This document provides further statistical analysis related to the cross-version evaluations of ICD-10.

1. Genral

2. Non-automatic transition

   • Forward
   • Backward

3. Comparison of ICD-10 codes with version 2024

The German Federal Institute for Drugs and Medical Devices (BfArM) publishes so-called crosswalk tables. The crosswalk tables, in the form of text files, contain information on how a code in two subsequent versions behaves, for example:

K20 ; K20.0 ; X ; A

K20 ; K20.1 ; X ; X

T66 ; K20.1 ; X ; X

The first two columns show codes from the older version (blue) and the newer version (green). The remaining columns display transition information: from older to newer version (forward, orange) and from newer to older version (backward, red). An automatic transition is expressed by A in the tables. Non-automatic transition is represented by an empty entry in the tables, which, in this work, will be represented by X to improve readability. In total, 21 ICD-10-GM versions are considered between the years 2004 and 2024. Based on the provided Neo4J database and ConceptMaps R4 and R5 the analysis were carried out.

Here it was analyzed how many codes cannot be automatically transitioned between two successive versions. In addition, five reasons were identified for the non-automatic transitions. These as well as examples and frequencies are also shown.

On average across the ICD-10-GM versions of 2004 and 2024, 0.89 % or 122 codes per crosswalk table cannot be transited automatically.

The following five causes for a non-automatic transition were identified:

The causes and their frequencies per crosswalk table are as follows:

Two further examples of reorganization special (with chapter change):

On average across the ICD-10-GM versions of 2004 and 2024, 0.48 % or 66 codes per crosswalk table cannot be transited automatically.

The following five causes for a non-automatic transition were identified:

The causes and their frequencies per crosswalk table are as follows:

Two further examples of reorganization special (with chapter change):

In the following section, the ICD-10 codes of the previous versions will be compared with those of the version 2024. This involves checking whether the codes of individual annual versions are still terminating in the version of 2024 or whether they still exist. In summary, the analysis yielded the following average results for the versions considered from 2004 to 2023:

• Still terminating in version 2024: 98.48 % (13,279 codes)
• No longer terminating in version 2024: 0.89 % (119 codes)
• No longer existing codes in version 2024: 0.63 % (84 codes).

The results for the individual annual versions are as follows:

Two examples of codes of version 2023 that can no longer terminating in version 2024: