Skip to content

Latest commit

 

History

History

ICD10fi

README.md

ICD10fi

Intro

ICD10fi is an extension of the ICD10who standard. It largely follows the ICD10WHO classification. The most significant differences are related to the classification of external causes of injuries, illnesses and deaths (Chapter XX), where the Finnish version of the classification has been kept significantly narrower than the WHO version to ensure usability. In addition, new national, more accurate or better differentiated diagnoses reported at the five-character level have been included in the classification.

An other characteristic of ICD10fi it combines codes to describe a more accurate condition. The meaning of these depend on the union-mark used:

  • Classification codes: decrive ranges of codes other than the conventional ICD10 herarchy
    • Code1-Code2 : from Code1 to Code2
  • Reason codes: combine codes to add more info on what caused the diagnose, there are 4 marks
    • Code1*Code2 : “Oirekoodi”, Code2 indicates an additional symtom
    • Code1+Code2 : “Syykoodi”, Code2 indicates the reason for Code1
    • Code1#Code2 : ATC-koodi, Code2 is and ATC code indicating the medicine that caused Code1
    • Code1&Code2 : “Kasvainkoodi”, Code2 is and endocrinological disorder code that caused Code1

source

Formating source vocabulary to OMOP

The list of the icd10fi codes have been downloaded from the official source: kodistopalvelu(7.5.2020).

This table contains all the single codes, some composed-codes, and classification. The last only used for hierarchy and not diagnose. In addition, the table also contains the rules to generated other valid composed-codes (in column A:Huom).

We generate new composed-codes based on the “1 code” codes that contain generation rules in column A:Huom.

tmp_code_class n
1 code 13346
2 code 1037
2 code generated 54099

For the how the English name was translated: THL if both were translated by THL, Google if both were translated by google-translator, and THL+Google if both.

tmp_name_en_source n
THL 51484
THL+Google 14635
Google 2363

Details in ./1_source_vocabulary/README.md

Mapping the source vocabulary to the standard vocabularies

In short, ICD10fi codes were match to ICD10who by code and English name ICD10who code and name_en (English name with out a full match were reviewed by @helmisuominen). Several codes in the ICD10fi table have the same name_fi. We made sure that if any of these was mapped to ICD10who, the rest mapped to the same code ICD10who name_fi. These ICD10fi that didn’t match ICD10who were mapped to a standard vocabulary using USAGI by @helmisuominen (USAGI). The “2 code generated” are mapped to the mappings of both composing codes (fully mapped if both codes have a mapping, missing one if not).

Following table summaries the codes by mapping type and code class.

tmp_code_class tmp_mapping_type n
1 code ICD10who code and name_en 10676
1 code ICD10who name_fi 15
1 code USAGI 842
2 code ICD10who name_fi 95
2 code USAGI 64
2 code generated fully mapped 31150
2 code generated missing one 19880

This process is detailed in 2_mapping_to_standard/README.md

Progess in number of codes

From 68 482 codes 62 722 have been approved.

This makes 92% of codes approved.

Assessing coverage of databases

Database finngen

How many codes labeled as ICD10fi in finngen are not in the atc standard?

There are 911 codes not found in the standard

Top10 sorted by freq:

code1 code2 freq freq_per
Z038 NA 8015 0.046%
E660 NA 1557 0.009%
K0401 NA 1162 0.007%
-1 NA 993 0.006%
-2 NA 984 0.006%
E890 NA 901 0.005%
F0019 G301 827 0.005%
H062 E050 721 0.004%
M073L405 NA 640 0.004%
N0832 E102 605 0.003%

The full list can be found in ./3_freq_of_source_codes/finngen_not_in_ICD10fi.csv

Status of the ICD10fi codes infinngen

status n_codes per_codes n_events per_events
mapped 7803 63.070% 16924333 96.356%
not_mapped 3658 29.567% 582380 3.316%
not_found 911 7.363% 57676 0.328%

Database tays

How many codes labeled as ICD10fi in tays are not in the atc standard?

There are 238 codes not found in the standard

Top10 sorted by freq:

code1 code2 freq freq_per
Pää NA 209109 2.078%
H02AB06 NA 848 0.008%
L01CD02 NA 299 0.003%
L01XX24 NA 200 0.002%
V03AX NA 131 0.001%
R518 NA 95 0.001%
M05BX04 NA 91 0.001%
V06DB NA 82 0.001%
E660 NA 81 0.001%
H360 NA 78 0.001%

The full list can be found in ./3_freq_of_source_codes/tays_not_in_ICD10fi.csv

Status of the ICD10fi codes intays

status n_codes per_codes n_events per_events
mapped 8959 84.455% 10110937 96.365%
not_mapped 1411 13.301% 168729 1.608%
not_found 238 2.244% 212620 2.026%

NOTES on missing codes

NOTES:

  • Z038, E660, E890 : many are missing the last digit, should it be 0 ??

  • K0401 : just don’t exists

  • F0019 G301 : not specify what can be include ?? TOFIX

  • H062 E050: invalid ?? “Etiologinen koodi valitaan ryhmästä B74”

  • H3603 E109 : invalid ?? In group H36* “Etiologinen koodi valitaan tässä ryhmässä ryhmistäE10-E14. Neljäs merkki on .3”

  • M073L405 : should be fixed before or after in the ELT process

  • N0832 E102 : may be valid. In N08.3* “Etiologinen koodi valitaan ryhmistä E10-E14.Neljänneksi merkiksi tulee valita .2” do i have to include subcodes ??

  • T36 N05BA: code2 is atc code which is correct but supose to be in code3 !!

  • many code1=NA what to do ??