Events conditions episodes project group activities Bruce Goldberg

Events, conditions, episodes project group activities Bruce Goldberg, MD, Ph. D

ECE Goals – 2016 and onward ▪ Update the editorial guide to clarify the decision process for assigning the various modeling patterns for combined disorders ▪ Develop a rigorous testing approach to evaluate the combined disorder model ▪ Introduce additional temporal intervals/sequences to SNOMED CT using new roles or expanding the range of values for existing roles with respect to combined disorders and procedure complications ▪ Develop a plan for retrospectively remediating existing combined disorder content ▪ Address complex combined disorders (>2 conditions) ▪ Update allergy model ▪ Review model for allergic condition (due to allergic reaction) ▪ Add allergic condition and autoimmune disorder under 414029004 |Disorder of immune function (disorder)| ▪ Add new classes of adverse sensitivity and intolerance

Current discussions ▪ Guidelines for modeling combined disorder patterns ▪ Addition of during, during AND/OR after to associated with role hierarchy ▪ Application of ECE guidance retrospectively maintenance of existing descriptions that may be ambiguous ▪ Syndromes in which constituent conditions are variably present

Guidelines for modeling combined disorder patterns ▪ Current model for combined disorders is based on time intervals (co-occurrence, following) and causality ▪ Interpretation of causality and time interval between two related disorders proving to be difficult even using a very liberal definition of causality (indirect as well as direct), leading to violation of the reproducibility rule

Test of reproducibility of assigning causal relationship between two disorders Concept 709471005 |Periodontitis co-occurrent with leukemia (disorder)| Causation BG Causation EC Role BG Role EC Indirect Co-occurrent and due to 41931000119102 |Sinusitis co-occurrent with nasal polyps (disorder)| 367171000119100 |Hematuria co-occurrent and due to cystitis (disorder)| 709147009 |Gingivitis co-occurrent with diabetes mellitus (disorder)| 366961000119106 |Albinism co-occurrent with hematologic disorder (disorder)| 706913006 |Varicose veins of lower limb co-occurrent with edema (disorder)| 714203003 |Acute bronchitis co-occurrent with bronchiectasis (disorder)| None Direct Indirect None Indirect Co-occurrent and due to Co-occurrent Co-occurrent and due to 708030004 |Pulmonary emphysema co-occurrent with fibrosis of lung (disorder)| 125011000119100 |Mild persistent asthma co-occurrent with allergic rhinitis (disorder)| None Co-occurrent 285821000119105 |Major laceration of heart co-occurrent with hemopericardium (disorder)| 713844000 |Dementia co-occurrent with human immunodeficiency virus infection (disorder)| 19448008 |Optic atrophy associated with retinal dystrophy (disorder)| 420543008 |Anemia associated with acquired immunodeficiency syndrome (disorder)| 281004 |Dementia associated with alcoholism (disorder)| 68241007 |Glaucoma associated with ocular trauma (disorder)| Direct Indirect Direct Indirect None Indirect due to Co-occurrent and due to Co-occurrent and due to Co-occurrent and due to after 43742007 |Pericarditis associated with severe chronic anemia (disorder)| 230685009 |Myasthenia gravis associated with thymoma (disorder)| 422183001 |Skin ulcer associated with diabetes mellitus (disorder)| 421132004 |Platelet dysfunction associated with uremia (disorder)| 425390006 |Dementia associated with Parkinson's Disease (disorder)| 791000119109 |Angina associated with type II diabetes mellitus (disorder)| 230571009 |Neuropathy associated with endocrine disorder (disorder)| 403560008 |Port-wine stain associated with spinal dysraphism (disorder)| Indirect Direct Indirect Unknown None Indirect None Indirect None Co-occurrent and due to Co-occurrent and due to N/A Co-occurrent and due to Co-occurrent due to Co-occurrent

Proposed approach to achieving reproducibility for modeling combined disorders ▪ Suggested resolution to the aforementioned issue is to assign causality and time interval based on consensus, to combined disorder concepts associated with specific domains and provide guidelines for modeling these domains in the editorial guide ▪ Domains would initially represent high impact, high usage areas ▪ Examples ▪ Infections ▪ Diabetic complications ▪ AIDs related disorders ▪ Additional domains to be considered based on authors input

Infections, co-occurrence, causation, sites and morphology “Covert co-occurrence” • • Review of NHS GP/FP refset reveals that many concepts due to the way they are modeled with role groups, may inherit more than parent and thus be interpreted as consisting of more than one CLP which may not be apparent from the FSN Example: Infective uveitis (may be recast as Uveitis co-occurrent and due to eye infection) Stated view Inferred view

Infective (infectious) vs infected ▪ Difference between infections directly resulting in a morphologically abnormal structure ( e. g. inflammation) and infections of a morphologically abnormal structure due to another cause (e. g. infected blister) and infections of a previously inflamed site (infected eczema). ▪ If infections cause and occur at the same time as inflammation they should be modeled as inflammation co -occurrent and due to infection Y ▪ Alternatively, if inflammation is considered to be the inevitable outcome of infections, then inflammation cooccurrent with infection Y would suffice

Standard modelling patterns (with preferred role group combinations) required for several infectious disease variants (which are currently inconsistently applied) ▪ ▪ ▪ infection (including asymptomatic carrier states) Infection at site Infection by organism causing morphologic abnormality at site (e. g. tuberculous ascites, infective corneal ulcer) ▪ Infection by organism in morphologic abnormality at site (e. g. infected ascites, infected varicose ulcer) ▪ Infection by organism in morphologic abnormality, causing other morphologic abnormality, at site

Diabetic complications ▪ FSN Patterns ▪ ▪ ▪ X in diabetes X of diabetes X due to diabetes Diabetic X X associated with diabetes X complication of diabetes ▪ Possible modeling patterns ▪ X due to diabetes ▪ X co-occurrent and due to diabetes ▪ Modeling pattern might be dependent upon nature of X ▪ X = coma (coma co-occurrent and due to diabetes) ▪ X = disorder of retina (disorder of retina due to diabetes) ▪ Allows for possibility of persistence of retinal disorder after treatment of diabetes with islet cell transplant ▪ Alternate approach – apply single modeling pattern (e. g. x co-occurrent and due to diabetes) to all diabetic complications ▪ In the majority of cases by the time long term complications appear diabetes is chronic and incurable and will thus be necessarily present during the entire course of the complication ▪ Advantage of capturing diabetes as a second parent and thus potentially increasing the documentation and reported prevalence of diabetes

Workplan ▪ Identify domains ▪ Based on frequency data of SNOMED from NHS GP/FP refset and KP CMT ▪ Input from modelers (authors) ▪ Develop modeling guidelines for each domain/subdomain ▪ Evaluate agreed upon model in test environment ▪ Update editorial guide

Addition of during, during AND/OR after to associated with role hierarchy ▪ Original goal of project was to develop model for intraoperative complications ▪ Model for post operative complications exists ▪ Model to be extended to include not only intraoperative and postoperative complications but also perioperative complications ▪ Although perioperative commonly refers to the period of time prior to, during and after surgery, in terms of perioperative complications, we considered perioperative as occurring during surgery, after surgery or a time interval beginning during surgery and continuing in the post operative period. ▪ Thus intraoperative and postoperative complications would be considered subtypes of perioperative complications. ▪ The preoperative period was not considered for this exercise as a complication (consequence) of surgery would not be expected to occur prior to the surgery.

Surgical complications ▪ Proposed model for intraoperative complications is to approve use of 371881003 |During (attribute)|with a range limited to procedures ▪ Concern expressed that extending range of during to include clinical findings/disorders would overlap with the co-occurrence model for combined disorders ▪ Proposed model for perioperative complication is to create a new role, during AND/OR after as a descendant of associated with and an ancestor of both during and of after ▪ Above models applicable to other procedure related complications

Revised associated with role hierarchy

Intraoperative complication • Intraoperative X or X during surgery • Model Defined concept ≡ X Complicatio n During Surgical Procedure

Intraoperative complication

Injury to nerve during surgery

Injury to viscus/spleen during surgery

Intraoperative fracture

Intraoperative hemorrhage

Intraoperative complication hierarchy

Perioperative complications

Remaining issues ▪ We did not consider approval of 288556008 |Before (attribute)| at this time. ▪ It is appreciated that additional mechanisms for representation of temporal intervals in SNOMED CT such as the use of occurrence or another role with 307158003 |Temporal periods relating to physiological functions (qualifier value)| may be required in addition to before, during and/or after and after in order to capture other temporal relationships such as perinatal, perimenopausal and pregnancy-associated

Syndromes in which constituent conditions are variably present ▪ An agreement with INSERM to add Orphanet rare diseases to the International release of SNOMED CT. ▪ 1700 rare conditions that are variably named for the primary clinical manifestations and the eponymous name of the people who identified the condition. ▪ E. g. Hydrocephalus-agyria-retinal dysplasia syndrome Walker-Warburg syndrome ▪ The issue with these terms is that the clinical manifestations in the name of these diseases are variably present and often not co-occurrent. Therefore, they cannot be modeled as always and necessarily being true as having those conditions. ▪ Questions ▪ Which term should be the FSN (eponym or clinical term) ? ▪ If the latter, what is the proper structure of the FSN? Currently Orphanet uses dashes to separate the different clinical manifestations

FSN Naming ▪ Current editorial favors expansions but ”…allow[ing exceptions] when the full description is exceptionally long and unwieldy. . ". ▪ Follow the current guidance to use expanded forms but accepting that there is room for judgment. ▪ Use eponym when commonly associated with usage ▪ E. g. Tetralogy of Fallot as opposed to RVH, VSD, pulmonary artery stenosis and overriding aorta ▪ On the actual form of the expanded term, don’t use the full 'cooccurrent with' connective between various components. ▪ Preference is to use a comma separated list, a final 'and' and a closing 'syndrome’ ▪ Eg. Hydrocephalus, agyria and retinal dysplasia syndrome' rather than 'Hydrocephalus-agyria-retinal dysplasia syndrome' simply ▪ Might make parsing & word tokenization more predictable.

General modelling ▪ Building on the established genetic/congenital approach (www. snomed. org/eg? t=findings_disorder_types_congen ital_genet_devel_acq ) many of the new additions will need modelling judgments on how they are added: ▪ Hereditary/genetic disorders ▪ Add as primitively asserted subtypes of 32895009 | Hereditary disease (disorder) ▪ Congenital disorders ▪ Model with occurrence=congenital

Chromosomal abnormalities ▪ Model using the most suitable finding site=chromosome structure refinement and associated morphology value. ▪ Example Ring chromosome 20 syndrome ▪ Finding site = 25610001 |Chromosome pair 20 (cell structure)| ▪ Associated morphology = 23345003 |Ring chromosome (morphologic abnormality)| ▪ Observable model required for representing specific mutations (see e. g. artf 227316 Malignant neoplasm with gene mutation)

Specific modelling ▪ The preferred approach is to identify from the experts/literature the essential/cardinal features of each named syndrome. ▪ Ideally each of these can be included as a role grouped 'condition' (and thus acquires the appropriate supertype by classification), but where this is not feasible essential features can only be included by directly asserting as supertypes. ▪ Variant features (e. g. encephalocele for Walker-Warburg) would not be included in the formal definition, but if important then variants that 'include' another feature can be added as subtypes of the base concept. ▪ Naming convention ▪ X syndrome with Y (Walker-Warburg syndrome with encephalocele) ▪ For those syndromes where only a handful of 'essential/cardinal' features are included in the definition, consider an FSN of the type “X syndrome including phenotypic variants” to indicate there may well be other features but they are not included here.

Application of ECE guidance retrospectively maintenance of existing descriptions that may be ambiguous ▪ Presentation at Montevideo meeting outlined plan for remediating existing combined disorder content based on identifying concepts modeled with associated with and its descendants and attempting to reconcile the model with the FSN ▪ Based on the current approach to providing guidelines for modeling combined disorder concepts as outlined on slide 6, we will initially concentrate on remediating content for specific high frequency, high impact domains such as infectious disease, diabetic complications, AIDS related disorders

Revised work plan ▪ Identify relevant domains ▪ Usage analysis of NHS GP/FP refset and KP CMT donation ▪ Feedback from authors/modeling team ▪ Identify naming patterns and models within each domain ▪ Search for inferred parents that might indicate “covert” cooccurrence ▪ Determine temporal sequence and causation ▪ Apply correct model and/or FSN if required based on domain-specific guidelines to be developed