Canadian Vital Death Statistics Database (CVSD) linked to Discharge Abstract Database (DAD) and National Ambulatory Care Reporting System (NACRS)

Canadian Vital Statistics Deaths Database (2008-2014) linked to Discharge Abstract Database (2004/05-2014/15), National Ambulatory Care Reporting System (2004/05-2014/15)

The objective of this project was to create a linked dataset that can be used to examine a national cohort (except Quebec) of persons who died (for any age group of interest) in relation to the characteristics and intensity of end-of-life care and to identify patient, disease and healthcare factors associated with variations in care intensity and location of death.

To achieve this objective, death records in the Canadian Vital Statistics Death Database (CVSD) from 2008 to 2014 were linked to patient records in the Discharge Abstract Database (DAD) and the National Ambulatory Care Reporting System (NACRS) from 2004/2005 to 2014/2015. Statistics Canada does not have Quebec hospitalization data as part of its data holdings and thus hospitalizations that occurred in the province of Quebec are not included in the linked datasets.

The Data

Canadian Vital Statistics Death Database (CVSD)

The Canadian Vital Statistics Death Database (CVSD) is a census of all deaths occurring in Canada each year. Deaths are reported by the provincial and territorial Vital Statistics Registries to Statistics Canada; the information provided includes demographic and cause of death information. Cause of death information is coded using the version of the International Classification of Diseases (ICD) in effect at the time of death. Records eligible for record linkage were deaths that occurred from January 1, 2008 through December 31, 2014.

In addition to the variables from the CVSD, the file includes variables from the Postal Code Conversion File+ (PCCF+) for each linked record. The PCCF+ was generated using the CVSD variable DEA_Q150 (usual residence of the deceased: postal code).

Discharge Abstract Database

The DAD includes administrative, clinical and demographic information on hospital discharges (including in-hospital deaths, sign-outs and transfers) from all provinces and territories, except Quebec. Over time, the DAD has also been used to capture data on day surgery, long-term care, rehabilitation and other types of care. DAD data for fiscal years 2004/2005 to 2014/2015 were included in the linkage.

In the DAD, jurisdiction-specific instructions for collection of data elements evolve over time. Collection of each data element may be mandatory, mandatory if applicable, optional or not applicable. Collection requirements can vary by jurisdiction and by data year. Researchers will find the listings of DAD data elements under the heading "Data Elements" at the DAD Metadata website. The documents on the website include information on mandatory versus optional collection status for each data element by jurisdiction, which is key to understanding coverage of data elements in the DAD.

National Ambulatory Care Reporting System (NACRS)

The NACRS contains data for hospital-based and community-based ambulatory care including day surgery, outpatient and community-based clinics, and emergency departments. Client visit data is collected at time of service in participating facilities from several jurisdictions. NACRS data for fiscal years 2004/2005 to 2014/2015 were included in the linkage.

For details on the provincial data coverage, please refer to the Data Quality Documentation, available under the "Data Quality" section of the NACRS Metadata website.

In NACRS, jurisdiction-specific instructions for collection of data elements evolve over time. Similar to DAD, collection of each data element may be mandatory, mandatory if applicable, optional or not applicable. Collection requirements can vary by jurisdiction and by data year.

Researchers will find the listings of NACRS data elements under the "Data Elements" section of the NACRS Metadata website. The documents on the website include information on mandatory versus optional collection status for each data element by jurisdiction, which is key to understanding coverage of data elements in NACRS.

File structure & layout

A single cohort file was produced of all the individual CVSD records between 2008 through 2014, including those that linked and those that did not link to the DAD or NACRS. A random, unique identifier (variable name: STC_ID) was generated for each record on the CVSD. Each record with a valid postal code has additional information added from the PCCF+. Names and other personal identifiers have been removed from the file.

Separate output files were created for each year of the DAD and for each year of the NACRS. Only records that linked to the CVSD are included in these outcome files. Since the DAD and NACRS are transactional files, the STC_ID from the CVSD cohort file was included on all records to identify those individuals with multiple transactions within a dataset and across datasets. Merging the STC_ID across datasets (i.e. CVSD to DAD and NACRS) will allow the larger picture of health interventions for an individual to be analysed.

Researchers can choose to use the outcome files as event-based (each row of data represents a hospitalization) or person-based (each row of data represents an individual). In order to use a file as a person-based file, the researcher must transform the data to include all hospital information for one person as one record (one row on the data file).

The linked data should not be used to produce official mortality statistics. Official counts and rates of mortality are available on the Statistics Canada website or can be generated by requesting use of the Canadian Vital Statistics Death Database which is accessible through the RDC or by requesting a custom tabulation from Statistics Canada (statcan.hd-ds.statcan@statcan.gc.ca).

The linked data cannot be used to produce statistics related to institutions and any outputs at the institution level will be restricted as per the vetting rules.

Institution information can be used as a method to generate other information (e.g., the postal code of the institution can be used to determine distance to a care facility) but cannot be used as an outcome of interest.