Equivalences to ICD-10-ES were attempted for the ORPHAcodes of the 5,775 disorder level entities from the Orphanet catalogue of RDs included at the MF, following the criteria agreed (Fig. 1) by the consortium.
To illustrate the different situations and how they were solved according to these criteria, six representative examples have been selected and are described below. In the standard workflow, the first criterion was to prioritise codes with matching descriptors in both classifications, ICD-10-ES and ORPHAcode (example 1).
Example 1:
ORPHAcode: 111 → Descriptor: Barth syndrome
ICD-10 code proposed by Orphanet: E71.1 → Descriptor: Other disorders of branched-chain amino acid metabolism
ICD-10-ES code proposed by this study: E78.71 → Descriptor: Barth syndrome
In the cases where matching descriptors were not found, the second criterion applied was that the ICD-10-ES code derived from the ICD-10 code proposed by Orphanet would be chosen if available (example 2).
Example 2:
ORPHAcode: 2045 → Descriptor: FLOTCH syndrome
ICD-10 code proposed by Orphanet: L60.8 → Descriptor: Other disorders of nails
ICD-10-ES code proposed by this study: L60.8 → Descriptor: Other disorders of nails
When those criteria could not be fulfilled, the ORPHAcodes were submitted along with their potential ICD-10-ES codes to the mapping team in order to agree upon the best equivalence to ICD-10-ES.
When more than one ICD-10-ES code seemed a suitable correspondence to an ORPHAcode, the extended workflow was activated (Fig. 2).
In the cases were a multi-systemic disease or syndrome was classified under a disorder level ORPHAcode, but there was no ICD-10-ES code fully matching the disorder features, the criterion was to prioritise the ICD-10-ES code derived from the ICD-10 code proposed by Orphanet (example 3).
Example 3:
ORPHAcode: 2585 → Descriptor: Ataxia-pancytopenia syndrome
ICD-10 code proposed by Orphanet: D61.0 → Descriptor: Constitutional aplastic anemia
ICD-10-ES code proposed by this study: D61.09 → Descriptor: Other constitutional aplasias
Complementary (not selected) ICD-10-ES code: G11.8 → Descriptor: Other hereditary ataxias
Secondly, in the cases where the main descriptor and one or more of its synonyms listed at Orphanet for one ORPHAcode match different ICD-10-ES codes, the criteria were to prioritise the ICD-10-ES code which descriptor matches the main descriptor associated to the ORPHAcode or in its absence, the one that derives from the ICD-10 code proposed by Orphanet (example 4).
Example 4:
ORPHAcode: 86879 → Descriptor: Extranodal nasal NK/T-cell lymphoma
Synonym: Lethal midline granuloma
ICD-10 code proposed by Orphanet: C86.0 → Descriptor: Extranodal NK/T-cell lymphoma, nasal type
ICD-10-ES code proposed by this study: C86.0 → Descriptor: Extranodal NK/T-cell lymphoma, nasal type
Alternative (not selected) ICD-10-ES code: M31.2 → Descriptor: Lethal midline granuloma
Anytime these criteria failed to produce a reasonable equivalence, the third step was to submit these ORPHAcodes along with their potential (if any) ICD-10-ES codes to the mapping team.
When multiple ICD-10-ES codes were correlated to the same ORPHAcode, they were clustered under a generic code constructed with their common “prefix” followed by a plus “+” symbol. This was made to avoid overstating univocal (group 1) equivalences when several ICD-10-ES codes were matching the same ORPHAcode (example 5).
Example 5:
ORPHAcode: 1163 → Descriptor: Aspergillosis
ICD-10 code proposed by Orphanet: B44+ → Descriptor: Aspergillosis
ICD-10-ES code proposed by this study (for data treatment): B44+ → Descriptor: Aspergillosis
Equivalent ICD-10-ES code(s) (for use at registries): B44.0; B44.1; B44.2; B44.7; B44.8 (B44.81; B44.89); B44.9
In addition, when an ORPHAcode equivalent ICD-10-ES code(s) or a subset of them was included within the multiple ICD-10-ES equivalences already clustered for another ORPHAcode(s), the ICD-10-ES code(s) of these new ORPHAcode(s) were also converted to the same condensed ICD-10-ES {common prefix followed by “+”} code (example 6).
Example 6:
ORPHAcode: 1164 → Descriptor: Allergic bronchopulmonary aspergillosis
ICD-10 code proposed by Orphanet: B44.1+ → Descriptor: Other pulmonary aspergillosis
ICD-10-ES code proposed by this study (for data treatment): B44+ → Descriptor: Aspergillosis
Equivalent ICD-10-ES code(s) (for use at registries): B44.81 → Descriptor: Allergic bronchopulmonary aspergillosis
Examples 5 and 6 illustrate different situations in which ICD-10-ES code(s) were converted to their condensed form for data treatment. In the first, the ORPHAcode 1163 had equivalences to 8 different ICD-10-ES codes that share the root “B44”. The other example shows ORPHAcode 1164, which had the ICD-10-ES code B44.81 as its sole equivalence. Nonetheless, B44.81 is common to both, 1163 and 1164 ORPHAcodes, making necessary to homogenize the equivalence in order to not count this equivalence as unique. By applying this strategy, we minimize the bias resulting from multiple ICD-10-ES equivalences to one ORPHAcode over the calculations. A total of 834 equivalences were adjusted by these means, and around 70% of them were similar to that exposed in example 5.
After all this process, 4,987 ORPHAcodes were mapped to ICD-10-ES by applying the standard and extended workflows meanwhile equivalences for 677 were solved only after revision by the mapping team, reaching over 98% of the target ORPHAcodes. The remaining 111 ORPHAcodes (< 2%) lacked of ICD-10-ES code equivalence after exhausting the resolution framework applied in this study, including 94 of the initial 484 without ICD-10 code proposed by Orphanet.
1,700 different ICD-10-ES codes were selected to produce 5,735 equivalences for these 5,664 ORPHAcodes successfully correlated between coding systems. This mismatch between the number of ORPHAcodes and the number of equivalences is due to the fact that 65 ORPHAcodes kept more than one corresponding ICD-10-ES code (making a total of 136 equivalences), because they did not share a common prefix. Moreover, 33 out of these 65 ORPHAcodes established equivalences to more than one ICD-10-ES chapter (See additional file 1).
Distribution of ORPHAcodes among the ICD-10-ES chapters
The 5,664 ORPHAcodes mapped to ICD-10-ES codes were grouped depending on the ICD-10-ES chapter to whom their assigned ICD-10-ES codes belonged. Twenty out of twenty-one ICD-10-ES chapters (See additional file 2) got at least one matching ORPHAcode and the distribution of the equivalences was not proportional among them. Three ICD-10-ES chapters gathered over two thirds of the equivalences: “Endocrine, nutritional and metabolic diseases” (E00-E89) with around 12%, “Diseases of the nervous system” (G00-G99) with over 14% and “Congenital malformations, deformations and chromosomal abnormalities” (Q00-Q99) with over 40% (Fig. 3a). The use of ICD-10-ES codes, although followed a similar trend, showed a reduction in the relative weight of these groups. Almost halving in the cases of chapter G with just over 8% and chapter Q with only 21% of the total ICD-10-ES codes used in this study (Fig. 3b).
Global overview of ICD-10-ES codes availability to map ORPHAcodes
Once stated that the overall distribution of ORPHAcodes per ICD-10-ES chapter was uneven, the next step was to assess the ratio of available ICD-10-ES codes per ORPHAcode, both globally and within chapters. Data analysis showed dissimilar results among the chapters, 7 chapters showed over 0.7 ICD-10-ES/ORPHAcode (A00-B99: 0.93; F01-F99: 0.73; J00-J99: 0.76; O00-O09A: 0.78; R00-R99: 0.71; S00-T88: 0.90 and Z00-Z99: 1.00) and 2 chapters showed below 0.2 ICD-10-ES/ORPHAcode (G00-G99: 0.17 and Q00-Q99: 0.15) (Fig. 4). Remarkably, the seven chapters with the highest ratio comprise altogether around 5% of the total ORPHAcodes and A00-B99 compiles around 3% on its own. On the other hand, the two chapters with the lowest ratios (G and Q) represent more than half of the total ORPHAcodes included in this study with 4 out of 10 being equivalent to codes from the Q chapter. All these circumstances made the average ratio to fall to around 0.3 ICD-10-ES codes available per ORPHAcode (Fig. 4).
Global and chapter level analysis of ICD-10-ES codes redundancy
So far there was an uneven distribution of ORPHAcodes among ICD-10-ES chapters and, moreover not all of the chapters had (enough) specific codes to match the ORPHAcodes assigned to them. Therefore, to deepen in the analyses, the patterns of the equivalences established to individual ICD-10-ES codes were studied. The results showed that ≈ 18% (1,055) of the total equivalences belonged to group 1 (1 ORPHAcode to 1 ICD-10-ES) while ≈ 23% (1,322) fell into group 3 (≥ 50 ORPHAcodes to 1 ICD-10-ES), leaving ≈ 59% (3,358) in group 2 (2 to 49 ORPHAcodes to 1 ICD-10-ES). The in-chapter distribution was studied to further characterize the capabilities of each ICD-10-ES chapter to absorb its share of ORPHAcodes. Yet again, chapters G (9.1%) and Q (7.6%) showed the lowest group 1 ORPHAcodes relative content and were also the only ones containing group 3 ORPHAcodes (31.7% and 45.6% respectively) (Fig. 5).
Section level assessment of the G chapter codes of the ICD-10-ES
Taking a deeper look at the G chapter of the ICD-10-ES, the distribution of ORPHAcodes into the different sections was assessed. Three of its eleven sections contained more than half of the 834 ORPHAcodes assigned to this chapter (“G10-G14 – Systemic Atrophies primarily affecting the central nervous system”: 240 ORPHAcodes; “G60-G65 – Polyneuropathies and other disorders of the peripheral nervous system”: 135 ORPHAcodes; and “G70-G73 – Diseases of myoneuronal junction and muscle”: 183 ORPHAcodes) (Fig. 6a). Group 3 ORPHAcodes were also limited to these 3 sections. Between one and two of every three ORPHAcodes assigned to each of them (G10-G14: 37.1%; G60-G65: 65.9% and G70-G73: 47.0%) fell into the group 3 of ICD-10-ES codes redundancy (Fig. 6b).
Regarding specific ICD-10-ES codes, “G11.4 – Hereditary spastic paraplegia”: 89 ORPHAcodes; “G60.0 – Hereditary motor and sensory neuropathy”: 89 ORPHAcodes; and “G71+ (in particular G71.0 – Muscular dystrophy- and G71.8 – Other primary disorders of muscles-)”: 86 ORPHAcodes, were the only ones with a redundancy of at least 50 ORPHAcodes each. Therefore, the 264 ORPHAcodes belonging to group 3 in the G chapter can be traced down to these 3 ICD-10-ES codes (See additional file 3).
Section level assessment of the Q chapter codes of the ICD-10-ES
Similarly to what happened with the G chapter of the ICD-10-ES, the distribution of ORPHAcodes into the different sections of the Q chapter was studied. Three of its eleven sections contained more than half of the 2,320 ORPHAcodes assigned to this chapter (“Q65-Q79 – Congenital malformations and deformations of the musculoskeletal system: 434 ORPHAcodes; “Q80-Q89 – Other congenital malformations”: 1,072 ORPHAcodes and “Q90-Q99 – Chromosomal abnormalities, not elsewhere classified”: 294 ORPHAcodes) (Fig. 7a). Group 3 ORPHAcodes were also limited to 3 sections, but in this occasion a much smaller section (Q00-Q07 – Congenital malformations of the nervous system) with just 128 ORPHAcodes assigned had 49.2% of them into group 3. Up to 3 out of 4 of the ORPHAcodes assigned to the other 2 sections (Q80-Q89: 76.4% and Q90-Q99: 59.9%) fell into the group 3 of ICD-10-ES codes redundancy (Fig. 7b).
Once again, when specific ICD-10-ES codes were analysed, just a few of them: “Q04.3 – Other reduction deformities of brain”: 63 ORPHAcodes; “Q82.8 – Other specified congenital malformations of the skin”: 96 ORPHAcodes; “Q87.0 – Congenital malformation syndromes predominantly affecting facial appearance”: 127 ORPHAcodes; “Q87.1 – Congenital malformation syndromes predominantly associated with short stature”: 68 ORPHAcodes; “Q87.2 – Congenital malformation syndromes predominantly involving limbs”: 57 ORPHAcodes; “Q87.89 – Other specified congenital malformation syndromes, not elsewhere classified”: 471 ORPHAcodes; “Q92.2 – Partial trisomy”: 68 ORPHAcodes; and “Q93.5 – Other deletions of part of a chromosome”: 108 ORPHAcodes showed redundancy of over 50 ORPHAcodes each. Therefore, the 1,058 ORPHAcodes belonging to group 3 in the Q chapter can be traced down to these 8 ICD-10-ES codes (See additional file 4).