Methodological Notes – Mortality Module

Updated: March 2012

Concepts and Definitions

Measuring the pattern of mortality in a population is an important aspect of assessing the health status of a population, although it should be emphasized that death represents only the severest consequence of ill health. Information on deaths is relatively easy to obtain, and death is final, unequivocal and occurs only once in a person.

The most basic information required is the number of deaths, which when divided by the size of the population, produces the crude death rate.

Crude death rate = (number of deaths during year X) / (mean population of year X)

The rate is expressed as per 1, 000 persons. Rates for the 2000-2004 and 2005-2009 periods are calculated by dividing the sum of deaths during the 5-year period by the sum of the 5 annual mean populations. This differs slightly from averaging the 5 single year rates.

As with births, deaths also do not occur evenly across different age groups. From the number of deaths by age and sex are obtained the age-sex-specific mortality rates. The data are aggregated by 5-year age groups (0-4, 5-9, 10-14, etc).

Age-specific mortality rate for males = (number of deaths among males in age group i during year X) / (mean number of males in age group i during year X)

Age-specific mortality rate for females = (number of deaths among females in age group i during year X) / (mean number of females in age group i during year X)

The rate is expressed as per 1, 000 males or per 1, 000 females.

A widely used health indicator is life expectancy at birth. Life expectancy at birth summarizes the mortality experience of a population that prevails across all age groups. It can be defined as the average number of years that a newborn is expected to live if current mortality rates continue to apply. Note that life expectancy (LE) can be computed for other ages, eg. at age 65. The computation of LE is complex, and requires the construction of a life table, which presents the probability of dying, the death rate and the number of survivors for each age or age group. LE at birth is heavily influenced by the extent of deaths among infants and the very young, but is not affected by the different age structures of the populations being compared.

The method for constructing life tables is described in most textbooks of demography. Life tables using 5-year intervals of data [except <1 and 1-4 years] are termed abridged life tables. Templates for computing abridged life tables are available from a variety of sources. CircHOB uses the one produced by the United Kingdom Office of National Statistics, based on the report on methodological options for small populations by B. Toson and A. Baker, National Statistics Methodological Series No.33, 2003.

Because infancy, defined as age under 1 year, is a particularly vulnerable period in a person’s life, the number of infant deaths is also recorded, from which the infant mortality rate is constructed. Infant mortality rate (IMR) is a health indicator that is also widely used in international comparisons.

Infant mortality rate = (number of deaths under 1 year of age during year X) / (number of livebirths during year X)

The rate is expressed as per 1, 000 livebirths. Rates for the 2000-2004 and 2005-2009 periods are calculated by dividing the sum of deaths during the 5-year period by the sum of the livebirths. This differs slightly from averaging the 5 single year rates.

The IMRs reported by most statistical agencies do not actually track a cohort of newborn infants for one year after birth, although such an approach is certainly feasible using electronic data linkage between births and deaths databases. Note that among infant deaths during year X, some were born in year (X-1), and among livebirths in year X, some may go on to die in year (X+1).

Infancy can be broken down into the neonatal and postneonatal period, with the corresponding neonatal mortality and post-neonatal mortality rates. To avoid confusion, if the first day of life is designated as Day 0, then the neonatal period extends from the beginning of Day 0 to the end of Day 27 (or, put in another way, 28 completed days):

Neonatal mortality (NNM) = deaths during day 0 to day 27
Postneonatal mortality (PNM) = deaths during day 28 to day 364
Infant mortality = NNM + PNM = deaths during day 0 to day 364

Beyond the overall number (and crude rate) of deaths, examining the causes of deaths can provide more detailed information on the relative importance of different health problems, at least those that result in death. Because populations differ in their age-sex composition, and that death is very much age-dependent, simple comparisons of crude death rates among populations can be misleading. A more meaningful comparison utilizes age-standardized mortality rates by cause.

Age-standardized mortality rates (ASMR) adjust for differences in the age distribution of the population by applying the observed age-specific mortality rates for each population (the study population) to a standard population. The age-standardized mortality rate is thus a weighted average of the age-specific mortality rates per 100, 000 persons, where the weights are the proportions of persons in the corresponding age groups of the standard population. The so-called “direct” method is used, according to the formula:

Age-standardized mortality rate = Σ(ri Ni) / ΣNi
Where ri is the age-specific mortality rate for the ith age group of the study population
Ni is the number of persons in the ith age group of the standard population

The rate is usually expressed as per 100, 000 persons. Because of the small number of deaths for some causes, the numerator of ri is the combined total of deaths in the 5 years, divided by the sum of the mean population of each of the 5 years.

Any population can serve as a standard. When comparing regions within a country, the national population tends to be used as the standard. For international comparisons, there are hypothetical or artificially constructed populations. For ASMR, the standard population used in CircHOB is the European Standard Population used by Eurostat. Other world standard populations such as those of WHO and the International Agency for Research on Cancer (IARC) are also available. The latter is most commonly used for computing age-standardized cancer incidence rates.

The age distribution of the European Standard Population is as follows:

Age group Population
0-4 8, 000
09-May 7, 000
14-Oct 7, 000
15-19 7, 000
20-24 7, 000
25-29 7, 000
30-34 7, 000
35-39 7, 000
40-44 7, 000
45-49 7, 000
50-54 7, 000
55-59 6, 000
60-64 5, 000
65-69 4, 000
70-74 3, 000
75-79 2, 000
80-84 1, 000
85+ 1, 000
Total 100, 000

Source: Eurostat (2009). Health Statistics: Atlas on mortality in the European Union. Annex 1 – Standard European population

The causes of death are coded according to the International Classification of Diseases, 10th edition (ICD-10). This is a system of alphanumeric codes, organized into chapters (I to XX). Each country, however, uses a slightly different system of grouping individual codes into causes and groups of causes. The European Short List consists of 65 causes [available from Annex 2 of the European mortality atlas]. A few causes and groups of causes are selected for inclusion in CircHOB. These causes, with their corresponding ICD-10 codes are as follows:

Causes ICD-10 Code
All Causes A00-Y89 [Chapters I-XX]
Infectious and parasitic diseases (eg.
tuberculosis, viral hepatitis, AIDS, malaria)
A00-B99 [Chapter I]
Neoplasms C00-D48 [Chapter II]
> Malignant neoplasms (eg. cancer of
the lungs, breast, prostate)
C00-C97
Endocrine, nutritional and metabolic
diseases
E00-E90 [Chapter IV]
> Diabetes mellitus E10-E14
Diseases of the nervous system and sense
organs (eg. meningitis, Alzheimer’s)
G00-H95 [Chapters VI-VIII]
Diseases of the circulatory system I00-I99 [Chapter IX]
> Ischemic heart disease I20-I25
> Cerebrovascular disease I60-I69
Diseases of the respiratory system J00-J99 [Chapter X]
> Influenza and pneumonia J10-J18
> Chronic lower respiratory diseases
(eg. chronic bronchitis, asthma)
J40-J47
Diseases of the digestive system (eg.
peptic ulcer, appendicitis)
K00-K93 [Chapter XI]
> Chronic liver diseases and cirrhosis K70, 73, 74
> [Some countries report all liver
diseases
K70-K76]
Diseases of the genitourinary system N00-N99 [Chapter XIV]
Certain conditions originating in the
perinatal period
P00-P96 [Chapter XVI]
Congenital malformations Q00-Q99 [Chapter XVII]
External causes of mortality (i.e.
injuries)
V01-Y89 [Chapter XX]
> Accidents (eg. transport accidents,
falls, accidental poisoning)
V01-X59
> Intentional self-harm (i.e. suicide) X60-X84
> Assault (i.e. homicide) X85-Y09

Data Sources and Limitations

All statistical agencies maintain mortality databases which contain information relating to the number and causes of death and also basic demographic data of the decedents. As the ages of the decedents are not known in all deaths, the sum of all deaths with known causes is generally less than the total number of deaths.

For data on population age distribution used in calculating crude, age-specific and age-standardized mortality rates, refer to the Population Module in CircHOB. Data on livebirths for use in calculating infant mortality rates can be found in the Fertility Module, unless otherwise specified.

United States

United States national and Alaska data for all deaths and infant deaths are available from the CDC Wonder website.

LE at birth for the US nationally are reported in the annual report Health United States from the National Center for Health Statistics. Abridged life tables are specially constructed for the State of Alaska based on age-specific mortality rates for the 2000-04 and 2005-09 periods.

Canada

Canada data on the number of deaths by age and sex are from Statistics Canada’s CANSIM Table 102-0504.

CANSIM Table 102-0507 provides data on infant deaths. Deaths included those that occurred in Canada or the United States, and are attributed to the decedents’ usual territory/province of residence.

Life expectancy for Canada is reported in CANSIM Table 102-0511 for 2000-06 and Table 102-0512 from 2007 on. The three northern territories are combined in Table 102-0512. CircHOB calculates separate values for each territory based on 5 years’ data for 2000-04 and 2005-09.

Greenland

Statistics Greenland publishes the number of deaths by age and sex (Statbank table BEEBBDM1) and life expectancy (table BEEDT) based on 5 years’ data. The number of infant deaths and their components are requested from the Greenland mortality database maintained at the National Institute of Public Health in Copenhagen.

Faroe Islands

The number of deaths by age and sex are from Statistics Faroe Island, Table DEYALDAR. Life expectancy can be found in Table LIVSAVI based on 5 years’ data.

Infant deaths data are from the Chief Medical Officer’s report Medicinalberetning for Færøerne 2000-2005, renamed Sundhedsberetning for 2006-09. For the denominator, the number of livebirths from the same source is used, rather than from Statistics Faroe Islands.

Iceland

The number of deaths by age and sex and the number of infant deaths are from Statistics Iceland, Table MAN05211 and Table MAN05321, respectively. Life expectancy data are available from Eurostat (demo_mlexpec).

Denmark

Data on the number of deaths by age and sex are from Statistics Denmark’s Statbank tables FOD2 for 2000-2005 and FOD207 for 2006-09.

Eurostat provides data on infant, neonatal and postneonatal deaths (demo_minf) and life expectancy [demo_mlexpec] (demo_mlexpec) in Denmark.

Norway

Statistics Norway reports the total number of deaths in Table 8426 and the number of deaths by 10-year age group in Table 5377 > Population > Vital statistics and migration >. Breakdown of deaths by 5-year age group are available by special request from Statistics Norway.

Life expectancy for the country is available from Statistics Norway Statbank Table 5375. Regional data are calculated from 5 years’ data.

For infant deaths in Norway, national and regional data are available by special request from the Medical Birth Registry of the Norwegian Institute of Public Health. The number of livebirths from this source differs slightly from Statistics Norway’s, and are used in the calculation of infant mortality rates. The Norgeshelsa website does provide infant, neonatal and postneonatal mortality rates by regions but not the number of deaths.

Sweden

The number of deaths by age and sex is available from Statistics Sweden. The “age during the year” (i.e. age at the time of death) is selected rather than “age by year of birth” (i.e. age attained by the end of the year).

Life expectancy of the whole country is available from Eurostat (demo_mlexpec). For the regions, LE is calculated based on 5 years’ data.

Infant deaths and their components need to be accessed from Statistics Sweden’s annual report Belfolkningsstatistik 2000-2003, and Tabeller över Sveriges befolkning for 2004 to 2009.

Finland

The number of deaths by age and sex for Finland and its regions, and life expectancy for the whole country, are available from Statistics Finland’s Statfin website. Note that regional data are available only in the Finnish-language website. Life expectancy for the regions are calculated based on 5 years’ mortality data.

For infant mortality and its components, national data are from the Eurostat website. Regional data are by special request from Statistics Finland.

The number of deaths with all ages combined is available from the Federal State Statistics Service Russian-language interactive database. Deaths by age and sex for Russia nationally are available from the Demographic Yearbook, relevant years. Regional data are not available in 5-year age groups.

The number of infant deaths for Russia are available from the Federal State Statistics Service Russian-language interactive database. However, breakdown of IMR into neonatal and postneonatal mortality rates is not reported for Russia nationally or regionally.

Life expectancy in Russia and its regions is available from the Federal State Statistics Service.