose were eligible to link to annual employment (T4) and social assistance (T5007) records for a maximum of five years preceding their overdose if there was a valid SIN available for them. For each calendar year, 99% of people had a valid SIN. Among them, over 40% linked to the T4 and/or T5007 files.
| Data source and year | Cohort members eligible to link | Cohort members with a valid SIN | Number of OD cohort members linked to given source year | Percent of overall OD cohort linked to given source, given year |
|---|---|---|---|---|
| T4 forms (from the Longitudinal Worker File) | ||||
| 2009 | 3,329 | 3,294 | 1,432 | 43.0 |
| 2010 | 7,379 | 7,311 | 3,100 | 42.0 |
| 2011 | 13,318 | 13,184 | 5,771 | 43.3 |
| 2012 | 13,318 | 13,184 | 5,812 | 43.6 |
| 2013 | 13,318 | 13,184 | 5,798 | 43.5 |
| 2014 | 13,318 | 13,184 | 5,625 | 42.2 |
| 2015 | 9,989 | 9,890 | 4,102 | 41.1 |
| 2016 | 5,939 | 5,873 | 2,333 | 39.3 |
| T5007 forms | ||||
| 2009 | 3,329 | 3,294 | 1,251 | 37.6 |
| 2010 | 7,379 | 7,311 | 2,943 | 39.9 |
| 2011 | 13,318 | 13,184 | 5,388 | 40.5 |
| 2012 | 13,318 | 13,184 | 5,515 | 41.4 |
| 2013 | 13,318 | 13,184 | 5,799 | 43.5 |
| 2014 | 13,318 | 13,184 | 6,006 | 45.1 |
| 2015 | 9,989 | 9,890 | 4,756 | 47.6 |
| 2016 | 5,939 | 5,873 | 3,061 | 51.5 |
|
Notes: SIN: social Insurance Number; OD: overdose. Sources: Longitudinal Worker File, Longitudinal Employment Analysis Program, Statistics Canada; Statistics Canada British Columbia Opioid Overdose Analytical File, Statistics Canada; Social Asssistance, Canada Revenue Agency. |
||||
Individuals who experienced opioid overdoses were linked to justice data for the two years prior to their index overdose event: results are provided in Table 6.
Approximately 7% of cohort members linked to the IMDB and 23% linked to the Canadian Vital Statistics Death Database.
| Data source and year | Cohort members eligible to link | Number of records linked to OD cohort in given source year | Number of OD cohort members linked to given source year | Percent overall of OD cohort members linked to given source, given year |
|---|---|---|---|---|
| UCR microdata, police contacts in the 24 months prior to the first overdose event, by year of charge date, 2012 to 2016Table 6 Note 1 | ||||
| 2012 | 3,329 | 1,213 | 509 | 15.3 |
| 2013 | 7,379 | 4,277 | 1,493 | 20.2 |
| 2014 | 13,318 | 7,330 | 2,637 | 19.8 |
| 2015 | 9,989 | 7,362 | 2,512 | 25.1 |
| 2016 | 5,939 | 3,400 | 1,326 | 22.3 |
| UCR microdata, all police contacts, by year of charge date, 2010 to 2016Table 6 Note 2 | ||||
| 2010 | 13,318 | 7,947 | 2,999 | 22.5 |
| 2011 | 13,318 | 8,227 | 3,071 | 23.1 |
| 2012 | 13,318 | 8,647 | 3,082 | 23.1 |
| 2013 | 13,318 | 9,781 | 3,344 | 25.1 |
| 2014 | 13,318 | 11,274 | 3,475 | 26.1 |
| 2015 | 13,318 | 12,151 | 3,601 | 27.0 |
| 2016 | 13,318 | 13,019 | 3,598 | 27.0 |
Sources: Uniform Crime Reporting Survey, Statistics Canada British Columbia Opioid Overdose Analytical File, Statistics Canada. |
||||
6 Discussion
The collaboration between partners in British Columbia and Statistics Canada has led to the development of the BC-OOAF—a unique dataset providing information on individuals experiencing opioid overdoses in the province of British Columbia. The creation of the BC-OOAF cohort and its analysis has additionally integrated information from federal data sources using the secure platform and linkage methodologies developed at Statistics Canada. Information on employment, income, social assistance, health care use, immigration and contacts with police have been linked successfully to a cohort of individuals who experienced fatal and non-fatal overdoses in British Columbia between 2014 and 2016.
As indicated, the identification of opioid overdose events was conducted according to the protocol developed by MacDougall et al. (2019) from the BCCDC. Results based on the analytical data file created as part of this project (i.e., BC-OOAF) were similar to those published by the BCCDC. The total number of opioid overdose events were within a reasonable range: the BCCDC identified 10,456 people and 14,292 overdose events between January 1, 2015, and November 30, 2016, compared with 13,318 people and 19,125 overdose events between January 1, 2014, and December 31, 2016, in the BC-OOAF. In both cases, the majority of opioid overdose cases were identified via BCEHS (ambulance) data. The overall characteristics of the cohort were similar, with a greater representation of opioid overdoses among males, particularly among fatal cases; the proportion was 79% and 80% in the Statistics Canada BC-OOAF and the BCCDC’s British Columbia provincial overdose cohort, respectively. Age distributions among fatal and non-fatal overdoses were also similar. These similarities provide evidence of the validity of the BC-OOAF created by Statistics Canada.
7 Limitations
Censoring for deaths using the Canadian Vital Statistics Death Database was not conducted. Therefore, people who did not link to ancillary files would not be counted if they had died in the reference period as a result of non-overdose causes. Although the overdose case ascertainment methods used sources covering much of the spectrum of healthcare services within two years of observation, overdoses are likely undercounted since some overdoses may not have had any medical health services contact. By one estimate for a region of British Columbia within 2015-2017, this was the case for 44% of overdoses that had received naloxone via bystander(s). (Karamouzian, Kuo, Crabtree, Buxton 2019) For the present study, while the index overdose event was the first observed within the observation period, it may not have been peoples’ first overdose. Furthermore, the analysis did not include a comparative group of people, among whom no overdose was detected. Work by others has shown that in British Columbia First Nations people are over-represented among people experiencing overdoses and overdose deaths. (First Nations Health Authority 2017. Overdose data and First Nations in BC: Preliminary findings) Information to identify Indigenous peoples or other ethno-cultural groups as a demographic characteristic for members of the BC-OOAF was not available for this study. Housing circumstance is another important characteristic relating to health generally that was not available for this study, yet could be developed in future. (Statistics Canada n.d.h) Finally, since this study’s observation period, there may have been changes to overdose circumstances in the overdose crisis.
8 Conclusion
Data integration provides tremendous opportunities to bring together existing health and non-health databases to better understand complex sub-populations. The BC-OOAF developed by Statistics Canada in collaboration with Government of British Columbia stakeholders can be used to further understand the heterogeneity of individuals experiencing opioid overdose events, which can hopefully lead to more targeted policy and program responses to more effectively reduce the burden of this crisis. The project serves as model that could be replicated in other jurisdictions across Canada and for other (illicit) drug-related events.
References
British Columbia Ministry of Health [creator] (2011): Consolidation File (MSP Registration & Premium Billing). BC Ministry of Health [publisher]. Data Extract. MOH (2011). (2011a).
British Columbia Ministry of Health [creator] (2011): Medical Services Plan (MSP) Payment Information File. British Columbia Ministry of Health [publisher]. Data Extract. MOH (2011). (2011b).
British Columbia Ministry of Health [creator] (2011): Mental Health. British Columbia Ministry of Health [publisher]. Data Extract. MOH (2011). (2011c).
British Columbia Ministry of Health [creator] (2011): Home & Community Care (Continuing Care). British Columbia Ministry of Health [publisher]. Data Extract. MOH (2011). (2011d).
BC Ministry of Health [creator] (2011): PharmaNet. British Columbia Ministry of Health [publisher]. Data Extract. MOH (2011). (2011e).
British Columbia Ministry of Health [creator] (2011): PharmaCare. British Columbia Ministry of Health [publisher]. Data Extract. MOH (2011). (2011f).
Canada Revenue Agency. n.d. Line 14500-Social assistance payments. (accessed July 23, 2019).
Canadian Institute for Health Information [creator] (2011a): Discharge Abstract Database (Hospital Separations). British Columbia Ministry of Health [publisher]. Data Extract. MOH (2011).
Canadian Institute for Health Information. (2011b). Discharge Abstract Database metadata (DAD). (accessed September 16, 2018).
Canadian Institute for Health information. (2011c). National Ambulatory Care Reporting System. (accessed September 16, 2018).
Canadian Institute for Health Information. (2015). Data Quality Documentation: National Ambulatory Care Reporting System, Current year information 2014 to 2015. Ottawa: Canadian Institute for Health Information. (accessed September 16, 2018).
Canadian Institute for Health Information. (2017). Data Quality Documentation: National Ambulatory Care Reporting System, Current year information 2016 to 2017. Ottawa: Canadian Institute for Health Information. (accessed September 16, 2018).
First Nations Health Authority. 2017. Overdose data and First Nations in BC: Preliminary findings. August. (accessed September 9, 2020).
Government of British Columbia. n.d. Introduction to the Client Registry – ECR.01. (accessed July 23, 2019).
Government of British Columbia. n.d. Medical Services Plan. (accessed July 23, 2019).
Government of British Columbia. n.d. Patient Care Records. (accessed July 23, 2019).
Government of Canada. 2018. National report: Apparent opioid-related deaths in Canada (released March 2018). (accessed February 27, 2019).
Government of Canada. 2020. Opioid-related harms in Canada (September 2020). (accessed Oct 30, 2020).
Karamouzian M, Kuo M, Crabtree A, Buxton JA. Correlates of seeking emergency medical help in the event of an overdose in British Columbia, Canada: Findings from the Take Home Naloxone program. International Journal of Drug Policy 2019; 71:157-63.
MacDougall, L., K. Smolina, M. Otterstatter, B. Zhao, M. Chong, D. Godfrey, A. Mussavi-Rizi, J. Sutherland, M. Kuo, and P. Kendall. 2019. “Development and characteristics of the Provincial Overdose Cohort in British Columbia, Canada.” PLoS One 14 (1): e0210129.
Statistics Canada. 2018a. Canadian Vital Statistics-Death Database (CVSD). February 23. (accessed August 8, 2018).
Statistics Canada. 2018b. Surrey Opioid Data Collection and Community Response Project: Linking Surrey. Opioids data with income, health, and immigration data. (008-2018). (accessed July 2019).
Statistics Canada. 2018c. Social Data Linkage Environment (SDLE) Methodology Report. Linkage between the Landing file (2014 to 2016) and the SDLE Derived Record Depository (Version 11). January 2018. Ottawa: Statistics Canada. Methodology report.
Statistics Canada. 2018d. Medical Services Plan – Social Data Linkage Environment (SDLE) Methodology Report. External linkage between the British Columbia Ministry of Health file (2011 to 2017) for the Surrey Opioids project and the SDLE Derived Record Depository (Version 21). December 2018. Ottawa: Statistics Canada. Methodology report.
Statistics Canada. 2018e. British Columbia Emergency Health Services – Social Data Linkage Environment (SDLE) Methodology Report. External linkage between the British Columbia Centre for Disease Control file (2014 to 2017) for the Surrey Opioids project and the SDLE Derived Record Depository. August 31, 2018. Ottawa: Statistics Canada. Methodology report.
Statistics Canada. 2018f. BC Coroner’s Database – Social Data Linkage Environment (SDLE) Methodology Report. Linkage between British Columbia Coroner’s Database (2007 to 2017) and the SDLE Derived Record Depository (Version 16). May 23, 2018. Ottawa: Statistics Canada. Methodology report.
Statistics Canada. 2018g. Discharge Abstract Database – Social Data Linkage Environment (SDLE) Methodology Report: Linkage between the DAD (Discharge Abstract Database, 1994 to 2016) and the SDLE Derived Record Depository (versions 13 and 14). May 1, 2018. Ottawa: Statistics Canada, Methodology report.
Statistics Canada. 2018h. National Ambulatory Care Reporting System – Social Data Linkage Environment (SDLE) Methodology Report. External linkage between the National Ambulatory Care Reporting System (2002 to 2016, versions 1 and 2) and the SDLE Derived Record Depository (versions 13 and 14). May 1, 2018. Ottawa: Statistics Canada, Methodology report.
Statistics Canada. n.d.a Uniform Crime Reporting Survey (UCR). (accessed July 23, 2019).
Statistics Canada. n.d.b Longitudinal Worker File. (accessed July 2019).
Statistics Canada. n.d.c Longitudinal Employment Analysis Program. (accessed July 2019).
Statistics Canada. n.d.d Longitudinal Immigration Database. (accessed September 16 2018).
Statistics Canada. n.d.e Social Data Linkage Environment. (accessed August 1, 2018).
Statistics Canada. n.d.f G-Link Version 3.0 User Guide. Ottawa: Statistics Canada. Internal document.
Statistics Canada. n.d.g Directive on Microdata Linkage (accessed July 2019).
Statistics Canada. n.d.h Pathfinder projects to lead to our modernization initiative. Canadian Housing Statistics Program. Ottawa: Statistics Canada. (accessed July 2019).
- Date modified: