### Mortality

About 500 people die annually out of a population of about 56,000. So far, only very few Greenlandic-born people have managed to reach the age of 80, and even fewer have celebrated their 90th birthday. And the number of people born in Greenland who have ever reached 100 years can be counted on two hands.The small population and the almost non-existent basis of experience for describing the mortality of the elderly cast a veil of uncertainty on the mortality targets calculated in this section. A further element of uncertainty is that part of the Greenlandic-born population emigrates during their lifetime, mainly to Denmark, to later die there.

After emigration, Statistics Greenland has severely limited access to information in the Danish administrative registers. Since 2016, Statistics Greenland has once a year received a data set from the CPR, with status information on everyone who at one time or another has resided in Greenland. From here, their date of death is known or whether they are still alive on the extraction date.

Statistics Greenland’s population model and data collections contain information about the population at individual level, which thus enables counting based on the exact number of days. These options have not yet been used. Instead, the information is converted into tables that are publicly accessible through the Statistics Bank of Greenland. All calculations in this section are made on the basis of these tables, particularly important is the Population Accounts.

#### Population size

Proportionally, the Greenlandic population corresponds to approximately 1 per cent of the population in Denmark, where 54,645 died in 2020 among the 5,825,337 calculated as the mid-year population 2020.

The summary death quotient, per 1000 inhabitants is astonishingly close: 520/56367*1000 = 9.22 compared to 9.38 in Denmark. Judging from the summary death quotient, mortality is the lowest in Greenland.

But it is NOT fair for mortality differences between the 2 countries, but is attributed solely to large differences in the countries’ gender and age distributions.

Furthermore, rates calculated for small populations are significantly more uncertain than for larger populations. In 1997, the Norwegian National Statistics Bureau (SSB) published a method description ‘Standardised rates’ ^{1} , in which Equation 26 shows that a 95% confidence interval is calculated as:

\[ (1000/middelfolketal) * (antalDødsfald + (1,96 * \sqrt{antalDødsfald})) \]

\[ > (antalDødsfald/middelfolketal) < \]

\[ (1000/middelfolketal) * (antalDødsfald - (1,96 * \sqrt{antalDødsfald})) \]

, where 1.96 is the value for a normally distributed variable, with expected value=0 and variance=1.

I figur 1 ses, at særligt for helt små befolkninger er den statistiske usikkerhed ekstrem.

Summarisk dødskvotient

“The simplest mortality measure for that type of comparison is the summary death rate, which is defined as the number of deaths per 1,000 inhabitants. Since the mortality rate is highest in the oldest age groups, the use of this measure requires that the age distribution of the populations be equal.”

Matthiessen, Poul Chr.; Nielsen, Vøgg Løwe: mortality in Den Store Danske on lex.dk. Retrieved 2 September 2021

Due to the very limited information required for calculation, the target is available for virtually all countries in the world, see e.g.: World Bank

Statistics Greenland recommends that this mortality measure **NOT be used** for comparisons of mortality. In fact, the summary mortality quotient cannot be used for comparisons of the same area over time either, as the age distribution is not the same.

In 2021, the summary mortality rate is calculated for 9,85 for persons born in Greenland. When the target is calculated and disseminated through the Statistics Bank, it is to satisfy the demand from international organ stations. In a different way, the Max Planck Institute collects and disseminates information about mortality.

People born in Greenland live shorter lives on average than in many countries we normally compare ourselves to. This is explained by many suicides and accidents that particularly affect the male part of the population and often at a young age.

Due to the small population, significant random fluctuations are seen from year to year, and the uncertainty of the calculated mortality measures is significantly higher than in larger populations. In Chart 4, uncertainty is marked by gray bands, the wider the band, the smaller the population, the greater the uncertainty.

But also for those born in Greenland, the mortality rate has fallen and the population is living longer, compared to 20 years ago. Women thus live r Grl_HMDf0-Grl_HMDf20 years longer and men r Grl_HMDm0-Grl_HMDm20 years longer.

## Calculation method

The Greenland population census contains a detailed count of the demographic events: birth, death, immigration and emigration. For the calculation of mortality, they are required information is simply the number of deaths calculated by age, time and year of birth, as well as status information for calculating the time lived. This is visualized in a Lexis diagram, which precisely shows the 3 dimensions: age, time and year of birth. In English, this type of analysis is referred to as APC- analysis (Age, Period and Cohort).

The age-specific death rates are calculated as the ratio between the number of deaths in each age group and the time lived by the risk population in the same age groups. In a given calendar year, all generations have been at age x before their birthday and at age x+1 afterwards. The triangle after the birthday and until the next turn of the year is called the ‘lower’ (lower) triangle and the one before the year’s birthday as the ‘upper’ (upper) triangle.

Due to Greenland’s very small population, considerable random variation is seen in the calculated death rates for a single calendar year. A commonly used trick is instead to calculate the mortality for subsequent 2- or 5-year periods, as a better estimate of the underlying mortality.

In the figure below, the areas a4:5 and b4:5 are first calculated (for 2-year averages), after which the death rate at age x is calculated as the average of the 2.

For 0-year-olds, the death rate is calculated as the death quotient number of deaths at age 0 from 2(5) subsequent cohorts.

### 2-års gennemsnit

### 5-års gennemsnit

Her vist med eksempel for 1994-fødselsårgangen af mænd i 2018.

\[
P_{(t+1)} \equiv P_{(t)} + B_{(t)} - D{(t)} + I{(t)} - E{(t)} + c{(t)}\\
P_{(2019)} \equiv P_{(2018)} + B_{(2018)} - D{(2018)} + I{(2018)} - E{(2018)} + c{(2018)}\\
431 \equiv 432+0-1+27-25+0\\
431 \equiv 432 - (0+1) + (15+12) - (17+8) + 0\\
\] Figure 4 shows the death rates (5-year) for the years 2000 and 2020 respectively for women and men. The curve for `r CONST_taar`

is clearly below the curve for `r CONST_taar-20`

, which means that mortality has decreased in virtually all ages.

Based on the calculated death rates, these are converted to the survival table targets using the R package Mortalitylaws (Parametric Mortality Models, Life Tables and HMD)

Mortalitylaws was written by Marius D. Pascariu and documented in his PhD thesis from 2018

The calculated targets are transferred to Statistikbanken’s table BEDBBDTB

Standardiserte rater, side 18, lining 26 (https://www.ssb.no/a/histstat/not/not_9722.pdf)↩︎