Conversation regarding file structure of mapping table: (11:18:29) jeff.lusted@…: The main structure would have to include something like: path, code and onyx variable (11:19:52) jeff.lusted@…: I could build a lot up just from a list of data like that (11:21:14) jeff.lusted@…: So each code had a path (snomed or local) and a corresponding onyx variable (11:22:26) jeff.lusted@…: We may need to work on the name to depict an onyx variable. It will probably need to be structured in some way so I can navigate the onyx questionnaires to find it within an export (11:23:54) jeff.lusted@…: I think the latter is probably suck it and see. If I have sufficient, I can find it given the xml we already have for metadata from onyx (11:24:13) jeff.lusted@…: or report an ambiguity (11:24:44) Nick: OK, that makes sense. Should we aim for one file for each questionnaire stage? And are we talking about an excel file or something else? (11:25:41) jeff.lusted@…: I've never read from an excel file (yet), but a comma separated file would be OK. I think you can emit them from excel? (11:26:34) Nick: Yeah, easily enough. But would something else be better? csv / excel is easy to work on manually because of the imposed cell-structure. But if you prefer xml or something we can look at that... (11:27:36) jeff.lusted@…: A csv file with one line per code should be OK (11:28:32) jeff.lusted@…: Perhaps one file for each questionnaire stage would make naming the onyx variable easier (11:28:40) jeff.lusted@…: But I'm easy. (11:29:48) jeff.lusted@…: If it makes more logical sense to have them all in one file, logically grouped (not necessarilly by qu' stage), I'm sure that would be OK

15.8.2011

Data Submission stage

Syncope - use 'syncope' code, not 'loss of consciousness'. Leg Pain on Walking - add "(Claudication)" to clarify questionnaire. Use 'claudication' SNOMED CT code. BUT --- do we make any effort to verify a cause at this point? If not, a much more general 'pain on walking' 426899007 might be appropriate. Left or Right sided weakness - 50582007, can't distinguish which side. Slurred speech - 289195008 Other - need mechanism to review the free text entries and consider adding to q'aire and also to ontology retrospectively

Diagnosis STEMI - 401303003 (because it's a disorder code, rather than the more specific 'ECG finding' codes) ACS - 394659003 (with additional code - 444931001 - if troponin positive) * note this is relevant only for as long as the test we use is trop I. There's an alternative UK specific code for 'cardiac troponin positive' - 372851000000101 - in TB's spreadsheet, which might be better. There's also a 'cardiac troponin negative code - 372901000000105. Stent thrombosis - 421327009 - not a clear mechanism for differentiating between acute and chronic, and no mention in q'aire of the 'immediate' phase. Perhaps best to just group all thromboses and leave the differentiation for the study Intermittent claudication - 63491006 Arterial occlusion - 2929001 (very general code) Cardiac arrest - 410429000 Done as far as ruptured aortic aneurysm. Start next meeting at ventricular ectopics.

23.8.2011 Vent ectopics: 17338001 - disorder, rather than finding Atrial ectopics: 406461004 Infective endocarditis is 233850007, need additional codes post-coordinated for specific structure: aortic valve (34202007 - structure - or 8722008 - disorder), mitral (91134007 or 11851006) or pulmonary (39057004 or 76267008). So, structure or disorder? Refer to John May for advice. Also, worth considering adding 'infective endocarditis - unknown location' to the questionnaire. Viral pericarditis 70189005 Bacterial pericarditis 233883000 post-infarction pericarditis 233885007 - i.e. early post-infarction not Dressler's syndrome other pericarditis 3238004 SVT: 6456007 VT: 25569003 VF: 71908006 Thrombotic Stroke or CVA 371040005 Haemorrhagic stroke 230706003 Restrictive cardiomyopathy 415295002 Ischaemic Cardiomyopathy 426856002 Dilated Cardiomyopathy 20529002 Valve surgery 73544002 TAVI - 174988008 PPCI: 414089002 Pacemaker: 307280005 ICD implantation: 395218007 thrombolysis: 426347000 ablation of lesion of heart: 10683007 coronary angiography 33367005 height: 50373000

Med Hx Stage Question: Separate heirarchy / ontology for interview Med Hx and notes Med Hx? So that we can review both sets of data separately? Note: no need for separate date facts, associate dates with the primary fact. chol: 13644009 chronic renal failure: 236425005 cabg: 232717009 valve surgery: 73544002 TAVI - 174988008 PPCI: 414089002 Other PCI 415070008 Pacemaker: 307280005 ICD implantation: 395218007 thrombolysis: 426347000 ablation of lesion of heart: 10683007 bare metal stent: 36969009 (parent code of below, maybe needs an additional child code for 'bare metal') drug eluting stent: 414509005 heart transplant: 32413006

Risk factor stage

228512004 for 'never chewed tobacco' if NO to first question (i.e. add two codes: never smoked AND never chewed) Don't add codes for the 'which' questions. just move on to categories for each smoking type cigs on most days ever - no - 266920004 (trivial cigaretter smoker) cigs on most days ever - yes - no answer, move on currently smoking - no - 281018007 (or, possibly, use the 'how many' answer below, to categorise from the children of this code) currently smoking - yes - no code, move on how many - number - 1-9: light 160603005, 10 - 20 moderate 160604004, 20-39 heavy 160605003, 40+ very heavy 160606002 how many when smoking the most - categories as above, but no date to attach. oops. maybe use this for pack year calculation? or not? age at starting smoking 228488005 (doesn't specify cigarettes, does it need to?) age at stopping smoking - not a snomed code for that, but there is a code for 'date stopped smoking' - 160625004 so could calculate that MAYBE SUBMIT A SNOMED REQUEST FOR 'age stopped smoking'?? Could then add in a number of other snomed codes / facts such as "total time smoked", "pack years" and so on, from the answers provided. Time of last cigarette? NO SMOMED code relevant to this.

Currently smokes cigars - 59978006 Ex-cigar smoker 160621008 How many? No SNOMED codes for differentiating quantities of cigar smoking 230057008

Diet: 1 Vegetarian 5940000 2 Vegan 24930006 3 Pescatarian - NO CODE YET? 4 Omnivore 41449007 - "General Diet" rather than 'omnivore' which is a description of an organism 9998 PNA x 9999 DK x

Next two questions - work-related and non-work-related activity need to be combined to give a single code, the higher of the two categories, plus the 'sedentary job' code if that is appropriate. 0 No_work 73438004 1 No_physical 160923008 - sedentary job (finding) 2 Light_physical 160631001 3 Heavy_physical 160633003 9998 PNA x 9999 DK x

0 NONE 228445002 1 Light_activity 160631001 2 Moderately_vigorous 160632008 3 Heavy_vigorous 160633003 4 Extreme_exercise 102533007 9998 PNA x 9999 DK x

# of brothers 224096003 # of sisters 224097007 How many children do you have?

Family history of conditions is unlikely to be valuable in SNOMED - the questions won't be asked the same way in other questionnaires. However, a basic heirarchy of SNOMED coded FH by diagnosis (regardless of which family member was affected). So represent the data twice - once in a SNOMED heirarchy and once in a more detailed, BRICCS-heirarchy.

FH dates? Not dates as such, but each fact will need a modifier or attribute, representing the 'age at onset' for that relative.

Alive / dead: SNOMED codes for father, mother, brother, sister. But children are recorded in SNOMED only as sons or daughters. This will have to be locally coded.

Death? Age at death of the relative is most important. No evidence of 'cause of death' in SNOMED, so local codes for all this.

Sudden death is 430283008 - but specific per relative not available. Double enter - SNOMED code and specific local questionnaire codes.