Technical Guide on Demographic Estimates at Statistics Canada
Chapter 5
Deaths

Introduction

Vital statistics for Canada have been published by the Dominion Bureau of Statistics (later Statistics Canada) since the first annual report for vital statistics in 1921. Prior to 1920, vital statistics in Canada were collected in an inconsistent manner by some provincial governments. The Dominion Bureau of Statistics was created following The Statistics Act, 1918 with an order-in-council in 1919 establishing a cooperative national vital statistics system including eight provinces.

Other provinces and territories joined the cooperative national reporting system over time. By 1956, all provinces as well as Yukon and the Northwest Territories were included in regular tabulations with Nunavut being separate from the Northwest Territories following the establishment of the territory in 1999.

In order to produce demographic estimates according to the cohort-component method, the Demographic Estimates Program (DEP) subtracts deaths from the base population as part of the calculation. For the DEP, the term death excludes stillbirths.

Definitions and concepts 

Data for deaths comes from Statistics Canada’s Canadian Vital Statistics - Death database (CVSD). The registration of all deaths is mandatory according to the Vital Statistics Acts (or equivalent legislation) of each of the provinces and territories in Canada. A death is the permanent disappearance of all evidence of life at any time after a live birth has taken place. The DEP does not include stillbirths in the demographic estimates, which are births (after 20 weeks of gestation or with a weight of over 500 grams) that do not show any evidence of life, such as breathing, beating of the heart, pulsation of the umbilical cord, or definite movement of the voluntary muscles.

The central vital statistics registry in each province and territory provides Statistics Canada with the deaths registrations which are then used in the demographic estimates.

The conceptual universe, target universe, and actual (study) universes of the CVSD are deaths occurring in Canada of Canadian residents and any non-residents.

The DEP excludes deaths of visitors to Canada from the demographic estimates. Deaths of Canadian citizens that occur abroad are not included in the demographic estimates, as a death certificate from a Canadian province or territory is required to be included in the CVSD.

The DEP counts deaths according to the place of residence of the decedent. This is because the death affects the population of the place of residence and not the province in which the death occurred.

Final estimates

Death estimates for all levels (preliminary, updated and final) are produced using the CVSD. For the final estimates, the numbers of deaths are taken directly from the annual deaths data in the CVSD. This usually happens within two to three years of the reference date, at the point at which the data in the CVSD is deemed by the DEP to be sufficiently complete.

In recent years, the duration of data collection for the CVSD has been shortened, such that fewer deaths may be captured at the time of each annual release of data. The data for the CVSD is revised and updated each year and the DEP includes the revised numbers until the point at which the estimates become final. The deaths estimates may also be revised at the time of the rebasing of the population estimates after each quinquennial census, when the final postcensal population estimates become of final intercensal level. This would happen if the death data from the CVSD had been further revised after they became final for the DEP.

The DEP cannot finalize the deaths estimates until they are considered to be sufficiently complete. Following the recent changes in release timing for the CVSD, the time between the reference date and the finalization of the deaths estimates may be lengthened. Data received from the CVSD are continually being reviewed by the DEP for completeness.

Data sources

The deaths data for the final estimates comes directly from the Canadian Vital Statistics - Death database (CVSD) which is an administrative survey that collects information on deaths on an annual basis from all provincial and territorial vital statistics registries. The form for the registration of a death consists of personal information, supplied to the funeral director by an informant, and the medical certificate of cause of death, completed by the medical practitioner last in attendance, or by a coroner, if an inquest or enquiry was held. 

The central vital statistics registry in each province and territory provides data from the death registrations to Statistics Canada for inclusion in the CVSD on a continual basis. Beginning with deaths occurring in 2008, the provinces have been able to send data electronically to Statistics Canada via the National Routing System (NRS) although the effective date for this method of transmission varies by province. The territories send paper information which is converted to electronic format at Statistics Canada.

Subsequent changes to registrations due to errors, amendments or omissions are transmitted to Statistics Canada when this information becomes available. However, changes received after a cut-off date are not reflected in published data.

Error detection and imputation are conducted for the CVSD following set rules, however due to the high quality of the data received, few records are imputed annually. Since the registration of deaths is a legal requirement in each Canadian province and territory, reporting is virtually complete. Undercoverage is thought to be minimal but is being monitored. Undercoverage may occur because of late registration, but this is not common as death registration is necessary for the legal burial or disposal of a body, as well as for settling estate matters. Late or missing registrations may also occur with unidentified bodies. By long-standing practice, the date of death for unidentified remains is defined as the date of discovery.

Undercoverage is also present for deaths of serving members of the Canadian Armed Forces. Deaths of Canadians who died overseas while serving in the Armed Forces are not included in the Statistics Canada databases because they are not registered by the provinces and territories.

Overcoverage is minimal. Deaths of non-residents in Canada are registered but are excluded from most tabulations. Duplicate death registrations are identified and removed as part of the regular processing operations on each provincial and territorial subset, as well as by additional interprovincial checks. Possible duplicate registrations are verified against microfilmed registrations or optical images, or by consulting with the provinces and territories. Age and date of birth outliers are annually reviewed for capture errors.

Method

To create the final estimates, the number of deaths that occurred between the date on which the base population was established and the reference date are subtracted from the base population.

Preliminary estimates

Since the final deaths from the CVSD are not available at the time that the preliminary demographic estimates are produced (approximately three months after the reference date), the preliminary deaths are estimated by applying mortality rates based on the last year of final data to the population, according to age group and gender.

Data sources

The mortality rates by age group and gender are produced from the last year of final data, which is supplied by the CVSD for most provinces and territories. However, preliminary and updated data for Quebec, British Columbia, and Yukon are provided by the Institut de la statistique du Québec (ISQ), BC Stats, and the Yukon Bureau of Statistics (YBS), respectively. The final estimates for all provinces and territories (including Quebec, British Columbia, and Yukon) are taken from the CVSD, when they are available and complete.

Method

The following formulas are used to generate the quarterly preliminary estimates of deaths at the provincial and territorial levels.

By quarter, for each province and territory, by age at the beginning of the quarter, for each gender, the equation is:

Equation 5.1:

$$D_{(t,t+i)}^{a,g}=\left(\frac{{}^{f}D{}^{a,g}}{{}^{f}P{}^{a,g}}\right)x{P}_t^{a,g}$$

Where:

Then the estimate of the total number of deaths is calculated as follows: 

Equation 5.2:

$$D_{(t,t+i)}={\displaystyle \sum _g^{}}{\displaystyle \sum _a{D}_{(t,t+i)}^{a,g}}$$

Updated estimates

Until 2020, when the COVID-19 pandemic increased the number of deaths which required adjustments to the typical method, all updated estimates were calculated in the same manner as the preliminary estimates, using mortality rates applied to the population. The data from the CVSD would replace the updated estimates for all provinces and territories when they became available and the estimates would become final.

However, as the CVSD collection period has shortened, the DEP has sometimes observed that the data in the CVSD are complete for some provinces and territories and incomplete for others. In such cases, the updated death estimates are produced by directly using the CVSD death counts for provinces and territories where the data are considered complete. For those where the data are considered incomplete, the death estimates are produced using mortality rates from the last year of final data, special adjustments, or a combination of both. Death estimates become final only once the CVSD death counts are considered complete for all provinces and territories and are used directly for all jurisdictions.

Death estimates by gender and sex at birth

The DEP currently publishes all its tables according to gender, rather than sex at birth, consistent with changes introduced in the 2021 Census of Population. The data provided to the DEP from the CVSD as well as from the ISQ, have information on sex at birth and not gender. Even though the sex at birth variable is collected, gender may in fact be recorded for an undetermined proportion of deaths among individuals whose gender differs from their sex at birth. The sex at birth variable for deaths is used by the DEP and released as gender, consistent with current guidelines.

Final estimates of deaths by age and gender are produced by using the age and sex at birth distribution of deaths in the CVSD directly. The updated and preliminary estimates by age and gender are produced using the age and gender distribution from the last two years of final data in the DEP (the death estimates for the two most recent years produced entirely using the CVSD – those that are considered complete for all provinces and territories). There are also special methodological treatments to model deaths in the youngest and oldest age groups.

Subprovincial death estimates

For the final subprovincial estimates, the number of deaths at the census division (CD), census metropolitan area (CMA) and census agglomeration (CA) levels are taken directly from the CVSD. They are then adjusted to the provincial and territorial totals using two-way raking to ensure their concordance. The postal codes of individuals are geocoded using PCCF+^Note so that they are assigned a CD and a CMA or a CA geographic code.

For the preliminary and updated subprovincial estimates, the preliminary and updated provincial or territorial estimates are distributed according to the most recent known subprovincial distribution, derived from the CVSD, to produce estimates by region. They are then adjusted to the provincial and territorial totals using two-way raking to ensure their consistency.

For Quebec, British Columbia and Yukon, the subprovincial preliminary and updated estimates are the total death supplied by the ISQ, BC Stats, and YBS respectively, which are distributed according to the most recent final subprovincial breakdown. The ISQ provides their subprovincial estimates by age and sex and also by subprovincial region, whereas BC Stats and the YBS provide counts for total deaths at the province or territory level only. As with the provincial and territorial estimates, final subprovincial estimates for Quebec, British Columbia, and Yukon are taken from the CVSD when they are available and complete.

Limitations

The preliminary estimates (and some updated estimates) are based on the mortality rates by age group and gender from the last year of final estimates, provided by the CVSD. However, the longer the interval between the last year of final estimates and the preliminary estimates, the more likely that the mortality rate has changed. This may lead to an underestimation or overestimation of the number of deaths in the preliminary or updated estimates calculated by applying mortality rates.

For example, mortality rates during a period with higher numbers of death (such as during the COVID-19 pandemic or the opioid crisis^Note ) used to estimate deaths three years later (after the elevated number of deaths had receded and at a time when the aftereffects of mortality displacement could occur^Note ), might potentially overestimate the number of deaths for the later year. Conversely, mortality rates from a non-pandemic period may not be able to accurately capture such an abrupt increase in mortality, as was the case with the COVID-19 pandemic. The DEP measures and evaluates the difference between the preliminary and final estimates (known as precocity error) to ensure the preliminary estimation technique continues to be appropriate.

Deaths are received through the NRS on a continual basis and the CVSD is updated annually with a new year of data and with historical updates to past data. It is possible that some of the data not included in the first release of data in the CVSD are non-random in distribution. This may occur due to such issues as outages in the NRS affecting only one jurisdiction or staffing shortages among one or more provincial or territorial registers. To ensure the quality and completeness of the deaths estimates in the DEP, all data received from the CVSD are reviewed. Methodological adjustments are applied in the case that missing data would affect the quality of the population estimates.^Note