WHO mortality data

WHO mortality data is a study by WHO to collect information about mortality rates in different countries. See the model file.

Scope

What are the mortality rates per country, sex, age group, and diagnosis?

Definition

The mortality data is actually quite complex. One could assume that is it country*ICD code*age group*year, but

• different ICD code groupings have been used in different countries
• different age group categories have been used in different countries
• different observation years.

Therefore, this is not a nice 4D cube. Instead, there are lots of merged and empty cells in the cube. There should be a plan for how this is organised. the current idea:

• Analyse the data for each country to identify the age, icd, and year locations used.
• Create indices for all different variations.
• On the database level,
  • describe which locations in which indices are equal.
  • describe which locations are mutually exhaustive subsets of another location.

The data contains the following locations:

Country (104 countries available): 1125 1300 1360 1365 1400 1430 2005 2010 2020 2025 2030 2040 2045 2050 2070 2085 2090 2110 2120 2130 2140 2150 2160 2170 2180 2190 2210 2230 2240 2260 2270 2300 2310 2320 2340 2350 2360 2370 2380 2385 2400 2410 2420 2430 2440 2445 2450 2455 2460 2470 3020 3030 3080 3090 3150 3160 3190 3255 3320 3325 3380 4010 4012 4018 4038 4045 4050 4055 4070 4080 4084 4085 4150 4160 4180 4182 4184 4186 4188 4190 4200 4210 4220 4230 4240 4260 4270 4272 4273 4274 4276 4280 4290 4300 4303 4308 4310 4320 4330 4335 4350 5020 5105 5150

Admin1, Subdiv: None

Year: 1996..2007

List: 101 103 104 10M

ICD 10 codes: 10584 different codes. Some of these are combinations of several ICD codes, listed as numbers.

Sex: 1,2,9 (Male, Female, Unknown)

Format (for age groupings): 0,1,2,4,7

IM format (for infant mortality): 1,2,8

Structuring the data in the database

For one country, it is straightforward to use indices that have only rows that contain data for that particular country. This will lead to 1 sex index, several year indices, 3 infant mortality indices, 5 non-infant mortality indices, and 4 different ICD indices. However, the whole study will be extremely complex with >= 14 dimensions and a huge number of empty cells.

What we need is a system that is able to aggregate and disaggregate data from one index to another. Aggregation is straightforward, but disaggregation requires data from other countries; this is used if no disaggregation data is not available for the particular country. Should we use Dirichlet for disaggregation?

Can the aggregation and disaggregation be done at the Base level? Or maybe we need an Analytica model that is uploaded to AWP, and that takes care of the (dis)aggregation. This sounds better.

• http://www.who.int/healthinfo/statistics/mortality/en/index.html
• http://www.who.int/whosis/mort/download/en/index.html
• http://www.who.int/healthinfo/morttables/en/index.html
• http://www.who.int/classifications/icd/en/
• http://www.who.int/classifications/icd/icdonlineversions/en/index.html

Result

{{#opasnet_base_link:Op_en2778}}

See also

• WHO mortality data.ANA
• Year 2003 version of the model.